In Silico and In Vitro Screening of Natural Compounds as Broad-Spectrum β-Lactamase Inhibitors against Acinetobacter baumannii New Delhi Metallo-β-lactamase-1 (NDM-1)

Abstract

Antibiotic resistance is one of the significant problems globally; there is an increase in resistance with introducing every new class of antibiotics. Further, this has become one of the reasons for arising of new resistance mechanisms in Acinetobacter baumannii. In this study, we have screened natural compounds as a possible inhibitor against the NDM-1 β-lactamase enzyme from A. baumannii using a combination of in silico methods and in vitro evaluation. The database of natural compounds was screened against NDM-1 protein, using Glide docking, followed by QM-polarised ligand docking (QPLD). When the screened hits were validated in vitro, withaferin A and mangiferin had good IC₅₀ values in reducing the activity of NDM-1 enzymes, and their fractional inhibitory concentration index (FICI) was ascertained in combination with imipenem. The withaferin A and mangiferin-NDM-1 docking complexes were analyzed for structural stability by molecular dynamic simulation analysis using GROMACS for 100 ns. The molecular properties of the natural compounds were then calculated using density functional theory (DFT). Withaferin A and mangiferin showed promising inhibitory activity and can be a natural compound candidate inhibitor synergistically used along with carbapenems against NDM-1 producing A. baumannii.

This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1. Introduction

The worldwide increase in carbapenem resistance among Gram-negative bacteria has become a significant clinical concern. Acinetobacter is one such species that showed high resistance after just four years of its identification in 1971 and as early as the 1990s. They were resistant to imipenem, the trusted drug of choice [1] [2]. Resistance to carbapenems in A. baumannii (carbapenem-resistant A. baumannii (CRAB)) can itself make them highly resistant. Apart from the intrinsic resistance determinants, oxacillinases (Ambler Class D) and Metallo β-lactamases (Ambler Class B) are major contributors to resistance against carbapenems [3].

There are three subclasses under the Metallo β-lactamases, B1, B2, and B3; the New Delhi Metallo-β-lactamase 1 (NDM-1) belongs to B1. NDM-1 is an emerging concern among the heterogeneous group of carbapenemases, first described from Klebsiella pneumoniae, Escherichia coli, Pseudomonas aeruginosa, A. baumannii, and, more recently, Enterobacteriaceae [4] [5] [6].

NDM-1-positive strains have been rapidly and widely spreading in many countries among A. baumannii strains [7]. NDM-1-positive strains are resistant to fluoroquinolones, aminoglycosides, and β-lactams (especially carbapenems) while being susceptible to colistin and sometimes to tigecycline. NDM-1 resistance dissemination occurs by transferring resistance plasmids with the bla_NDM-1 gene, making them a severe clinical and public health concern [7].

The crystal structure of NDM-1 reveals two zinc ions, which are essential for cleaving the C-N bond in the β-lactam ring of the antibiotic, thereby inactivating them. NDM-1 structure comprises lysine-rich residues that help protonate lactam nitrogen in β-lactam antibiotics [8]. MBLs are resistant to commercially available β-lactamase inhibitors, such as clavulanic acid, sulbactam, tazobactam, and avibactam, and there are no commercially available MBL inhibitors meant for clinical use [9]. Increased resistance to carbapenem in A. baumannii is often seen with a high mortality rate. The current management of infections and outbreaks due to MDR A. baumannii requires a combination of medical interventions, antibiotic stewardship, environmental, and prevention of resistance dissemination. There is an urgent need for newer ways to combat the infections; one is to synergistically enhance the bioactivity of the available drugs with inhibitors of resistance determinants [10]. The purpose of this research is to search for natural compounds that can inhibit acquired β-lactamases, which contribute to carbapenem resistance in A. baumannii, specifically NDM-1 Metallo-lactamase. The workflow for the study is given as in Figure 1. A preliminary computational method was used to identify a subset of compounds from a database of natural compounds by predicting their binding mode against the target proteins from in silico molecular docking experiments and density functional theory (DFT). The top hits were validated by molecular dynamic simulation studies and in vitro enzyme inhibition assay, and fractional inhibitory concentration and in silico ADME properties were predicted.

[figure omitted; refer to PDF]

The enzyme inhibition assay with β-lactamase and NDM-1 enzyme was performed with 8 compounds, mentioned in Table 1. Of the eight, only two compounds (withaferin and mangiferin) showed the highest activity, and two compounds showed moderate activity (rutin and mangostin) (explained in Section 3.5). The four compounds were further analyzed for docking interaction with NDM-1. Four compounds, rutin, mangiferin, withaferin A, and mangostin, were found to have good Glide score for NDM-1 (-9.44, -9.12, -5.12, and -6.54 kcal/mol, respectively) and the Glide score for the known inhibitor D-captopril (-5.52), respectively. Molecular docking analysis showed that all four selected compounds interacted with NDM-1 catalytic amino acid residues through hydrogen bonding and hydrophobic interactions.

Withaferin A, a terpenoid containing four cycloalkane ring structures, three cyclohexane rings, and one cyclopentane ring, is isolated from Acnistus arborescens and Withania somnifera [29]. Recent research has shown that withaferin A has anticancer, adaptogenic, antistress, anticonvulsant, immunomodulatory, and neurological effects. It is a natural proteasome inhibitor, an apoptosis inducer by inhibiting the topoisomerase-I DNA complex, and a mitotic poison with antiangiogenic properties [30]. Withaferin A showed a Glide score of -5.12 and -6.21 kcal/mol and Glide energy of -36.50.15 and -43.02 kcal/mol from QPLD, respectively, and binding affinity with NDM-1. In both XP and QPLD docking analyses, withaferin interacted strongly with ASP124 (1.8 Å) and Zn+ and formed hydrophobic interactions with ILE35, VAL73, MET67, and TRP93 (Figure 3 and Suppl. Figure 2). A pharmacophore of withaferin A involves the 4-hydroxy-5,6-epoxy-22-en-1-one moiety, and its unsaturated lactone has been identified as essential for cytotoxic activity by structure-activity relationship studies [31]. In this study, we found that (6S)-6-ethyl-3-(hydroxymethyl)-4-methyl-5,6-dihydropyran-2-one region of withaferin A formed hydrogen bond with active site amino acids in NDM-1. This showed that (6S)-6-ethyl-3-(hydroxymethyl)-4-methyl-5,6-dihydropyran-2-one region might be important in the biological activity of the withaferin A against β-lactamase (Figure 4).

[figure omitted; refer to PDF]

Mangiferin is a xanthone found in higher plants and various parts of the mango crop, including the peel, stalks, leaves, barks, kernel, and stone. It is a promising antioxidant with a long list of health benefits, including antiviral, anticancer, antidiabetic, antioxidative, antiaging, immunomodulatory, hepatoprotective, and analgesic properties [32].

