Introduction

A healthy cell cannot survive without regulating the expressions of its proto-oncogenes or tumor suppressor genes. In brief, proto-oncogenes modulate the cell cycle by initiating cell division, and growth, and in the mutated form, they become oncogenes. On the other hand, the tumor suppressor proteins arrest the progression of the cell cycle to regulate the activities of these proto-oncogenes. However, when oncogenes are overexpressed or tumor suppressor proteins are under-expressed, the phenomenon would indicate that tumor cells will emerge. Therefore, it is possible to treat cancer by downregulating and upregulating the activities of oncogenes, and tumor suppressor genes, respectively.

It has been established in the literature that restoring the lost functions of the tumor suppressor protein p53 in tumor cells, by disrupting its interactions with the oncogene MDM2 or MDMX is a promising therapeutic strategy against various human cancer [1,2,3,4,5]. One currently employed approach for effectuation is the development and validation of small molecule inhibitors. Interestingly, a list of small molecule inhibitors of p53-MDM2 interaction signaling has been published by Estrada-Ortiz et al. [1]. Similarly, Fang et al. [5], reported documentation of small molecule inhibitors of MDM2. Furthermore, Lemos et al. [6], have reported a list of p53-MDM2 interaction inhibitors. While there are numerous reports of small molecule MDM2 inhibitors since the disruption of p53-MDM2 interaction signaling has been identified as an ideal drug target in cancer therapy, research on p53-MDMX inhibition is still evolving. Initially, we have a hypothesis that compounds that are MDMX-specific or optimized for dual inhibition of p53-MDM2/MDMX interaction signaling could be a promising therapeutic strategy for cancer to retain the wild-type p53 (Barakat et al. [7]). However, most potential inhibitors are highly selective for MDM2 than for its structural homolog MDMX [2, 3, 5]. Interestingly, Fang et al. [5], have discussed some of the challenges involved in designing or optimizing potent MDM2 inhibitors for disruption of p53-MDMX interactions. To this end, most recent efforts to optimize potential MDM2 candidates’ inhibitors for disruption of p53-MDMX interaction signaling have adopted computational-based approaches, particularly those allowing these molecules to be fitted into the confined region on targeted proteins called the catalytic site. Computer-based approaches in drug discovery and development are cost-effective and often less cumbersome, thus, allowing researchers to rapidly screen and identify MDM2 inhibitors that can be optimized for MDMX inhibition. Additionally, the effectiveness of these molecules for cancer therapy is elevated, and the problem of cancer drug resistance is alleviated. Herein, we build upon these facts and apply various computational techniques to account for the potentials of a 1, 8-naphthyridine containing small molecule ligand which was identified in previous our study as a potential blockade of p53-MDM2 interaction, to act as dual-inhibitor of p53-MDM2/X interaction signaling [8]. It is worth noting that this molecule was selected for this study because it showed the most remarkable predicted pharmacological activities against p53-MDM2 interaction in our previous study compared to the other identified lead molecules from the same [8]. More also, the dual-inhibitory potentials of this small molecule ligand were compared to standard dual inhibitor of p52-MDM2/X interactions (RO-2443) as a control.

Computational methods

All the calculations performed on the titled compound CPO were carried out using various computational methods.

Quantum mechanical calculation

We employed the MOPAC 2016 software program to implement all quantum mechanical (QM) calculations [9]. These calculations were performed using the PM7 semi-empirical Hamiltonian [10], with an implicit COSMO solvation model [7]. Noteworthy, prior to the theoretical calculations, the geometry of the titled compound, was pre-optimized using the molecular mechanics force field (MMFF94) implemented in Avogadro v1.2.0 software program [11], after protonating the structure at a pH of 7.4. The pre-optimized geometry was the input for QM calculations [12]. Geometry optimization of structures at semi-empirical theory was achieved using the Broyden–Fletcher–Goldfarb–Shanno (BFGS) geometry optimizer. The keywords “DIPOLE” and “MULLIK” were used to compute the dipole moments and Mulliken atomic charges, respectively. The Time-Dependent Hartree Fock (TDHF) was used to calculate the molecular polarizabilities using the “POLAR” keyword as implemented in MOPAC 2016. Jmol software program was used to visualize the charge distribution diagram of frontier molecular orbitals (FMOs) and molecular electrostatic potential (MEP) of the studied compounds [13]. All the quantum chemical reactivity descriptors were computed from the energies of the highest occupied and lowest unoccupied molecular orbitals (EHOMO−LUMO).

Molecular docking studies

Protein and ligand preparation

The crystal structures of the titled targets MDM2 (PDB: 3JZK) and MDMX (PDB: 7C3Q) were obtained from the protein data bank (https://www.rcsb.org/) and were prepared using the Dock prep module implemented in UCSF chimera v1.10.2 software program [14]. The details of protein preparation have been reported by [8]. Subsequently, the 3D structure of CPO and RO2443 with PubChem IDs 2102154 and 136683437 respectively were retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov) in structure data format (SDF). The structures of cognate ligands of targets were extracted from each protein. The geometries of all ligands were subjected to a two-phase optimization technique to obtain the conformation at global minima. The details for geometry optimization are described above. The prepared proteins and fully optimized geometry of ligands were used as input for molecular docking.

