A computational odyssey: uncovering classical β-lactamase inhibitors in dry fruits

Abstract In the antibacterial arsenal, β-lactams have held a prominent position, but increasing resistance due to unauthorized use and genetic factors requires new strategies. Combining β-lactamase inhibitors with broad-spectrum β-lactams proves effective in combating this resistance. ESBL producers demand new inhibitors, leading to the exploration of plant-derived secondary metabolites for potent β-lactam antibiotics or alternative inhibitors. Using virtual screening, molecular docking, ADMET analysis, and molecular dynamic simulation, this study actively analyzed the inhibitory activity of figs, cashews, walnuts, and peanuts against SHV-1, NDM-1, KPC-2, and OXA-48 β-lactamases. Using AutoDock Vina, the docking affinities of various compounds for target enzymes were initially screened, revealing 12 bioactive compounds with higher affinities for the target enzymes compared to Avibactam and Tazobactam. Top-scoring metabolites, including Oleanolic acid, Protocatechuic acid, and Tannin, were subjected to MD simulation studies to further analyze the stability of the docked complexes using WebGro. The simulation coordinates, in terms of RMSD, RMSF, SASA, Rg, and hydrogen bonds formed, showed that these phytocompounds are stable enough to retain in the active sites at various orientations. The PCA and FEL analysis also showed the stability of the dynamic motion of Cα residues of phytochemical-bound enzymes. The pharmacokinetic analysis of the top phytochemicals was performed to analyze their bioavailability and toxicity. This study provides new insights into the therapeutic potential of phytochemicals of selected dry fruits and contributes to future experimental studies to identify βL inhibitors from plants. Communicated by Ramaswamy H. Sarma


Introduction
The spread of antimicrobial resistance (AMR) on a global scale not only endangers millions of lives annually, but also seriously compromises human health, food security, and societal development (Vasudevan et al., 2022).Understanding the resistance mechanisms employed by bacteria is crucial for the development of novel antibacterial drugs.In addition to inherent resistance, bacteria have progressively evolved and acquired diverse antibiotic resistance mechanisms.These mechanisms can be categorized as: (a) modification of bacterial target proteins, (b) reduction of antibiotic penetration into bacterial cells, and (c) direct modification of antibiotics themselves (Wang et al., 2021).The most prevalent way for bacteria to hydrolyze b-lactam antibiotics and develop resistance is by producing b-lactamases (bL) (Tooke et al., 2019).b-lactamases enzymatically cleave the amide bond in the b-lactam ring, rendering the antibiotics inactive and serving as essential defense mechanisms, particularly in Gram-negative bacteria (Bahr et al., 2021;Carcione et al., 2021).
Based on protein homology and mode of action, Ambler divides b-lactamases into four classes (A, B, C, and D) and two major families: metallo b-lactamases (MBLs) and serine b-lactamases (SBLs) (Akhtar et al., 2022).As the names imply, SBLs (Ambler's classes A, C, and D) have serine residues at the catalytic center, while MBLs (Class B) are distinguished by zinc ions, making them more concerning for human health due to their broad-spectrum activities (Joji et al., 2022).Also, based on the categorization by Bush and Jacoby, b-lactamases are classified into three groups (1, 2, and 3) according to their functionality.Cephalosporinases are in group 1, oxacillinases, extended spectrum b-lactamases (ESBLs) and serine Carbapenemases are in group 2 while metallic Carbapenemases are in group 3 (Akhtar et al., 2022;Castanheira et al., 2021).
KPC-2, classified as a class A enzyme and named after its initial isolation from K. pneumoniae, represents a predominant Carbapenemase frequently found in Gram-negative pathogens.Resistance to aminoglycosides and third-generation cephalosporins, as well as widely used bL inhibitors such as Avibactam, etc., is observed in different KPC-2 strains (Findlay et al., 2021;Lebreton et al., 2021;Tooke et al., 2021;Wang et al., 2022).OXA-48, a prevalent class D oxacillinase in Gram-negative pathogens, particularly Enterobacteriaceae, is responsible for Carbapenem hydrolysis (CHDL) and is associated with severe infections (Loqman et al., 2021;Stewart et al., 2018).The catalytic activity of OXA-48 demonstrates increased efficacy towards Imipenem and exhibits limited susceptibility to Sulbactam and Clavulanic acid (Guzm� an- Puche et al., 2021).New Delhi-metallo b-lactamase (NDM-1), a member of Class B (subtype B1) b-lactamases, can hydrolyze a wide range of b-lactam antibiotics.NDM-1 was initially identified in India in 2009; however, the occurrence of pathogenic strains that produce NDM-1 is now being reported across the world (Qamar et al., 2019;Venkata et al., 2021).
It is imperative to discover novel b-lactam antibiotics and bL inhibitors due to the high prevalence of b-lactamases, particularly extended-spectrum b-lactamases (ESBLs) and Metallo b-lactamases (MBLs) (Makabenta et al., 2021).To control the corresponding high drug resistance in pathogens, new promising therapeutic models, such as nanodrugs, bacteriophages, monoclonal antibodies, herbal products, etc., are being developed.Plant chemicals are emerging as potent therapeutic agents for a variety of ailments, including bacterial infections.Secondary metabolites are to be credited for rapid development in plant research due to their significance in the fields of drug design, the food industry, pharmaceuticals, etc.
Dry fruits such as raisins, cashews, almonds, figs, walnuts, and many others are abundant sources of many salubrious products such as proteins, minerals, fibers, and essential nutrients (Sohaib et al., 2017).Extensive studies have been conducted on dry fruits to explore their numerous medicinal and pharmacological properties.For instance, almonds and peanuts have demonstrated inhibitory activities against colon cancer (Dhiman et al., 2014).Walnuts are known for their richness in omega-3 fatty acids, which contribute to their antioxidant and anti-inflammatory properties (Mishra et al., 2010).Furthermore, dried apricots and figs have a low glycemic index digestion rate, which can contribute to the prevention of diabetes (Alasalvar & Shahidi, 2013).Stilbenoids extracted from peanuts have shown strong inhibitory effects against Methicillin-resistant Staphylococcus aureus (MRSA) (de Bruijn et al., 2018).Our research focused on the screening of bioactive metabolites from four dry fruits, namely, Cashew (Anacardium occidentale), Walnut (Juglans regia), Fig (Ficus carica), and Peanut (Arachis hypogaea), to investigate their potential inhibitory activity against KPC-2, OXA-48, NDM-1, and SHV-1 b-lactamases.
But screening of these phytochemicals is both time and energy consuming because of their production in limited amounts and the severe effects of stress conditions (Seca & Pinto, 2019).In silico technologies like computational docking, MD simulation, and artificial intelligence, provide a promising new alternative to this problem (Singh & Bharadvaja, 2021).Computational approaches based on structural knowledge have been used to identify possible b-lactamase inhibitors from selected dry fruits.Herein, stateof-the-art bioinformatic studies such as molecular docking and molecular dynamic simulation and pharmacophore studies namely absorption, distribution, metabolism, excretion, and toxicity (ADMET) analysis have been adopted to explore the potential relationships between various natural products and bL inhibitory action.