Mangiferin docked with NDM-1 protein gave a Glide score of -9.12 and -10.62 kcal/mol, and Glide energy of -57.25 and -52.76 kcal/mol was observed by XP and QPLD analysis. Mangiferin formed hydrogen bond between 9H-xanthen-9-one and D-glucitol region and active site amino acids ASP223 (2.0 Å), GLU152 (2.4 Å), ASP124 (2.5 Å), and Zn+ in XP docking and GLU152 (1.9 Å), GLN123 (1.9 Å), and ASN220 (2.0 Å) in QPLD docking (Figure 3 and Suppl. Figure 2). The 1,5-anhydroglucitol region of mangiferin formed hydrogen bonds with active site amino acids of NDM-1, indicating the possible biological activity of mangiferin (Figure 4).

Mangostin is a significant xanthone found in the Garcinia mangostana Linn (mangosteen tree’s) fruit pericarps, bark, and dried sap. Mangostin has a wide range of pharmacological effects, including antioxidant, anticancer, and anti-inflammatory effects [33]. From docking analysis with NDM-1, Glide scores of -6.54 and -7.76 kcal/mol and Glide energy of -40.17 and -44.45 kcal/mol were observed in the XP and QPLD. The hydrogen bonds were formed between 9H-xanthene region of mangostin and active site amino acids GLU152 (2.2 Å) in XP docking and ASN220 (2.1 Å) and GLU152 (1.7 Å) in QPLD docking (Figure 3). In QPLD docking, 9H-xanthene region of mangostin formed PI–PI cation interaction with HIE122 and a strong interaction between Zn+ in the active site (Figure 3 and Suppl. Figure 2).

A flavonol glycoside, rutin (quercetin-3-rutinoside), is between quercetin and α-l-rhamnopyranosyl-(16)—d-glucopyranose [34]. Rutin contains flavonolic aglycone quercetin and disaccharide rutinose [35], the hydroxy (-OH) groups in the sugar moiety aid in the formation of H-bond interactions between amino acid residues. Docking results of NDM-1 and rutin are shown in Table 1 and (Figure 3 and Suppl. Figure 2). Rutin with NDM-1 showed a Glide score of -9.44 and -9.31 kcal/mol and Glide energy of -59.59 and -88.67 kcal/mol in XP docking and QPLD docking methods. In XP docking, rutin interacted with GLU152 (1.6 Å), ASP124 (2.4 Å), GLU152 (2.4 Å), ASP223 (2.0 Å), and Zn+ ions in QPLD. In both docking methods, rutin strongly binds to the two Zn+ ions. And it also formed PI–PI interaction between its quercetin region and HIS250 residue in the active site.

The Structural Interaction Fingerprints (SIFt) algorithm helps find the types of interactions in a binding site by locating amino acids around bound ligands and residue types. SIFt was analyzed using residue features such as backbone, side chain, hydrophobic, aromatic, acceptor, donor, polar, and charged groups in the neighbourhood of the ligand’s binding site. Withaferin A interacts with most of the side chains in the active site except GLY219 (Table 2). Two hydrophobic interactions were found in the ILE35 and TRP93. It also forms polar interactions with HIS122, ASP124, HIS189, SER217, ASN220, and HIS250. Withaferin A forms aromatic (TRP93), acceptor (ASP124), and charged (ASP124) interaction with NDM-1 binding site. GLY219 and HIS250 form backbone interaction with withaferin A. In SIFt analysis, mangiferin shows side chain and polar interaction with most of the amino acids in the binding site in the NDM-1 (Table 3). It forms hydrophobic interaction with ILE35 and forms acceptor and charged interaction with ASP124, GLU152, and ASP223. The backbone amino acid interaction between mangiferin is found in the HIS122, GLN123, and GLY219, respectively.

Table 2

SIFt analysis of NDM-1 binding site amino acid interaction with withaferin A.

Amino acid	Side chain	Hydrophobic	Aromatic	Polar	Acceptor	Charged	Backbone
ILE35	1	1	0	0	0	0	0
TRP93	1	1	1	0	0	0	0
HIS122	1	0	0	1	0	0	0
ASP124	1	0	0	1	1	1	0
HIS189	1	0	0	1	0	0	0
SER217	1	0	0	1	0	0	0
GLY219	0	0	0	0	0	0	1
ASN220	1	0	0	1	0	0	1
HIS250	1	0	0	1	0	0	0

Table 3

SIFt analysis of NDM-1 binding site amino acids interaction with mangiferin.

Amino acids	Side chain	Hydrophobic	Polar	Acceptor	Charged	Backbone
ILE35	1	1	0	0	0	0
HIS122	1	0	1	0	0	1
GLN123	1	0	1	0	0	1
ASP124	1	0	1	1	1	0
GLU152	1	0	1	1	1	0
HIS189	1	0	1	0	0	0
LYN211	1	0	1	0	0	0
GLY219	0	0	0	0	0	1
ASN220	1	0	1	0	0	0
ASP223	1	0	1	1	1	0
HIS250	1	0	1	0	0	0

3.3. Validation of XP Docking Protocol

The validity of the docking protocol was performed by redocking the X-ray crystal structure ligand (i.e., D-captopril) at the active site of NDM-1. RMSDs and amino acid interactions between the docked pose and the crystal structure pose of D-captopril are as shown in the suppl. Figure 3. We found that D-captopril redocked into the active site of NDM-1, which is similar to the crystal structure.

3.4. Binding Free Energy Calculation

The MM-GBSA was used to determine the effect of solvent and free energy on the interaction between natural compounds and NDM-1 proteins. From Glide XP and QPLD docking, ligand-bound complexes were obtained, and MM-GBSA calculation was performed using surface area energy, solvation energy, and energy minimization of the protein-ligand complexes. The MM-GBSA of 4 natural compounds and known inhibitors are shown in Table 4. And for NDM-1 docked poses, Prime MM/GBSA (DGBind) ranges from -32.20 and -33.49 kcal mol (withaferin A), -23.92 and -21.05 kcal mol (mangiferin), -15.74 and -13.92 kcal mol (mangostin), -24.32 and -27.18 kcal mol (rutin), and -11.93 and -18.57 kcal mol (sulbactam) in XP and QPLD docking. The results showed that nonsolvation terms (Glipo), van der Waals (GvdW), and covalent energy (Gcovalent) are more promising contributors for ligand binding. The binding free energies of these known inhibitors do not change significantly.

Table 4

Binding free energy calculation of natural compounds and known inhibitor with NDM-1 protein.