Molecular docking

The vina module in PyRx v 0.8 freeware program was used to carry out the protein–ligand docking [15]. The active site of the targets was obtained from the perusal of literature [2, 3]. A grid box of size x = 22.2999 Å, y = 24.4569 Å, and z = 33.9603 Å, with a center dimension of x = 7.4603, y =  − 9.1234, and z = 23.7660 were set to define the active site of MDM2 (PDB: 3JZK). Similarly, a grid box of size x = 20.3312 Å, y = 25.3334 Å, and z = 28.6278 Å, was centralized at position x = 14.4383, y =  − 11.3765, and z =  − 4.5618 to define the hotspot residues of MDMX (PDB: 7C3Q). An exhaustiveness of 10 was used throughout the entire simulation. The output of the docking studies was visualized using PyMOL v2.4.1 Academic freeware program [16]. The 2D interaction profile of protein–ligand complexes was generated using the LigPlot v2.2.4 freeware program [17]. The reliability of the docking protocol was validated by superimposing unprocessed co-crystalised ligand structures and re-docked ligand structures of proteins (Fig. S1).

Molecular dynamics simulation

Molecular dynamics (MD) simulations of all protein–ligand complexes were performed for 50 ns with CHARMM36 force field [18], using GROMAC version 2019.6 [19]. Each system was prepared with the TIP3P water model [20], in periodic Dodecahedron boxes that extend 10 Å beyond any protein atom, and Na+ and Cl− ions in 0.15 mm concentration were added to neutralize all systems. Subsequently, the solvated systems were subjected to the steepest descent energy minimization process by using the Verlet cut-off scheme at a total minimization cycle of 50,000. Then, the equilibration of systems for molecular dynamics simulations of the complexes was completed in two-dependent processes consisting of an isothermal-isochoric ensemble and an isothermal-isobaric ensemble, respectively. The canonical ensemble (NVT) was accomplished with a constant temperature of 310 K for 0.1 ns using the Berendsen thermostat, and then isothermal–isobaric ensemble (NPT) was conducted on complexes for 100 ps with a constant pressure of 1 bar. The V-rescale and Parrinello–Rahman methods were employed for temperature and pressure coupling, respectively [21]. Leonard–Jones potentials and Particle Mesh Ewald (PME) method were used to handle van der Waals and long-range electrostatic interactions respectively [22]. The complexes were subjected to a final MD simulation production run for 50 ns.

Drug-likeness, and ADMET evaluation

The drug-likeness of the compounds was evaluated based on Lipinski’s rule of five (RO5) using the SwissADME online server (https://www.swissadme.ch.index.php/) [23]. Subsequently, the pkCSM webserver (http://biosig.unimelb.edu.au/pkcsm/) [24] was employed to study the ADME properties of the compounds. Ultimately, a comprehensive study of the toxicity profile of the compound was carried out using the ProTox-II webserver (http://tox.charite.de/protox_II/) [25].

Bioactivity predictions

PASS (Prediction of Activity Spectra) online server (https://www.pharmaexpert.ru/passonline/predict.php) was used to predict the biological and pharmacological activities of the compounds [26]. PASS (prediction of activity spectra), employs the structural information of molecules to make predictions of the pharmacological influences, mechanisms of action, and toxicity exhibited by these molecules when they interact with biological macromolecules. Based on the anticancer properties, a compound with a Pa value greater than the Pi value would have a higher tendency to exhibit anticancer activity and vice-versa.

Results and discussion

Frontier molecular orbital analysis (FMO) and global reactivity descriptors

The FMOs, including the HOMO and LUMO, are important utility tools for investigating the electronic properties of molecules. While the HOMO energy corresponds to the electron donating potentials of a molecule and is correlated to the ionization potential of that molecule, the LUMO energy represents the first empty innermost orbital unfilled by electron and is directly related to the electron affinity of the molecule signifying the electron-withdrawing tendency. The energy gap between the HOMO and LUMO energy decodes information about the molecular chemical stability or reactivity of a molecule predominantly revealing how electrons in the molecule transit from the ground state to its excitation state. In addition, several other characteristics including the chemical hardness, softness, electronegativity or polarizability of a molecule can be directly accessed from the HOMO–LUMO energy [27, 28]. For instance, the lower energy gap between two frontier orbitals indicates lower kinetic stability, higher polarizabilities and reactivity of a molecule which implies the softness of the molecule and vice-versa.

In this study, the FMO and energy gap of the compounds CPO and RO2443 were analyzed by using the PM7-based semi-empirical quantum mechanical method in COSMO implicit water model as shown in Fig. 1. It is clear from Fig. 1 that CPO underwent an intermolecular charge transfer from the alkylated phenyl moiety to the naphthyridine nucleus as it excited from the ground state (S0) to the first excitation state. On the other hand, the HOMO of RO2443 was localized mainly on the imidazolidine-2, 4-dione ring with partial restriction on the indole ring moiety while the LUMO was confined mainly on the indole ring and methylidine centre. The FMOs showed no localization on the difluorophenyl ring of RO2443 in spite flooding almost the entire structure. The molecular illustration of CPO and RO2443 along with their predicted HOMO, LUMO and gap energies are presented in Table 1. From the results, CPO showed a higher energy gap (ΔE) value between HOMO and LUMO than RO2443 which indicates higher chemical stability and lower chemical reactivity of the molecule. To fully understand the reactivity of the studied compounds with other chemical species, the energy of the HOMO and LUMO was used to estimate other reactivity parameters including; ionization potential, electron affinity, chemical hardness (η), chemical softness (ζ), electronic chemical potential (µ), Electrophilicity index (ω), and electronegativity (χ). The expression for the aforementioned reactivity parameters has been described according to Koopman’s theorem [29], and can be calculated by the accompanying mathematical statements;