Prediction of active binding sites of target proteins
The AutoDock tool v1.5.7 was used to convert PDB files of enzymes into PDBQT format, which is a requirement for docking by AutoDock Vina (Etehadpour, 2021).The structures of all these proteins contain previously bound ligands (in cocrystallized form), aiding in the prediction of the active sites of the target proteins (Ziani et al., 2020).Based on manual visualization, the grid boxes around the active sites were modified using the AutoDock tool v1.5.7.

Retrieval and preparation of phytochemicals (or ligands)
The selection of Walnuts (Juglans regia), Figs (Ficus carica), Peanuts (Arachis hypogaea), and Cashews (Anacardium occidentale) for this study was based on extensive literature review (Benmaghnia et al., 2019;de Bruijn et al., 2018;Nicu et al., 2018;Nguyen & Vu, 2023;Prakash et al., 2018).The Dr. Duke Phytochemical and Ethnobotanical Database (www.phytochem.nal.usda.gov)served as a valuable resource in identifying the phytochemicals present in the selected dry fruits, known for their antibacterial activity.This online repository offers extensive information on the diverse biological aspects of dry fruits and their metabolites (Lans & van Asseldonk, 2020).
To acquire the 3D structures (in SDF format) and information about the category, classification, and related medicinal properties of these phytochemicals, we accessed the PubChem compound database (www.pubchem.ncbi.nlm.nih.gov) (Supplementary Table S2).Moreover, 3D structures of Avibactam and Tazobactam (commercially available synthetic bL inhibitors) were also obtained from the PubChem database and considered as positive controls or standards in this study.The SDF format of the ligands was converted to the PDBQT format using AutoDock tool v1.5.7 (DeLano, 2002;Trott & Olson, 2010).Hydrogens were added, and nonpolar hydrogens were removed from ligands as well.

Molecular docking
The grid spacing was set at 1 Å for all target proteins to ensure comprehensive coverage of the geometric space for docked compounds.Furthermore, the dimensions (x, y, and z) of the grid map were set at 25 Å x 25 Å x 25 Å for SHV-1, 23 Å x 23 Å x 23 Å for KPC-2, 27 Å x 27 Å x 27 Å for OXA-48 and 22 Å x 22 Å x 22 Å for NDM-1, respectively (Surya Ulhas & Malaviya, 2022).The central coordinates around grid box were adjusted to x ¼ 8.967, y ¼ 7.881 and z ¼ 10.225 for SHV-1, x ¼ 17.983, y ¼ 41.761 and z ¼ 32.101 for KPC-2, x ¼ 12.997, y ¼ 10.784 and z ¼ 8.244 for OXA-48 and x ¼ 10.342, y ¼ 46.561 and z ¼ 57.761 for NDM-1.Using AutoDock Vina (Lamarckian genetic algorithm with default parameters), molecular docking experiments were run for each complex at 10 distinct sites (Aydemir et al., 2022).Each experiment used up to 25 million energy evaluations, with an elitism of 1, and the output data were clustered using a 2 Å tolerance (Vakayil et al., 2022).The binding positions of ligands with target proteins were sorted according to their binding energies, and their 3D structures were obtained in PDB format.

Molecular dynamic simulation study
The physical movement of atoms or molecules can be determined using molecular dynamics (MD) simulation studies.WebGro for macromolecular simulations (https://simlab.uams.edu/) was used to run the MD simulation to detect the stability of docked complexes (Jha et al., 2021;Tumskiy & Tumskaia, 2021).The topology of the docked complexes of ligands and target enzymes was also generated using the GlycoBioChem PRODRG2 Server (http://davapc1.bioch.dundee.ac.uk/cgi-bin/prodrg/) (Sch€ uttelkopf & Van Aalten, 2004) prior to the simulations.The SETTLE algorithm was employed to restrain the water molecules, while the lengths of bonds between the protein and ligands were constrained using the LINCS algorithm (Moharana et al., 2023).
The same parameters were set for running all simulations.The steepest descent algorithm was used not only to minimize the systems but also to attain the target volume (Jaidhan et al., 2014).Leapfrog stochastic dynamics algorithm was used to calculate the trajectories with 2fs time duration (Chiang et al., 2020).Using the Particle Mesh Ewald methods with a cutoff of 1.2 nm and a Fourier grid spacing of 1.2 nm, the long-range electrostatics were determined (Patra et al., 2007).The Single Point Charge (SPC) water model was used to solvate the whole system (Mukherjee et al., 2021).Also, the GROMOS96 43a1 force field setting was used (Halder et al., 2022).The system was neutralized by introducing 0.15 M NaCl salt (de Azevedo & Nascimento, 2022); the octahedron box type of 52 Å � 70 Å � 75 Å size (Tumskiy & Tumskaia, 2021) and the two equilibration types selected were constant number of particles, constant volume, and constant temperature (NVT), as well as constant pressure and temperature (NPT) (Dalal et al., 2021;Khairy et al., 2022).The pressure and temperature for the 50 ns simulation time and the 5000 simulations per frame were maintained at 1.0 bar and 300 K, respectively (Haq et al., 2017;Lu et al., 2022).Radius of Gyration (Rg), Hydrogen bonds, Root Mean Square Fluctuations (RMSF), and Root Mean Square Deviations (RMSD) coordinates for each MD simulation were requested (Amanat et al., 2022).

Solvent accessible surface area (SASA)
The output trajectories were further analyzed for Solvent Accessible Surface Area (SASA) for target enzymes in bound and unbound forms (Dalal et al., 2021).It is often used to study the interactions between molecules and their environment, as well as to predict the structural and functional properties of molecules (Kim & Na, 2022;Laddach et al., 2018).The SASA value is a measure of the amount of surface area of the atom exposed to solvent molecules and can be calculated by multiplying 4pR 2 with the ratio of the number of accessible points on the protein surface to the total number of points on the protein surface, where R equals the sum of the van der Waals radius and the solvent radius (Cavallo et al., 2003;Durham et al., 2009).The SASA values of protein ligands complexes along with proteins alone were plotted to further analyze the structural stability of protein.

MM-GBSA-based free energy analysis
Based on Molecular Mechanics-Generalized Born and Surface Area (MM-GBSA) method, the binding free energy (DG bind ) of the protein and ligand complexes was also calculated using the 'Prime' module of Schrodinger (Dalal et al., 2021;Kollman et al., 2000).In this study, the phytocompounds were evaluated for their ability to bind to the target protein.
To quantify the strength of binding, the relative binding free energies of the phytocompounds were calculated and analyzed (Hayes et al., 2011).This allowed for a comparison of the binding affinity of different compounds and provided insights into their potential as inhibitors of the target protein (Temel et al., 2023).For the task at hand, the Python script 'thermal_mmgbsa.py'was used to carry out the necessary calculations or analyses (Nandhini et al., 2023;Parida et al., 2021).The free binding energy of all frames in the simulated trajectories of protein-ligand complexes was calculated using the following equation; where DG bind ¼ Binding Free Energy, G complex ¼ Free Energy of the protein-ligand complex, G protein ¼ Free Energy of the target protein only, and G ligand ¼ Free Energy of the ligand only.Further analysis was performed on the MMGBSA output trajectories to identify any structural modifications that occurred after the dynamic simulation (Mukerjee et al., 2022).

