Pharmacophore-based virtual screening & molecular docking studies on selected plant constituents of Plantago major

Phytochemicals are a striking source to discover new leads for the expansion of novel compounds for several diseases. In this study, various in silico techniques are used to showcase the multitarget inhibitors of selected plant constituents of Plantago major . Five plant components having an anti-inflammatory activity are used to build a pharmacophore model with “PharmaGist webserver” which generated a four-point hypothesis. The best model with a score of 12.402, was used to screen the National Cancer Institute database of the Pharmit web server to obtain similar pharmacophore hits. Subsequently, molecular docking was performed on the Cyclooxygenase-2 (PDB ID: 4COX) protein by using Autodock Vina, to prioritize top lead molecules. Among all the hits, four compounds have the best dock scores than the standard Celecoxib (−9 kcal/mol). From our result, compound NSC86473 has the highest potential as an anti-inflammatory agent with binding energy (−10 kcal/mol) and may act as a powerful inhibitor against Cyclooxygenase 2 as it has the lowest binding energy than the standard with specified pharmacophoric features according to developed pharmacophore model 1 model. In accordance with earlier findings, it can provide a few insights to research scholars in the future to identify and design new lead molecules with effective anti-inflammatory activity.


INTRODUCTION
Natural products can serve as budding resources for the development of anti-inflammatory drugs due to better pharmacological activities and lower toxicity (Deng et al., 2010). Presently, almost all drugs are derived from plant origins. Plant constituents are attractive agents with low cost, effectiveness, and biocompatibility, which makes phytochemicals a prominent cause for the development and identification of the lead compounds that provide support to the pharmacological activity of existing drugs (Samuelsen, 2000). Plant extracts have been reported with enormous biological activities like anti-inflammatory, antioxidant, wound healing activity, weak antibiotic, analgesic, immunomodulating, and antiulcerogenic activity (Mozaffarian, 2012) Plantago major is a perennial plant that belongs to the family Plantaginaceae. It consists of biologically potent compounds like alkaloids, polysaccharides, flavonoids, iridoid glycosides, lipids, caffeic acid derivatives, terpenoids, and organic acids (Samuelsen, 2000) (Fig. 1). These compounds can be found in almost all parts of the plant such as the seeds, leaves, flower, and roots (Adom et al., 2017;Wang et al., 2015). Earlier studies reported that P. major is used in different parts of the world for the treatment of numerous conditions like skin diseases, infectious diseases, digestive problems, respiratory abnormalities, glitches related to reproduction, circulation, tumors, and inflammation (Azab et al., 2016). Owing to the tradition of employing P. major for wound healing, it is worth exploring this plant further for antiinflammatory activity (Mozaffarian, 2012;Samuelsen, 2000) In recent years, great advancement has been made in the expansion and resolution mechanisms of chronic inflammatory diseases, and the use of phytoconstituents to lighten inflammatory diseases (Isailovic et al., 2015). We are persistent in exploring the anti-inflammatory targets in the context of chronic inflammation, establishing a screening technique, and identification of potent leads with strong anti-inflammatory activity (Deng et al., 2010;Newman and Cragg, 2007) Some of the structures of the major constituent are illustrated in Figure 1.
Molecular modeling (MM) is a significant improvement in the design and discovery of new leads (Langer and Hoffmann, 2001). At present, MM is considered a necessary tool in drug discovery and optimizing the prevailing prototypes and the rational design of lead candidates, in which virtual screening plays a prominent role (Lionta et al., 2014). Due to that, numerous procedures, such as the pharmacophore model development, molecular dynamics, etc., have been recognized (Saikiran Reddy et al., 2018).
The ligand-based pharmacophore method is used to develop a pharmacophore model. The current research work encompasses various phytoconstituents to identify and design new molecules with potent anti-inflammatory activity. . Phytoconstituents are an attractive source in developing the lead molecules for various diseases.
The basis of the perception is that molecules that share few structural similarities may possess the same activity. Some approaches use existing active ligands as a query to retrieve structurally analogous compounds from huge databases (Lionta et al., 2014).
In present computational chemistry, the vital topographies of ligands with the same pharmacological activity were well-defined using pharmacophore (Kaserer et al., 2015). It is defined as the three-dimensional alignment of features required for a ligand to interact with a particular protein. These particular traits are substantial to accomplish the finest pharmacophore model. In this research, PharmaGist, a web-based software that is a ligand-based method was used to build a pharmacophore model (Schneidman-Duhovny et al., 2008;Usha, 2016). Molecular interactions and their predictions are a prominent step in rational drug design. The spatial alignment of features that is crucial for a molecule to interact with a definite target is a pharmacophore. The ligand-based screening method is employed here (Dror et al., 2009).
The screening process includes choosing the finest pharmacophore employing the Pharmit server's inbuilt database (Sunseri and Koes, 2016) and validating through the docking studies. Pharmit offers an online, user-friendly environment for the virtual screening of large databases using pharmacophores, energy minimization, and molecular shape. It also provides easy access to large chemical datasets. Docking studies were performed to identify the screened molecules' affinity toward the protein target, which is one of the major steps in the rational drug design process (Saikiran Reddy et al., 2018).