$${\text{Energy gap} \Delta E ={\text{HOMO}}}_{\varepsilon }- {\text{LUMO}}_{\varepsilon },$$
(1)
$$\text{Ionization \,potential }I = - {E}_{\text{HOMO}},$$
(2)
$$ {\text{Electron affinity }}A{ } = { } - E_{{{\text{LUMO}}}} , $$
(3)
$$\text{Chemical hardness }\eta = \frac{1}{2}\left(\frac{{\partial }^{2}E}{\partial {N}^{2}}\right)V= \frac{1}{2}\left(\frac{\partial \mu }{\partial N}\right)V=(I-A)/2,$$
(4)
$$\text{Chemical potential }\mu = \left(\frac{\partial E}{\partial N}\right)V= -(I+A)/2,$$
(5)
$$\text{Electronegativity }\chi = -\mu = -\left(\frac{\partial E}{\partial N}\right)V=(I+A)/2,$$
(6)
$$\text{Softness }\zeta =\frac{1}{\eta },$$
(7)
$$\text{Electrophilicity index }\omega = \frac{\mu 2}{2\eta }.$$
(8)
Fig. 1
figure 1

Frontier molecular orbitals of: A. CPO. B. RO2443

Full size image
Table 1 Calculated quantum reactivity descriptors of CPO and RO2443 using PM7 Hamiltonian method
Full size table

Note that the ionization potential of a molecule describes the amount of free energy required to remove the electron of an atom while the electron affinity represents the amount of energy required to accept an electron when it is supplementary to a neutral atom. Lower ionization potential energy often indicates higher reactivity or lower stability of a compound, and it provides a great insight for inhibition. In contrast, a superior electron affinity value may be ascribable to a high electron-withdrawing ability of a compound. From the observation of the results in Table 1, it is evident that CPO had higher chemical stability and electron-withdrawing potentials compared to RO2443. This observation is consistent with the gap energy between the HOMO and LUMO FMOs. It is worth noting that the stability of compounds increases with hardness and decreases with softness. The higher hardness value of CPO over RO2443 is in support of the inferences of the ionization potential and energy gap respectively. The value of µ provides insight into the tendency of an electron to escape from a stable molecule. A higher negative chemical potential (µ) implies that a compound will not spontaneously decompose into elements which specify its stability. CPO had more negative chemical potential than RO2443. The Electrophilicity index (ω) describes the electron-withdrawing behavior of a compound in response to an electron from an external environment. Noteworthy, a good electrophile can be described by a higher ω value. In contrast, a lower value of ω would indicate the existence of a reactive nucleophile. Drawing from the observations from Table 1, it is evident that CPO is a better electrophile than RO2443. This is in good agreement with the calculated value of electronegativity (χ).

Molecular electrostatic potential (MEP)

MEPs have been helpful to identify the relative polarity of compounds and providing crucial information about the patterns of molecular charge distribution. Consequently, it is plausible to gain insight into the electrophilic and nucleophilic centers of the studied molecules by analyzing their MEP. It is worth noting that the electrostatic potential data of molecules can be categorized using classical color codes. A red color scheme represents the most negative electrostatic potential and it indicates the electron-rich centers. In contrast, a blue color region codifies the electron-deficient areas (i.e. the most positive electrostatic potential). The light blue, yellow and green color moieties denote the region of slightly electron-deficient centers, marginally electron-rich areas and zero electrostatic potential potions of a molecule, respectively. By observing the different color schemes, we could say that the potential of a molecule decreases in the following order: blue > light blue > green > yellow > red. Figure 2, displays the molecular electrostatic potential maps of CPO and RO2443 respectively. From Fig. 2, we can identify that the areas where there is a maximal concentration of electrons are located at the carbonitrile and oxygen atoms attached to the 1, 8-naphthyridine ring of CPO, and the oxygen atom of the carbonyl stretch that connects with the 2-hydroxy-5-methylbenzoyl moiety. In contrast, the region of greatest positive potentials of CPO is located at the extremities formed by the hydrogen atoms of the methyl group, fluorophenyl group and 2-hydroxy-5-methylbenzoyl moiety. The most negative potential of RO2443 is located on the oxygen atom of the imidazolidinedione (hydantoin), nitrogen atom and carbon atoms of the indolemethylidine ring, while the region of positive electrostatic potential can be ascribed to the hydrogen atoms of the molecule.

