Identification of potential biogenic chalcones against antibiotic resistant efflux pump (AcrB) via computational study

Abstract The cases of bacterial multidrug resistance are increasing every year and becoming a serious concern for human health. Multidrug efflux pumps are key players in the formation of antibiotic resistance, which transfer out a broad spectrum of drugs from the cell and convey resistance to the host. Efflux pumps have significantly reduced the efficacy of the previously available antibiotic armory, thereby increasing the frequency of therapeutic failures. In gram-negative bacteria, the AcrAB-TolC efflux pump is the principal transporter of the substrate and plays a major role in the formation of antibiotic resistance. In the current work, advanced computer-aided drug discovery approaches were utilized to find hit molecules from the library of biogenic chalcones against the bacterial AcrB efflux pump. The results of the performed computational studies via molecular docking, drug-likeness prediction, pharmacokinetic profiling, pharmacophore mapping, density functional theory, and molecular dynamics simulation study provided ZINC000004695648, ZINC000014762506, ZINC000014762510, ZINC000095099506, and ZINC000085510993 as stable hit molecules against the AcrB efflux pumps. Identified hits could successfully act against AcrB efflux pumps after optimization as lead molecules. Communicated by Ramaswamy H. Sarma


Introduction
Antibiotics are the metabolic products of microorganisms that are utilized for the prevention or inhibition of the growth of other microorganisms (Mann et al., 2021).Since the discovery of Penicillin in 1928, antibiotics are the first choice of therapeutics against several bacterial and fungal infections (Hutchings et al., 2019).Due to their wide spectrum of activity and potency, antibiotics have been extensively used against microbial infections (C.Reygaert, 2018).Past incidences such as the introduction of methicillinresistant Staphylococcus aureus (MRSA) with reduced susceptibility to Vancomycin witnessed that with the great power of controlling the life-threatening infections, misuse, and overuse of antibiotics are responsible for the increasing rate of antibiotic resistance in microorganisms (Mogasale et al., 2021).Antibiotic resistance is the change in the response of the microorganism agents toward antibiotics, which might lead to the development of antibiotic-resistant organisms (Gil-Gil et al., 2021).The overexpression of antibiotics in humans, agricultural, and veterinary practices led to the increasing emergence of antibiotic resistance in clinically important bacteria (Kasimanickam et al., 2021).Antibiotic resistance has been declared a major health threat of the twenty first century by several prominent regulatory agencies, including the International Monetary Fund (IMF), the World Bank, the World Health Organization (WHO), and the G8.This recognition highlights the urgent need to address this issue, as antibiotic resistance poses a serious risk to global public health (Bloom et al., 2017).Antibiotic resistance is spreading globally and has reached a dangerous level.This is reflected by the growing list of infections like tuberculosis (MDR, XDR, and TDR), and pneumonia, which is very difficult to treat (Gould, 2009).Antibiotic resistance can affect human life, as it leads to increased treatment costs, prolonged hospital stays, and ultimately increased mortality rates (Sivagami et al., 2020).Another reason behind this scenario is the growing gap between the clinical need and the rate of the research and development of new antibiotics, which now a day have become more challenging (Belete, 2019).While there are numerous suggested solutions to address the challenging and critical problem of antibiotic resistance, the practical implementation of these approaches remains a significant issue that must be addressed.It is imperative to find feasible ways to implement these solutions to combat antibiotic resistance (Gao et al., 2021).Preservation of the current antibiotics is one of the easiest ways to overcome the evergrowing problem of antibiotic resistance (Frieri et al., 2017).This preservation strategy is one of the key routes to minimizing the gap between the clinical need and the development speed of antibiotics (Bartlett et al., 2013).
The increasing emergence of antibiotic resistance to clinically available antibiotics has intensified the search for new compounds to potentiate efficacy by inhibiting the drug resistance mechanism (Watkins & Bonomo, 2016).Microorganism shows antibiotic resistance via different mechanisms such as modification of utilized antibiotics, receptor alteration, enzymatic degradation, and antibiotic efflux/efflux pumps (EP) by integral membrane transporters (Matamoros-Recio et al., 2021).However, the antibiotic efflux type mechanism is the fundamental cause of antibiotic resistance in gram-negative bacteria (Kobylka et al., 2020).The fundamental role of antibiotic efflux is to recognize the harmful agents of microorganisms and efflux them out before they get penetrated the protective cell wall of the microorganism (Petchiappan & Chatterji, 2017).Antibiotic efflux carries out the active transport of utilized small molecules to the extracellular medium of the bacterial cell for their protection of themselves (Cacciotto et al., 2018).Antibiotics must cross the hurdles of EP to produce a therapeutic effect.Hence, the inhibition of the EP is necessary to achieve therapeutic effects, and to do so, potential antibiotic adjuvants or new is required, which will bind to the EP and inhibit its working mechanism (Pulingam et al., 2022).Inhibition of the efflux mechanism could be a promising strategy to develop new antibiotics or antibiotic adjuvants to conventional antibiotics (Melander & Melander, 2017).Various strategies are discovered over the years to combat antibiotic resistance, from which potentiation of antibiotic activity of commercially available antibiotics by inhibiting the drug resistance mechanism in microorganisms are tested in this project (Aslam et al., 2018).
The polyphenolic family, which is made up of a huge amount of naturally occurring molecules, includes a significant group of molecules known as chalcones (Orlikova et al., 2011).The beginning of the 1800s and continuing over the next centuries were the initial few attempts to explore the biogenic and synthetic chalcones (Jasim et al., 2021).Chalcones are a, b-unsaturated ketones that are held between two aromatic rings with various substituent arrangements (Salehi et al., 2020).Many naturally occurring plant products, such as spices, vegetables, fruits, and beverages, include the basic chemical scaffold of chalcone (Ouyang et al., 2021).It is possible to develop naturally occurring and synthetic chalcone compounds to serve as lead compounds for the development of antioxidant, antiinflammatory, anticancer, or anti-infective agents, owing to their promising biological activity and safety features (Aboukhatwa et al., 2023;Ni et al., 2004).Previously, some synthesized chalcones are reported to have EP inhibitory activity, leading to increased accumulation of drugs within the cells and overcoming drug resistance (da Cunha Xavier et al., 2021;Ferraz et al., 2020;Freitas et al., 2021;Rocha et al., 2021;Silva et al., 2022).Moreover, there are already reports of EP inhibitors that have a similar structure to chalcones.Resveratrol is a well-known natural polyphenol compound found in grapes and red wine and it has a similar chemical structure to chalcones, with two aromatic rings connected by a three-carbon a, b-unsaturated carbonyl bridge and resveratrol has the EP inhibitory potential (Menezes & Diederich, 2019;M. Santos et al., 2022;Singkham-In et al., 2020).The study reported by Hwang, D. (Hwang & Lim, 2019) indicates that resveratrol also have the ability to controls E. coli growth by inhibiting the AcrAB-TolC EP.Considering the fact that chalcones have the potential to inhibit EPs and share structural similarities with the reported EP inhibitor Resveratrol, a computational screening study has been proposed.This study aims to identify potential hits from a library of diverse biogenic chalcones, which could contribute to the development of novel therapeutic agents with the ability to effectively combat antibiotic resistance.Briefly this study involves systematic exploration of biogenic chalcones against AcrB EP from E. coli through utilizing various computer-aided drug discovery tools like molecular docking, in silico ADMET analysis, pharmacophore modeling, molecular dynamics (MD), and density functional theory (DFT) calculation.With consideration of the medicinal potentials of the chalcones, the current works aim to virtually explore the biogenic chalcone structures against the antibacterial-resistant EP.This attempt will result in the identification of hits or medicinally important ideal fragments, which will help medicinal chemists to develop potent antibacterial agents to control bacterial resistance.

