Targeting dysregulation signatures of p53-MDM2 interactions to identify newer small molecule inhibitors for cancer therapy: Insight from computational direction

Dysregulation of the p53-MDM2 interactions has been implicated in majority of human tumors presenting a target for nding small molecule inhibitors. In this study, a training set of 17 experimentally tested inhibitors of MDM2 was used to develop series of pharmacophore models among which a four-featured (AHRR_1) model with one hydrogen bond acceptor, one hydrophobic group and two aromatic ring features and characterized by a survival score of 4.176 was considered signicant among the top ranked generated hypothesis. Further, the model was validated by an external set of actives and decoy molecules and was found to exhibit encouraging statistical attributes (such as AUC > 0.7, BEDROC > 0.5 and EF > 1.0 etc). The model was used to screen the ZINC compound database, from the database, the top best 1375 hits satisfying the pharmacophore model was were docked to MDM2 protein to identify the likely interactions of the compounds as well as their binding anity with MDM2. Further, druglikeness and pharmacokinetic properties screening on top-ranked compounds with higher binding anity than reference inhibitors revealed four compounds (ZINC02639178, ZINC38933175, ZINC77969611, and ZINC06752762) with suitable pharmacological properties including low ligand toxicity. Investigation of the dynamic behaviour of each candidate inhibitors in complex with MDM2 via molecular dynamic simulation suggested ZINC02639178 and ZINC06752762 as the most potential inhibitors. Thus, these compounds may emerged as therapeutic option for cancer treatment after extensive in vitro and in vivo studies. RIE: robustness initial enhancement; BEDROC: Boltzmann-enhanced discrimination of receiver operating characteristics.


Introduction
Regulating cellular activities for normal tissue homeostasis is a complex process dependent on numerous signalling cascade involving proteins. However, a number of this proteins have been compromised at expressional level resulting in several cellular anomalies. Ostensibly, in most types of human cancer the expression of the p53 tumor suppressor protein is diminished affecting cells that rely on its pathway for modulating tumor growth, proliferation, transition and progression. The tumor suppressor protein p53 is a requisite modulator of most delicate cellular activities including cell proliferation, death and reprogramming [1,2], and is negatively regulated by the oncogene MDM2 in a self-orchestrated feedback loop activating the proteasome pathway, as revealed by models. Mdm2 possesses E3 ubiquitin ligase activity that allows it to prime p53 through its hydrophobic cleft to polyubiquinate p53 and tagging it for degradation by cytoplasmic proteases [3]. Ostensibly, studies have shown that the rate of cellular proliferation under conditions such as stress, genetic aberration and DNA injury is often dysregulated in cells with the genetic lesions resulting in supernumerary masses [4]. As such, p53 is upregulated to induce apoptosis-mediated cell cycle arrest to ultimately abrogate tumor initiation. In contrast, it repression is a signature for MDM2 hyper-activation and so far, has been the most widely contemplated hallmark of human tumors.
Disrupting the p53-MDM2 interactions using small molecule inhibitors has gained recognition in oncology. To date, a number of small molecule ligands generally the analogues of cis-imidazole (e.g. nutlins) [5], spiro-oxindales (such as MI-63 and MI-219) [6], chalcones [7], sulfonamides [8], quilinols [9], isoquinolines [10], terphenyls [11], pyrrolidones [12], indoles [13], piperidinones [14], morpholinnones [15], imidazothiazoles [16] and benzodiazepinedione [17] have been reported with inhibitory activities against p53-MDM2. These compounds could subvert hyper-activation of MDM2 oncogenes in malignant tumors serving as a rational to safe treatments for cancer. Unfortunately, none of this compounds have been approved worthy as anti-cancer drugs, although a number of them are currently in preclinical and clinical studies. The inability of the majority of these compounds to establish desirable pharmacokinetic pro les in spite exhibiting interesting binding a nity with MDM2 as target in bioassays is a major problem. Furthermore, p53 is highly susceptible and could be easily compromised thus, its reactivation might fail. Considering the high rate of mutation of the TP53 gene and its in uence as a mechanistic for tumor survival, there is need for more treatment option. So far, computational method have been helpful to identify potential and promising inhibitors of MDM2 from diverse chemical scaffolds [18], some of which have been substantiated to display su ciently high binding a nity, desirable pharmacokinetic and medicinal chemistry properties based on pharmacological predictions. Computational based-approaches involving virtual high throughput screening has proved to be an intrinsic component of drug discovery, and has enabled accelerated screening of chemical repositories to nd newer hit compounds. Besides being a quicker approach for developing drugs, this technique takes the bene t of cost compared to traditional drug development. Moreover, identi ed hits compounds are already approved therefore, the need to re-evaluate their safety pro les such as toxicity and carcinogenicity may not necessarily apply and thus, labour is explicitly minimized. Realistically, the appropriate use of this method in drug discovery/development process, would continually improve the ability to not only identify hits with likeliness to serve as potential scaffolds for producing new therapeutics but also optimize these compounds for better selectivity and speci city.
In this study, we applied pharmacophore based virtual screening, molecular docking techniques, and molecular dynamic simulations to establish newer small molecular ligands against the generally recognised regulator of the p53 pathways (i.e. MDM2).

