The 3D structure of InlA protein from L. monocytogenes with resolution of 1.60 Å (PDB ID: 1O6T) (Figure 1) was obtained in PDB format from the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (https://www.rcsb.org/, accessed 12 February 2024) (Berman et al., 2000). This protein was then subjected to preparation prior to docking using the AutoDock Tools 1.5.6 by eliminating water molecules, heteroatoms and adding polar hydrogens and Kollman charges. Finally, the prepared target was saved in pdbqt format (Morris et al., 2009).

In the case of ligands, 80 terpenes from different plant species with a range of medicinal properties were selected based on ‘Lipinski’s rule of five’ (Table 1) and the 3D structures of these compounds were obtained in SDF format from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/, accessed12 February 2024) (Benet et al., 2016; Kim, 2021). The compounds were energy minimized using the steepest descent algorithm of Avogadro software and Open Babel program was further utilized to generate their pdbqt files (Hanwell et al., 2012; O'Boyle et al., 2011).

AutoDock Vina program was used to dock all the 80 ligands against InlA protein inorder to gain an insight into the various interactions involved during the binding process (Trott and Olson, 2010). This program employs an efficient optimization technique that depends on a special scoring function (combination of empirical and knowledge-based approaches) and a gradient-based local search genetic algorithm to predict the binding modes thereby producing reliable information concerning the receptor and ligand interactions (Jaghoori et al., 2016; Sarkar et al., 2024). The binding affinities were estimated for each InlA-ligand complex. The protein was kept rigid and the ligands were flexible throughout the docking study. With the grid box centered at x = −15.289, y = −1.522 and z = 26.283 and dimensions set at points 126 × 82 × 126 separated by 1 Å, all the input files needed for docking were created in ADT 1.5.6. Based on the best docking scores, the top ten compounds from among the 80 terpenes were chosen for further interaction analysis. A compound with greater negative binding affinity value revealed its stronger binding to the target receptor (Rolta et al., 2022).

Different interactions and the amino acid residues involved during the binding of InlA protein with the top 10 compounds were examined using the BIOVIA discovery studio (DS) visualizer and Ligplot+ software (Studio, 2008; Laskowski and Swindells, 2011). In particular, Ligplot+ was used to study the two-dimensional (2D) protein-ligand interactions involving hydrogen and hydrophobic interactions, and DS visualizer was used to investigate the different forms of hydrogen and hydrophobic bonding that were present in the complexes. Additionally, PyMOL software was also utilized for the three-dimensional (3D) visualization of receptor-ligand complexes (Yuan et al., 2017).

The pharmacokinetic properties and the efficacy of the top 10 compounds obtained upon docking were computed by the admetSAR server (http://lmmd.ecust.edu.cn/admetsar2/, accessed 15 February 2024). The server predicts the carcinogenicity, blood-brain barrier penetration (BBB), Human Intestinal Absorption (HIA), subcellular localization, human oral bioavailability, toxicity and various other parameters of drugs or drug-like compounds that play a crucial role in the drug development process (Cheng et al., 2012). Additionally, the PASS (Prediction of Activity Spectra for Substances) online server (https://www.way2drug.com/passonline/, accessed 18 February 2024) was used to further explore the possible therapeutic activities of these compounds (Filimonov et al., 2014). This server estimates a compound’s expected activity spectrum as ‘probability of activity (Pa)’ and ‘probability of inactivity (Pi)’. Pa and Pi have values ranging from 0.000 to 1.000 and the compound activity is considered possible only if Pa > Pi (Chakraborty et al., 2016).

MD simulation of protein-ligand complexes with the best docking scores and the protein alone (InlA) was carried out using the GROMACS software (version 2023) for a 100 nanosecond (ns) timescale to gain an insight into their stability as well as the conformational behaviour upon ligand binding (Bauer et al., 2022). CHARMM27 forcefield of GROMACS and SwissParam server were used to generate the protein and ligand topologies (Bjelkmar et al., 2010; Bugnon et al., 2023). Both the protein and complex systems were further solvated in cubic box and neutralized by adding Na⁺/Cl⁻ ions. Post solvation and neutralization, using steepest descent algorithm the systems were energy minimized to eliminate the steric clashes and then equilibrated utilizing the NVT and NPT ensembles so as to maintain the temperature (300K) and pressure (1atm) of the systems (Berendsen et al., 1984; Andersen, 1980; Petersen, 1995). Ultimately, MD run was initiated for 100 ns and the root mean square deviation (RMSD), root mean square fluctuation (rmsf), radius of gyration (Rg), solvent accessible surface area (SASA) of the protein-ligand and protein systems were determined. Furthermore, the protein-ligand complexes were subjected to principal component analysis (PCA). Xmgrace software was used to obtain the 2D plots of all the aforementioned parameters (Turner, 2005).