Protein selection, retrieval, and preparation
The 3D crystal structure of the multidrug transporter AcrB with PDB 4DX5 having 1.90 Å resolution was selected from previously reported literature and downloaded in PDB file format from the RCSB Protein Data Bank (Available at https:// www.rcsb.org/)(Berman et al., 2000;Eicher et al., 2012).The downloaded protein structure was then prepared for further study by removing all the water molecules (Al-Sehemi et al., 2021;Basha et al., 2022;Gaikwad et al., 2022).The previously bound co-crystallized ligand groups from the protein structure were separated and saved for further use.The protonation of the cleaned protein structure was done by adding polar hydrogen atoms to define the correct tautomeric and ionization states of all the residues (Al-Sehemi et al., 2022;Rathod et al., 2023).The protein preparation protocol was done with the help of BIOVIA Discovery Studio software (Dassault Syst� emes, 2020).The prepared protein was finally validated using the ProSA-Web and SAVESv6.0 web servers.

Ligand library preparation
In the current study, SMILES and structures of 280 biogenic chalcones were acquired from the online Zinc15 Database (Available at https://zinc.docking.org/substances/home/)and used to prepare the ligand library (Irwin & Shoichet, 2005).The co-crystallized ligand groups separated from the protein structure during the protein preparation protocol were added to the prepared ligand library.The ligand structures from the prepared ligand library were subjected to energy minimization using MMFF94 force field with the steepest descent optimization algorithm and then converted as AutoDock ligand (pdbqt) files (Bagal et al., 2023).The ligand preparation protocol was carried out using the OpenBabel plugin of PyRx 0.8 (O'Boyle et al., 2011).