Fig. 2
figure 2

Molecular electrostatic potential map of: A CPO. B RO2443

Full size image

Mulliken population analysis

Table 2 presents the atomic charge distribution of CPO and RO2443 obtained from Mulliken population analysis using the PM7-based semi-empirical Hamiltonian calculations. The estimation of partial atomic charges of any compound is crucial for describing the charge distribution since atomic charges affect the molecular and electronic properties of compounds. Additionally, the adsorptive centers of any small molecule ligand can be determined by calculating their atomic charges. From Table 2, we observed that the oxygen and nitrogen atoms of the studied structures are the electron-rich chemical species (i.e. they possess the most negative electronic charges) which may be attributed to their molecular relaxation. On the other hand, the carbon and hydrogen atoms of the studied structures predominantly cover the positive charges and are considered electron-deficient atoms. Although some carbon atoms of the studied compounds had shown negative atomic charges. Precisely, the atoms N2, N3, C22, O1, O2 and O3 of CPO have the most negative atomic charge values Table 2, while the O1, O2, N1, N2, N3, and C17 are observed with the highest negative atomic charges for RO2443.

Table 2 Calculated Mulliken atomic charges of CPO and RO2443
Full size table

Nonlinear optics (NLO) analysis

To date, NLO materials play crucial roles in recent technologies with several industrial and medical benefits some of which have been described in previous studies [28, 30]. From a more interesting perspective on chemical methods and applications, the most notable attributes of evaluating NLO properties are their tendency to provide significant insights into how subtle changes in molecular structures can affect NLO responses. The various NLO responses and their components for CPO and RO2443 calculated using the PM7 semi-empirical Hamiltonian calculations in MOPAC 2016 are presented and summarized in Tables 3 and 4, respectively. The dipole moment (µ) provides insight into the ionic character state in a bond or a molecule [28]. Generally, molecules having larger dipole moment values will have more ionic character. In addition, the value of dipole moments is important in predicting the shape and reactivity of a molecule. The results of Table 4 specify that CPO is less reactive than RO2443 which is consistent with the inferences made from the FMOs results. The computation of polarizability (α0) and hyperpolarizability (β0 and γ0) are useful to describe charge delocalization, and measure NLO effects in molecular systems [31]. More interestingly, they have been applied in drug design [30]. The first hyperpolarizability (β0) and related properties (µ, α0 and γ0) of the titled compound CPO and RO2443 are designated as the coefficients in the Taylor series expansion based on the energy in the external electric field [31, 32]. For a weak homogeneous external electric field, the expansion can be expressed as;

$$E= {E}_{0}- \sum {\mu }_{i}{F}^{i}-\frac{1}{2}\sum {{\alpha }}_{ij}{F}^{i}{F}^{j}- \frac{1}{6}\sum {\beta }_{ijk}{F}^{i}{F}^{j}{F}^{k}+ \frac{1}{24}\sum {\gamma }_{ijkl}{F}^{i}{F}^{j}{F}^{k}{F}^{l}+\cdots .$$
(9)
Table 3 The non-linear optics (NLO) measurements of CPO and RO2443
Full size table
Table 4 The molecular electric dipole moment (µ), static polarizability (α0), static first order hyperpolarizability (β0), static second order hyperpolarizability (γ0), of CPO and RO2443
Full size table

Note, E0 describes the energy of the unperturbed molecules, Fi represents the field at the origin, µi, αij, βijk and γijkl correlate to the dipole moment, static polarizability, first-order hyperpolarizability and second-order hyperpolarizability. The total dipole moment µ, static mean polarizability α0, the mean first-order hyperpolarizability β0 and second-order hyperpolarizability γ0 can be estimated by the equations below;

$$\text{Dipole moment }\mu = \sqrt{{\mu }_{x}^{2}+{\mu }_{y }^{2}+{\mu }_{z}^{2}},$$
(10)
$${\text{Static mean polarizability }{\alpha }_{0} = (\alpha }_{xx}+{\alpha }_{yy}+{\alpha }_{zz})/3,$$
(11)
$$\text{Static first-order hyperpolarizability} \beta = \sqrt{{\beta }_{x}^{2}+{\beta }_{y }^{2}+{\beta }_{z}^{2}}$$
(12)

where

$${\beta }_{x}=3/5\left({\beta }_{xxx}+ {\beta }_{xyy}+ {\beta }_{xzz}\right),$$
(13)
$${\beta }_{y}=3/5\left({\beta }_{yyy}+ {\beta }_{yzz}+ {\beta }_{yxx}\right),$$
(14)
$${\beta }_{z}= 3/5\left({\beta }_{zzz}+ {\beta }_{zxx}+ {\beta }_{xyy}\right),$$
(15)
$${\beta }_{\text{Total}}=\sqrt{{({\beta }_{xxx}+ {\beta }_{xyy}+ {\beta }_{xzz})}^{2}+({\beta }_{yyy}+ {\beta }_{yzz}+ {\beta }_{yxx}{)}^{2}+({\beta }_{zzz}+ {\beta }_{zxx}+ {\beta }_{xyy}{)}^{2}},$$
(16)
$$\gamma = 1/5 [{\gamma }_{xxxx}{\gamma }_{yyyy}{\gamma }_{zzzz}+2({\gamma }_{xxxx}+{\gamma }_{yyyy}+ {\gamma }_{zzzz})].$$
(17)

It is crucial to note that the higher value of first-order hyperpolarizabilities of a molecule over the classical standard NLO material Urea, suggests that such molecule is NLO active and would act as better NLO material. Table 4 shows that the calculated hyperpolarizability of CPO (8.3307 × 10–30 esu) is 8 times greater than that of Urea (0.3728 × 10–30 esu) and is nearly 6 times lower than that of RO2443 (14.2880 × 10–30 esu) based on a relative scale. However, this study has come to an agreement that CPO could be a suitable material for NLO-based technology despite the lower NLO properties than its counterpart RO2443.

