Integrated virtual screening and molecular dynamics simulation approaches revealed potential natural inhibitors for DNMT1 as therapeutic solution for triple negative breast cancer

Abstract Triple negative breast cancers (TNBC) are clinically heterogeneous but mostly aggressive malignancies devoid of expression of the estrogen, progesterone, and HER2 (ERBB2 or NEU) receptors. It accounts for 15–20% of all cases. Altered epigenetic regulation including DNA hypermethylation by DNA methyltransferase 1 (DNMT1) has been implicated as one of the causes of TNBC tumorigenesis. The antitumor effect of DNMT1 has also been explored in TNBC that currently lacks targeted therapies. However, the actual treatment for TNBC is yet to be discovered. This study is attributed to the identification of novel drug targets against TNBC. A comprehensive docking and simulation analysis was performed to optimize promising new compounds by estimating their binding affinity to the target protein. Molecular dynamics simulation of 500 ns well complemented the binding affinity of the compound and revealed strong stability of predicted compounds at the docked site. Calculation of binding free energies using MMPBSA and MMGBSA validated the strong binding affinity between compound and binding pockets of DNMT1. In a nutshell, our study uncovered that Beta-Mangostin, Gancaonin Z, 5-hydroxysophoranone, Sophoraflavanone L, and Dorsmanin H showed maximum binding affinity with the active sites of DNMT1. Furthermore, all of these compounds depict maximum drug-like properties. Therefore, the proposed compounds can be a potential candidate for patients with TNBC, but, experimental validation is needed to ensure their safety. Communicated by Ramaswamy H. Sarma