Molecular docking
The AutoDock Vina package of PyRx 0.8 software was implemented to perform the docking study (Eberhardt et al., 2021;Forli et al., 2016;Rathod et al., 2023).The prepared protein structure of multidrug transporter AcrB (PDB 4DX5) was first imported in PyRx 0.8 and energy was minimized to convert as an AutoDock macromolecule.The energy-minimized structure of ligands and protein were selected in AutoDock vina.The maximized grid box was selected in the Vina workspace to carry out blind docking.Dimensions for the maximized grid box were 65.3510 Å � 130.7687 Å � 88.88903 Å, and with a center of 29.7239 � À 15.0819 � À 16.533 as x � y � z coordinates and exhaustiveness was set at 8 (Forli et al., 2016).The best pose of docked ligands with the highest negative binding affinity was saved and interactions were visualized via BIOVIA Discovery Studio (Dassault Syst� emes, 2020).Moreover, co-crystallized ligands from PDB 4DX5 were re-docked and binding affinities were compared with the identified hits.

Cross-docking study
The cross-docking protocol was performed by docking the identified hits against PDB 4DX5 along with co-crystallized ligand groups from the AcrB proteins of different organisms (Ram� ırez & Caballero, 2018;Rathod et al., 2023;Shen et al., 2021;Thilagavathi & Mancera, 2010).Recently resolved AcrB protein structure (PDB 8FFS) from K. pneumoniae was used for a cross-docking study.The results obtained from the docking of hits and co-crystallized ligands with PDB 8FFS were compared with the results of PDB 4DX5.The crossdocking protocol used was modified to align with the current study, thereby increasing the reliability.

Drug-likeness and in-silico ADMET assessment
Pharmacokinetic (ADMET) profiling of biogenic chalcones having the highest negative binding affinity was computed to investigate their chances of being a drug-like candidate.Lipinski's rule of five (RO5), Veber's rule, Ghose's rule, Egan's rule, and Muegge's rule were applied to find the drug-like molecules.Lipinski's RO5 has been a commonly utilized filter to sort the ligands groups according to lead-likeness and it helps to predict the oral absorptivity of medicinally important chemical structures (Lipinski et al., 1997).SwissADME server was used to determine the drug-likeness properties and the ADMET properties of selected chalcones showing the highest negative binding affinity were predicted using the pkCSM server (Daina et al., 2017;Pires et al., 2015).

Pharmacophore mapping
3D pharmacophore model was generated via PharmaGist and the determination of pharmacophoric features of docked ligands was generated as per the provided input ligand structures (Chen, 2015;Dror et al., 2009;Joshi et al., 2018;Schneidman-Duhovny et al., 2008a, 2008b).One of the ligands was auto-assigned a pivot on which the other ligands were aligned (Schneidman-Duhovny et al., 2008b).Results of PharmaGist were obtained in a tabular form containing information on the number of aligned ligands with their score in decreasing order.Out of all the resulting models, the ligandbased pharmacophore of the model exerting the highest score was used further for analysis (I.V. F. dos Santos et al., 2022;Rathod et al., 2023).

Density functional theory calculation
The DFT method was performed using the Orca 4.2.1 package to theoretically calculate the quantum chemical properties of identified hits (Buddensiek et al., 2021).Input files of the chemical structure were prepared via the Orca-enhanced version of Avogadro (Snyder & Kucukkal, 2021).The ligand structures were optimized using the B3LYP correlation functional method (Amer et al., 2021;Becke, 1988;Lee et al., 1988) and the def2-SVP basis set was applied for calculations (Kausar & Nayeem, 2018).The frontier molecular orbital (FMO) analysis and global chemical reactivity descriptors (electronic properties) were calculated according to the equations of Koopmans' theory (Elkaeed et al., 2022;Flurry, 1968).All output geometries and FMOs were visualized using the orca-enhanced Avogadro software.

Molecular dynamic simulation
The docked complexes having the highest negative binding affinity were subjected to the MD simulation study.The Desmond program from Schr€ odinger Drug Design suite 2017 was used to perform the MD simulation (Abdalla et al., 2022;B. S. Kumar et al., 2022;Shivanika et al., 2022).Each complex was solvated using an orthopodic box with the periodic boundary conditions in the TIP4PEW model.The Na þ or Cl-ions were added to neutralize the overall charge of the complex system.The NPT ensembles were utilized for the minimization and relaxation of the simulated complex system.Subsequently, the  minimization protocol based on the steepest descent method was subjected to all the protein-ligand complexes and then heated from 0 to 310 K with the annealing steps of 2000.Further, the system normalized in an equilibrium state by 1000 steps before the production run (Thangavel et al., 2020).A simulations study was done for 100 ns with a recording interval of 100ps.The temperature and pressure were kept constant throughout the simulations at 310K and 1.01325 bar, respectively.MD trajectory was used to analyze the root mean square deviation (RMSD), root-mean-square fluctuation (RMSF), protein secondary structure elements, protein-ligand contacts with amino acids, and ligand torsion profile.