Principal component analysis (PCA)
Principal component analysis (PCA) is a common analytical method for simplifying large datasets by identifying the most relevant patterns of variability and reducing the dimensionality of the data (Kushwaha et al., 2021).To facilitate the analysis of the configuration space of anharmonic motion with a restricted number of degrees of freedom, the molecular dynamics simulation data were subjected to dimensionality reduction using PCA, also called essential dynamics (ED) (Moharana et al., 2023).This process involved transforming the atomic positions into a covariance matrix after converting them into Cartesian coordinates via linear conversion (Talukder et al., 2022).The covariance matrix was diagonalized to verify the flexibility of target enzymes and the eigenvalue and eigenvectors were solved, which allowed the derivation of the principal components of the target proteins (Mesentean et al., 2006).The eigenvectors describe the orientation of the motion, while the eigenvalues provide information about the extent of motion along those orientations.Each eigenvector and its corresponding eigenvalue represent a specific motion or vibrational mode of the biomolecule (Robert Frost, 2022;Tsoulfidis & Athanasiadis, 2022).The matching of the eigenvalues and eigenvectors describes the energy contribution of each component to the overall motion.By superimposing the molecular trajectory onto a specific eigenvector, it is possible to observe the timedependent motion of that component in the chosen vibrational mode (Jolliffe & Cadima, 2016).This can help gain a deeper understanding of the motion and behavior of biomolecules, which is important for studying various biological and chemical processes (Haider et al., 2008).OSIRIS DataWarrior software v 5.5.0 was used to compute the principal components and trajectory coordinates (L� opez-L� opez et al., 2019;Papaleo et al., 2009).

Gibbs free energy landscape (FEL)
By employing a conformational sampling method, such as MD simulation, that enables exploration of conformations in the vicinity of the native state structure, it is possible to derive the free energy landscape (FEL) of a protein (Papaleo et al., 2009).The FEL is a visualization of the potential conformations that a protein can adopt during a molecular dynamic simulation, along with their corresponding Gibbs free energy (Pathak et al., 2022;Pradeep et al., 2022).It is a two-dimensional or three-dimensional plot that displays the distribution of protein conformations in a low-dimensional space obtained through PCA (Al-Khafaji & Tok, 2020;Granata et al., 2015).To calculate the FEL, the probability distribution obtained from the essential plane composed of the first two eigenvectors was utilized (Gupta et al., 2021;Kar et al., 2023).The Origin2021b software was used for the construction of the FEL of target proteins and their ligands (Talukder et al., 2022).

Results
Virtual screening is a crucial and quick method for mining molecular databases to identify bioactive chemical components.To find effective compounds against bL, several wellknown phytochemicals of selected dry fruits were taken into consideration.Figure 1 illustrates the various stages of the current study.

Screening of phytochemicals having antibacterial activity
A pool of 59 bioactive compounds that show antibacterial activity was selected from four dry fruits: Fig ( Ficus carica), Walnut (Juglans regia), Cashew (Anacardium occidentale), and Peanut (Arachis hypogaea).These phytochemicals consist of various chemical properties, including, but not limited to, terpenes, phenols, alkaloids, flavonoids, saponins, etc., (Table S2 in the supplementary file).For standard or positive control, the commercially available synthetic inhibitors Avibactam and Tazobactam were also retrieved from publicly available chemical databases.

Molecular docking analysis
Molecular docking study was performed to explore the binding pattern of phytochemicals at the active site of the target enzymes.The quality of these interactions was evaluated in terms of their docking scores, i.e. binding affinity (kcal/mol).
The binding energies of the standards (Avibactam and Tazobactam) were considered as cut-off values to screen out phytochemicals for further studies.Of the total of 59 phytochemicals, 12 compounds were found to have binding affinities higher than the standards (Table 1).The bioactive compounds, including Oleanolic acid, Tannin, and Protocatechuic acid, showed the highest binding affinities for the target enzymes.These compounds were selected for further evaluation to better understand their potential as bL inhibitors.Table 2 presents the amino acid residues involved in various interactions with these ligands.

OXA-48 b-lactamase
Docking results showed that Oleanolic acid, a triterpenoid occurring naturally in figs, has the highest binding affinity for OXA-48 (-9.8 kcal/mol) followed by Protocatechuic acid (-8.8 kcal/mol).Also, Tannin, Naringenin, Kaempferol, Rutin, etc. showed higher affinities (-8.6 kcal/mol, À 8.6 kcal/mol, À 8.4 kcal/mol, and À 8.3 kcal/mol, respectively) compared to Avibactam (-6.2 kcal/mol) and Tazobactam (-6.6 kcal/mol).It was also observed that all these phytochemicals interacted with the catalytic site residues of OXA-48 bL.Oleanolic acid, however, interacted with OXA-48 through hydrophobic and Van der Waals interactions.Ser70, Ser118, Ile102, Trp105, and Tyr211 were found to be common residues of OXA-48 that interacted with Oleanolic acid, Avibactam as well as Tazobactam (although via different types of bonds).Leu158, Val120, and Arg214 were some other common residues that interacted with Avibactam and Oleanolic acid.While Avibactam formed hydrogen bond-Oleanolic acid showed Van der Waals interactions with Arg214.Oleanolic acid showed some peculiar interaction as it did not form any conventional hydrogen bond with any amino acid residue, although it interacted with several residues present at the catalytic site of OXA-48.Conversely, both Avibactam and Tazobactam interacted with amino acids via hydrogen as well as hydrophobic bonds.Oleanolic acid has a strong affinity for the amino acids of OXA-48 bL, forming a complex that is significantly more stable than those observed in other ligands, Figure 2. The 2D interactions of Avibactam and Tazobactam with OXA-48 bL are shown in Figure S1 (supplementary file).

SHV-1 b-lactamase
Both Protocatechuic acid and Tannin exhibited a binding affinity of À 9.4 kcal/mol to the target SHV-1 bL, which was higher than the controls (Avibactam; À 6.7 kcal/mol and Tazobactam; À 6.7 kcal/mol).Protocatechuic acid is a phenolic acid compound found in both peanuts and walnuts.Ser70, Ser130, Asn132, and Ala237 were identified as common residues involved in the interaction with Protocatechuic acid, Avibactam, and Tazobactam.Apart from these, Glu166, Gly238, Ala237, and Tyr105 were found to be responsible for Van der Waals interactions.Ser70, Asn170, Lys73, Ser130, and Asn132 were involved in the making hydrogen bonding interactions between Protocatechuic acid and the catalytic site of SHV-1, Figure 3.In the case of Tazobactam, Ser70, Ser130, and Lys234 were responsible for hydrogen bonding, and Met272 and Ala237 were involved in hydrophobic  interactions.On the contrary, Glu240, Asn170, Ala237, and Asn132 formed hydrogen bonds while residues Thr167, Glu166, Tyr105, Lys73, Ser130, Ser70, Met69, and Gly238 showed Van der Waals interactions with Avibactam.The 2D interactions and bonding between Avibactam and Tazobactam with the amino acids of SHV-1 are shown in Figure S2 (supplementary file).
Tannin, an astringent polyphenolic biomolecule, is a commonly found phytochemical in cashews, figs, and walnuts.The catalytic site of SHV-1 bL was stabilized by the involvement of several amino acid residues with Tannin.Ser70, Asn170, Ser130, Asn132, Ala237, and Gly238 were among the common residues interacting with Tannin, Avibactam, and Tazobactam through various types of bonds.While Tannin did form hydrogen bonds with four amino acids, its interaction with the ligand-protein complex was primarily mediated through hydrophobic interactions and Van der Waals interactions.Tannin exhibited a distinct behavior compared to the standards, as it formed a hydrogen bond with Arg275 and Thr267, which were absent in the Avibactam-SHV-1 or Tazobactam-SHV-1 complexes.Figure 4 illustrates the interactions of the Tannin and SHV-1 bL complex.
In the case of Protocatechuic acid, four hydrogen bonds were observed with amino acid residues, i.e.Ser130, Lys73, Ser70, and Asn132 of the catalytic site of KPC-2, Figure 6.Trp105 was also observed to be a common residue that interacts with Tannin, Protocatechuic acid, and Tazobactam through a hydrophobic bond, but strikingly, this interaction The highest docking affinities among the compounds.
was missing in the Avibactam-KPC2 complex.Avibactam and Tazobactam formed hydrogen bonds with polar amino acid Asn170 but Protocatechuic acid showed Van der Waals interaction with this residue.