Ligand selection for pharmacophore modeling
The 3D structures of selected ligands were retrieved from PubChem in SDF format (Kim et al., 2016). Later, the Open Babel server was used to convert all five ligand structures into ( luteolin, alpha-linoleic acid, oleanolic acid, ursolic acid) were used to build a pharmacophore model with "PharmaGist webserver."

PharmaGist
The best open-source web server used for searching a pharmacophore using a group of ligands that bind to a protein is PharmaGist. For every input ligand, this web server lists the number of atoms and three-dimensional and physicochemical features like aromatic rings, hydrophobic regions, hydrogen bond donors, and hydrogen bond acceptors as given in Table 1. This method expeditiously searches for all possible pharmacophores, generates three-dimensional visualization of detected pharmacophores by multiple alignments of input ligands, and executes virtual screening based on the detected pharmacophore model.

Virtual screening
The built pharmacophore model is used as a query in Pharmit to screen libraries of small molecules. Load pharmacophore features and selects a database for screening (e.g., National Cancer Institute (NCI) Open chemical repository database). The number of hits was displayed, to reduce the hits. The flow chart is depicted in Figure 3.

Pharmit
It is an online open-source platform for performing structure-based virtual screening with pharmacophore and shape queries (http://pharmit.csb.pitt.edu). It allows users to interactively search libraries of millions of compounds as part of structure-based drug discovery. It enables us to explore ligands grounded on their chemical and structural similarity to other ligands. The compounds which exhibited extreme similarity to query pharmacophore were filtered.

Molecular docking studies
Dataset ligands and ligand optimization ACD/Chemsketch was used to generate the 2D ligand structures. The generated ligands were subjected to 3D optimization and saved in Molecular Dynamics Language Molfile format. Later, by using Open Babel, optimized ligands were converted to a Protein Data Bank, Partial Charge (Q), & Atom Type (T) format (PDBQT) file format.
The downloaded crystal structure of the protein (PDB ID: 4COX) was optimized by removing water molecules, hydrogen atoms were added to satisfy the valences, charges were included, further missing amino acids were added to stabilize side chains, and energy of the complete structure was minimized with AUTODOCK suite of Molecular Graphics Laboratory Tools.
Molecular docking studies were performed with Autodock Vina (Muni Sireesha et al., 2022). A grid was created around the co-crystallized ligand. The coordinates (x = 22.51, y = 22.24, z = 16.199) were produced with the help of Pharmit server (http://pharmit.csb.pitt.edu/). Protein target and ligand PDBQT files were prepared. In the absence of water molecules, docking was performed for all 10 compounds (9 + 1 standard). Subsequently, all the molecules were analyzed and visualized in the discovery studio for the interactions with the active site (MuniSireesha et al., 2021). Further, the efficiency of the binding and its interactions was calculated in terms of dock score, which is a blend of hydrophobic, hydrophilic, metal binding groups, freezing rotatable bond, Vander Waals energy, and polar interactions with receptors.