Introduction
Triple negative breast cancers (TNBC) are among the most aggressive and deadly breast cancer subtypes, with high rates of tumor recurrence and poor overall survival (Carey et al., 2010;Foulkes et al., 2010).Triple negative phenotypes account for 15-20% of all breast cancer cases (Yao et al., 2017).TNBC is more common in breast cancer patients with BRCA1 mutation as compared to those breast cancer patients with BRCA2 mutation (Chen et al., 2018;Pogoda et al., 2020).TNBC shares striking similarities with basal-like breast cancer (BLBC), a subgroup of breast cancer defined by gene-expression profiling (Milioli et al., 2017).TNBC tumors are histologically heterogeneous and may include infiltrating ductal carcinoma and other subtypes including apocrine, medullary, and squamous (Bianchini et al., 2016;Pareja et al., 2016).In comparison with other molecular subtypes of breast cancer, where early pregnancy has been recognized as a protective factor against breast cancer, gestation seems to increase the risk for TNBC phenotype (Boere et al., 2022;Feng et al., 2018).Distant metastasis, lack of novel prognostic markers, and limited treatment options make the early diagnosis of TNBC extremely challenging.Thus, the identification of novel biomarkers that can block the development and pathogenesis of the TNBC is unmet needs.
DNA methylation is essential for silencing retroviral elements, regulating tissue-specific gene expression, genomic imprinting, and X chromosome inactivation (Moore et al., 2013).Importantly, DNA methylation in different genomic regions may exert different influences on gene activities based on the underlying genetic sequence.DNA methyltransferases (DNMT) are a group of enzymes that establish methylation of CpG (cytosine-guanine) dinucleotides and inhibit gene transcription by blocking accessibility to transcriptional activators Recent studies have found that gene expression abnormalities caused by alterations in DNMT activity and function are closely connected with the occurrence and development of various cancers (Jurkowska et al., 2011;Yanagisawa et al., 2002).There are four known members of the DNMT family: DNMT1, DNMT3A, DNMT3B, and DNMT3L.In mammals, DNMT1, DNMT3A and DNMT3B, the generally recognized three types of DNA methyltransferases (DNMTs), execute the genomic methylation process (Okano et al., 1999;Xu et al., 2020).These proteins are highly conserved and have similar amino acid sequences with N-terminus contains a regulatory domain, which allows DNMTs to anchor in the nucleus and recognize nucleic acids or nucleoproteins, and the C-terminus possesses a catalytic domain, which is responsible for the enzymatic activity (Zhang & Xu, 2017).More importantly, DNMT1 inhibited the transcription of tumor suppressor genes which is needed for the normal progression and proliferation of the cell cycle (Cai et al., 2020;Garinis et al., 2002).DNMT1 is involved in tumorigenesis of several cancer types including hematological cancers (leukemias and lymphomas) (Loo et al., 2017(Loo et al., , 2018) ) and multiple solid tumors (breast, liver, gastric, bone) (Hu et al., 2017;Qadir et al., 2014) including TNBC.Altered epigenetics regulation including is implicated as a root cause of TNBC tumorigenesis (Lu et al., 2020;Wong, 2021).Although DNMT1 is key drug targets in TNBCs, but potential phytocompound inhibitors that target DNMT1 in TNBS are of great interest and are yet to be discover.
Conventional drug designing is a long, costly, and highrisk process, as there is a high degree of uncertainty that whether a drug will succeed or not (Rehman et al., 2022;Tahir Ul Qamar et al., 2019).The toxicities and adverse effects of traditional drug designing can be avoided by using computational approaches which lend a helping hand in uncovering the potential drug candidate and ultimately reduce the risks of side effects (Noor, 2021).Computational drug designing is an emerging theme in the pharmaceutical industry.The advent of incredibly efficient and widely accepted approaches for biological data analysis has opened up innovative and exciting avenues for discovering more fascinating and effective diagnostic and therapeutic methods (Singh & Pathak, 2020;Tiwari & Singh, 2022).Past few years witnessed the wave of sequencing breakthroughs which enables the scientist to make groundbreaking contributions in the area of rational drug designing.Rational drug designing has thus become an effective approach in the long-term process of drug discovery and development.Almari et al. (2022) integrate in silico screening with molecular docking and simulation and uncovered phytocompounds that might assists in the rational designing of novel inhibitors against rift valley L protein.Qamar et al. (Tahir Ul Qamar et al., 2019) employed docking analysis with molecular dynamic (MD) simulation to screen medicinal plant phytochemicals to uncover novel and effective pan-serotype inhibitors against dengue virus.Their study introduced novel scaffolds namely Canthin-6-one 9-O-beta-glucopyranoside, Kushenol W and Kushenol K against dengue virus serotypes to serve as lead molecules for further optimization and drug development against all dengue virus serotypes with equal effect against multiple disease-causing dengue virus proteins.In the similar vein, several other studies reported handful numbers of phytochemical entities against different diseases including diabetes (Arif et al., 2021;Maurya et al., 2020), breast cancer (Ahmed et al., 2014;Rasul et al., 2022), COVID-19 (Alamri et al., 2021b;Omrani et al., 2021), MERS-CoV (Alamri et al., 2021a), and many others (Alamri et al., 2020;Muhseen et al., 2020).
With all these revolutions, computational methods are on the way to efficiently control the pathogenesis of different diseases and disorders.This study undertakes virtual screening using molecular docking along with MD simulation to identify putative phytochemicals capable to control the pathogenicity of diseases by using a multi-target approach.Regarding this, the target proteins were screened with the library of phytochemicals through molecular docking for the identification of novel compounds with drug-like potential.Initially, hits were selected based on their physicochemical, absorption, distribution, metabolism, excretion, and toxicity (ADMET), and other drug-like properties.Afterwards, the obtained results were complemented by all-atom MD simulation for 500 ns, followed by MMGBA/PBSA analysis to investigate the conformational changes, stability, and interaction mechanism of DNMT1 in-complex with the proposed compounds.Our results are promising and may be considered for both in vitro and in vivo clinical trials for inhibition of TNBC.

Structure analysis and evaluation of target protein
The biological activity of proteins is determined by their overall three-dimensional (3D) structure.Changes in the structure of proteins may alter the function of the protein and unfortunately cause the deadliest disease.The structural analysis of the target protein is the preliminary step in rational drug designing.In the current study, the 3D structure of the target protein was collected from Protein Data Bank (PDB) (https://www.rcsb.org/)(Kouranov et al., 2006).The X-Ray predicted structure of DNMT-1 (1256 amino acid residues) was retrieved from the PDB using PDB id 4WXX.The 3D structure of target proteins was then refined using UCSF chimera (Pettersen et al., 2004), a molecular graphics software for analysis and graphical visualization of molecular structures.