Compounds ID	Binding free energy (kcal/mol) (XP docking)							Binding free energy (kcal/mol) (QPLD)
	ΔG bind	ΔG bind Coulomb	ΔG bind Covalent	ΔG bind Hbond	ΔG bind Lipo	ΔG bind SolvGB	ΔG bind vdW	ΔG bind	ΔG bind Coulomb	ΔG bind covalent	ΔG bind Hbond	ΔG bind Lipo	ΔG bind SolvGB	ΔG bind vdW
Withaferin A	-32.20	-42.92	6.82	-0.70	-14.07	78.07	-32.57	-33.49	-39.72	6.82	-0.88	-14.96	75.30	-33.21
Mangiferin	-23.92	-53.47	6.72	-2.91	-12.04	89.30	-40.44	-21.05	-29.32	7.27	-1.26	-12.79	88.86	-47.54
Mangostin	-15.74	-18.50	3.41	-0.91	-15.24	65.68	-42.97	-13.92	-32.57	3.27	-2.28	-15.42	63.13	-46.23
Rutin	-24.32	-50.66	3.41	-2.72	-10.54	104.10	-44.10	-27.18	-51.41	2.27	-5.48	-15.97	46.60	-48.59
D-captopril	-11.93	-67.53	5.68	-0.45	-4.93	83.25	-22.25	-18.57	-12.76	6.82	-0.35	-5.002	102.46	-22.91

3.5. Determination of the Inhibitory Activity of Candidate Natural Compounds against NDM-1

The IC₅₀ values were calculated for natural compounds and known inhibitors under controlled experimental conditions. The percent inhibitory curves for the candidate natural compounds with efficacy for NDM-1 and β-lactamase enzyme are given in Figures 5(a) and 5(b). The natural compounds were incubated with NDM-1 enzyme, and the substrate, nitrocefin, and hydrolysis of the substrate were monitored at 490 nm OD.

[figures omitted; refer to PDF]

3.5.1. NDM-1 Enzyme Inhibitory Activity Assay

Of the eight natural compounds (at concentrations 10-100 μM) tested for the NDM-1 enzyme inhibitory activity assay, only four compounds showed 50% maximum inhibitory activity suggesting they could be potential inhibitors. The addition of withaferin A, mangiferin, rutin, and mangostin to the enzyme assay reduced the NDM-1 enzyme activity (reduction in the hydrolysis of nitrocefin) significantly with IC₅₀ of $24.03\pm 2.92$ μM for withaferin A, mangiferin $30.6\pm 4.41$ μM, and rutin $48.94\pm 3.74$ μM and minimal reduction with mangostin $97.48\pm 5.59$ μM. The IC₅₀ values of other compounds were more than 100 μM, indicating that these compounds can have minimal or no inhibitory effect on NDM-1 and the IC₅₀ of the known NDM-1 enzyme inhibitor EDTA was $2.081\pm 0.296$ μM.

3.5.2. β-Lactamase Enzyme Inhibitory Activity Assay

The eight candidate natural compounds identified by in silico analysis against NDM-1 were also tested for β-lactamase inhibition assay. Of the eight, the addition of compounds withaferin A, mangiferin, rutin, and mangostin in the assay decreased β-lactamase activity significantly with IC₅₀ of $32.83\pm 2.59$ μM for withaferin A and mangiferin $58.02\pm 2.08$ μM. A very minimal inhibition of the enzyme activity was observed in the presence of rutin $127.4\pm 15.9$ μM and mangostin $136.1\pm 37.1$ μM, while that of the known β-lactamase inhibitor sulbactam was $7.244\pm 0.392$ μM.

3.5.3. Augmentation of Antimicrobial Effect of Imipenem by the Withaferin A and Mangiferin

Every new development in drug discovery is the challenge of resistance that competes equally. Making a comeback for the already available drugs differently is a more approachable method to combat the resistance development. Combination therapy is not a new avenue; combination drugs have been routinely used to treat life-threatening illnesses. Plant natural compounds are weak antimicrobials; however, as seen in in vitro studies, the synergizing effect has proven effective when used together.

The compounds, withaferin A and mangiferin, showed good enzyme inhibition activity against NDM-1 and were checked for their mechanism of action (potential augmentation effect) with imipenem against the carbapenem-resistant A. baumannii strain. The subinhibitory concentrations of the natural compounds were used to test the synergy along with imipenem, and the results were given as fractional inhibitory concentration values.

The checkerboard method was used to design the dilution panel, with twofold concentrations above the MIC of the antibiotics and MIC concentration and fivefold below the MIC. The 96-well microtitre plate was used for the above study, and the effects of the compounds on the MIC of imipenem are classified as synergistic, indifferent, or additive based on the FIC index as given in Table 5. Four natural compounds shortlisted from the enzyme inhibition assay were tested for synergy effect with imipenem, of which withaferin A and mangiferin had a potentiating effect on imipenem.

Table 5

Fractional inhibitory concentration (FIC) and FIC indices (FICIs) of combination of withaferin A and mangiferin with imipenem against A. baumannii MDR strain.

A. baumannii MDR strain	FIC	FIC index	Interpretation
Imipenem with withaferin	0.0625	0.3125	Synergy
Withaferin with imipenem	0.25
Imipenem with mangiferin	0.125	0.625	Synergy
Mangiferin with imipenem	0.5

Withaferin A (128 mg/L) exhibited highly significant synergistic/potentiating effect with imipenem tested against carbapenem-resistant A. baumannii clinical strain (FIC index of 0.3125) followed by mangiferin, (128 mg/L) with an FIC index of 0.625. It is not surprising to see the synergistic potential of withaferin A, which has proven potentiating effect with anticancer drugs [36]. Withania somnifera, the source plant, has numerous therapeutic benefits. A previous structure-activity study by Moujir et al. has explored the antimicrobial properties and the functional groups involved in their potency [37]. Kannan and Kulandaivelu reported the antibacterial activity of Withaferin A activity towards both Gram-negative and Gram-positive bacteria, including Bacillus subtilis, E. coli, and Staphylococcus aureus [38]. Anemarrhena asphodeloides, a plant used in traditional Chinese medicine and whose main active component is mangiferin, has been found to have antiviral and antibacterial activities [39]. Mangiferin has many properties and is traditionally used in many countries; in Cuba, they are available with the brand name Vimang® and in Sri Lanka as Salaretin® [40]. Mangiferin has also been previously reported to be a good potentiator, along with many antibiotics proving to be effective for therapy [41]; this coincides with the findings from our current view of mangiferin in aiding the efficacy of available antibiotics. Mangiferin has been shown to have antibacterial activity against two S. aureus and Salmonella typhi [42] [41]. Mangiferin and its derivatives showed antibacterial and antifungal action against Bacillus pumilus, Bacillus cereus, Salmonella virchow, and two fungal species, Thermoascus aurantiacus and Aspergillus flavus [43].