Molecular docking studies

Structural analysis of p53-MDM2/MDMX interaction

The structural details of p53-MDM2/MDMX interactions have been reported in the literature [2, 3]. Summarily, the site of interaction between p53 and MDM2 or co-regulator MDMX is located in the N-termini of both proteins and is composed of three sub-pocket extensively forming a hydrophobic cavity on MDM2/MDMX. The hydrophobic cavity of MDM2 is similar to MDMX in the Phe19 sub-pocket, corresponding to Ile61, Met62, and Val93 of MDM2 (PDB ID: 3JZK), and Ile59, Met60, and Val91 on MDMX (PDB ID: 7C3Q), and Trp23 sub-pocket which have been observed to correspond to amino acid Gly58, Leu57, and Ile99 of MDM2, and Gly56, Leu55, and Ile97 on human MDMX. In contrast, MDMX is more compact than MDM2 in the Leu26 sub-pocket due to amino acids Pro94, Ser95 and Pro96 in MDMX (PDB ID: 7C3Q), which have been substituted by His96, Arg97, and Lys98 in MDM2 (PDB ID: 3JZK) [2]. It is important to note that these amino acid substitutions are located on Helix a2′ at the interface of the p53-binding site within the two proteins. Consequently, the proline residues (Pro94 and Pro96) in MDMX displace this helical domain in MDMX relative to MDM2, which ultimately causes the side chain groups of Met52 and Tyr98 (which are substituted by Leu54 and Tyr100 in human MDM2) to protrude into the p53-binding cleft on MDMX. This phenomenon results in the partial occlusion of the Leu26 cavity, making it less accessible to selective MDM2 inhibitors (Fig. 3). As a matter of fact, it has been reported in literature that the Leu26 sub-pocket is unoccupied by MDM2 inhibitors that are optimized for MDM2-inhibition [5]. However, it is possible to achieve improved selectivity by designing molecules that can completely occupy the Leu26 sub-pocket of MDMX. More interestingly, this unique binding mode opens the door for the discovery of a novel small molecule agonist of p53 that is either MDMX-specific or that can be optimized for dual-inhibition of MDM2 and MDMX. Noteworthy, the Leu26 sub-pocket of MDM2 comprises amino acids Leu54 and His96, corresponding to Met52 and Pro94 on human MDMX (PDB ID: 7C3Q).

Fig. 3
figure 3

Overview of superimposed structure of MDM2 and MDMX showing structural variation in amino acid composition of Leu26 sub-pocket

Full size image

Protein-inhibitor interaction analysis

To study the potency of CPO to inhibit MDM2 and MDMX, a simulation of the interaction of the compound with both oncogenic proteins was carried out and the results were compared with RO2443 (which is a standard dual inhibitor of p53-MDM2/MDMX interaction) and cognate ligands. The results of the molecular docking studies are given in Table 5.

Table 5 Interaction profile between MDM2/MDMX and CPO, RO2443, and co-crystalized ligands with their binding affinity
Full size table

Interactions between MDM2 and CPO, RO2443, and co-crystallized ligand

It is obvious from the observation of the results in Table 5 that the binding affinities of the studied compounds varied significantly. CPO showed the highest binding affinity for MDM2 (ΔG =  − 9.4 kcal/mol), which is about 1.2 times higher than RO2443 (ΔG =  − 8.0 kcal/mol). Like RO2443, CPO also exhibited species-specific binding with amino acids corresponding to Phe19, Trp23, and Leu26 residues in p53 to occupy the p53 binding sites on MDM2 (Figs. 4, 5). Major interactions in the complexes of CPO, RO-2443 and cognate ligand of MDM2 (YIN) to MDM2, were through hydrophobic interactions with Leu54, Ile61, His96, Ile99, and Tyr100 (Fig. 6). The same is conserved amino acid residues mediating the inhibition of p53-MDM2 interactions by most MDM2 inhibitors [2, 8, 33]. The methylbenzoyl group of CPO stably filled into the Leu26 sub-pocket and was strongly linked through hydrophobic interactions to the Trp23 sub-pocket to mimic the binding mode of p53. The Phe19 sub-pocket was weakly occupied by the carbonitrile group attached to the naphthyridine core of CPO (Fig. 4). Although the ring groups of RO2443 and YIN were tightly maintained in the Phe19 and Leu26 sub-pocket (Fig. 5), nonetheless, they bound weakly to the Trp23 sub-pocket of p53. This difference in the binding mode of compounds-MDM2 complexes may be due to the favorable edge-to-edge aromatic-aromatic contact between the methylbenzoyl group of CPO and Tyr100, which may indicate the existence of a proximal hydrogen bonding interaction. Interestingly, the oxygen atom of the methylbenzoyl moiety in CPO formed a bifurcated hydrogen bonding network with Gln24 explaining the localization of CPO in the Trp23 and Leu26 sub-pockets in the binding cavity of MDM2, compared to RO2443 and YIN respectively. Hydrogen bonding interaction with Gln24 or Tyr100 is crucial for the overall binding of selective MDM2 inhibitors [5]. This interaction which was completely absent in the binding mode of YIN was retained through hydrophobic interactions in the binding mode RO2443. Thus, it explained the lower binding affinity of YIN and RO-2443 to MDM2 (Fig. 6).