pharmacophore modelling
The modelling of pharmacophores is an integral aspect in drug discovery and drug development, since it can aid in identifying selective hits against various biological targets implicated in diseases. In this study, a chemically diverse set of 17 experimentally tested MDM2 inhibitors with pIC50 values spanning between 5.91nM and 10.0nM were taken for 3D pharmacophore model development using the PHArmacophore Search Engine (PHASE) of Schrodinger. The list of compounds, SMILES, and activity in pIC50 are summarized in Table 1. Dataset preparation: 3D structures of the compounds were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) repository of small molecule ligands and was prepared using the Ligprep module of Schrodinger software. During dataset preparation, the OPLS-2005 force eld implemented in the mentioned software was used with default settings for geometry optimization and energy minimization.
Generation of pharmacophore hypothesis: The selected compounds for pharmacophore generation were bifurcated into categories comprising of active and inactive molecule sets with activity threshold value of pIC50 ≥ 7.50 nM taken for active and pIC50 7.50 nM for inactive compounds category which resulted in 8 actives whereas the rest were considered as inactive dataset. Further, the dataset was used for generating a series of pharmacophore models with different descriptor combinations among which the model AHRR_1 was selected among the top best ranked pharmacophore hypothesis based on the initial statistical attribute characterized by the corresponding survival score.

Pharmacophore validation and virtual screening
Especially noteworthy in pharmacophore modelling is the model's robustness and capability to distinguish between active from inactive or less active compounds. Since the selected pharmacophore model would be employed for virtual screening, it is essential to validate this model. Here, the validation metrics considered for the selected model among the presently employed statistical criteria for evaluating the performances of a prevailing pharmacophore hypothesis are: the area under the receiver operating characteristics curve (AUC-ROC), the area under the accumulation curve (AUAC), the enrichment factor (EF), the robustness initial enhancement (RIE) and the Boltzmann-enhanced discrimination of receiver operating characteristics (BEDROC). Subsequently, the model was used to screen the ZINC database of commercially available compounds through the ZincPharmer server (https://zincpharmer.cbs.pitt.edu/) [19]. Noteworthy, to obtain the best similarity hits from the ZINC database, we applied the following constraint which include a maximum root mean square deviation of 0.2 (RMSD = 0.2), number of rotatable bond cut-off between 1 and 10 (i.e. 1≤nRB≤10), and a molecular weight range of 200-500 (200≤MW≤500) Daltons. The potential hit compounds obtained were regarded as candidate inhibitors against the MDM2 protein.

Protein structure and ligand dataset preparation
The crystal structure of MDM2 (PDB: 3JZK) [20] was obtained from the protein data bank accessible from (https://www.rcsb.org/) [21]. Prior to molecular docking, hetatoms co-crystallized with the protein were removed. The protein was further subjected to a preparation step using the Dock prep module implemented in the UCSF chimera software program (University of California, USA) [22]. In the Dock prep preparation protocol, all co-crystallized solvent molecules were deleted and the 3D protonation module implemented in chimera was used to add hydrogen atoms. Further, incomplete side chains were replaced using Dunbrack 2010 rotamer library [23] and AM1-BCC was used to de ne charges. Next to the preparation protocol, energy minimization was performed on the protein structure using AMBERff14SB force eld. The minimization was done with default parameters which include: 100 steepest descent steps at a step size of 0.02 Å, conjugate gradient steps at 10 and an update interval of 10.
In the ligand preparation step, the top 1375 hits obtained from the ZINC database were imported into PyRx through the open babel module implemented in the software. An energy minimized step was carried out on all the ligands using the universal force eld (UFF) geometry optimization with default settings.