Post MD simulation, MM/GBSA (Molecular Mechanics/Generalized Born Surface Area) binding free energy of the protein-ligand complexes with the best docking scores were estimated for the last 20 ns MD trajectory using the gmx_MMPBSA program (v1.6.0) (Valdés-Tresanco et al., 2021). In MM/GBSA method, the total binding free energy is determined by the following equation: ∆ G bind = G complex   −   G protein   −   G ligand ∆ G bind = ∆ G gas + ∆ G sol   −   T ∆ S where G_complex is the free energy of the protein-ligand complexes, while G_protein and G_ligand are the free energies of the protein and ligands. The conformational entropy upon ligand binding is represented by T∆S. ∆G_gas is the sum of internal energies (ΔE_int - bond, angle, dihedral), van der Waals energy (ΔE_vdW) and electrostatic component of internal energy (ΔE_ele). ∆G_sol is the combination of polar (∆E_GB) and non-polar (∆E_SURF) components of solvation energy.

All the selected compounds obeyed ‘Lipinski’s rule of 5’ without any violation suggesting their possibility of oral administration. According to this rule, a drug or a drug-like compound can be orally administered only if it satisfies the following criteria: molecular weight ≤500 g/mol, HBD ≤5, HBA ≤10 and LogP ≤ 5 (Walters, 2012; Saini et al., 2021). The molecular weight of all the terpenes varied from 136.23 to 398.8 g/mol. Alpha-Fenchene and Isoterpinolene had the lowest molecular weight of 136.23 g/mol while the highest value of 398.8 g/mol was shown by Chlorojanerin. Moreover, the HBDs and the HBAs of all the compounds ranged from 0 to 3 and 0 to 7, respectively. The lowest logP value of 0.1 was noted for Janerin whereas Cespitularin A had the maximum logP value of 4.4. These observations collectively demonstrate the potential of the chosen terpenes in oral delivery.

All the 80 terpenes were blindly docked against the InlA protein with the grid box covering the entire protein surface. Additionally, a reference ligand Ampicillin, which is one of the most widely used antibiotic to treat illnesses caused by L. monocytogenes, was also docked against the virulence protein in order to determine whether the terpenes were more or less effective compared to the drug. We observed exceptional binding affinity values for each protein-ligand complexes and the top 10 compounds (Figure 2) with the highest scores were finalised for further analysis. The important amino acid residues and the different interactions involved in the binding of the best 10 ligands with the InlA protein from DS visualizer and Ligplot+ software are displayed in table 2. The binding affinity values obtained for all the 80 terpenes and Ampicillin are highlighted in table 3.

According to numerous studies, higher negative binding affinity values often imply a ligand’s increased capacity to inhibit the disease-causing target protein (Deepasree and Subhashree, 2023; Sahu and Pattanayak, 2019; Islam et al., 2021; Correa-Basurto et al., 2015). The top 10 terpenes had varied binding affinities ranging from −8.8 to −9.5 kcal/mol. Among all the selected terpenes, Bipinnatin (lig.13) and Epispongiadiol (lig.54) were found to have displayed the best interaction with InlA protein with a binding affinity value of −9.5 kcal/mol followed by Artemisinin and Bonducellpin G showing the next highest docking score of −9.1 kcal/mol. In DS visualizer, Bipinnatin displayed 2 hydrogen bonds with Ser173 and Asp213 amino acid residues of InlA along with 3 hydrophobic interactions (2 pi-alkyl and 1 pi-sigma bonds) with Phe150 (Figure 3A). Similarly, the ligand13-InlA interaction analysis with Ligplot+ revealed a maximum of 3 hydrogen bonds to Ser172, Ser173 and Asp213 residues formed at a distance of 2.95 Å, 3.02 Å and 3.23 Å. Here, Asn129, Phe150 and Asn151 had hydrophobic interactions with Bipinnatin (Figure 3B). The 3D representation of ligand13-InlA complex is shown in Figure 3C. In the case of Epispongiadiol, the ligand showed 1 conventional and 1 pi-donor hydrogen bond with Asn129, 1 carbon hydrogen bond with Ser172 and 3 pi-alkyl bonds with Phe150 in DS visualizer (Figure 4A) while 2 hydrogen bonds to Asn129 and Glu 170 (formed at a distance of 3.29 Å and 3.04 Å) were noted in Ligplot+ (Figure 4B). Moreover, the compound had hydrophobic interactions with Asn151, Phe150 and Ser172 residues of the virulence protein. Figure 4C depicts the 3D view of ligand54-InlA complex.

Artemisinin (ligand 27) that exhibited the next best affinity value (−9.1 kcal/mol) had 1 hydrogen bond with Ser172, 1 pi-alkyl and 1 pi-sigma bond with Phe150 residue of the protein (DS visualizer- Figure 5A). No hydrogen bonds were revealed upon the visualization of ligand 27-InlA complex in Ligplot+ and only hydrophobic interactions to Ser173, Ser172, Phe150, Glu170 and Asp213 were identified (Figure 5B). Also, Bonducellpin G (ligand 75) sharing the same value as that of ligand 27 possessed 5 hydrogen bonds with residues including Asn151, Asn129, Ser173, Ser215 and Asp213 as well as 1 pi-alkyl bonding with Phe150. The ligand also had an additional pi-pi stacked bonding with Phe150 residue (DS visualizer- Figure 6A). Figure 6B (Ligplot+) shows that ligand 75 formed hydrogen bonds with Asn129 and Ser173 each at a distance of 3.15 Å and 3.23 Å. Moreover, this compound had hydrophobic interactions with residues, namely, Ser215, Ser216, Ser172, Phe150 and Glu170.