Compound library preparation
Phytochemicals are the splendid gifts of nature and have served as a valuable source due to their pleiotropic effect on target molecules.Pharmacotherapy consisting of natural products is considered as a potential and effective approach for TNBC.In the framework of current study, the active compounds were retrieved from Medicinal Plants Database for Drug Designing (MPD3) database (https://mpd3.com/)(Mumtaz et al., 2017).MPD3 is a curated collection of phytochemicals obtained from medicinal plants, especially for virtual screening.Currently, MPD3 database contains more than 5000 phytochemicals from more than 1000 herbal plants with around 900 literature references.The chemical structures of active compounds were downloaded from MPD3 in SDF format.Later, all ligands were added in a single easy-to-use library for molecular docking analysis.

Structure-based virtual screening
Virtual screening lies at the heart of rational drug designing.Virtual screening is an outstanding cornerstone in any drug discovery pipeline as it figures out the potential therapeutic compounds which can be used as novel inhibitors to halt the pathophysiology of disease (Shoichet, 2004).Structurebased virtual screening accurately predicts the interactions among both compounds and target proteins as well as the amino acid residues associated with the ligand binding (Lyne, 2002).In the current study Autodock vina 1.1.2in PyRx 0.8 (Dallakyan & Olson, 2015) was employed to perform the structure-based virtual screening of candidate compounds against the X-ray structure of DNMT1.The ligand molecules in SDF format were submitted to OpenBabel, available in PyRX, and were exposed to energy minimization.For stable conformation, the conjugate gradient descent was employed as an optimization algorithm while the Universal Force Field (UFF) was considered as the energy minimization parameter.
Further, 2000 steps were set for energy minimization and the minimization was set to stop at an energy difference of <0.01 kcal/mol.The energy-minimized ligands were converted to .pdbqtformat for virtual screening.Later, the blind docking approach was used in Autodock vina 1.1.2available in PyRx 0.8 (Dallakyan & Olson, 2015) to calculate binding energies of ligand molecules with DNMT1.Autodock vina used an empirical scoring function to calculate the affinity of protein-ligand binding by summing up the contributions of a number of individual terms.Dockings were run in triplicates to ensure absolute consistency of the results.With a box size of (x ¼ 20 Å, y ¼ 20 Å, z ¼ 20 Å) and dimensions of (x ¼ 122.005Å, y ¼ 105.326Å, and z ¼ 137.824Å), the grid box was configured to completely enclose the binding site of DNMT1.Further, selective side-chain residue flexibility has ability to improve AutoDock vina docking score accuracy, without a significant increase in processing time.Therefore, selective side-chain residue flexibility option was used in AutoDock Vina which provide a more realistic ligand-protein interaction environment, without an unmanageable increase in computer processing time.Finally, visualization of docked complexes was performed using Discovery Studio (Studio, 2008), PyMOL (Yuan et al., 2017), and ChimeraX (Goddard et al., 2018) programs.

Prediction of drug-like properties
The top five compounds were further evaluated for their physiochemical properties and toxicity analysis by SwissADME online bioinformatics tool (http://www.swissadme.ch/index.php)(Daina et al., 2017).For drug-likeness, Lipinski-rule of five was considered as the parameter.This rule estimates drug-like properties by looking into four different parameters, i.e.Hydrogen Bond Donor (HBD), Molecular Weight (MW), Hydrogen Bond Acceptor (HBA), and lipophilicity (logP), with acceptable values of MW <500Da, HBD <5, HBA <10, and logP <5.Furthermore, a comparison among drug-likeness and non-drug-likeness was predicted using molsoft online tool (https://molsoft.com/).The compounds were than screened for Pan-Assay Interference Structural (PAINS) alert to determine their toxicity.This assay is also called toxicophores because of the presence of some group elements that affect the biological process by interference with DNA or proteins which lead to a fatal condition such as carcinogenicity and hepatoxicity (Baell & Holloway, 2010).All compounds with 0 PAINS structural alert were selected for further analysis.Lastly, an appraisal of synthetic accessibility of compounds was performed, which gives an idea to the easiness of synthetic possibility.The synthetic accessibility is of high importance in manufacturing processes of drug molecules.Regarding this, the synthetic accessibility was measured in terms of a score on a scale from 1 (very easy to synthesize) to 10 (complex and challenging to synthesize) by using the SYLVIA-XT 1.4 (Boda et al., 2007).