Molecular docking simulation
Having prepared the working structures which included the ligand library and the target (receptor), the binding hotspot of the target was de ned using the native ligand. The hotspot residues are located in the hydrophobic cavity on the protein structures composed of 14 key amino acids that could actively facilitate the interaction of the protein with small molecule compounds. The docking simulation was performed using the vina module [24], integrated in PyRx v 0.8 software [25] and ran at an exhaustiveness of 8. Native ligand, as well as two known experimentally tested inhibitors of the protein were used as standard inhibitors throughout the docking simulation. To de ne the search space, a grid box resolution was centred at 7.6748, -9.7283, 25.8077 along the x, y and z axes respectively with a grid dimension of 19.7982 Å × 20.0006 Å × 34.6640 Å, thereby allowing the entire coverage of the hotspot residues. These residues included Leu54, Leu57, Ile61, Met62, Tyr67, Gln72, Val75, Phe86, Phe86, Phe91, Val93, His96, Ile99, Tyr100, and Ile101 [26]. At the end of the docking study, the crystallographic pose of each ligand were examined for on the basis of their binding a nity and those with the highest binding a nities were selected for further analysis.
The 2D interactions of the protein-ligand complexes were evaluated using the LigPlot + 2.2 software [27], and the visualization of the binding mode performed in PyMOL (https://www.pymol.org/) [28]. The docking protocol was validated by re-docking of cocrystalized MDM2 (3jzk) ligand and the resulted pose superimposed to the binding mode of un-processed ligand x-ray structure. The docking validation resulted in a root mean square deviation (RMSD) of 0.485 Å which is way below the maximum recommend value (i.e. 2 Å) thus, indicating the reliability of the docking criteria.

Drug-likeness prediction and pharmacokinetic properties evaluation
In the early stage of drug discovery, risk assessment is important to determine the safety pro le and behavioural mechanism of small molecule ligands on biological systems. To date numerous servers and software are available to evaluate the safety pro le of drugs using pre-de ned experimental models. Here, those compounds that exhibited higher binding a nity compared to the standard inhibitors were screened for drug-likeness and ADMET properties respectively. Drug-likeness and ADMET properties have been the most used methods to evaluate the pharmacokinetic properties of drugs, and have been helpful to speed up the approval of drugs.

Molecular dynamics simulations
Molecular dynamics simulations were carried out on energy minimized receptor-ligand complexes using the Desmond package [31]. The OPLS3e force eld was used in the system builder to model the protein interactions. The complexes were immersed in solvation molecules that have been prede ned using the Simple Point Charge (SPC) water model [32], in a rectangular periodic box allowing for an equidistance of 10 Å buffer region between protein atoms and box sides. The systems were neutralized with appropriate number of sodium (Na + ) and chloride (Cl -) counter ions for stability. Particle-mesh Ewald (PME) method [33], with a grid spacing of 0.8 Å was applied for calculating long-range electrostatic interactions. The simulation temperature and pressure were controlled using Nose-Hoover thermostats and Martyna-Tobias-Klein barostats respectively [34,35]. A multiple time step RESPA integration algorithm was used throughout the simulations with an inner time step of 2.0 fs for bonded and non-bonded interactions within the short-range cut-off, and an outer time step of 6.0 fs for non-bonded interactions beyond the cut-off. Periodic boundary conditions (PBC) were applied on all the systems. Thereafter, a 50 ns production run in the NPT ensemble was performed for all the complexes.

Results And Discussion
It has been demonstrated that the activity of the MDM2 oncogenes determines the expressional status of p53. MDM2 interacts with p53 through its hydrophobic cleft to modulate p53 hyper-activation. Nonetheless, upregulation of MDM2 could be lethal and is a hallmark of human tumor. Therefore, targeting MDM2 hyper-activity has been an option in cancer therapy.