Detection of pharmacophore model
PharmaGist helps in detecting pharmacophore, which is the three-dimensional alignment of various topographies that allows a ligand to interact with the protein in its binding pocket. The input of a group of ligands in PharmaGist will highlight Pharmacophore candidates by superimposing the three-dimensional ligand conformations. The selected ligands are phytoconstituents of P. major with anti-inflammatory activity. Further, the ligands were imported into Marvin viewer and saved in Tripos mol2 format. Later, the individual mol2 files are compressed as a zip file and submitted as a query in PharmaGist to identify the pharmacophore candidates as given in Table 2. Based on the score retrieved by the pairwise alignment in the PharmaGist server, the best ligands were selected as mentioned in Table 3. The pairwise alignment and structures of the ligands, the bond angles, and bond distances are given in Figure 4.

Virtual screening
The output pharmacophore model 1 (PM1) obtained as a mol2 file through PharmaGist software which has the highest score of 12.402 was submitted as a query to the Pharmit web server to visualize the alignment of the standard with the ligands (Fig. 5). It is a four-point pharmacophore having features of one hydrogen bond donor and three hydrogen bond acceptors. From over 52,237 ligands, the virtual screening process retrieved 11,341 hits, which are matched with the features in the query pharmacophore model. Then by applying filters of the Lipinski rule, root mean squared distance (RMSD), and rotatable bonds, finally we selected the top-nine hits. Further, these nine hits were subjected to molecular docking analysis to prioritize the best ligand for the anti-inflammatory target.

Molecular docking
Molecular docking was performed with the X-ray crystal structures of the target (PDB ID: 4COX), pivot molecule ursolic acid, and four input ligand molecules (aucubin, luteolin, alpha-linoleic acid, oleanolic acid, ursolic acid). Subsequently, the screened nine hit molecules from the NCI database were sketched in Chemdraw and saved in mol2000 format. All the ligand molecules' energy was minimized and converted into PDBQT format using the Open Babel server. All the compounds' low energy conformations were docked into the active site of the target, i.e., 4COX using AUTODOCK VINA. It is a technique for assessing the accuracy of the docking process to determine how Ursolic acid 13 9 3 0 4 6 * It is taken as a pivot molecule for developing a pharmacophore model.   Continued closely the lowermost binding configuration is predicted by the object score function. For all these 14 molecules, the docking scores above nine were taken into consideration. During this exercise, the obtained nine hits having NCI IDs NSC86071, NSC86473, and NSC111636, have binding energy scores nearer and more than the standard molecule, i.e., Celecoxib (−9 Kcal/mol). Finally, NSC86473 (−10 Kcal/mol) and NSC111636 (−9.3 Kcal/mol), have binding energy scores more than a standard molecule. Consequently, NSC86473 can behave as a powerful inhibitor against Cyclooxygenase 2. The binding energy scores and interactions are shown in the following tables and figures.
The results of docking are shown in Tables 4 and 5.

Interaction studies of screened hits
A ligand interaction diagram was created with discovery studio to analyze the interactions of hits attained from screening as shown in Figure 6. Hydrogen bond interactions were highlighted with green-colored lines. The ligand interactions for a target protein with top energy score hits are illustrated in Figure  6 As per our results, we have identified the lead molecule that has a higher docking score than the standard Celecoxib and was found to satisfy drug-likeliness properties. We are planning to extend our work further by performing animal studies with ethical approval in our future projects.

CONCLUSION
This study was designed to identify the herbs that are active in treating various diseases and is dynamic for future disease control programs. To attain this purpose, different computational tools like pharmacophore-based virtual screening, molecular docking, and binding free energy analysis were employed. A four-point pharmacophore hypothesis was recognized using five phytoconstituents of plant P. major and developed a PM1 in which all the input ligands were properly aligned and were feasible to screen the NCI database. The screened compounds were docked into the active site of 4COX and further Absorption, Distribution, Metabolism, Excretion properties were calculated. The results displayed that the compound NSC86473 may act as a powerful inhibitor against Cyclooxygenase 2 as it has the lowest binding energy than the standard and also had specified pharmacophoric features according to the developed PM1 model. In accordance with earlier findings, it can provide a few insights to research scholars in the future to identify and design new lead molecules with effective anti-inflammatory activity.