Molecular dynamics simulations
Molecular Dynamics (MD) Simulations is considered as an important methodology to examine the stability and compactness of the docked complex (Case et al., 2008).In this research work the AMBER20 software was opted to conduct MD simulations experiment on all top 5 docked complexes.Initially, the water molecules were used to perform solvation step for docked complexes, followed by the establishment of neutral system by the addition of counter ions.A water box of 08 Å thickness was developed by considering TIP3P as a solvent model (Amin et al., 2021;Man et al., 2021).This water box was used to encircle the protein-ligand complex structures (Salomon-Ferrer et al., 2013) to further simulate them via periodic boundary restrictions.For the first 500 cycles, water molecules were deliberately minimized, followed by the minimization of the whole system for 1000 rounds.Once system minimization was achieved, a sudden rise in temperature (up to 300K) was applied to it.With the rise of temperature, the system equilibrium was rapidly achieved by using NPT ensemble for 100ps, however, during this equilibrium condition, the solutes were limited to 50ps only with an immediate relaxation of side chains of proteins after 50ps.MD simulations were then run for 500 ns through NPT ensemble at 300K temperature and 1 atm pressure.Systems temperature was maintained by Langevin dynamics, whereas covalent and hydrogen bonds were restrained through SHAKE algorithm.Root Mean Square Deviation (RMSD) plot was developed through the CPPTRAJ (Roe & Cheatham, 2013) of AMBER20 by using initial structure as the baseline to confirm the convergence of the system for MD simulations (Kouznetsova et al., 2020).RMSD value was then used to predict the flexibilities of ligands in complex form, while Root Mean Square Fluctuation (RMSF) was calculated to measure the mean root mean square distance among a specific atom and its average geometric position in a particular dynamics range (Alamri et al., 2021a).Finally, the RoG was calculated to estimate the compactness and packaging at 3D level of ligands in protein's pockets.

MMGBA/PBSA analysis
MMGBA/PBSA (molecular mechanics energies combined with the Poisson-Boltzmann or generalized Born and surface area continuum solvation) analysis is conceivably a highly effective and successful way for binding free energy calculation of protein-ligand complexes as is more specific than the majority of scoring functions used for protein-ligand docking analysis.MMGBA/PBSA has now been frequently used in various domain including protein-protein interaction, protein-ligand interaction, protein folding etc.In the current study, Assisted Model Building and Energy Refinement 20 (AMBER20) was employed to perform MMGBSA/MMPBSA procedure to analyze binding free energies ðDG tol Þ of docked complexes (Homeyer & Gohlke, 2012).In short, 10,000 snapshots were captured with 2ps interval in the last 20 ns to determine the DG tol by using the following equations: where DG sol is the solvation-free energy and DEMM is molecular mechanics binding energy.The values of these energies can be calculated by different other forms of energies or components.DEMM is also equals to the sum of van der Waals energy ðDE vdw Þ, internal energy ðDE int Þ, and electrostatic energy ðDE ele Þ while DG sol is equal to the sum of polar ðDG p Þ and non-polar components.

Structural evaluation of target proteins
The X-ray crystal structure of DNMT1 (PDB IB: 4WXX) was first downloaded from PDB.The PDB structures contain ligands, excluding water molecules, in macromolecular complexes (Shao et al., 2022).The 4WXX PDB ID (chain A) was used as a model for DNMT1.Final DNMT1 structure was 600 residues in length (ranging from 1000 to 1600 residues).Nearly, all protein structures available in PDB lack hydrogens due to the low resolution in crystal.It is important to add the hydrogens to the 3D structures before performing docking or simulations.Therefore, the downloaded structure of DNMT1 was refined using chimera to prepare the protein for further analysis.The refinement of 3D structure is an important step as it brings the 3D models closer towards the experimental accuracy for further computational studies (Adiyaman & McGuffin, 2019).Later, the structure of target protein was evaluated using Ramachandran plot which revealed that 12.0% residues located in additional allowed regions, 87.0% residues located in most favored regions, 0.4% residues located in disallowed regions, and 0.6% residues located in generously allowed regions (Figure 1).It is expected that no more than 2% of residues should belong to the allowed region and no residue should reside in the disallowed or outlier region (Agnihotry et al., 2022).Therefore, the structure of DNMT1 it is a near to good quality model.