Protein selection and preparation
Bacterial efflux pumps localized and embedded in the plasma membrane play a major role in the formation of antibiotic resistance (Sun et al., 2014).EPs recognize harmful agents that have penetrated through the protective cell wall of the microorganism and efflux them out before they reach their intended targets (Amaral et al., 2014).Bacterial efflux pumps are divided into five different families in which the EPs from the resistance-nodulation-cell division (RND) family play a major role in drug resistance (Venter et al., 2015).AcrAB-TolC multidrug efflux machinery belongs to the RND family and is present at the outer membrane to periplasm to the inner membrane and also works as a bridge for drug transport (Du et al., 2014;Shi et al., 2019).Acriflavine resistance protein B (AcrB) from the tripartite AcrAB-TolC efflux system is well known for its presence in gram-negative bacteria (Anes et al., 2015).Murakami, S., et al. first reported the resolved symmetrical crystal structure for the AcrB protein (Murakami et al., 2002).Basically, AcrB is a homotrimeric structure located on the inner membrane of gram-negative bacteria (Reading et al., 2020).AcrB is mainly involved in substrate recognition and energy transduction of the EP (Pos, 2009).Inhibition of AcrB will also result in the inactivation of the entire tripartite AcrAB-TolC efflux machinery and could decrease the resistance of previously available antibiotics (Baquero & Alenazy, 2022).In the current work, the 3D crustal structure of multidrug transporter AcrB (PDB 4DX5) resolved by Eicher, T., et al. at a 1.90 Å resolution was selected for the identification of leads against the AcrB EP (Eicher et al., 2012).The retrieved AcrB (PDB 4DX5) structure was prepared and validated before the high-throughput computational screening.The Ramachandran plot was computed for the prepared AcrB (PDB 4DX5) structure to estimate the number of amino acid residues present in the favored regions of the structure.through the generated Ramachandran plot, it is identified that 90.5% of the amino acid residues are present in favored regions of the plot, as represented in Figure 1(a).The ProSA-web server was utilized to estimate the z-scores of all chains present in prepared PDB 4DX5 and z-score was found to be À 7.22.Figure 1  represents the overall model quality of PDB 4DX5 with the X-ray crystallography (light blue) dots and NMR spectroscopy (dark blue) dots, while the z-score of PDB 4DX5 is indicated with a red star.Figure 1(c) represents the local model quality of PDB 4DX5 in that a thick, dark green line denotes the average energy, while a faint green line indicates the average energy.The overall quality factor for the prepared 4DX5 structure estimated via ERRAT for prepared 4DX5 structures was found to be 98.89% (Figure 1d).