Pharmacophore model generation and validation
Series of pharmacophore model with different combination of physiological features have been developed with 8 actives (pIC50 ≥ 7.50) and10 inactive compounds (pIC50 < 7.50) from selective and experimentally tested MDM2 inhibitors using the PHArmacophore Search Engine (PHASE) module of Schrodinger. The model AHRR_1 which was characterized by one hydrogen bond acceptor, one hydrophobic group and two aromatic centroids was selected for use in further investigation on the basis of adjusted survival score. The statistical details of this pharmacophore hypothesis is shown in in Table 2. Figure 1 depicts the spatial arrangement of pharmacophore features, along with the alignment of most active compound to the pharmacophore model. Although adjusted survival score could be a reliable criterion to select signi cant pharmacophore hypothesis nonetheless, useful models can also have low adjusted survival scores which is consistent with the observation by Bahera et al [36]. To circumvent this limitation, we employed the techniques of discriminating active compounds from false binders (decoys). As such, we could truly extrapolate the robustness and sensitivity of the model to selective MDM2 inhibitors rather than naïve molecules lacking activity. To do achieve this, a dataset containing 40 MDM2 inhibitors and 1200 false positive molecules obtained from DEKOIS database [37], was employed in ROC curve generation (see Figure 2) and to analyse the calculated area under curve (AUC). Further, the Boltzmann-Enhanced discrimination of ROC curve, the enrichment factor (EF), the area under accumulation curve (AUAC) and the robust initial enhancement (RIE) was used to appraise the predictive power of the model to active compounds among the loch molecules in the dataset. Table 3 contains the statistical validation criteria and the minimum recommended value of the parameters in view [38,39]. According to Table 3, the early enrichment factor (EF 10 % ) at 8.5 and AUC-ROC value of 0.85 indicated that the selected pharmacophore hypothesis was rational for virtual screening as it was able to retrieve 34 active molecules from 40 active compounds in the dataset which approximate to about 85 % actives at 10 % of the dataset. Besides, the model exhibited BEDROC, AUAC, and RIE value well above the critical values as evidenced in Table 3, suggesting the usefulness of model to earlier and ordered recovery of active compounds [38]. Based on this statistical merits it can be inferred that the model was better than a randomly generated model and has not been generated by chance, thus adequate for virtual screening of external molecules

Pharmacophore-based virtual screening
Once the AHRR_1 model have been substantiated to be reliable and adequate for use, the model was used to screen the ZINC compound database through the ZINCPharmer server. Among the 206, 433, 075 compounds from the screened database, 563, 879 compounds had showed molecular group that t the pharmacophore hypothesis. The top ranked 1375 hit compounds within the speci ed search tolerance of the model as earlier described in the method section, were selected to molecular docking studies.

Molecular docking simulation based virtual screening
Virtual screening presents a strategic approach to identify druggable compounds that could be use as treatment option against various diseases. The screening method could be either ligand-based or structure-based [40][41][42], nonetheless, both of this methods have been deemed suitable for nding potential lead compounds against disease signatures (target) and have been synergistically employed to improve screening precisions and the success rate of drug discovery and development. Ligand-based virtual screening uses ensemble of the molecular characteristics in multiple ligands to identify potential hit compounds from combinatorial scaffolds while the latter select hits based on likely interactions of the compounds with receptors as well as their binding a nity against these target, by selectively docking the ligands into a con ned region within the target called hotspot. In this study, a constraint-speci c pharmacophore screening was employed to identify compounds that exhibited similar features with the selected pharmacophore hypothesis yielding a total top hit list of 1375 compounds based on RMSD preferences.
Interestingly, molecular docking of all the 1375 hit compounds retrieved from the ZINC database against MDM2 (PDB: 3JZK) resulted in binding a nities ranging from -5.3 to -9.9 Kcal/mol. The validation of the docking methodology which was performed by re-docking of cognate ligand to it receptor (MDM2) resulted in a conformation with binding energy of -8.9 Kcal/mol and RMSD value of 0.485 Å upon alignment to x-ray structure (Figure 3). The docked ligands could be categorized as highly selective, moderately selective, and selective on the basis of their binding a nities compared to three control inhibitors of MDM2 (i.e. Nutlin-3a, NVP-CGM097 and co-crystallized ligand) having binding a nities of -7.5, -8.6 and -8.9 Kcal/mol respectively.
According to our observation, a total of 18 compounds exhibited higher binding a nities compared to that of the native ligand. 49 compounds showed higher binding a nities than NVP-CMG097, while a total of 541 of the ligands had binding a nities surpassing that of Nutlin-3a. The ligands with binding energies ≥ -8.9 Kcal/mol were considered for further investigations among which 25 ligands had met the criterion. (Table 4) presents the results of the 25 selected hit compounds, their binding energies, and interactions with MDM2 (PDB: 3JZK) after the docking calculations. According to the result, it is apparent that these compounds adapted similar binding mode as the reported standard inhibitors within the active site of MDM2 receptor. More also, the docked ligands form extensive hydrophobic interactions with the hydrophobic residues embedded within the catalytic domain which is consistent with the hydrophobic interactions made control inhibitors. Importantly, the aromatic ring of almost all candidate molecules could interact with Leu54, Leu57, Gly58, Ile61, Val93, His96, and Tyr100 amino acid residues which supports the observation by Atatreh et al [43]. Especially noteworthy is ZINC06670015 (-9.9 Kcal/mol) and ZINC02639178 (-9.6 Kcal/mol) which made hydrogen bond with the -COO termini and -NH side chain of Leu54 and Gln24 amino acids of MDM2 (see gure 4), which are absent in the reported control inhibitors and are uncommon interactions exhibited by most native and experimentally tested MDM2 inhibitors. This hydrogen bonding interaction to either of the mentioned residues may have contributed to their increased binding speci city as evidenced in (Table 4), and hence enhanced their selectivity against MDM2 over others except in the case of ZINC24394575 (-9.8 Kcal/mol) which is poorly understood. Nonetheless, ZINC24394575 was found to interact through hydrophobic interactions with both amino acids which is believed to be the reason for the observed high binding a nity against MDM2 (Figure 4).