Molecular docking analysis
One of the key methodologies in the field of computational drug discovery is molecular docking.This approach assists in estimating the biomolecular interactions of drugs with a specific binding site of a peculiar target protein (Pantsar & Poso, 2018;Sivakumar et al., 2020).It predicts the binding potential of small molecules by placing them on the target site in a non-covalent fashion and predicts whether it forms stable complex with high specificity or not.As a result of obtained information from docking experiment, it becomes relatively easier to predict binding affinity, stability, and free energy of a ligand with the active site of the target protein (Guedes et al., 2014;Williams-Noonan et al., 2018).Currently, the approximation of binding free energy is an important objective of docking protocols, which is revealed in terms of hydrogen bonds, total internal energy, energy of dispersion and repulsion, torsional free energy, electrostatic force, energy of desolvation, and unbound system's energy (Luo et al., 2022).
In the current study, the phytochemical database was docked against the DNMT1 and docked compounds were ranked based on the binding energies (kcal/mol).Only those compounds were selected which showed stronger binding energy than native ligand.The top five compounds having highest binding energy were selected for further analysis (Figure 2).The Beta-Mangostin from the Cratoxylum arborescens is on the top of the list owing to its highest affinity and binding score (Table 1).Beta-Mangostin formed two hydrogen bonds with Gly 1223, Cys 1148 having binding energy of À 11.5905 kcal/mol while Gancaonin Z formed two hydrogen bond with Asn 1578, Gly 1223 having binding energy of À 10.7033 kcal/mol.Gly 1223 were found to be involved in forming hydrogen bond with the Beta-Mangostin through hydrogen atoms of hydroxyl group.On the other hand, 5hydroxysophoranone formed two hydrogen bonds with Glu 1226, Pro 1225 with a binding energy of À 10.5337 kcal/mol.In this case, the hydrogen bond formed between hydrogen of hydroxyl group and Glu1266 residue of DNMT1 protein.
Sophoraflavanone L formed one hydrogen bond with Glu 1266 having binding energy of À 10.1437 kcal/mol.In terms of Sophoraflavanone L-DNMT1 complex, Glu 1266 formed hydrogen bond with oxygen atoms of hydroxyl group, while the Arg 1310 formed hydrogen bond with hydrogen atoms of hydroxyl group.While, Dorsmanin H formed three hydrogen bond with Glu 1266, Gly 16577, ARG 1310 residues with binding energy of À 10.0501 kcal/mol.In Dorsmanin H-DNMT1 complex, Arg 1310 formed hydrogen bond with hydrogen atoms of hydroxyl group, while Glu 1266 and Gly 16577 residues formed bond with the oxygen atoms of hydroxyl group.
Recently, a list of different studies identified compounds for TNBC treatment using computational approaches.Assumpc¸ão et al. (2020) unveiled the effects of propolis and phenolic acids on TNBC cell lines.Their findings uncovered that S-adenosylhomocysteine, caffeic acid, hydrocinnamic acid, p-coumaric acid, (À )-epigallocatchin À 3-gallate has binding affinity of À 8.3 kcal/mol, À 6.7 kcal/mol, À 5.2 kcal/mol, À 6.0 kcal/mol, and À 10.4 kcal/mol with the DNMT1 protein.If we compare our findings with their study, it is worth noting that compounds explored in current study has comparable binding affinities with DNMT1 therefore, selected compounds can act as promising candidates when used for the treatment of TNBC.Additionally, the ligand binding site atoms of each complex were also presented in Supplementary File 1 (Figures S1-S5).