NDM-1 b-lactamase
Among all the phytochemicals subjected to docking, Protocatechuic acid displayed the highest affinity (-9.2 kcal/mol) to NDM-1 bL, followed by Oleanolic acid and Tannin; both showing À 8.9 kcal/mol.The docking analysis revealed that the molecular interactions of Protocatechuic acid formed hydrophobic bonds with the aromatic amino acid Tyr64 residue, as well as the aliphatic amino acid Ala72 residue.Some Van der Waals interactions were also observed between Protocatechuic acid and the Asp66, Asp48, and Val50 residues.Although Avibactam and Tazobactam exhibited several hydrogen and hydrophobic interactions, they displayed very poor binding affinity, i.e.À 5.8 kcal/mol and À 6.2 kcal/mol, Figure S4 (supplementary file).On the contrary, Protocatechuic acid showed better docking stability at the catalytic site of NDM-1 bL, interacting mainly with pocket atom residues.His250 residue was a common residue interacting in same nature with both standards.Interestingly, Tyr64 was found to interact with both Avibactam and Protocatechuic acid, albeit through different types of bonds.However, no such coherence was observed in Tazobactamenzyme and Protocatechuic acid-enzyme interactions.Figure 7 illustrates the position and binding nature of Protocatechuic acid with the crucial active site residues of NDM-1 bL.

Molecular dynamic simulation
The top three hit phytochemicals (Protocatechuic acid, Oleanolic acid, and Tannin), as well as both standards (Avibactam and Tazobactam) in complexed with respective target enzymes (NDM-1, SHV-1, OXA-48, and KPC-2) were subjected to a 50 ns dynamics simulation using the GROMOS96 43a1 forcefield via WebGro.MD simulation is one of the best methods to examine the changes in the behavior of ligands in-vivo and the pattern of resulting conformational changes in the target enzymes at the nanosecond level.Although, these conformational changes in proteins are normal up to some extent, if beyond the threshold, can disrupt the normal structure and function of proteins.Thus, to validate the interactions, trajectories coordinate of macromolecules such as Root Mean Square Deviation (RMSD), Root Mean Square Fluctuation (RMSF), Number of Hydrogen Bonds, Radius of Gyration (Rg), etc., were analyzed for each docking complex.

OXA-48 b-lactamase
The number of hydrogen bonds, RMSD, RMSF, and Rg values of the enzyme backbone and ligand complexes were plotted for the simulation time of 50 ns and are shown in Figure 8.
For the protein backbone alone, RMSD remained stable for  Moreover, the number of hydrogen bonds was also determined for each docked complex during the simulation period.The number of hydrogen bonds formed by ligands during the simulation was also recorded.The maximum

SHV-1 b-lactamase
For SHV-1 bL and its complexes with ligands, the ranges of RMSD values were found to differ slightly in each case, shown in Figure 9.For protein alone, the RMSD ranges near 0.23 nm, Figure S5 (supplementary file).But for each ligand, the value varied.The RMSD of Tannin showed the stability of the ligand in the protein backbone after docking since it fluctuated only slightly in a 10 ns period.Tannin's RMSD values range from 0.12 nm to 0.2 nm.The RMSD of Avibactam becomes highly fluctuated at 20 ns but becomes stable toward the end, with values ranging from 0.05 nm to    Tazobactam with active site residues of SHV-1 bL was also analyzed, Figure 9. Protocatechuic acid, Tannin, Avibactam, and Tazobactam formed 6, 5, 4, and 3 hydrogen bonds, respectively.The Rg value of these complexes was also observed.The average Rg value of Tannin, Avibactam, and Tazobactam was near 1.82 nm whereas that of Protocatechuic acid was close to 1.85 nm, Figure 9. Furthermore, Figure S7 (supplementary file) presents the protein-ligand complex that was extracted to confirm the positions of Protocatechuic acid, Tannin, Avibactam, and Tazobactam within SHV-1 bL in multiple MD simulation trajectories.

KPC-2 b-lactamase
Complexes of KPC-2 bL with the top phytochemicals (Protocatechuic acid and Tannin), as well as the standards (Avibactam and Tazobactam) were subjected at a 50 ns simulation to better comprehend their stability under dynamic conditions.The RMSD of the protein backbone, which was calculated from each simulated complex, allowed access to conformational and structural changes.The RMSD (ranging from approximately 0.24 nm) of the protein backbone atoms alone is shown in Figure S5 (supplementary file).The RMSD all ligand-protein complexes ranged between 0.1 nm and 0.25 nm; however, after stabilizing RMSD values were less than 0.22 nm and thus were acceptable, Figure 10.The RMSD of Protocatechuic acid at the beginning of the simulation was observed near 0.12 nm, increasing to 0.2 nm towards the end.A similar behavior was observed in the case of Tannin complexed with protein backbone, with RMSD between 0.1 nm and 0.15 nm.The RMSD of both Avibactam and Tazobactam fluctuated greatly until the 25 ns time span, but both stabilized toward the end, i.e.50 ns.The RMSD of Tazobactam was recorded between 0.1 and 0.22 nm, while that of Avibactam was between 0.1 nm and 0.24 nm.The average RMSF was observed to be in the acceptable range with no significant fluctuations or any major structural changes, ranging from 0.05 nm to 0.25 nm for all ligand-protein complexes, Figure 10.However, the amino acid regions in which fluctuations were recorded were different for each complex.
To better understand the interaction of the ligand-KPC-2 bL complexes, the number of hydrogen bonds formed during the simulation was also analyzed; Figure 10.Herein, Tannin and Avibactam showed 8 maximum hydrogen bonds, while Protocatechuic acid formed 3 and Tazobactam formed 4. The Rg value of all complexes was also observed; Figure 10.The Rg values of both Tazobactam and Avibactam were near 1.82 nm toward the end of the simulation.Conversely, the Rg values of Protocatechuic acid and Tannin started from 1.8 nm and gradually decreased very close to 1.7 nm at the end.Figure S8 (supplementary file) depicts the protein-ligand complexes that were extracted in various time periods (0, 10, 20, 30, 40, and 50 ns) to verify the locations of Protocatechuic acid, Tannin, Avibactam, and Tazobactam during the MD simulation.