Drug-likeness prediction and pharmacokinetic properties evaluation
Reportedly, several molecules have showed interesting biological properties that tentatively render them druggable at the preliminary stage of drug establishment process, some even shows desirable binding a nities against their biological targets, increasing the precision for their approval. Majority of these molecules could also abscond rejection until the penultimate phase of clinical trials and later fail. So far, the most signi cant reasons for these failures are due to late identi cation of side effects. Therefore, practicing the predictions of pharmacokinetic and toxicological pro ling of any potential lead molecule at the initial stage of drug discovery process is an effective paradigm against futuristic menaces. Here, we investigated the likeliness of the top 25 hits from the virtual screening process to establish desirable pharmacokinetic pro le. In this regard, we employ the admetSAR and SwissADME web servers to help with these predictions which include the generally recognized ADMET (i.e. absorption, distribution, metabolism, excretion, and toxicity) and drug-likeness properties in order to screen out potential nuisance and meticulously decrease drug redundancy before further investigating's. The results of the predictions are presented in (Supplementary Table 1 and 2). All the 25 potentially selective drug candidates obeyed the Lipinski rule of ve. According to this rule, a compound is considered druggable if it ts in the following constraints; LogP ≤ 5, MW≤ 500, no of rotatable bond (nRB) ≤ 10, Hydrogen bond donor (HBD) ≤ 5 and Hydrogen bond acceptor (HBA) ≤ 10. In contrast, compounds with more than a single violation of the mentioned molecular properties is likely to exhibit poor absorption and oral bioavailability [44]. In spite obeying the aforementioned rule, majority of the compounds showed unfavourable ADMET status. However, the compound ZINC06752762, ZINC02639178, ZINC38933175, and ZINC77969611 exhibited the most desirable pharmacological properties.
The result of the predicted pharmacokinetic properties of these compounds are reported in (Table 6). According to Table 6, it can be inferred that all four compounds indicated satisfactory toxicity status due to their non-carcinogenic and non-mutagenic properties as well as their inability to inhibit the hERG gene encoding the potassium ion channels necessary for the normal electric activity in the heart. Interestingly these compounds exhibited acceptable LD50 (lethal dose at 50) value with rat models ranging from (2.1785-3.0322) mol/kg and were found to be in class III on the basis of their oral toxicity which is quite desirable. Furthermore, the compounds showed good metabolism pro le with the CYP450 iso-enzymes (the key enzymes in drug metabolism) which is characterized by the non-inhibition of the CYP450 2D6, CYP450 3A4 and CYP450 2C19 enzymes. The CYP450 2C9 also show non-inhibitory property for ZINC06752762 and ZINC38933175 whereas it was the enzyme was inhibited by ZINC02639178 and ZINC77969611. Similarly, the CYP450 1A2 enzyme was also inhibited by ZINC38933175, and ZINC77969611. In contrast, the ZINC06752762 and ZINC02639178 had shown non-inhibitory property for these drug metabolizing enzyme. Additionally, all the four selected ZINC compounds also exhibited non-substrate alert for CYP450 2C9 and CYP450 2D6. All except ZINC38933175 considered substrate for CYP450 3A4. Oral bioavailability is the hallmark of any potential drug candidate. Conversely, for a compound to be orally bioavailable, it must possess several unique functions that enables it to pass through cellular membrane. Otherwise, they may be trapped within these barriers and might constitute serious health threat.
These functions are desolvation, diffusion, resolvation and physicochemical properties including lipophilicity [45]. In the context of absorption, distribution and excretion of drugs, admetSAR provides several ways to identify this pharmacokinetic models including Human Intestinal Absorption (HIA), blood-brain barrier permeability, cell permeability based on caco-2, aqueous solubility (Absorption); p-glycoprotein inhibition (distribution); and renal organic cationic transporters inhibition (excretion). It is important to identify the aforementioned pharmacokinetic parameters because they provide additional knowledge for drug acceptability or rejection in any drug development process. Apparent in Table 6, all the four hit molecule are permeable to the human intestinal membrane and Blood-Brain Barrier. All except ZINC38933175 lacks the Caco-2 cell permeability. The four ZINC compounds also showed good aqueous solubility. The p-glycoprotein serves as a relevant drug transporter [46], as such investigating the possibility of the lead compounds to act as a substrate or inhibitor of p-glycoprotein is clinically relevant. The pglycoprotein showed inhibitory alert for all the lead compounds. Regardless of this, the inability of ZINC02639178, ZINC38933175 and ZINC77969611 to act as substrate of the p-glycoprotein provided the likelihood for insigni cant inhibition.
Moreover, the four hits are non-inhibitor of the renal organic cationic transporter 2 which implies the overall clearance of the compounds. From our analysis, we can infer that the selected hits could be a safe to prescribe treatment option against cancer. In order to take account into the behavioural mechanism of the candidate inhibitors on protein stability and exibility, the docked complexes of ZINC02639178, ZINC06752762, ZINC38933175, and ZINC77969611 bound MDM2 protein was studied in a molecular dynamics simulation in an explicit water solution for 50 ns using the Desmond package. The trajectory plots representing the protein-ligand root mean square deviation (RMSD) and the protein root mean square uctuation (RMSF) is illustrated in (Figure 6). Figure 6 clearly indicated that ZINC02639178, ZINC06752762, and ZINC77969611, each complexed with MDM2 receptor, are more stable than the ZINC38933175 bound MDM2 complex. Importantly noteworthy, the complexes have comparable stability pattern and convergence time as the cognate ligand-MDM2 complex. From our observation of the trajectory plot, it was evident that major uctuation of each complexes was till about 18 ns after which they attain plateau around 20 ns of the simulation time and showed no major deviation till the end of the simulation whereas, the ZINC38933175-MDM2 complex uctuated signi cantly between 0.8 and 1.7 Å till 25 ns and then comes to 1.2 Å round 28 ns, beyond which the complex experienced only slight deviations after 30 ns of the simulation time.
To investigate the structural changes occurring in each complexes due to ligand binding, the dynamic mobility of the residues of each docked structures were analysed by calculating the root mean square uctuations (RMSF) (Figure 7). RMSF characterizes the local changes along a protein chain. An inspection of the RMSF trajectory plot showed that ZINC06752762, ZINC77969611, ZINC38933175 bound MDM2 complexes were somewhat similar uctuation pattern to co-crystallized ligand-MDM2 complex. The Gln18 residues in the loop region of MDM2 had the highest uctuation in all complexes with RMSF value between 2.0 and 2.75 nm. In contrast, major uctuation in the ZINC02639178-MDM2 complex were observed between position 50-55 of the calphas atom of MDM2 residues. Especially noteworthy, is the ZINC77969611-MDM2 complex which showed the highest uctuation pattern suggesting less stability during the simulation (Figure 7).