Molecular docking
Prepared protein and ligand structures were subjected to a molecular docking study in order to estimate the binding orientation of the ligands with the targeted protein.AutoDock Vina module of PyRx 0.8 was utilized to perform the docking study.The co-crystalized ligand (ERY) group from the PDB 4DX5 and 280 biogenic chalcones were used in the docking study.Redocked co-crystalized ligand (ERY) showed a binding affinity of À 8.2 kcal/mol and compared with this five out of all docked chalcones showed equal or higher binding affinity than redocked ligand.Those five ligands were considered as hits and selection were done on the basis of the binding affinity compared to the co-crystallized ligand group and the quality of interactions (number of H bonds) with the targeted protein.ZINC000004695648, ZINC000014762506, ZINC000014762510, ZINC000095099506, and ZINC000085510993 were the five chalones having the highest negative binding affinity and the strong binding interactions with targeted AcrB protein structure.The chemical structures of identified hits is represented in Figure 2 and Table 1 represents the binding affinity with the binding interactions, the distance of interaction, and the type of interaction observed between the identified five hits and the AcrB structure.ZINC000004695648 showed a binding affinity of À 8.5 kcal/mol and the molecule formed three conventional hydrogen bonds with the amino acid residues such as GLU273, TYR772, and ARG620.ZINC000004695648 formed hydrophobic interactions with the ALA777, TRP187, TYR275, and MET774, respectively.In hydrophobic interactions, TRP187 and TYR275 showed p-p stacked and p-p T shaped bonds.There is a possibility that docked complex 4DX5-ZINC000004695648 will show stable interactions as the compound showed strong interactions.The 2D and 3D binding interactions between 4DX5 and ZINC000004695648 are represented in Figure 3.
ZINC000014762506 showed a binding affinity of À 8.5 kcal/mol and the molecule formed four conventional hydrogen bonds with the residues such as HIS505, LEU359, LYS509, and THR524, respectively.The molecule showed hydrophobic interactions with amino acid residues such as GLU521, ILE500, ALA421, ARG418, and HIS525.In hydrophobic interactions, GLU521 showed p -Anion type of interaction, while ILE500, ALA421, ARG418, and HIS525 showed Alkyl and p -Alkyl interactions.Figure 4 represents the 2D and 3D binding interactions of the 4DX5-ZINC000014762506 complex.
ZINC000014762510 showed a binding affinity of À 8.4 kcal/mol and formed one conventional hydrogen bond, which is ARG185 amino acid residue.ZINC000014762510 observed Pi-interactions like ARG780, TYR275, TRP187 and PRO50, respectively.In Pi-interactions, ARG780 and TYR275 showed p-cation and p-sigma, and TRP187 and PRO50 showed p-Alkyl interactions.Figure 5 showed 2D and 3D interactions between 4DX5 and ZINC000014762510.
ZINC000095099506 showed a binding affinity of À 8.4 kcal/mol and the molecule formed five conventional hydrogen bonds with the THR463, GLY460, MET862, ALA670, and LEU564 amino acid residues while observed carbonhydrogen bond with SER462.LEU674 and ALA677 showed hydrophobic interactions with ZINC000095099506.The 2D and 3D binding interactions between 4DX5 and ZINC000095099506 are represented in Figure 6.
ZINC000085510993 exerted a binding affinity of À 8.2 kcal/mol and the ligand showed a conventional hydrogen bond with the GLN34.The hydrophobic interactions were formed with ALA299, ILE38, and ILE671.The 2D and 3D binding interactions between 4DX5 and ZINC000095099506 are represented in Figure 7. Performed docking study provided initial confirmation of strong binding between the identified hits and the AcrB (PDB 4DX5) of E. coli.In addition to the docking study, we performed a cross-docking study of the identified hits with the AcrB (PDB 8FFS) EP of K. pneumonia.This was done to compare the binding affinities and assess the binding potential of the hits with the same EP from a different bacterium.The results of performed cross-docking study indicated that the identified hits also have a strong affinity with the AcrB (PDB 8FFS) EP of K. pneumonia.The binding affinity of the hits with AcrB (PDB 8FFS) ranged between À 8.3 and À 10.3 kcal/mol, and we observed good binding interactions as represented in Table 1.ZINC000004695648 exhibited the highest negative binding affinity with the cross-docked AcrB (PDB 8FFS) EP.We also used co-crystallized ligands from each targeted protein in our cross-docking study, and all hits showed higher binding affinity than the co-crystallized ligands (ERY, MIY).These findings suggest that the identified hits have binding potential with AcrB EP of multiple bacterial species, including E. coli and K. pneumonia.The cross-docking study provided additional evidence of strong binding affinity and interaction of the hits against the AcrB EP, indicating their potential for use as effective therapeutic agents.Though the docking study indicated the formation of strong bonds, the docking protocol has many limitations, as it does not consider the flexibility of the protein backbone and the absence of explicit water molecules (Rathod et al., 2023;Vargiu & Nikaido, 2012).Previously, Vargiu, A. V. (Vargiu & Nikaido, 2012) reported the role of flexibility and explicit water molecules in the case of docking of the AcrB pump with its inhibitors.Further, the MD simulation study of the identified hits is necessary due to the limitations of docking studies to predict the binding interaction and energy in the presence of water molecules and the effect of mobility on the flexibility of protein structure during the binding.