Fig. 4
figure 4

Simulated interaction profile revealing the 3D binding mode of: A CPO, RO2443, and Co-crystallized ligand with the receptor MDM2. B CPO, RO2443, and Co-crystallized ligand with the receptor and MDMX

Full size image
Fig. 5
figure 5

Binding mode of proposed dual inhibitor, standard dual inhibitor, and cognate inhibitors in the p53-binding site of MDM2 and MDMX. A CPO-MDM2 complex. B RO2443-MDM2 complex. C YIN-MDM2 complex. D All inhibitors-MDM2 complexes. E CPO-MDMX complex. F RO2443-MDMX complex. G NUT-MDMX complex. H All inhibitors-MDMX complexes. Noteworthy, Blue = co-crystallized inhibitors, Yellow = RO2443, and Orange = CPO in the All inhibitors bound MDM2/MDM2 complexes

Full size image
Fig. 6
figure 6

2D interaction plots of: A CPO, RO2443, and Co-crystallized ligand with the receptor MDM2. B CPO, RO2443, and Co-crystallized ligand with the receptor and MDMX

Full size image

Interactions between MDMX and CPO, RO2443, and co-crystallized ligand

Since we are only interested in the dual-inhibition of MDM2 and MDMX, the titled compound CPO was docked the second time into the p53-binding site on MDMX, and the results compared with the results of RO2443, and that of the cognate ligand of MDMX (NUT) (Table 5). According to Table 5, it is glaring that CPO (ΔG =  − 7.8 kcal/mol) showed the highest binding affinity to MDMX compared to the agonist RO2443 (ΔG =  − 7.6 kcal/mol) and NUT (ΔG = − 6.8 kcal/mol) respectively. However, CPO showed higher binding affinity to MDM2 (ΔG =  − 9.4 kcal/mol) than structural analogue MDMX (ΔG =  − 7.8 kcal/mol). The lower binding affinity of CPO bound MDMX complex compared to CPO bound MDM2 complex, is in agreement with experimental findings [4, 34]. On the other hand, NUT showed the lowest binding affinity to MDMX compared with the aforementioned agonists. Therefore, we have attempted in this section to compare and discuss only the results of CPO and RO2443. Based on the inhibitor–protein interaction analysis Table 5, we observed that most of the residues in the CPO-bound MDM2 complex, and RO2443-MDM2 complex were maintained through hydrophobic interactions with MDMX (Fig. 5). By further analyzing the binding mode of CPO, it was evident that the fluorophenyl group and carbonitrile group present on the naphthyridine ring of CPO were well directed towards the Phe19 and Leu26 sub-pocket of the p53-binding interface on MDMX. Interestingly, a similar observation has been noted in the binding mode of RO-2443 with the same sub-pockets through the coplanar indolemethylidine group and hydatoin moiety (Fig. 5). The delocalization of atoms in this sub-pocket may explain the proximity in the binding affinity of both compounds to MDMX. Furthermore, the methylbenzoyl group of CPO was found to be deeply rooted in the Trp23 sub-pocket of MDMX. In contrast, the difluorophenyl group was partially restricted from occupying the Trp23 sub-pocket and is suspected to be caused by the lack of interaction with key amino acid residues Leu55 and Gly56 in the Trp23 cavity. The interaction with these residues may be associated with a local increase in the binding affinity of CPO than RO2443. In conclusion, the higher binding affinity of CPO for MDM2 and MDMX respectively, than RO2443 suggests that the existence of methylbenzoyl moiety in the ligand structure may enhance the pharmacological activity of CPO.

Molecular dynamics simulation analysis

A comparative MD simulation has been performed to evaluate the effect of CPO and RO2443 respectively on the structural characteristics of MDM2 and MDMX. The root mean square deviation (RMSD) as a measure of the overall structural fluctuation for protein–ligand complexes for 50 ns is illustrated in Fig. 7. In regards to Fig. 7A, it is clear that the CPO-MDM2 complex reached equilibrium after 5 ns of the whole MD production run, while the RO2443 bound MDM2 complex remained unstable until about 23.6 ns (Table 6). Consistent with the observation made in Fig. 7A, the CPO-bound MDMX complex also showed a lower average RMSD value compared to RO2443-bound MDMX complex (Table 6). Of note, the lower average RMSD value of CPO with the backbone atom of MDM2 or MDMX indicates that the stability of the proteins is more complex with CPO than RO2443. However, the average RMSD value increases upon ligand binding to MDMX indicating lower stability compared to ligand binding to MDM2.