Conclusion
Considering the high rate of mutation resulting in several forms of malignancy, there is need for consistent approach for remedy. The noteworthy tumor suppressor protein p53 modulates majority of the activities undergoing in cells and are regulated by the oncogenic protein MDM2 to maintain the normal tissue functions. Dysregulation in the expression of either p53 or the cellular regulators could be lethal resulting in the overall tissues detriment. Conversely, MDM2-p53 interaction is compromised in relatively all types of human tumors. The overexpression of MDM2 is a hallmark of tumor, and a mechanistic by which cancerous cells abscond cell division regulation, allowing their inde nite proliferation into malevolent cells capable of invading and reprogramming normal tissues to cancerous tissues (in a process called metastasis). Therefore, there is an urgent need to develop potent arsenals to combat both current and futuristic threats [47]. The upregulation of the p53 tumor suppressor autoregulators MDM2 serves as an effective signature to nd potent anticancer drugs.
Consequently, we have identi ed in this present study four small molecule inhibitors (ZINC06752762, ZINC02639178, ZINC38933175, and ZINC77969611) against the p53-MDM2 interactions using a pharmacophore model of clinically tested MDM2 inhibitors. Apparently from our studies, these compounds exhibited interesting binding a nities of -9.2, -9.6, -8.9 and − 8.9 Kcal       interaction predicted using SwissADME server.