NDM-1 b-lactamase
Data obtained from simulated MD trajectories were used to analyze the conformational transition and the degree of stability of ligands-NDM-1 bL docked complexes.The RMSD of protein backbone atoms was also graphically analyzed and is shown in Figure S5 (supplementary file).For Protocatechuic acid-NDM-1 bL complex, RMSD remained stable throughout the simulation with no major fluctuations, and the value ranged near 0.2 nm, Figure 11.This shows that Protocatechuic acid was stable enough in the protein backbone during the simulation period of 50 ns.The RMSD of Avibactam was near 0.1 nm at the beginning, increasing to 0.15 nm until 30 ns, then decreasing again toward the end.The RMSD of Tazobactam fluctuated highly until the 30 ns time span, then stabilized at 0.15 nm towards the end of the simulation.The RMSF data of the protein backbone atoms showed local changes in the protein chain during the simulation of the complexes, Figure 11.RMS fluctuations, however, demonstrated acceptable ranges and were observed in the critical docking regions.These fluctuations were recorded in The number of hydrogen bonds formed between the ligands and the protein, obtained from the simulation trajectories, demonstrated that both Avibactam and Tazobactam have 6 maximum hydrogen bonds, while Protocatechuic acid has 4, Figure 11.The Rg of each complex decreased gradually towards the end of the simulation, Figure 11.The Rg values of Protocatechuic acid, Avibactam, and Tazobactam all ranged from 1.85 nm at the beginning to 1.79 nm at the end.Several protein-ligand complexes were obtained from the MD simulation trajectories and are presented in Supplementary Figure S9.

MM-GBSA-based free energy analysis
To further evaluate the docking, MM-GBSA-based free energy of the simulation trajectories of each complex was calculated.The major contributing components include covalent bond (DG bind _COV), Coulomb (electrostatic) interaction (DG bind _COUL), Solvation (DG bind _SOL), Hydrogen bond (DG bind _Hbond), Van der Waals interaction (DG bind _VWE) and lipophilic interaction-based (DG bind _LIPO) free energies were all considered for calculating total free energy (DG bind ).The average DG bind calculated for trajectory frames of each protein and ligand complex is represented in Table 3.
In the case of OXA-48 bL, the highest relative binding affinity was observed for Oleanolic acid, i.e.À 65.99 kcal/mol, with DG bind _VWE of À 51.82 kcal/mol influencing the most of total energy.Both standards have relatively low binding affinities, with Avibactam at À 41.59 kcal/mol and Tazobactam at À 35.7 kcal/mol.For SHV-1 bL, Protocatechuic acid showed the highest binding affinity, with the Average of DG bind ¼ À 68.7 kcal/mol, followed by Tannin (-60.32 kcal/mol).As observed from Table 3, the highest contribution for both phytochemicals was Van der Waals interactions with À 63.25 kcal/mol for Protocatechuic acid and À 56.76 kcal/mol for Tannin.Conversely, Avibactam and Tazobactam were observed to have low binding affinities (-57.9 kcal/mol and À 58.8 kcal/mol respectively).Solvation-based free energy (DG bind _SOL) was observed to be positive for all SHV-1 bLligand complexes except Avibactam.Similarly, Avibactam exhibited the highest positive contribution of Coulomb interaction-based energy (DG bind _COUL).
In the case of KPC-2 bL, Tannin exhibited the highest binding affinity, at À 64.94 kcal/mol, followed closely by Protocatechuic acid, at À 63.1 kcal/mol.The energy breakdown for the KPC2 bL-Tannin complex showed that the source of highest energy contribution was Solvation-based energy (DG bind _SOL), at À 60.31 kcal/mol.However, for Protocatechuic acid, the highest contribution came from Van der Waals interactions, at À 55.81 kcal/mol.Both standards showed low affinities in the case of KPC-2 bL as well, with Avibactam at À 51.69 kcal/mol and Tazobactam at À 47.52 kcal/mol.The most significant difference in DG bind was observed for NDM-1 bL-Protocatechuic acid and NDM-1 bL-Avibactam complexes.Protocatechuic acid exhibited average DG bind ¼ À 68.26 kcal/mol, whereas Avibactam showed only À 29.36 kcal/mol.Tazobactam also showed an average DG bind of À 36.32 kcal/mol.The major difference in energy of all these NDM-1 bL complexes was attributed to Van der Waals interactions based free energy (DG bind _VWE), as depicted in Table 3.

Principal component analysis (PCA)
To observe the collective motion of Ca atoms of target enzymes in the presence and absence of phytochemicals, principal component analysis was performed.The eigenvectors and eigenvalues of each protein-phytochemical complex were compared with those of unbound protein and protein-standards complexes to understand the conformational motion of Ca atoms.

OXA-48 bL
Figure 14 shows the eigenvalues plotted for unbound OXA-48 bL as well as OXA-48 bL in complexed with Oleanolic acid,  Avibactam, and Tazobactam.It can be observed from Figure 14 that unbound protein occupies a wider space region as compared to bound complexes.The conformational space occupied by Avibactam-bound OXA-48 bL was somewhat like an unbound protein, followed by Tazobactam-bound complex.In contrast, the reduction in essential dynamics was observed in the case of Oleanolic acid bound OXA-48 bL, as shown in Figure 14.

SHV-1 bL
The dynamic motion of bound and unbound protein along PC-1 and PC-2 is projected in Figure 15.It has been shown that Protocatechuic acid bound SHV-1 bL showed less correlated motion as compared to other complexes as well as SHV-1 bL alone.The SHV-1 bL-Avibactam and SHV-1 bL-Tazobactam showed an increase in the correlated motion compared to the SHV-1 bL alone or other complexes.

KPC-2 bL
The analysis of the collective motion of Ca atoms of KPC-2 bL in the presence and absence of ligands is represented in Figure 16.The covariance matrix of overall simulation trajectories showed that the motion of KPC-2 bL residues was significantly increased in the presence of Avibactam and Tazobactam.Tannin bound protein complex also occupies a somewhat wider space region as compared to unbound protein.However, the residues of KPC-2 bL, when complexed with Protocatechuic acid, showed lesser degree of dynamic motion, as shown in Figure 16.

NDM-1 bL
The essential dynamic behavior of NDM-1 bL in ligand-bound and unbound form is represented in Figure 17.The NDM-1 bL-Avibactam complex showed an increased collective motion as compared to unbound protein or other bound complexes, representing a higher degree of flexibility and changes in the protein residues over the course of 50 ns simulation.The lesser phase space occupied by NDM-1 bL in the presence of Protocatechuic acid shows better stability and less dynamic shift in the protein residues.The motion of protein residues in the presence of Tazobactam was also observed to be wider as compared to unbound protein.