Drug-likeness and in-silico ADMET assessment
Drug-likeness prediction of identified hits is a key consideration as it gives information related to the impact of the physicochemical properties of the ligands on their bioavailability (Bickerton et al., 2012).Various sets of rules such as Lipinski's rule of five, Veber's Rule, Ghose's Rule, Egan's Rule, and Muegge's Rule were applied to the identified hits in order to study the drug-likeness of those hit molecules (Lipinski et al., 1997;Pathania & Singh, 2021;Veber et al., 2002).All five hits satisfied the drug-likeness prediction with acceptable values for predicted physicochemical properties.Every hit molecule showed zero violations in all the implemented drug-likeness rules indicating the good bioavailability of the studied hits.The predicted physicochemical properties and drug-likeness profile of the identified hits are represented in Table 2.The physicochemical properties also influence the pharmacokinetic profile of the ligands.Lipophilicity is one of the important parameters while studying the passive diffusion of compounds across the cell membranes, solubility, drug-receptor interactions, metabolism, and toxicity (van de Waterbeemd & Gifford, 2003;Wan, 2013).Along with the lipophilicity, other physicochemical properties include molecular size, molar refractivity, number of rotatable bonds, number of hydrogen bond acceptors/donors and total polar surface area of the ligands also influence the absorption, distribution, metabolism, elimination, and toxicity profile (Karlgren & Bergstr€ om, 2015).The predicted pharmacokinetic profile of the hit molecules is represented in Table 3. Identified hits showed intestinal absorption of more than 80%.The prediction of the volume of distribution (VDss) of all the hits was found to be acceptable except for ZINC000095099506 as it failed to satisfy the acceptable range for VDss.The blood-brain barrier (BBB) permeability prediction indicated that ZINC000004695648, ZINC000014762506, ZINC000014762510, and ZINC000085510993 will be poorly distributed to the brain as they showed logBB < À 1.While ZINC000095099506 showed logBB more than À 1 indicating readily permeation in the BBB.Similar to the BBB permeation study, CNS permeability (log PS) was predicated and results indicated that all the hits except the ZINC000095099506 failed to satisfy the CNS permeability.The metabolism profile of all five hit molecules was found to be acceptable.The excretion profile of the hits was also found in an acceptable range.It is well known that chlcones are a class of organic compounds having wide range of biological activities, including anti-inflammatory, anti-cancer, anti-microbial, and anti-diabetic properties.However, they are also known to have some toxicity aspects that may make them difficult to transform into drugs and one of the main toxicity aspects of chalcones is their potential to cause liver damage (Zhuang et al., 2017).Previously reported studies have shown that some chalcone derivatives can induce hepatotoxicity by causing oxidative stress and inflammation, and by interfering with liver enzymes that are involved in drug metabolism (Escribano-Ferrer et al., 2019;Narayanapillai et al., 2014;Niu et al., 2016).Moreover, they have also been found to be cytotoxic, meaning they can damage or kill cells (Gomes et al., 2017;Kuber Banoth & Thatikonda, 2020;Mahapatra et al., 2015;Ouyang et al., 2021;Rozmer & Perj� esi, 2016).This cytotoxic potential of chalcone can be beneficial in certain contexts, such as in the treatment of cancer, where chalcones have shown promise as anti-cancer agents (Bandgar & Gawande, 2010;Orlikova et al., 2011;Salehi et al., 2020).However, the cytotoxicity of chalcones can also be a concern in drug development, as it may limit their therapeutic window and increase the risk of side effects including mutation in DNA.Although the chalcones have previously shown great promise as therapeutic agents, their toxicity aspects must be carefully considered in the drug development process.Hence, the AMES toxicity and hepatotoxicity of the identified hits were screened using insilico models.All hits passed the in-silico toxicity profiling indicating no AMES toxicity.Hence, these hits do not have mutagenic in nature and possibly do not cause cancer.In addition to that all the hits never showed hepatotoxicity.Drug-likeness and in silico ADMET profiling provided valuable insights regarding the bioavailability of the identified hits.Results for all the studied parameters were found to be in acceptable regions.

Pharmacophore mapping
Pharmacophore features including hydrogen bond acceptor, hydrogen bond donor, lipophilic and aromatic contacts play a major role in the biological response after happening protein and ligand interactions (Katiyar et al., 2021).Hence, pharmacophore modeling helps to map the potential structural features present in ligands (Dror et al., 2009).In the current study, pharmacophore modeling of docked chalcones was done to understand the structural features of the ligands having the highest negative binding affinity.The scored sets of pharmacophore feature present in the screened chalcone structures were generated via the online PharmaGist server.Initially, PhrmaGist aligned all the input chalcone structures together and one form all kept as a rigid (pivot) ligand.Generated pharmacophore model of aligned ligands with their score was obtained from the PharmaGist server.The pharmacophore model with the highest-ranked score of 24.875 was retrieved in mol2 file format for spatial feature   obtained pharmacophore models were identified according to the quantitative characteristics like pharmacophore features and their geometric distance as shown in Figure 8.