Drug-like properties of top hits
In modern drug discovery pipelines, the potential of a new compound is often investigated initially without making it or testing it (Alves et al., 2021;Shultz, 2019).A set of drug-likeness rule are applied on the compound to check their druglike properties.Drug-likeness rules are set of guidelines for the structural properties of compounds, used for fast calculation of drug-like properties of a molecule (Ahire et al., 2021).The drug like properties of selected compounds were then analyzed using SwissADME software (Table 2).SwissADME provides an open-access and rapid prediction of pharmacokinetics, physiochemical, drug-like, and other similar parameters for molecules of interest.Lipinski's rule of five and Veber's criterion was used to determine if the compounds were appropriate for oral bioavailability.These results showed that none of the proposed compounds violates Lipinski's rule of five and Veber rule.This suggests that all these inhibitors possessed good attributes of drug-like or pharmacological properties which could make them orally bioavailable.5-hydroxysophoranone was the only compound which did not follow Egan rule.Further, the PAINS analysis indicates the possibility of a molecule to be toxic, although, all compounds have 0 PAINS structural alerts which signify their non-toxic nature.Lastly, Beta-Mangostin, Gancaonin, Z 5-hydroxysophoranone, Sophoraflavanone L, and Dorsmanin H depicts good synthetic accessibility scores 4. 07, 3.84, 4.71, 4.27, and 4.56 respectively.The synthetic accessibility scores range from 1-10, where score 1 represents easy synthetic formation while score 10 represents its difficult synthetic route.These findings revealed that these selected compounds have drug-like properties, therefore can be considered as potential natural inhibitors for TNBC.

Molecular dynamics simulations
MD Simulations have been conducted in many previously published studies to investigate the validity of docked complexes (Hasan et al., 2022;Mahmud et al., 2021).MD Simulation is a strong biophysical technique as it uncover accurate binding conformations and other vital dynamic values of ligand-protein interactions, therefore, it is an important technique in computer-aided drug designing (Liu et al., 2019;Michael et al., 2021).MD simulations have been very successful in recent years for optimizing the docked hits (Anwar et al., 2020;Guterres & Im, 2020;Vanmeert et al., 2019).This data set can guide novel drug design, making MD simulation a valuable tool in modern drug discovery.In the current study, all five top-hit compounds were considered for MD simulations experiment for 500 ns, and various trajectories were obtained to analyze the results, as recommended.Firstly, RMSD values of complexes were determined to estimate the structural distance between different coordinates, i.e. it represents the mean distance between atoms which are being existed at a peculiar site on the protein.In other words, RMSD values codes for the assessment of structural integrity of a protein-ligand docked complex.Overall, in the current experiment, RMSD of all studied complexes was reported to be stable in nature (Figure 3).The RMSD value was found to be increased from 0 to 500 ns with little fluctuations in all of the studied complexes.The average RMSD value of Sophoraflavanone L, Dorsmanin H, Gancaonin Z, 5-hydroxysophoranone, and Beta-Mangostin was found to be 7 Å, 4.5 Å, 9 Å, 6.5 Å, 5 Å, respectively and the maximum value of 8 Å, 5 Å, 12.5 Å, 9 Å, and 7.5 Å, respectively.According to these results, the docked complex of Dorsmanin H remained most stable throughout the time period of 500 ns, followed by Beta-Mangostin, Sophoraflavanone L, 5-hydroxysophoranone, and Dorsmanin H.As the intermolecular docked conformation of complexes was observed in stable dynamics during the simulation, the chemical interactions network including hydrogen bonding and van der Waals contacts pattern was seen the same however with minor distance variations.Briefly, the RMSD value of all studied complexes was within the acceptable range.
Furthermore, the RMSF values of protein-ligand complexes were generated for further assessment of flexibility at residual level of protein.RMSF predicts the mean fluctuation of a particular amino acid residue from its time-averaged position over time.The mean RMSF values of DNMT1 -Sophoraflavanone L, DNMT1 -Dorsmanin H, DNMT1 -Gancaonin Z, DNMT1 -5-hydroxysophoranone, and DNMT1 -Beta-Mangostin was found to be 3.0 Å, 2.5 Å, 4 Å, 3.2 Å, and 3.5 Å, respectively (Figure 3B), which clearly demonstrate the high level of stability in between the molecules of proteins in complex with ligands.The maximum level of fluctuation was shown by Gancaonin, particularly, in the first 100 ns.Once the RMSD and RMSF values were calculated, Rg analysis was carried out to evaluate the compactness of protein and structural equilibrium of protein-ligand complex throughout the time frame of whole simulation period (Figure 3C).The average Rg values were as follow: DNMT1 -Sophoraflavanone L 67 Å, DNMT1 -Dorsmanin H 55 Å, DNMT1 -Gancaonin Z 64 Å, DNMT1 -5-hydroxysophoranone 62 Å, and DNMT1 -Beta-Mangostin 58 Å.Overall, all of the protein-ligand complexes demonstrated significant compactness over the entire time of 500 ns of simulations.