3.6. ADME

In silico predicted ADME (absorption, distribution, metabolism, and excretion) properties of four natural compounds were done using the QikProp (Table 6). Drug kinetics and tissue exposure, which are essentially determined by their ADME properties, affect a drug’s pharmacological activity and efficacy. Natural compounds were predicted and analyzed for approximately 13 physically important descriptors and pharmacologically active properties. The analysis of predicted ADME properties shows QPlog QPlogPo/w QPpolrz, QPlogS, QPlogPw, QPlogPoct, QPlogKp, QPlogPC16, and Khsa in the allowed range. The physiochemical properties of the natural compounds were within the allowed ADME range. Molecular weights of these natural compounds range between 380 and 610 g/mol, and each comprises 1–6 H-bond donor groups, 2-10 H-bond acceptor groups, and 5-15 rotatable bonds. The identified compounds’ total polar surface area (PSA) was within the 92-273 Ų range. And drug-like properties like QPpolrz, aqueous solubility (QPlogS), hexadecane/gas (QPlogPC16), octanol/gas (QPlogPoct), water/gas (QPlogPw), octanol/water (QPlogPo/w), skin permeability (QPlogKp), and Khsa serum protein binding (QPlogKhs) were shown to be in the allowed region. All the values for natural compounds come under the desirable range making it a suitable drug candidate.

Table 6

ADME properties of four natural compounds.

Compounds	#rotor	mol MW	donorHB	accptHB	QPpolrz	QPlogPC16	QPlogPoct	QPlogPw	QPlogPo/w	QPlogS	QPlogKp	QPlogKhsa	PSA
Accepted range	(1-15)	(130.0–725.0)	(0.0–6.0)	(2.0–20.0)	(13.0–70.0)	(4.0–18.0)	(8.0–35.0)	(4.0–45.0)	(−2.0–6.5)	(−6.5–0.5)	(−8.0–−1.0)	(−1.5–1.5)	(7.0–200.0)
Mangiferin	11	438.344	6	13.75	34.718	14.346	30.63	26.29	-1.82	-2.282	-6.727	-1.013	190.756
Withaferin A	5	470.605	1	9.4	47.967	13.214	23.019	12.717	3.117	-4.997	-3.993	0.385	114.741
Mangostin	8	380.44	2	3.75	39.477	12.498	18.024	8.63	4.449	-5.672	-2.54	0.779	92.69
Rutin	15	610.524	6	18	48.459	17.033	41.62	35.966	-2.558	-2.242	-7.331	-1.298	273.511

3.7. Molecular Dynamic Simulation Results

3.7.1. RMSD

After the QPLD docking study, molecular simulations of all the protein-ligand complexes and apoproteins were carried out for a period of 100 ns. The stability and conformational distribution of the ligands in the receptor protein were calculated using generated RMSD trajectories. For NDM-1, at the initial simulation phase, the backbone of all complexes was increased between 0 and 3 ns. In the apo NDM-1 protein, an increased fluctuation was observed at 43 ns, RMSD of 0.27 nm; after this, a stable deviation was observed throughout the simulation time, with an overall RMSD of 0.25 nm (Figure 6(a)). NDM-1 bound withaferin and D-captopril had a maximum deviation at around 40 ns with RMSD of 0.2 nm and 0.25 nm, respectively, whereas mangiferin RMSD reached a maximum of 0.3 nm before 20 ns simulation. RMSD trajectories of NDM-1, when analyzed together with known inhibitors and natural compounds, showed that after an initial fluctuation, a successful steady confirmation with minimal deviation was maintained throughout the simulation.

[figures omitted; refer to PDF]

3.7.2. RMSF

The conformation and stability of Apo and protein-ligand complexes were determined by their amino acid residues. It was perceived from the RMSF plot that NDM-1 and ligand complex attained the overall low RMSF than apoprotein. The plot suggested the stable binding of all three ligands and residues interacting with all three ligands having low flexibility compared with apoprotein fluctuation data. The resultant RMSF profile of NDM-1 apo and ligand complex showed fluctuations in the range of 0.1 to 0.25 and 0.1 to 0.52 nm at the protein’s catalytic site and noncatalytic sites. We have calculated the B-factor for Apo protein and compared it with RMSF of a protein-ligand complex.

In NDM-1, high fluctuation (0.3 to 0.52 nm) was observed in the loop region of GLU30, ILE31, MET39, ASP66, MET67, PRO68, GLY69, PHE70, and ARG270 in apo and ligand-bounded complexes of NDM-1. These regions belong to the loop region in the NDM-1 and had fluctuations in both apo and ligand-bound complexes. In D-captopril–NDM-1 complex, high fluctuation was observed in the N-terminus loop region of GLU30, THR34, GLN37, and THR41 with RMSF around 0.25 to 0.46 nm. In ligand-bounded NDM-1 complex, slight fluctuations were observed in the loop region of ASN220, LEU221, GLY222, ASP223, ALA224, and ARG270 RMSF of 0.2 to 0.38 nm (Figure 6(b)). In this study, NDM-1 ligand-bounded complexes had no significant fluctuations at the ligand-binding site in the protein-ligand complexes compared with the Apo protein (Figure 6(b)). B-factor RMSF from the crystal structure and the simulated RMSF of NDM-1 were highly correlated throughout 100 ns.

3.7.3. The Radius of Gyration (Rg)

The radius of gyration (Rg) indicates the size and compactness of the protein. The Rg values of NDM-1 complexes were 1.7 nm at the initial state. The fluctuation of Rg of apo and ligand-bounded NDM-1 was observed initially, and it was stabilized after five ns. For apoprotein, Rg fluctuation was observed in 27 and 38 ns, and for withaferin A–NDM-1 complex, a slight increase in fluctuations was there at 43 and 48 ns, and the system was stable after that, still 100 ns. For D-captopril–NDM-1 complex, there was high fluctuation initially, and it was stabilized after 20 ns till 100 ns. For mangiferin–NDM-1 complex, overall, there was no fluctuation observed, and the system was stable till 100 ns (Figure 6(c)).

3.7.4. SASA

To better understand the impact of inhibitors on protein accessible surface area for solvent, we calculated solvent accessible surface area (SASA). The average SASA value of the NDM-1 ligand complex was calculated as 105-119 nm throughout 100 ns. The average SASA values of $118\pm 5$ nm for withaferin A, $121\pm 10$ nm for mangiferin, and $120\pm 9$ nm for D-captopril were observed in 100 ns (Figure 6(d)). All the protein-ligand complexes and apoproteins showed minimum fluctuations during 100 ns. The SASA results showed that active site residues were well exposed to the solvent and readily accessible.