FEL analysis
The FEL against the first two principal components PC1 and PC2 were generated to explore the energy minima landscapes of native and ligand-bound target enzymes.Figure 18 illustrates the color-coded Gibbs free energy landscapes of free and bound SHV-1 and KPC-2 bL, which indicated the DG values 2 to 14 kcal/mol.Each protein-ligand complex has a unique FEL pattern.The yellow color represents the unfavorable protein conformations, while dark blue spots show the energy minima and the favored conformations.The FEL plots of native, Protocatechuic acid, and Tannin bound SHV-1 bL in Figure 18(a-c) show that when phytochemicals are docked, the ligated form of SHV-1 bL also adopts a single energy minimal ensemble, like a native enzyme.This suggests that the binding of the ligand does not induce significant conformational changes in the protein structure, and the stability of the interaction could be attributed to specific intermolecular forces between the ligand and protein.The  FEL of SHV-1 bL when bound to standards (Avibactam and Tazobactam), is shown in Supplementary Figure S10.
Tazobactam achieved a single compact energy well whereas Avibactam showed two distinct energy minima during the 50 ns simulation when docked to SHV-1 bL.
In the case of KPC-2 bL (Figure 18(d-f), two distinct regions of energy minima were observed for both the free enzyme (Figure 18(d)) and the enzyme-bound to Protocatechuic acid (Figure 18(e)).These distinct minima in the energy landscapes correspond to metastable conformational states that are separated from one another by a negligible energy barrier.On the contrary, for the Tannin-bound KPC-2 (Figure 18(f)), only a single minimal energy landscape was observed.Avibactam and Tazobactam showed multiple smaller energy wells when ligated to KPC-2 bL for 50 ns (Supplementary Figure S10).
Figure 19 shows the free energy landscape (FEL) for obtaining the global minima of the Ca backbone atoms of the OXA-48 and NDM-1 proteins.A shift in the minima energy well was observed for free OXA-48 bL (Figure 19(a)) and OXA-48 bL bound to Oleanolic acid (Figure 19(b)).Single energy minima were observed for the native enzyme, whereas two energy wells were observed in the case of binding of Oleanolic acid to the enzyme.A similar scenario was also observed for NDM-1 bL as well, Figure 19(c,d).Two different energy minima were observed for NDM-1 bL bound to Protocatechuic acid, contributing to the stability of the interaction.The FELs of standards-bound OXA-48 and NDM-1 bL are represented in Supplementary Figure S10, each showing distinct energy wells indicating the stability of docked complexes.

ADME and toxicity profile analysis
To ensure that drugs work as intended, pharmacophore analysis is important in the drug development process.In order to predict the physiochemical and pharmacokinetic properties of the most effective phytochemicals as well as synthetic inhibitors, admetSAR (http://lmmd.ecust.edu.cn/admetsar2/) and PROTOX-II (https://tox-new.charite.de/protox_II/)were used.The results are shown in Tables 4 and 5. Lipinski's Rule of Five is a valuable criterion for quickly evaluating the druglike properties of a compound and was applied in our study as well.It is a guideline to assess the drug-likeness of a compound, focusing on its physicochemical properties such as molecular weight, hydrogen bond donors, and lipophilicity.The rule states that a drug-like molecule should have no more than five hydrogen bond donors, ten hydrogen bond acceptors, an octanol-water partition coefficient of less than 5, and a molecular weight of less than 500 Da.Table 4 presents the ADMET profiles of the phytochemicals, which displayed an acceptable range, demonstrating their potency as potential drug candidates.
The topological polar surface area (TPSA), a crucial parameter for the availability of drug transport of a compound, and its value must be in the range of 40-130 Å for the best outcome.Molinspiration (www.molinspiration.com)was used to determine the TPSA of the ligands.Oleanolic acid, Protocatechuic acid, Sakuranetin, Naringenin, Kaempferol, and Beta-sitosterol all fall within the normal range of TPSA.Since the blood-brain barrier (BBB) exists, many drugs cannot enter the brain.This makes it difficult to treat disorders or diseases that require drug delivery directly to the brain.Betasitosterol, however, was the only phytochemical in our study that showed the ability to cross this barrier.The compounds also showed good human intestinal absorption (HIA) except Tannin, Isoquercitrin, Avibactam, and Tazobactam.All compounds also showed inactivity for hepatotoxicity.Additionally, the AMES toxicity, Caco-2 permeability, and   Toxicity class of the compounds were also determined and are shown in Table 5.