Binding free energy calculations
Binding free energy calculation offer an attractive approach to analyze the ligand-receptor intermolecular interactions by determining the ligand-receptor energy values and intermolecular interactions.Binding free energies of the studied complexes was determined to measure MMGBSA and MMPBSA which assists in improved understanding of binding capabilities of ligands in the active sites of proteins.Table 3 shows the results of studied parameters of both MMPBSA and MMGBSA for all studied protein-ligand complexes in this study.The results of binding free energy calculations depict the favorable binding of ligands in the active sites of DMNT-1 protein.The phytocompound 5-hydroxysophoranone depicted the more favorable binding energies in terms of MMGBSA (DG total ¼ À 41.6376 kcal/mol) as compared to the other ligands, while for MMPBSA, compound 5-hydroxysophoranone demonstrated the most satisfactory binding energy with DNMT1 with DG total value of À 30.5382 kcal/mol.The major energy source for both MMPBSA and MMGBSA in all cases was contributed by DG gas except for that of  Dorsmanin H.These results demonstrated that the selected compounds showed promising results and therefore, these findings strongly support the ability of the compounds studied to act as drugs against TNBC.

Conclusions
Triple-negative breast cancer is considered as an invasive breast cancer that is progesterone receptor-negative, estrogen receptor-negative, and HER2-negative.It is an aggressive type of breast cancer as it grows and spreads quickly as compared to other types of breast cancer.There is no licensed or commercially available drugs are available for TNBC.Drug discovery based on natural products has a long successful history and can be essential framework for identifying novel inhibitors against TNBC.In the present research work, we have identified five phytochemicals namely; Sophoraflavanone L, Dorsmanin H, Gancaonin Z, 5-hydroxysophoranone, Beta-Mangostin, as the potential inhibitors of DNMT1, a potent target of TNBC.All of these lead hits demonstrate strong binding affinity with the active sites of DNMT1 as shown by molecular docking and MD simulations experiments.Furthermore, all of these compounds also showed acceptable drug-like properties, making them excellent lead compounds for targeting TNBC.Overall, the findings indicate that natural compounds fit into the same binding pockets of DNMT1, indicating that these compounds can be promising treatment options for TNBC.However, future research should look into these targets to see if they affect the lifetime and pathogenicity of TNBC.The findings of this study could be further exploited to develop novel and more potent DNMT1 inhibitors.

Figure 1 .
Figure 1.(A) 3D structure of DNMT1.Coils were represented with green color, strands represented with cyan color and helices represented with red color.(B) Ramachandran plot analysis.Ramachandran plot comprises of four quadrants.The upper left and bottom right quadrant represent the allowed region, while upper right and bottom left represent the disallowed regions.Blue dots in plot indicate the density of amino acid residues.

Figure 3 .
Figure 3. Molecular dynamic simulation.(A) RMSD value of top five docked complexes during molecular dynamics simulations of 500 ns.(B) RMSF value of top five docked complexes during MD simulations of 500 ns.(C) RoG values of top five docked complexes during MD simulations of 500 ns.

Table 1 .
Details of top five phytochemicals having maximum binding potential with DNMT1, based on the docking scores.

Table 2 .
Physiochemical properties, drug-likeness, and medicinal chemistry related properties of top 5 best docked compounds.