3.7.5. Hydrogen Bond

The number of hydrogen bonds was calculated during simulations for ligand-bound complexes of NDM-1. For NDM-1, an average of 2-3 hydrogen bonds in the NDM-1 ligand-bound complex were observed during 100 ns of simulations. For D-captopril–NDM-1 complex, 3-4 hydrogen bonds were observed in 90-95 ns. Overall, hydrogen bonds between the three ligands had stable interaction with the active site residues of NDM-1 (Figure 6(e)).

3.8. MMPBSA Analysis

The binding free energies were calculated using polar and nonpolar energy terms. During the 100 ns MD simulations, the energies associated with the binding of withaferin A, mangiferin, and D-captopril with NDM-1 were calculated. The following energies have been calculated: vdW interaction energy, electrostatic energy, polar solvation energy, SASA energy, and average binding energy (Table 7). NDM-1/ligand complexes and withaferin A/NDM-1 complex showed the lowest binding energy of -96.597 kcal/mol followed by mangiferin/NDM-1 and D-captopril/NDM-1 complexes -168.570 kcal/mol and -132.842 kcal/mol, respectively. The polar and nonpolar energies were distinct in each complex. The lowest polar solvation energy was observed in withaferin A–ligand complexes (39.594 kJ/mol), and the maximum energy was observed in the mangiferin–NDM-1 complex (62.730 kJ/mol). The withaferin A and mangiferin had good binding energies with NDM-1.

Table 7

Binding free energy and interaction energy of NDM-1–ligand complexes calculated using the MM-PBSA approach.

Target	ID	Binding energy (kJ/mol)	Van der Waal energy (kJ/mol)	Electrostatic energy (kJ/mol)	Polar solvation energy (kJ/mol)	SASA energy (kJ/mol)
NDM-1	Withaferin A	$$-96.597\pm 56.626$$	$$-112.747\pm 67.710$$	$$-12.105\pm 12.260$$	$$39.594\pm 38.035$$	$$-11.339\pm 6.804$$
	Mangiferin	$$-168.570\pm 48.904$$	$$-97.906\pm 44.098$$	$$-114.624\pm 77.901$$	$$62.730\pm 31.574$$	$$-10.878\pm 5.269$$
	D-captopril	$$-132.842\pm 35.026$$	$$-60.775\pm 19.936$$	$$-99.864\pm 55.580$$	$$101.701\pm 66.646$$	$$-7.948\pm 2.811$$

3.9. Density Functional Theory (DFT) Calculations

3.9.1. HOMO and LUMO Analysis

The properties of the molecular structure were explained using DFT calculations. DFT calculations at the level of B3LYP/6-31G$\ast \ast$ were optimized to produce these four molecules. In chemical reactivity, the molecules are sparkly, and the HOMO–LUMO distance increases charge transfer. The molecular orbital interactions with the other species and their energy difference (gap) assist in quantifying the chemical reactivity of the structure that participates in the chemical reactions. Naturally, the electron density is indicated by the intensity of the colour that would reflect the molecule’s characteristic feature. The chemical reactivity of HOMO, LUMO, and MESP has been used to analyze molecule parameters. The LUMO is commonly used as an electron acceptor, while the HOMO is used as an electron donor. The stability of the structure was used to analyze the HOMO and LUMO energy gaps. Energy gap levels in the HOMO and LUMO elucidate the fragile essence of reactivity since electrons can be transferred quickly between the energy levels. In HOMO and LUMO, energy gaps of the natural compounds and known inhibitors are 0.1865 eV (withaferin A), 0.1530 eV (mangiferin), and 0.2394 eV (D-captopril), respectively (Table 8). The lesser the energy gap, the more energetically favourable the compound is for chemical reaction. As a result, the values indicate that the molecules have good chemical reactivity. The stability of the compounds was influenced by total energy, dipole moment, HOMO, LUMO, and energy distance. The natural compounds were plotted in the HOMO and LUMO energy gaps (Figure 7). The colour-coding area was red, which indicated the most negative potential region, and blue, which indicated the most positive potential region in the natural compounds.

Table 8

HOMO, LUMO, HELG, ESP, and solvation energy parameter of the withaferin A, mangiferin, and D-captopril.

S. no	Compound IDs	HOMO (eV)	LUMO (eV)	HLG (eV)	QM dipole (debye)	Solvation energy (kcal/mol)	ESP min kcal/mol	ESP max kcal/mol
1	Withaferin A	-0.25442	-0.06792	0.1865	8.9441	-23.82	-62.87	63.53
2	Mangiferin	-0.21798	-0.06246	0.1555	7.6612	-15.80	-74.91	79.73
3	D-captopril	-0.2394	-0.0058	0.2394	11.2215	-71.45	-79.28	76.73

[figure omitted; refer to PDF]

In Figure 7, the frontier molecular orbital diagrams, the electrons are localized on 4-hydroxy-5,6-epoxy-22-en-1-one moiety in withaferin A and 2-hydroxymethyl-6methloxane-3,4,5-triol in mangiferin in the HOMO structures. The hydroxy groups in withaferin A and mangiferin are readily available to donate electrons to the interaction groups of amino acids at the binding site of NDM-1, based on the HOMO and LUMO structures. The DFT results agree well with the docking results, indicating that withaferin A and mangiferin are the study’s best inhibitors. Both withaferin A and mangiferin have the smallest HOMO-LUMO gap among the ligands studied, indicating that the inhibitor’s HOMO can transfer electrons to lower-energy LUMO of amino acid residues in the enzyme’s active site.

3.9.2. Electrostatic Potential and Dipole Moment

Furthermore, the molecule’s electrostatic potential and dipole moment have been proposed to determine the various intermolecular interaction properties and the most suitable regions between ligand and receptor [44]. Based on the electron density distribution on the respective molecular surface, the electrostatic map for withaferin A, mangiferin, and D-captopril is shown in Figure 8. The blue colour region on the electrostatic map indicates the strong electrophilic region of electrons in hydrogen atoms. This was observed for four natural compounds in this analysis. Similarly, the higher nucleophilic regions contributed by electron-rich molecules, such as hydroxy and carbonyl groups in the respective bioactive compounds, were coloured red on electrostatic maps. The measured electrostatic potential map for designated bioactive compounds indicated that highly electronegative atoms, such as oxygen, and highly electropositive atoms, such as carbon atoms bonded to oxygen in cyclic chains and hydrogen atoms, can introduce intermolecular interactions with the active residues of proteins in their vicinity during docking simulations [45]. The highly positive points were located at the hydroxy groups of oxygen atoms, and the highly negative points were located at the hydroxy group of hydrogen for the natural compounds (Figure 8). The natural compounds’ high electropositive and electronegative atoms interacted with active site amino acids of NDM-1. The electrostatic potential maximum and minimum ranges for natural compounds and known inhibitors are shown in Table 8, respectively.