Discussion
Innovative drug research and development continue to depend profoundly on the successful discovery of new lead compounds for a specific target protein (Dai et al., 2021).In order to achieve this goal, pharmaceutical companies have invested heavily in the latest Computer-Aided Drug Designing (CADD) techniques such as structure-based drug design, fragment-based lead optimization, high throughput screening (HTS), artificial intelligence (AI) etc. CADD involves the exploitation of virtual tools to analyze and optimize the pharmacophore profiles of new or hit drugs (Dai et al., 2021;Yu & Chen, 2021).Modern cheminformatic techniques viz.computational screening, docking, and MD simulations have  become effective and reliable approaches for identifying bioactive metabolites, and these techniques also help in overcoming some of the major cons of drug discovery, such as resource and time management (Adelusi et al., 2022;Murugan et al., 2022;Saban� es Zariquiey et al., 2022).Thus, these approaches were applied in this study to identify metabolites of selected dry fruits that have potential bL inhibitory activities.
Targeting b-lactamases has become a powerful approach to tackling increasing b-lactam resistance in Gram-negative pathogens (Annunziato, 2019;Richter et al., 2022).Several invitro studies have reported that the secondary metabolites of many plants have the potential to successfully block bL (Ayaz et al., 2019;Adekunle et al., 2022;Cheesman et al., 2017).Vakayil et al. (2021) have reported that Boswellia serrata resin extract has high inhibitory activity against K. pneumoniae and its phytochemicals also showed high affinity for AmpC bL.Maheshwari et al. (2019) described that bioactive compounds of Carum copticum can successfully inhibit ESBLs producing multidrug resistant bacteria.In addition to experimental studies, several computational approaches have been employed to identify bioactive compounds from plants that demonstrate inhibitory effects against b-lactamases (Bhat et al., 2021;Kongkham et al., 2022;Parida et al., 2021;Vasudevan et al., 2022).Current research investigated the inhibitory activities of phytochemicals several some dry fruits by implementing different in-silico approaches, including computational docking and molecular dynamics.Moreover, the pharmacodynamic and ADMET profiles of the top phytochemicals were also analyzed to predict their bioavailability and toxicity.SHV-1, NDM-1, KPC-2, and OXA-48 were the targeted enzymes in our study due to their documented role in enhancing resistance in clinical isolates of Gram-negative pathogens such as E. coli, K. pneumoniae, particularly in the context of hospital-acquired nosocomial infections (Cherak et al., 2022;Huang et al., 2023;Khalifa et al., 2019).
The docking of 59 total phytochemicals with the target enzymes was performed using AutoDock Vina and the docking scores were compared with standards, i.e.Avibactam and Tazobactam which are commonly used synthetic inhibitors (Brink et al., 2022;Gill et al., 2021;Wilson et al., 2021).We selected only the phytochemicals showing the highest binding affinity for the target enzymes for further analysis, as shown in Figure 20.The selected phytochemicals with high binding affinity, along with the standards, underwent MD simulation studies to analyze their binding characteristics.Protocatechuic acid, present in walnuts and peanuts, showed a high docking affinity for SHV-1 (-9.4 kcal/mol), KPC-2 (-9.5 kcal/mol), and NDM-1 (-9.2 kcal/mol) as compared to Avibactam and Tazobactam.Tannin, a phytochemical in cashews, figs, and walnuts, also displayed a high docking score for SHV-1 (-9.4 kcal/mol) and KPC-2 (-9.4 kcal/mol) than standards.Both Protocatechuic acid and Tannin are phenolic compounds that are found in several plants and have antimicrobial, anticancer, antioxidant, anti-neurodegenerative, and antifungal activities (Baer-Dubowska et al., 2020;Jing et al., 2022;Krzysztoforska et al., 2019;Song et al., 2020;Zhang et al., 2021).These compounds can play a significant role in protecting against microbial infections and decreasing inflammation associated with cancer, aging, and other illnesses, as well as improving overall health and wellness.Oleanolic acid, a pentacyclic triterpenoid compound found in figs, showed the highest binding affinity for OXA-48 (-9.8 kcal/mol) as compared to other ligands.Oleanolic acid and its derivatives have potent anti-inflammatory, immunomodulatory, cardioprotective, antimicrobial, antioxidant, and anticancer effects, which make them of great interest to modern medicine (Sun et al., 2019;Sen, 2020).
The top three phytochemicals were then subjected to MD simulations along with standards to predict the conformational stability of protein-ligand complexes.RMSD, RMSF, Rg, and the number of hydrogen bonds formed between ligands and proteins for 50 ns time span were some of the parameters requested from each simulation trajectory.The most useful physical parameter to express changes in atomic coordinates between different conformations of the same molecule is the root mean square deviation (RMSD) (Schreiner et al., 2012).RMSD is an important metric for MD simulations, as it indicates of the changes in protein structure and energy that occur when a molecule moves from one conformation to another (Vel� azquez-Libera et al., 2020).Small or moderate changes in RMSD values are acceptable, but substantial changes cause the target's conformation to become unstable.The computed temporal RMSD has shown several drifts, which may be a result of the inherent flexibility or the relaxation of the crystal structure of enzymes in solution (Davari et al., 2017).Such changes can also be attributed to the presence of a ligand, since steric interactions with the binding pocket of enzymes cause significant shifts in the coordinates of backbone atoms (Makepeace et al., 2020;Sharma et al., 2022).However, in each phytochemical-bL complex, the RMSD was observed to be lower than that of bL alone, showing more rigidity and stability of complexes (Schreiner et al., 2012).The root mean square fluctuation (RMSF) is a measure to describe the changes in flexibility between residues relating to the average conformation from MD simulation (Zia et al., 2022).This measure considers the mean-squared displacement of each residue from its average position, allowing for the evaluation of flexibility differences between residues (Mart� ınez, 2015).While the RMSD shows the overall flexibility of a complex, RMSF highlights the residue regions of the highest mobility.Notably, in this study, the RMS fluctuations of phytocompounds and target enzyme complexes were observed to be in the region of amino acids which were involved in the docking, and the maximum fluctuation shown by residues was also in an acceptable range (� 0.25 nm).
The specificity of ligand binding is largely analyzed by hydrogen bonds, and their stability over time can be determined by MD simulation.As they affect how drugs are metabolized and absorbed, they are regarded as a crucial parameter in drug design.These hydrogen bonds can be classified as conventional, which are formed by many electronegative atoms, or nonconventional, which are formed by others.Such interactions also help in the effective anchoring of the ligand in the binding pocket of target proteins (Kato et al., 2017;Ya et al., 2015).Surprisingly, the number of hydrogen bonds observed during the simulation was somewhat different from those formed during docking.This difference is due to the reason that MD simulation takes both conventional and non-conventional hydrogen bonding into account (Singh et al., 2022).To determine the compactness of bL in the presence and absence of ligands, the radius of gyration (Rg) was also evaluated.The average distance between the center of the molecule and all the scattering elements is the radius of gyration (Lobanov et al., 2008).This definition makes the radius of gyration an ideal measure for understanding the physical properties of a molecule and how it interacts with its environment.Interestingly, no major fluctuations were observed in the Rg values of the phytocompounds-bL complexes as compared to the standard complexes, indicating stability throughout the 50 ns simulation.
The solvation-free energy of a protein is influenced by the interplay between its polar and non-polar residues (Savojardo et al., 2020).The solvent accessible surface area (SASA), which represents the portion of the protein's surface accessible to solvent molecules, is determined by the exploration of the protein's Van der Waals surface by solvent molecules (Moharana et al., 2023).The area of a protein-ligand complex that is accessible to solvent can also reveal how compact and stable the complex is.A lower solvent accessible surface area (SASA) value corresponds to a more stable and tightly bound complex (Baammi et al., 2023;Dalal et al., 2021;Zaki et al., 2022).It is noteworthy that in this study, the SASA values of enzymes bound to phytochemicals were significantly lower than those of enzymes bound to standards, or even unbound enzymes.This suggests that the complexes formed by enzymes and phytochemicals were remarkably more stable and compact.Additionally, the simulation trajectories were also used to determine binding free energy analysis by the MM-GBSA method (Hayes et al., 2011;Parida et al., 2021;Temel et al., 2023).The findings indicate that the hydrophobic interactions play a dominant role in stabilizing ligands within the binding pocket, as evidenced by the larger contribution of Van der Waals energies over other energies such as lipophilic or Coulomb interactions, etc. (refer to Table 3).
A balance between flexibility and rigidity is crucial for proper protein function, particularly in the context of protein-ligand interactions.The residues in the protein-binding sites must be able to adjust to the ligand while maintaining the overall stability of the protein structure.Thus, to understand the essential dynamic behavior of residues of target enzymes in the presence of ligands, the principal component analysis was also studied (Das et al., 2021;Kushwaha et al., 2021).The analysis relies on examining the intense movement of the Ca atom within the protein by considering its movement about to eigenvectors, which represent the overall direction of atomic motion, and eigenvalues, which reflect the individual atomic contributions to that motion (Mishra et al., 2023;Papaleo et al., 2009).The key component analysis of the simulation trajectories of target enzymes showed a comparatively lower dynamic motion of the residues in the presence of phytochemicals than that of the standards.Reducing the motion of Ca atoms through interactions with a ligand can lead to increased structural stability and rigidity of the protein, which may enhance binding affinity and specificity (Al-Khafaji et al., 2021;Gupta et al., 2021).This is because reduced motion can decrease the conformational entropy of the protein, making it more likely to adopt a specific conformation that is compatible with the ligand (Maisuradze et al., 2009).Studying the Free Energy Landscapes (FEL) along the first two principal components helped to better understand the sub-conformational patterns of the target enzymes and phytochemical complexes (Qais et al., 2023;Moharana et al., 2023).The FEL depends on the configuration, duration, temperature, and quantity of simulation trajectories (Londhe et al., 2019).It allows for the identification of two key parameters: conformational variability and free energy.The conformational variability parameter measures the degree of diversity among the sampled conformations of the protein, while the free energy parameter reflects the relative stability and accessibility of these conformations (Maisuradze et al., 2010;Prada-Gracia et al., 2009).These two parameters provide valuable insights into the protein's conformational dynamics and can aid in understanding its biological function and the design of inhibitors or drugs (Baidya et al., 2022;Mishra et al., 2023).As evidence of the stability of protein-phytochemical interactions, all the SHV-1, KPC-2, OXA-48, and NDM-1 bL complexes in our study had well-defined and compact energy minima.
Several in-vitro studies have reported that different parts of selected dry fruits can inhibit bacterial growth due to the secondary metabolites they contain.Arslan et al. (2023) reported that the husk and leaves of walnuts exhibit significant inhibitory activities against Pseudomonas aeruginosa (P.aeruginosa) and Staphylococcus aureus (S. aureus).Figs have also shown the potential to not only the growth but also kill certain pathogens such as Enterobacter cloacae (Souhila et al., 2021).Peanuts and cashews contain a variety of biological benefits, including antifungal, antibacterial, and antioxidant properties (Mingrou et al., 2022).Sutrisno et al. (2021) reported that peanut oils and derivatives have potential inhibitory effects on E. coli and S. aureus.Because of their significant pharmacological activities and effectiveness as a natural medication, cashew nuts have a high potential as natural functional foods (Shahrajabian & Sun, 2023).Overall, the findings from these in-vitro studies highlight the diverse antimicrobial properties of selected dry fruits, providing valuable insights into their potential application as natural remedies against bacterial infections.Further investigations are warranted to explore the specific bioactive components and their mechanisms of action, paving the way for the development of novel therapeutics based on these natural sources.
The pharmacokinetic analysis of the top phytochemicals was also performed based on Lipinski's rule of five (Lipinski et al., 2001).This rule is used to evaluate the best drug-like phytochemical, but this rule was not strictly followed in our study to determine inhibitors for b-lactamases.Because many natural products have a somewhat different chemical domain than synthetic compounds and often remain bioavailable regardless of violating this rule (Parida et al., 2021).In addition to this, some other pharmacophore properties such as TPSA, Caco-2 permeability, and hepatotoxicity were also determined.Our study suggests that the phytochemicals of selected dry fruits have the potential to block the activity of b-lactamases commonly reported in pathogenic bacteria and these suggested compounds outperform currently marketed synthetic inhibitors.
Apart from these phytochemicals, many anti-microbial peptides derived from plants and animals can also be used to inhibit b-lactamases and other antibiotic-resistant enzymes (Basu et al., 2022;Lewies et al., 2019).These peptides are effective against a wide range of bacteria, including those that are resistant to conventional antibiotics (Thawabteh et al., 2023).Additionally, researchers are also exploring the use of nanoparticles to deliver antibiotics directly to infected cells, bypassing the need for systemic delivery and reducing the risk of side effects (Bahrani et al., 2023;Lagan� a et al., 2023).Other promising strategies include the use of bacteriophages, which are viruses that specifically target and kill bacteria, and the development of new antibiotics that target different aspects of bacterial physiology (Ruemke et al., 2023).A multi-pronged approach is needed to combat antibiotic resistance, including better stewardship of existing antibiotics, improved infection control measures, and continued investment in research and development.