Density functional theory (DFT)
DFT study was performed to estimate the FMOs present in the identified hits and to correlate the docked binding interactions with the FMOs.Higher occupied molecular orbital (HOMO) with the power to donate an electron and the lower occupied molecular orbitals (LUMO) having the ability to accept electrons present in the identified hits were studied with the DFT approach (Rochlani et al., 2023).HOMO energy, LUMO energy, and the energy gap between the HOMO-LUMO help to predict the mechanistic view of the protein-ligand interactions.Mumit, M.A., et al. previously reported that the compounds having a low energy gap between HOMO-LUMO lead to higher biological activity as well as higher chemical reactivity (Mumit et al., 2020).Here in this current study, the interactions identified after the molecular docking study was correlated with the HOMO and LUMO regions of the identified hits.The estimated FMOs of hit molecules and their energy gap are represented in Figure 9, while Table 5 represents the calculated global chemical reactivity descriptors of the studied hit molecules.The greater value of the HOMO-LUMO energy gap indicates lesser reactivity (more stability) of a molecule (Patial & Cannoo, 2021).ZINC000004695648 showed the lowest HOMO-LUMO energy gap (3.709 eV) compared to other hits, which indicates the higher chemical reactivity, may have good biological activity, and has less chemical stability.ZINC000085510993 showed the highest energy gap between HOMO-LUMO energy (4.332 eV) indicating lower reactivity and higher chemical stability among all other hits.The order of chemical stability of identified hits is ZINC000085510993> ZINC000095099506> ZINC000014762510> ZINC000014762506> ZINC000004695648.The HOMO region (as shown in Figure 9) of the ZINC000004695648 showed the maximum hydrogen interaction during the docking study (as shown in Figure 3) and hydrogen donors were present in that region to form interaction with the hydrogen acceptor part of the interacted residues of the targeted protein.The LUMO region of the ZINC000004695648 showed hydrophobic interaction with the amino acid residues of the targeted protein.In both the case of ZINC000014762506 and ZINC000014762510, most of the interactions (as shown in Figure 4 and Figure 5) were exerted from the LUMO and HOMO regions (as shown in Figure 9).Both HOMO and LUMO regions of the ZINC000095099506 (as shown in Figure 9) led to the formation of hydrogen bonds with targeted protein (as shown in Figure 6) indicating the involvement of hydrogen donors and acceptors of chalcone during the formation binding interaction.GLN34 residue of the targeted protein formed two hydrogen bonds with the oxygen (as shown in Figure 7) present in the HOMO regions of the ZINC000085510993 (as shown in Figure 9).The electrophilicity values of hits helped to predict the binding capacity with biomolecules as the ligands with higher electrophilicity index have a higher binding capacity with biomolecules (S.Kumar et al., 2018;Mumit et al., 2020).ZINC000004695648 and ZINC000014762506 showed higher electrophilicity values and the binding affinity of these molecules was also found to be higher.All hits showed good values for calculated global chemical reactivity descriptors.The performed DFT study aided in estimating the electronic structure of the hit molecules and guided the correlation of these electronic features with the  binding interactions observed between protein-ligand complexes.This insight provided valuable information regarding the role of electronic properties in the binding process, and may ultimately contribute to identify more effective therapeutically important molecules.

Molecular dynamic simulation
Five hit molecules were identified after the docking study based on the binding affinity and the binding interactions.Further, the identified hits were evaluated for their drug-like properties, and pharmacokinetic profile and obtained interactions were correlated with the computed FMOs of the hit molecules.The docking study provided insights regarding the binding orientation and helped to sort the hits from the subjected library.The docking study has some limitations as it fails to give information on the effect of atomic mobility or dynamic behavior of the docked complex (Bagal et al., 2023;Vargiu & Nikaido, 2012).Hence, the binding of the identified hits with the targeted AcrB (PDB 4DX5) structure was achieved using the MD simulation study in order to explore the effectiveness of the formed interactions under the mobile condition.This approach helped to study the dynamic nature of the Ca atoms of the targeted protein, bounded ligand, and overall docked protein-ligand complex.Deviation, fluctuations, and the number of hydrogen bonds formed between complexes were studied using over 100 ns MD simulation run.MD trajectory of each simulated complex was subjected to statistical analysis via estimation of RMSD and RMSF.In addition to that, the formation of hydrogen bonds between protein-ligand complexes with increasing time was also determined.Figure 10  RMSF estimation helped to explore the residual fluctuations during the MD simulation.Figure 10(f-j) represents the plotted graphs of RMSF of each simulated protein-ligand complex.The majority of the amino acid residues present in the utilized AcrB structure showed RMSF values less than �5 Å indicating negligible fluctuations except for residues such as PRO200, VAL201, ASP202, VAL203, ILE204, and THR205 showed the highest fluctuation during the simulation study.The deviation observed during the RMSD plot may occur due to the higher fluctuation of residue present at 200-205 during the simulation run.The fraction of formation of hydrogen bonds between ZINC000004695648-AcrB occurred with the GLN125, ARG185, and ARG780.Hydrophobic interactions and water bridges were found to be maximum in the case of ZINC000004695648-AcrB as represented in Figure 10(k).Hydrophobic interactions majorly contribute to the drug-receptor interactions (Ferreira De Freitas & Schapira, 2017).ZINC000014762506-AcrB complex showed the formation of hydrogen interactions with the residues such as GLN360, ASP504, GLY506, ALS509, ASN517, GLU521, ARG969, and ARG973 as represented in Figure 10(l).SER48, TYR49, GLY51, THR85, ARG185, and ARG620 were involved in the formation of hydrogen bonds between the ZINC000014762510-AcrB during the 100 ns MD simulation.Figure 10(m) represents the fractions of interactions between the ZINC000014762510-AcrB complex system.ZINC000095099506-AcrB showed the highest number of formation of hydrogen contacts and SER462, THR463, PRO565, ALA670, THR837, MET862, GLN865, and GLU866 were involved in the formation of hydrogen contacts as represented in Figure 10(n).Amino acids such as GLN34, SER134, SER135, GLN569, LEU668, PRO669, ALA670, and VAL672 were involved in the formation of hydrogen contacts between the ZINC000085510993-AcrB complex system as shown in Figure 10(o).
The protein-ligand contacts formed over 100 ns were estimated to identify the key interacting residues between the simulated protein-ligand complex.In the case of ZINC000004695648-AcrB, GLN125, ARG185, and TYR275 maintained binding with the ZINC000004695648 throughout the 100 ns simulation as shown in Figure 11(a).GLN125 and ARG185 were involved in the formation of hydrogen bonds with ZINC000004695648, while TYR275 formed hydrophobic interaction.This formed strong and consistent binding stabilized the complex system and ligand tightly bounded in the docked binding pocket of AcrB.ASP504 residue formed hydrogen bonds and showed consistent interaction throughout the MD simulation between the ZINC000014762506-AcrB complex as represented in Figure 10(l) and Figure 11(b).SER48, THR85, ARG185, and ARG620 showed consistent interaction and formed hydrogen bonds between the ZINC000014762510-AcrB complex during the MD simulation as shown in Figures 10(m) and 11(c).SER462, THR463, PRO565, ALA670, MET862, GLN865, and GLU866 showed consistent contacts between the simulated ZINC000095099506-AcrB complex as shown in Figure 11(d).Plotted protein-ligand contacts between the simulated ZINC000085510993-AcrB indicated that the formed hydrogen bond with the PRO699 has stable binding throughout the simulation study as shown in Figure 11(e).The snapshots were taken at 0, 25, 50, 75, and 100 ns for each simulated proteinligand complex system, and through this, it was identified that the docked hit molecules never shifted the binding pocket throughout the 100 ns simulation confirming the stable interactions in dynamic conditions (Supplementary Figure S1a-e).