[figure omitted; refer to PDF]

3.10. Molecular Docking Studies of Other Types of β-Lactamase with the Four Natural Compounds

The natural compounds withaferin A and mangiferin were also docked with other types of β-lactamases from different organisms to find their broad-spectrum β-lactamase activity. The β-lactamases, ADC-7 (PDB ID: 6PWL, A. baumannii), AmpC (PDB ID: 1KDW, E. coli), CTX-M-15 (PDB ID: 4HBT, E. coli), KPC-2 (PDB ID: 3RXX, K. pneumoniae), OXA-24/40 (PDB ID: 6MPQ, A. baumannii), SHV-1 (PDB ID: 3MXR, K. pneumoniae), TEM-1 (PDB ID: 1TEM, E. coli), and VIM-2 (PDB ID: 2YZ3 P. putida), were selected for the Glide docking studies. The previously used protein preparation and Glide protocol in Sections 2.1 and 2.7 were followed for this β-lactamase and four natural compounds.

Using XP Glide docking, the natural compounds were docked into their active site region of VIM-2, ADC-7, AmpC, CTX-M-15, KPC-2, SHV-1, and TEM-1 β-lactamase. The XP Glide score, Glide energy, and amino acid interactions are shown in Table 9. All four natural compounds had a good docking score and formed hydrogen bonds in the active site amino acids of all seven β-lactamases similar to NDM-1. These results indicate that withaferin A and mangiferin have good binding activity with the other types of β-lactamases.

Table 9

The binding interactions of natural compounds in the active site residues of other β-lactamase enzymes.

Compounds	Beta-lactamase	Glide score	Glide energy	Glide EvdW	Glide Ecoul	Glide emodel	Amino acid interactions
Withaferin A	AMPC	-4.606	-38.022	-23.74	-14.282	-50.842	ARG204, GLN120, ASN289, and ASN346
	TEM-1	-2.791	-31.98	-17.237	-14.743	-34.431	GLU110, SER106, and VAL216
	VIM-2	-3.848	-35.208	-25.42	-9.788	-39.222	ASP97, ASN190
	SHV-1	-4.145	-31.208	-23.642	-7.566	-36.326	MET129
	KPC-2	-3.665	-32.831	-28.41	-4.422	-42.998	CYS238
	OXA 24/40	-5.68	-38.15	-35.609	-2.543	-37.991	SER128, SER81
	CTX-M-15	-4.1	-34.691	-25.794	-8.896	-41.456	GLY238
	ADC-7	-3.562	-30.258	-23.526	-6.732	-35.533	ASN204, GLN120, and PHE121
Mangiferin	AMPC	-7.848	-49.667	-30.064	-19.603	-62.404	ASN343, VAL121, and ASP123
	TEM-1	-4.079	-29.904	-20.442	-9.462	-40.812	ALA237, GLU104
	VIM-2	-6.117	-46.288	-20.793	-25.494	-56.383	TRP67, GLU126, and ASP43
	SHV-1	-7.653	-43.892	-28.665	-15.228	-57.531	TYR105, SER130, ASN276, and VAL216
	OXA 24/40	-10.05	-50.28	-34.581	-15.707	-69.695	ARG261, LEU127, and ALA126
	KPC-2	-6.293	-36.214	-25.905	-10.309	-47.456	TYR129, CYS238
	CTX-M-15	-6.117	-55.386	-26.468	-28.918	-84.69	ASN132
	ADC-7	-4.738	-33.168	-15.139	-18.029	-40.601	GLU289, ASN287, ASN343, and GLU344

4. Conclusions

Half of the global occurrence of infections due to A. baumannii is attributed to resistance to antibiotics of choice, including CRAB. A. baumannii has a solid shield of intrinsic factors and the acquisition of resistance mechanisms, together with working in unison against all known antibiotics of choice for treatment. CRAB is panresistant, often referred to as extensively drug-resistant (XDR). This study focused on identifying potential inhibitors targeting β-lactamases with carbapenem as substrate, using molecular docking, molecular dynamic modeling, in silico pharmacokinetics, and quantum chemical analysis on a collection of over two hundred phytochemical compounds. The in silico screening was done with D-captopril (known inhibitors) against NDM-1 target proteins, respectively, using a comparison of molecular docking scores and hydrogen bond interactions between the active site amino acids of the protein and phytochemicals. The top eight compounds were selected for in vitro NDM-1 inhibition assay; withaferin A, mangiferin, mangostin, and rutin showed inhibitory activity against β-lactamase and NDM-1 enzymes. Based on the IC₅₀ values, withaferin A and mangiferin were subjected to fractional inhibitory concentration assay for ascertaining their mechanistic potential (synergy) with imipenem. The compounds had an FIC index in the range suggestive of possible synergistic mechanism of action potentiating the MIC of imipenem. Further refinement of the results was done using the in silico molecular dynamic simulation studies using GROMACS. The RMSD of NDM-1 and compounds and the RMSF and binding energy obtained from MD simulation trajectories strongly indicated that selected molecules have a good potential as β-lactamase inhibitors. Quantum chemical analyses revealed that the proposed molecules contain several reaction mechanisms to inhibit lactamase. According to pharmacokinetics analyses, all molecules have a strong absorption, distribution, metabolism, and distribution profile. As a result, it can be concluded that withaferin A and mangiferin could be safe natural potentiators and/or inhibitors substituting the available inhibitors against a wide spectrum β-lactamases contributing to carbapenem resistance with a possible use clinically.

Authors’ Contributions

KD and VA designed the study, performed experiments and analyses, and helped to draft the manuscript. LW critically reviewed and edited the manuscript. ZS helped in drawing figures, and SW revised for its integrity and accuracy. MK helped in designing of in vitro studies. WH helped in analyzing computational studies. HX designed the study and approved the final version of this manuscript. Vasudevan Aparna and Kesavan Dinesh Kumar are co-first authors.

References

[1] M. Asif, I. A. Alvi, S. U. J. I. Rehman, D. Resistance, "Insight into Acinetobacter baumannii: pathogenesis, global resistance, mechanisms of resistance, treatment options, and alternative modalities," Infect Drug Resist., vol. Volume 11, pp. 1249-1260, DOI: 10.2147/IDR.S166750, 2018.

[2] Y. Doi, G. L. Murray, A. Y. Peleg, "Acinetobacter baumannii: evolution of antimicrobial resistance—treatment options. Seminars in respiratory and critical care medicine," NIH Public Access, vol. 36, 2015.

[3] L. Poirel, P. Nordmann, "Carbapenem resistance in _Acinetobacter baumannii_ : mechanisms and epidemiology," Clinical Microbiology and Infection, vol. 12 no. 9, pp. 826-836, DOI: 10.1111/j.1469-0691.2006.01456.x, 2006.