Conclusion
A valuable strategy in the search for new drugs is identifying an inhibitor of a protein or enzyme using computational methods.Such an approach offers a cost-effective and efficient solution for the discovery of inhibitors, as it can reduce the time needed to discover potential drugs.Thus, pharmacoinformatics techniques, namely in-silico screening, molecular docking, molecular dynamic simulations, and pharmacophore profiling, were used in our study to identify potential bL inhibitors from selected dry fruits.The commercially available synthetic bL inhibitors, that is, Avibactam and Tazobactam, were used as standards.Initial screening of phytochemicals was done based on their docking affinities for target enzymes.The phytochemicals with binding affinities higher than those of standards were screened for further analysis by MD simulation.The molecular docking and simulation studies provided significant results that the phytocompounds viz.Protocatechuic acid, Oleanolic acid, and Tannin can inhibit the active sites of target enzymes.PCA and FEL also demonstrated the stability of protein and phytochemicals complexes.Most of these phytochemicals also yielded satisfactory results for pharmacophore analysis.To sum up, it can be inferred that phytochemicals driven by selected dry fruits may have chemical properties that, when put into wet lab test experiments, will successfully inhibit b-lactamases and may be potential drug candidates to combat microbial infections.

Figure 1 .
Figure 1.Flow diagram of the study.
Pi/ Pi-Sigma/Pi-Cation/ Pi-Anion/Pi-Amide) time during simulation; FigureS5(supplementary file).For Oleanolic acid, the average RMSD ranged between 0.15 nm and 0.2 nm, remaining stable for most of the simulation.The RMSD of Tazobactam ranged from 0.15 nm to 0.21 nm.It showed slight fluctuations from 38 ns to 41 ns but remained stable at the very start and end of the simulation.Avibactam, in contrast, showed more fluctuations at the start and then from the time period of 20 ns to 35 ns.It also stabilizes towards the end, with RMSD values ranging between 0.1 nm and 0.24 nm.RMSD values of ligands in complexed with OXA-48 bL were all, however, in acceptable range, i.e. � 0.2 nm (2 Å).Similarly, the RMSF of each ligand did not show any major deviations or, in simple words, the flexibility of the structural changes.RMS fluctuations of both Oleanolic acid and Tazobactam were recorded in amino acid regions of[270][271][272][273][274][275][276][277][278][279][280] Avibactam showed slight variation as its fluctuations were in the30-70, 100-150, 160-180, 200-210, and 250-280  residue regions.Most of the RMS fluctuations of each ligand were recorded in the docking regions.

Figure 8 .
Figure 8. RMSD plot of the OXA-48-ligand complexes.The Radius of Gyration plot of the OXA-48-ligand complexes.The RMSF plot of the OXA-48-ligand complexes.Hydrogen bonding pattern of the OXA-48-ligand complexes.

Figure 9 .
Figure 9. RMSD plot of the SHV-1-ligand complexes.The Radius of Gyration plot of the SHV-1-ligand complexes.The RMSF plot of the SHV-1-ligand complexes.Hydrogen bonding pattern of the SHV-1-ligand complexes.

Figure 10 .
Figure 10.RMSD plot of the KPC-2-ligand complexes.The Radius of Gyration plot of the KPC-2-ligand complexes.The RMSF plot of the KPC-2-ligand complexes.Hydrogen bonding pattern of the KPC-2-ligand complexes.

Figure 11 .
Figure 11.RMSD plot of the NDM-1-ligand complexes.The Radius of Gyration plot of the NDM-1-ligand complexes.The RMSF plot of the NDM-1-ligand complexes.Hydrogen bonding pattern of the NDM-1-ligand complexes.

Figure 13 .
Figure 13.The solvent accessible surface area (SASA) plot for SHV-1 bL and KPC-2 bL (unbound and bound complexes) for simulation period of 50 ns.

Figure 12 .
Figure 12.The solvent accessible surface area (SASA) plot for OXA-48 bL and NDM-1 bL (unbound and bound complexes) for simulation period of 50 ns.

Figure 14 .
Figure 14.Projection of the motion of the OXA-48 bL protein in bound and unbound forms in phase space along the PC1 and PC2.

Figure 15 .
Figure 15.Projection of the motion of the SHV-1 bL protein in bound and unbound forms in phase space along the PC1 and PC2.

Figure 16 .
Figure 16.Projection of the motion of the KPC-2 bL protein in bound and unbound forms in phase space along the PC1 and PC2.

Figure 17 .
Figure 17.Projection of the motion of the NDM-1 bL protein in bound and unbound forms in phase space along the PC1 and PC2.

Table 1 .
The binding affinities (Kcal/mol) of the best 11 phytochemicals and 2 commercially available bL inhibitors, interactions were analyzed using AutoDock tools and were visualized using PyMOL.
�Standard in use bL inhibitors.†

Table 2 .
The hydrogen/hydrophobic bonds and interacting residues of the complexes of best phytochemicals with target enzymes, along with distances between atoms, the interactions were analyzed by BIOVIA Discovery Studio Visualizer and PyMOL software.Complexes

Table 3 .
MM-GBSA-based free energy analysis of simulation trajectories of each docking complex.

Table 4 .
The Lipinski properties of the compounds screened for their potential as Beta lactamase inhibitors.