Conclusion
Performed high-throughput computational screening of the targeted protein i.e. bacterial AcrB EP, and the selected biogenic chalcones via molecular docking.Five hit molecules were identified from the results of the molecular docking and drug-likeness prediction.ZINC000004695648 showed the highest negative binding affinity.Identified hits were then subjected to pharmacokinetic profiling and all the hits satisfied ADMET requirements without showing (AMES) toxicity.Pharmacophore mapping revealed the geometric framework between the available pharmacophores in the hits.The DFT study of the hits was done in order to estimate the FMOs of the molecules and the results were correlated with the docked interactions.The HOMOs and LUMOs of the hits actively participated in the formation of the binding interactions with the AcrB (PDB 4DX5).The MD simulation of the hit molecules revealed the stable binding of all hit molecules in the binding pocket of the AcrB (PDB 4DX5) throughout the 100 ns.The ZINC000004695648 exerted lesser deviation with the formation of strong binding during the MD simulation.However, all the hits showed stable information with strong interactions and could be used as leads against AcrB EP.This work is a primary confirmation of the stability of the identified hits and further, these findings will guide medicinal chemists to design molecules to overcome the danger of antibiotic resistance caused by bacterial EPs.

Figure 1 .
Figure 1.Validation and quality evaluation of prepared AcrB (PDB 4DX) structure using (a) Ramachandran plot, (b) overall model quality, (c) local model quality observed, and (d) overall quality factor.

Figure 9 .
Figure 9. Frontier molecular orbital including their HOMO energy, LUMO energy, and HOMO-LUMO energy gap.
(a) represents the RMSD plot of the ZINC000004695648 fit to protein and it indicated the linearity

Table 1 .
Binding affinity and interactions along with the distance between targeted proteins and docked ligands.
analysis.The BIOVIA Discovery Studio was implemented to analyze the highest-scored pharmacophore model (Dassault Syst� emes, 2020).Three pharmacophore features were observed in which two aromatic rings and one hydrogen donor are present.The obtained quantitative characteristics of the highest-ranked pharmacophore model with respective scores are represented in Table4.Using the ZincPharmer, docked chalcones having similar spatial coordinates for

Table 2 .
Predicted physicochemical and drug-likeness properties of identified hits.

Table 3 .
Predicted pharmacokinetic properties of identified hits.

Table 4 .
Generated pharmacophore models of docked chalcone having good binding affinity from PharmaGist tool.

Table 5 .
Calculated orbital energy, HOMO-LUMO energy gap, and global chemical distributors for the hit molecules.