[4] K. Bush, G. A. Jacoby, "Updated functional classification of β -lactamases," Antimicrobial Agents and Chemotherapy, vol. 54 no. 3, pp. 969-976, DOI: 10.1128/AAC.01009-09, 2010.

[5] D. Yong, M. A. Toleman, C. G. Giske, H. S. Cho, K. Sundman, K. Lee, T. R. Walsh, "Characterization of a new metallo- β -lactamase gene, blaNDM-1, and a novel erythromycin esterase gene carried on a unique genetic structure in Klebsiella pneumoniae sequence type 14 from India," Antimicrobial Agents and Chemotherapy, vol. 53 no. 12, pp. 5046-5054, DOI: 10.1128/AAC.00774-09, 2009.

[6] K. Karthikeyan, M. Thirunarayan, P. Krishnan, "Coexistence of blaOXA-23 with blaNDM-1 and armA in clinical isolates of Acinetobacter baumannii from India," The Journal of Antimicrobial Chemotherapy, vol. 65 no. 10, pp. 2253-2254, DOI: 10.1093/jac/dkq273, 2010.

[7] K. K. Kumarasamy, M. A. Toleman, T. R. Walsh, J. Bagaria, F. Butt, R. Balakrishnan, U. Chaudhary, M. Doumith, C. G. Giske, S. Irfan, P. Krishnan, "Emergence of a new antibiotic resistance mechanism in India, Pakistan, and the UK: a molecular, biological, and epidemiological study," The Lancet Infectious Diseases, vol. 10 no. 9, pp. 597-602, DOI: 10.1016/S1473-3099(10)70143-2, 2010.

[8] Z. Liang, L. Li, Y. Wang, L. Chen, X. Kong, Y. Hong, L. Lan, M. Zheng, C. Guang-Yang, H. J. P. O. Liu, X. Shen, C. Luo, K. K. Li, K. Chen, H. Jiang, "Molecular basis of NDM-1, a new antibiotic resistance determinant," PLoS One, vol. 6 no. 8, article e23606,DOI: 10.1371/journal.pone.0023606, 2011.

[9] S. D. Lahiri, S. Mangani, T. Durand-Reville, M. Benvenuti, F. De Luca, G. Sanyal, J. D. Docquier, "Structural insight into potent broad-spectrum inhibition with reversible recyclization mechanism: avibactam in complex with CTX-M-15 and Pseudomonas aeruginosa AmpC β -lactamases," Antimicrobial Agents and Chemotherapy, vol. 57 no. 6, pp. 2496-2505, DOI: 10.1128/AAC.02247-12, 2013.

[10] M. J. Cheesman, A. Ilanko, B. Blonk, I. E. Cock, "Developing new antimicrobial therapies: are synergistic combinations of plant extracts/compounds with conventional antibiotics the solution?," Pharmacognosy Reviews, vol. 11, 2017.

[11] N. Guex, M. C. Peitsch, T. Schwede, "Automated comparative protein structure modeling with SWISS-MODEL and Swiss-PdbViewer: a historical perspective," Electrophoresis, vol. 30 no. S1, pp. S162-S173, DOI: 10.1002/elps.200900140, 2009.

[12] R. A. Laskowski, M. A. MW, D. S. Moss, J. M. Thornton, "PROCHECK: a program to check the stereochemical quality of protein structures," Journal of Applied Crystallography, vol. 26 no. 2, pp. 283-291, DOI: 10.1107/S0021889892009944, 1993.

[13] C. Colovos, T. O. Yeates, "Verification of protein structures: patterns of nonbonded atomic interactions," Protein Science, vol. 2 no. 9, pp. 1511-1519, DOI: 10.1002/pro.5560020916, 1993.

[14] J. U. Bowie, R. Luthy, D. J. S. Eisenberg, "A method to identify protein sequences that fold into a known three-dimensional structure," Science, vol. 253 no. 5016, pp. 164-170, DOI: 10.1126/science.1853201, 1991.

[15] US Department of Agriculture, ARS, Dr. Duke’s Phytochemical and Ethnobotanical Databases. Home Page, 1992-2016.

[16] D. S. Raj, C. P. D. Kottaisamy, W. Hopper, U. Sankaran, "Identification of immucillin analogue natural compounds to inhibit Helicobacter pylori MTAN through high throughput virtual screening and molecular dynamics simulation," In Silico Pharmacology, vol. 9, 2021.

[17] M. W. van der Kamp, A. J. Mulholland, "Combined quantum mechanics/molecular mechanics (QM/MM) methods in computational enzymology," Biochemistry, vol. 52 no. 16, pp. 2708-2728, DOI: 10.1021/bi400215w, 2013.

[18] K. D. Singh, K. Muthusamy, "Molecular modeling, quantum polarized ligand docking and structure-based 3D-QSAR analysis of the imidazole series as dual AT 1 and ET A receptor antagonists," Acta Pharmacologica Sinica, vol. 34 no. 12, pp. 1592-1606, DOI: 10.1038/aps.2013.129, 2013.

[19] S. Chinnasamy, G. Selvaraj, A. C. Kaushik, S. Kaliamurthi, A. S. Nangraj, C. Selvaraj, S. K. Singh, R. Thirugnanasambandam, K. Gu, D. Q. Wei, Identification of potent inhibitors against Aurora kinase A using molecular docking and molecular dynamics simulation studies, 2019. http://Preprints.org

[20] Q. Zhang, A. Khetan, S. Er, "Comparison of computational chemistry methods for the discovery of quinone-based electroactive compounds for energy storage," Scientific Reports, vol. 10, 2020.

[21] M. A. Jordaan, O. Ebenezer, N. Damoyi, M. Shapi, "Virtual screening, molecular docking studies and DFT calculations of FDA approved compounds similar to the non-nucleoside reverse transcriptase inhibitor (NNRTI) efavirenz," Heliyon., vol. 6 no. 8, article e04642,DOI: 10.1016/j.heliyon.2020.e04642, 2020.

[22] C. A. Lipinski, "Drug-like properties and the causes of poor solubility and poor permeability," Journal of Pharmacological and Toxicological Methods, vol. 44 no. 1, pp. 235-249, DOI: 10.1016/S1056-8719(00)00107-6, 2000.

[23] P. Das, R. Majumder, M. Mandal, P. Basak, "In-silico approach for identification of effective and stable inhibitors for COVID-19 main protease (Mpro) from flavonoid based phytochemical constituents of Calendula officinalis," Journal of Biomolecular Structure & Dynamics, vol. 39, pp. 6265-6280, 2020.

[24] R. Majumder, M. Mandal, "Screening of plant-based natural compounds as a potential COVID-19 main protease inhibitor: an in silico docking and molecular dynamics simulation approach," Journal of Biomolecular Structure & Dynamics, vol. 40, pp. 696-711, 2020.