Fig. 7
figure 7

A Backbone RMSDs of MDM2. B Backbone RMSDs of MDMX. C RMSF profile of MDM2. D RMSF profile of MDM2. E Radius of gyration trajectory analysis of MDM2. F Radius of gyration trajectory analysis of MDM2. The color-coding scheme is as follows: CPO-MDM2/MDMX (Red), RO2443-MDM2 (Blue), and RO2443-MDMX (Green)

Full size image
Table 6 Summary of molecular dynamics trajectory plots
Full size table

Since RMSD may not necessarily account for the mobility of structural elements, we computed the root mean square fluctuation (RMSF) plot to identify the flexible regions in both proteins upon ligand binding. The RMSFs of the protein backbone atoms are illustrated in Fig. 7C and D respectively. According to Fig. 7C, it is obvious that RO2443 and CPO-bound MDM2 complexes, showed an average RMSF value of 0.14 nm and 0.11 nm, respectively. This indicates that the backbone atom of MDM2 residues showed low fluctuation in complex with CPO than RO2443. Accordingly, the backbone atom fluctuation of MDMX residues when they interact with CPO and RO2443 has been evaluated by the RMSF data Fig. 7D. The analysis of Fig. 7D, showed similar levels of fluctuation in RO2443 bound MDMX, and CPO bound MDMX complexes with an average RMSF value of 0.23 nm and 0.24 nm respectively (Table 6). Consistent with our initial observation of the RMSD plot, we observed that the average RMSF values of CPO or RO2443 to MDMX also increased compared to ligands-bound MDM2 complexes. This phenomenon indicates more conformational change with MDMX than MDM2 in response to both ligands. More also, the major fluctuation was recorded around the loop region.

Furthermore, the influence of RO2443 and CPO binding on the compactness of MDM2 and MDMX was analyzed through the evaluation of the radius of gyration (Rg). The average Rg value for the RO2443 bound MDM2, and CPO bound MDM2 complexes were shown to be 1.31 nm and 1.32 nm, respectively (Table 6). This indicates that MDM2 attained a similar level of compactness after accommodating CPO and RO2443. On the other hand, the average Rg values of RO2443-bound MDMX, and CPO-bound MDMX complexes were found to be 1.45 nm and 1.51 nm, respectively. From the observation of the Rg profiles, it can be deduced that MDMX showed less compactness when it accommodated CPO (Fig. 7). However, the average Rg value increases slightly with MDMX upon inhibitor binding, which is in accord with experimental analysis [2].

By analyzing the solvent-accessible surface area (SASA) it is plausible to gain additional insight into the stability of the ligand-bound protein complexes. As shown in Table 6, the average SASA values for RO2443-MDM2, CPO-MDM2, RO2443-MDMX and CPO-MDMX complexes were enumerated as 62.16 nm2, 61.74 nm2, 67.69 nm2, 67.51 nm2, respectively. Since a low SASA value denotes the compression of protein, it can be concluded that MDM2 and MDMX were more stable in complex with CPO than RO2443 during the entire MD simulation. These results are consistent with those of the RMSD analysis, in which the proteins attained more stable conformation when bound with CPO.

Hydrogen bond interaction is crucial in the overall stability and activity of agonists when they are accommodated by protein. Consequently, Intermolecular hydrogen bonds were analyzed for all the complexes and the results are shown in Fig. 8. Our analysis showed that the complexes of RO2443-MDM2 and CPO-MDM2 exhibited two and three intermolecular hydrogen bonds, respectively, throughout the MD production run. Even though the CPO-MDM2 complex showed three hydrogen bonds, the number of stable hydrogen bonds was less compared to the CPO-MDMX complex, which is in agreement with the results of molecular docking analysis. However, we observed that the number of stable intermolecular hydrogen bonds in the RO2443-MDM2 complex was equally maintained in the RO2443-MDMX complex (Table 6). Based on the comparative analysis of the intermolecular hydrogen bonding in the complexes, we can conclude that CPO contributed more to the stability of the MDM2 backbone atom than RO2443. On the other hand, RO2443 contributed more to the stability of the backbone atom of MDMX than CPO.

Fig. 8
figure 8

A Hydrogen bond analysis profile of MDM2 over 50 ns MD simulation. B Hydrogen bond analysis profile of MDMX over 50 ns MD simulation. C Solvent accessible surface area profile of MDM2. D Solvent accessible surface area profile of MDMX. The color-coding scheme is as follows: CPO-MDM2/MDMX (Red), RO2443-MDM2 (Blue), and RO2443-MDMX (Green)

Full size image

Drug-likeness and ADMET evaluation

Noteworthy, late recognition of potential side effects of bioactive molecules especially undesirable ADMET properties is one of the main reasons for drug rejection. However, early investigation of drug pharmacokinetics accelerates the drug discovery and development process. Lipinski’s rule of five (RO5) is crucial to identify orally bioavailable drug candidates [8]. The drug-likeness properties of CPO and standard inhibitor RO2443 were evaluated using the swissADME webserver and presented in Table S1. The RO5 is completely satisfied by both compounds. ADME profile and Toxicity properties of the titled compound CPO and RO2443 were analyzed using the pkCMS and ProTox-II webservers and the results were compared in Table 7.