Ligand 9 (Inuchinenolide C), ligand 39 (Arbusculin C), and ligand 78 (Scabequinone) had a docking score of −9.0 kcal/mol respectively. Out of the best 10 compounds, Axinysone C (ligand 44) showed the lowest binding affinity value of −8.8 kcal/mol with InlA protein indicating its weak interaction with the target protein when compared to the other compounds in the top 10 list. In contrast to the top 10 terpenes with the best docking scores, Ampicillin displayed a remarkably low binding affinity value of −8.7 kcal/mol. This result reveals that the top 10 terpenes exhibited stronger interaction with InlA in comparison to Ampicillin, pointing to the terpenes’ potential to be more bactericidal than the drug. Figures 7A, B (2D images) and Figure 7C (3D images) depict the various interactions involved between Ampicillin and InlA. Therefore, based on the docking results, Bipinnatin and Epispongiadiol with the highest docking scores could possibly act as potential antibacterial agents that can suppress the virulence protein InlA of L. monocytogenes. Furthermore, the key residues that were observed predominantly in the docked complexes were Ser172, Ser173, Phe150 and Asp213. This may also highlight the vital role played by these residues in inhibiting the protein upon binding with the antibacterial drugs thereby turning the virulence protein inactive.

Post docking, the 10 terpenes were subjected to ADMET and biological activity analysis for a deeper understanding of their pharmacokinetic profiles and wide spectrum of therapeutic functions particularly the antibacterial activity. Some of the significant pharmacokinetic characteristics and therapeutic activities observed for the top 10 terpenes are shown in tables 4, 5.

One of the biggest challenges faced during oral drug development is assessing the drug’s capability to cross the intestinal epithelial barrier that estimates the rate and extent to which the drug is absorbed into the human body eventually impacting its bioavailability (Rudrapal et al., 2022; Aja et al., 2021; Qaddir et al., 2017). For this reason, both HIA and Caco-2 intestinal cell permeability of the compounds were estimated. Except for Tanacetin and Arbusculin C, the remaining terpenes exhibited positive response for HIA indicating that each of those terpenes can undergo intestinal absorption upon oral administration. Also, Inuchinenolide C and Bonducellpin G did not show caco-2 permeability while all other terpenes had positive response for the same. P-glycoprotein, an ABC transporter is a protein that actively engages in a number of biological processes including drug absorption, metabolism, distribution and excretion thereby detoxifying and protecting the body from harmful substances. It was discovered that none of the 10 compounds inhibited this protein. With the exception of Tanacetin and Arbusculin C, the remaining terpenes revealed BBB penetration. Moreover, all the terpenes were observed to be non-carcinogens. The subcellular localization of the terpenes were found to be in the mitochondria except for Furodysinin in plasma membrane. In the case of acute oral toxicity, the terpenes were categorized into four levels-mainly category I, II, III and IV based on their LD50 values. Category I consists of compounds with LD50 ≤ 50 mg/kg while category II has compounds with LD50 > 50 mg/kg but less than 500 mg/kg. Compounds with LD50 values higher than 500 mg/kg but less than 5,000 mg/kg are grouped into category III and those with LD50 value greater than 5,000 mg/kg comes under category IV. Generally, compounds that fall in category III and IV are often considered non-toxic while II is slightly toxic and category I compounds are highly toxic (Guan et al., 2019). In this study, none of the terpenes belonged to category I for acute oral toxicity. However, Inuchinenolide C and Bipinnatin displayed category II toxicity indicating their slightly toxic nature. Aside from Artemisinin and Tanacetin that belonged to category IV, all other terpenes were placed in category III. Thus, the top 10 compounds were predicted to have satisfactory ADMET findings for the major parameters.

The biological activities possessed by the 10 terpenes were further explored using the PASS web server. Antineoplastic, antiprotozoal, immunosuppressant, antiinflammatory, antifungal, antiviral and antileukemic activities were some of the predominantly expressed functions of the 10 terpene compounds obtained from the server and the Pa value was significantly higher than Pi in all the activity predictions. Furthermore, all the terpenes except for Inuchinenolide C, Furodysinin and Axinysone C exhibited antibacterial activity supporting the core aim of the study.

Bipinnatin (ligand 13) and Epispongiadiol (ligand 54) showcased the highest binding affinities (−9.5 kcal/mol) based on the docking data obtained and hence MD simulation was performed for the complexes of these compounds with InlA protein as well as the apo form of InlA over a 100 ns timescale. Here, the RMSD, RMSF, Rg and SASA of the complexes and the apo protein were computed as a function of time inorder to monitor the variations in their stability, conformation and the amino acid residue interactions (Joshi et al., 2021; Surti et al., 2020; Nagamalla et al., 2022).