[25] R. Kumari, R. Kumar, Open Source Drug Discovery Consortium, A. Lynn, "g_mmpbsa—A GROMACS tool for high-throughput MM-PBSA calculations," Journal of Chemical Information and Modeling, vol. 54 no. 7, pp. 1951-1962, DOI: 10.1021/ci500020m, 2014.

[26] A. M. King, S. A. Reid-Yu, W. Wang, D. T. King, G. De Pascale, N. C. Strynadka, T. R. Walsh, B. K. Coombes, G. D. Wright, "Aspergillomarasmine A overcomes metallo- β -lactamase antibiotic resistance," Nature, vol. 510 no. 7506, pp. 503-506, DOI: 10.1038/nature13445, 2014.

[27] N. Parvaiz, F. Ahmad, W. Yu, A. D. MacKerell, S. S. Azam, "Discovery of beta-lactamase CMY-10 inhibitors for combination therapy against multi-drug resistant Enterobacteriaceae," PLoS One, vol. 16 no. 1, article e0244967,DOI: 10.1371/journal.pone.0244967, 2021.

[28] V. Aparna, K. Dineshkumar, N. Mohanalakshmi, D. Velmurugan, W. Hopper, "Identification of natural compound inhibitors for multidrug efflux pumps of Escherichia coli and Pseudomonas aeruginosa using in silico high-throughput virtual screening and in vitro validation," PLoS One, vol. 9 no. 7, article e101840,DOI: 10.1371/journal.pone.0101840, 2014.

[29] F. Aqil, R. Munagala, A. K. Agrawal, R. Gupta, "Anticancer phytocompounds: experimental and clinical updates," Journal of Virological Methods, vol. 1, pp. 237-272, DOI: 10.1016/B978-0-12-814619-4.00010-0, 2019.

[30] S. K. Yousuf, R. Majeed, M. Ahmad, P. Lal Sangwan, B. Purnima, A. K. Saxsena, K. A. Suri, D. Mukherjee, S. C. Taneja, "Ring A structural modified derivatives of withaferin A and the evaluation of their cytotoxic potential," Steroids, vol. 76 no. 10-11, pp. 1213-1222, DOI: 10.1016/j.steroids.2011.05.012, 2011.

[31] G. Garg, A. Khandelwal, B. S. Blagg, "Anticancer inhibitors of Hsp90 function: beyond the usual suspects," Advances in Cancer Research, vol. 129, pp. 51-88, DOI: 10.1016/bs.acr.2015.12.001, 2016.

[32] M. Imran, M. S. Arshad, M. S. Butt, J.-H. Kwon, M. U. Arshad, M. T. Sultan, "Mangiferin: a natural miracle bioactive compound against lifestyle related disorders," Lipids in Health and Disease, vol. 16, 2017.

[33] Y. Akao, Y. Nakagawa, Y. Nozawa, "Anticancer effects of xanthones from pericarps of mangosteen," International Journal of Molecular Sciences, vol. 9 no. 3, pp. 355-370, DOI: 10.3390/ijms9030355, 2008.

[34] R. Jadeja, R. Devkar, "Chapter 47 - polyphenols and flavonoids in controlling non-alcoholic steatohepatitis," Polyphenols in Human Health and Disease, vol. 1, pp. 615-623, 2014.

[35] A. Ganeshpurkar, A. K. Saluja, "The pharmacological potential of rutin," Saudi Pharm J., vol. 25 no. 2, pp. 149-164, DOI: 10.1016/j.jsps.2016.04.025, 2017.

[36] F. A. Al Hassan Kyakulaga, R. Munagala, R. C. J. O. Gupta, "Synergistic combinations of paclitaxel and withaferin A against human non-small cell lung cancer cells," Oncotarget, vol. 11 no. 16, pp. 1399-1416, DOI: 10.18632/oncotarget.27519, 2020.

[37] L. Moujir, L. Araujo, G. Llanos, I. Jimenez, I. Bazzocchi, "Evaluation of withaferin-A and analogues as antibacterial agents: structure-activity relationship study," Planta Medica, vol. 80, 2014.

[38] N. Kannan, G. Kulandaivelu, "Novel method to isolate withaferin A from Withania somnifera roots and its bioactivity," Allelopathy Journal, vol. 20, pp. 213-220, 2007.

[39] T. Miura, H. Ichiki, N. Iwamoto, M. Kato, M. Kubo, H. Sasaki, M. Okada, T. Ishida, Y. Seino, K. TANIGAWA, "Antidiabetic activity of the rhizoma of Anemarrhena asphodeloides and active components, mangiferin and its glucoside," Biological and Pharmaceutical Bulletin, vol. 24 no. 9, pp. 1009-1011, DOI: 10.1248/bpb.24.1009, 2001.

[40] S. Singh, Y. Kumar, S. S. Kumar, V. Sharma, K. Dua, A. Samad, "Antimicrobial evaluation of mangiferin analogues," Indian Journal of Pharmaceutical Sciences, vol. 71 no. 3, pp. 328-331, DOI: 10.4103/0250-474X.56023, 2009.

[41] N. A. Mazlan, S. Azman, N. F. Ghazali, P. Z. S. Yusri, H. M. Idi, M. Ismail, M. Sekar, "Synergistic antibacterial activity of mangiferin with antibiotics against Staphylococcus aureus," Drug Invention Today, vol. 12, pp. 14-17, 2019.

[42] T. Biswas, A. Sen, R. Roy, S. Maji, H. S. Maji, "Isolation of mangiferin from flowering buds of Mangifera indica L and its evaluation of in vitro antibacterial activity," J Pharm Anal, vol. 4, pp. 49-56, 2015.

[43] S. K. Singh, R. M. Tiwari, S. K. Sinha, C. C. Danta, S. K. Prasad, "Antimicrobial evaluation of mangiferin and its synthesized analogues," Asian Pacific Journal of Tropical Biomedicine, vol. 2 no. 2, pp. S884-S887, DOI: 10.1016/S2221-1691(12)60329-3, 2012.

[44] E. J. Yearley, E. A. Zhurova, V. V. Zhurov, A. A. Pinkerton, "Experimental electron density studies of non-steroidal synthetic estrogens: diethylstilbestrol and dienestrol," Journal of Molecular Structure, vol. 890 no. 1-3, pp. 240-248, DOI: 10.1016/j.molstruc.2008.03.053, 2008.

[45] K. E. Lee, S. Bharadwaj, U. Yadava, S. G. Kang, "Computational and in vitro investigation of (-)-epicatechin and proanthocyanidin B2 as inhibitors of human matrix metalloproteinase 1," Biomolecules, vol. 10 no. 10,DOI: 10.3390/biom10101379, 2020.