Table 7 Pharmacokinetics profile of CPO and RO2443
Full size table

Human intestinal absorption (HIA) is the process by which drugs that are orally administered are absorbed from the gastrointestinal system into the bloodstream. The predictions of the HIA for both compounds were high. However, the titled compound CPO was predicted to be more absorbed compared to RO2443. Both compounds were also predicted to be water soluble with the titled compound CPO being more soluble based on the results of the prediction. It is worth noting that water-soluble drugs tend to be more absorbed than lipid-soluble ones [24]. Despite the permeability of the studied compounds across intestinal membranes which can be characterized by the positive Human Intestinal Absorption property, these compounds were predicted to pass through the Caco-2 cells medially with the compound titled CPO exhibiting higher passage (Table 7). CPO and RO-2443 were observed to function as inhibitors of the P-glycoproteins (P-gp) I and II, they were also observed to be substrates of the P-gp. Note that the inhibition of P-gp infers that a chemical substance will most likely cause the inhibition of drug efflux proteins. From the results of the prediction, it can be inferred that the administration of CPO may lead to potential drug-drug interactions with other substrates of P-gp due to competition for transport. Also, its ability to function as an inhibitor may increase exposure to other drugs that are substrates. Hence, leading to potential toxicity and altered pharmacodynamics. The aqueous solubility of compounds (LogS) describes the relative solubility of a molecule in water at 25 °C. A lower LogS value implies that a compound will be poorly absorbed into the membrane, especially in enteral routes. As presented in Table 7, CPO was found to have a lower value of LogS compared to RO2443.

With regards to the distribution of the compounds, the ability of the compounds to permeate the blood–brain-barrier (BBB) was quite low, hence, they would be poorly distributed to the brain. However, they were predicted to penetrant the central nervous system (CNS). The fraction unbound of a drug gives an insight into the amount of the drug that will not be bound to proteins in the blood, hence, making it to be available to interact with its target. As presented in Table 7, the fraction unbound for CPO was predicted to be relatively high compared to that of RO2443.

Based on metabolism, the cytochrome P450 isoforms are mainly responsible for drug metabolism in phase I reactions. It is therefore crucial to access the potential of a compound to act as a substrate or inhibitor. From the results in Table 7, it is clear that CPO could act as a non-inhibitor of CYP2D6 and was a substrate of CYP3A4. Furthermore, it could also inhibit CYP3A4, CYP1A2, CYP2C9, and CYP2C19. The inhibition of the aforementioned enzymes could lead to potential drug-drug interactions with other drugs that are metabolized by these enzymes [35]. Hence, only drugs selected carefully based on their metabolism profile can be co-administered with CPO.

Concerning toxicity, both compounds tested non-carcinogenic, non-mutagenic, and non-cytotoxic. However, CPO was predicted to be hepatotoxic while RO2443 was predicted to be immunotoxic (Table 7). It is worth noting that the predicted toxicity class of the compounds were predicted to be 4 while the LD50 was predicted to be 1000 mg/kg and 800 mg/kg for CPO and RO2443 respectively. Hence, a very high dose of the compounds will be needed to exert the observed toxic effects and to effectuate the death of 50% of the test organism population, with the dose being higher for CPO relative to RO2443. As presented in Table 7, both compounds were also predicted to be non-probable sources of toxicity in the critical toxicological pathways. Drawing from the observation of the pharmacokinetics profile of the studied compounds, we have concluded that the titled compound CPO has better possibilities in the drug discovery and development process than RO2443. However, further research is needed in this area.

Bioactivity prediction

As earlier mentioned in the methodology, the anticancer activity of CPO and standard dual inhibitor RO2443 has been analyzed using the PASS prediction server [26], and the results are presented in Table 8. According to the results in Table 8, it is clear that CPO could serve as a potential measure for uterine cancer, sarcoma, and leukaemia. These activities were absent in RO2443. Although RO2443 had also shown antineoplastic (solid tumor) activity and prostatic cancer treatment. More interesting to note is that CPO acted as an inhibitor of Hsp27 and Bcl-2. Hsp27 in phosphorylated form has been reported to inhibit the Daxx apoptotic protein [36], thus, it has been a reputable drug target in cancer therapy. Similarly, deregulated B-cell lymphoma 2 (Bcl-2) family of proteins have been a potential drug target in cancer treatments [37].

Table 8 PASS Bioactivity prediction of CPO and RO2443
Full size table

Conclusion

In this current study, we evaluated the tendency of a previously identified potential blockade of the notable p53-MDM2 interaction signaling in oncogene-induced tumor progression to serve as a dual inhibitor of p53-MDM2/MDMX interactions, and the results compared to a standard dual inhibitor RO-2443. From QM point of view, the studied compound, which is entitled CPO showed a relative chemical reactivity value compared to RO2443 and was found to be a stable molecule. In addition, CPO was found to be a potential NLO material. From molecular docking results, CPO showed a better potential to be developed as a dual inhibitor of p53-MDM2/MDMX interactions than the agonist RO2443. Furthermore, MD simulation was performed to understand the dynamics and interactions of compound-protein complexes, which showed that CPO was stable over 50 ns MD production run. The analysis of the pharmacokinetic properties of the ligands showed that CPO would be safe to use as a drug. CPO also showed bioactivity against uterine cancer, sarcoma, and leukemia. These activities were absent in RO2443. More interestingly, CPO showed bioactivity against Hsp27 and Bcl-2. The inhibition of the aforementioned is a promising rationale for cancer therapy. Collectively, we showed that a molecule containing 1, 8-napthyridine scaffold is a potential inhibitor of p53-MDM/X interaction. However, this compound should be further evaluated in experimental studies to validate our claim.