key: cord-0906125-nu2l1zbe
authors: Jordaan, Maryam A.; Ebenezer, Oluwakemi; Damoyi, Nkululeko; Shapi, Michael
title: Virtual screening, molecular docking studies and DFT calculations of FDA approved compounds similar to the non-nucleoside reverse transcriptase inhibitor (NNRTI) efavirenz
date: 2020-08-11
journal: Heliyon
DOI: 10.1016/j.heliyon.2020.e04642
sha: 0411d5131a66cc753f60eef7cde805ceb7b4dd36
doc_id: 906125
cord_uid: nu2l1zbe

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was confirmed as the causative virus of COVID-19 disease, which is currently a worldwide pandemic. Efavirenz, a non-nucleoside reverse transcriptase inhibitor (NNRTI), is one of the most potent chemical compounds proposed to treat COVID-19 infection. We, therefore, performed virtual screening on FDA approved drugs that are similar to the efavirenz moiety. Subsequently, the compounds were subjected to screening by analyzing their drug-likeness, such as Lipinski's rule of five and ADMET properties. Molecular docking study revealed that Met165, His41, His163, and Phe140 were important interacting residues for COVID-19 main protease receptor-ligand interaction. Five top-ranked compounds, podophyllotoxin, oxacillin, lovastatin, simvastatin, and gefitinib, were selected by virtual screening and docking studies. The highest occupied molecular (HOMO) orbital, lowest unoccupied molecular orbital (LUMO) and energy gap values was calculated using density functional theory (DFT). The results of the study showed that lovastatin and simvastatin might be considered as lead compounds for further development for COVID-19 main protease inhibitors.

The novel coronavirus (COVID-19)was identified in the Hubei Province of China in 2019, and, today, over half a million people are currently infected globally with more than 20 834 deaths [1, 2, 3] . Coronaviruses are positively stranded RNA viruses that cause respiratory, enteric, and central nervous system diseases. The recent novel COVID-19 virus is considered a betacoronavirus alongside severe acute respiratory syndrome coronavirus (SARS-CoV) and the Middle East respiratory syndrome (MERS-CoV). More also, additional sequence alignment revealed a 96.1% comparison to the sequence of the main protease between COVID-19 and SARS-CoV [4, 5] . The crystal structure of the COVID-19 main protease in complex with a peptidomimetic inhibitor (PDB code 6LU7) is presented in (Figure 1 ). Viruses like the COVID-19 mutate rapidly rendering it difficult to design an appropriate treatment. Previous studies demonstrated that the main protease of SARS-CoV is essential for the life cycle of the virus, and considered to be an attractive target for drug development [4, 6] . Drugs that target conservative protease are usually capable of preventing the replication and proliferation of the virus while reducing the risk of mutation mediated drug-resistance [2, 4] .

Recent positive data highlights the application of a cocktail of antivirals, including two antiretroviral components lopinavir and ritonavir, by doctors in Thailand who were able to cure a COVID-19 patient using this treatment. These two antiretroviral components are protease inhibitors designed to block HIV viral replication and holds that these drugs could do the same for COVID-19 [7, 8] .

Furthermore, a similar study conducted by Beck and co-workers (2019) highlighted the application of a Molecule Transformer-Drug Target Interaction (MT-DTI) model to identify possible existing compounds to treat COVID-19. They predicted antiretroviral medication for HIV as the best chemical compounds to treat COVID-19, i.e.: atazanavir, efavirenz, ritonavir, and dolutegraviran [9] . Shaha and coworkers also identified a range of ARVs, including efavirenz see Figure 2 as a potential treatment for COVID-19 [10] . Confirming ARVs as a potential therapy, Yan Li and his team at the Sichuan University and Army Medical University in China (2020) identified, via a large chemical screening, four molecular drugs with high-affinity to a coronavirus protein. Of these, two were ARVs, i.e. bictegravir and nelfinavir. The other two were Prulifloxacin, a chemotherapeutic antibiotic, and Tegobuvi, an antiviral drug used in the treatment of Hepatitis C infection [4] . Contini (2020) also used virtual screening and identified four ARVs to treat COVID-19, i.e. indinavir, lopinavir, and atazanavir and cobicistat [2, 11, 12, 13, 14, 15, 16] . Furthermore, three types of in silico DTI prediction methods are used, i.e. molecular docking, similarity-based, and deep learning-based [11, 17] . The selection of the efavirenz scaffold was based on the following factors: firstly, ARVs have been utilized as an antiviral regime for patients infected with the coronavirus and secondly, in silico DTI predictions identified antiretroviral (ARV), efavirenz, a non-nucleoside reverse transcriptase inhibitor (NNRTI) as one of the most potent chemical compounds with the inhibitory potency with Kd of 199.17 nM against the COVID-19 main protease [9] .

In this study, virtual screening, also called in silico screening, was chosen to provide a rapid and inexpensive method for the discovery of FDA approved active compounds exhibiting a scaffold similar to efavirenz, which binds to the active pocket of COVID-19 main protease. Virtual screening and molecular docking results revealed promising potential hit compounds for COVID-19 main protease inhibition. Density functional theory (DFT) was used further to calculate the orbital energy value and the energy gap.

The active compounds similar to efavirenz were retrieved from the zinc database [18] . Protocol constraints such as biogenic data, FDA approval, anodyne, and sell data were set to filter out the compounds. A total of 232 molecules were retrieved and prepared for docking by subjecting to energy minimization using the Open Babel module in PyRx program.

The crystal structure of COVID-19 main protease (PDB: 6lu7) and the native ligand was downloaded from RCSB with a resolution of 2. 16 Å [19] . Discovery Studio Visualizer software was used to prepare the receptor for docking. The native ligand and water molecules were deleted from the crystal structure of COVID-19 main protease. Molecular docking was performed with AutoDock 4.2 module implemented in PyRx 0.8 using the empirical free energy force field and Lamarckian genetic algorithm conformational search with the default parameters [20] . The grid on the ligand-binding site of the protein was centered at the binding site of X ¼ -20, Y ¼ 13, Z ¼ 47, and the grid dimensions were 40 Â 30 Â 62 Å 3 . For further analysis, 43 compounds with a lowest binding affinity (<-7.0 kcal/mol) were selected. Protein-ligands interaction was analyzed using Discovery Studio Visualizer software.

Selected compounds from the molecular docking analysis were evaluated for their drug-like behavior through analysis of pharmacokinetic parameters required for absorption, distribution, metabolism, and excretion (ADME). QikProp module and Lipinski's rule integrated into virtual screening workflow (filtering option) of Schr€ odinger software was employed for calculations. We discarded 20 compounds, which were predicted to violate Lipinski's rule of five and, also fall out of optimum range for partition coefficient (QPlogPo/w), critical for estimation of absorption within the body; cell permeability (QPPCaco), a key factor governing drug metabolism and its access to biological membranes; QPPMDCK and percentage human oral absorption. Hence, 27 compounds were observed to have physicochemical and pharmacokinetic parameters within the acceptable range.

Density functional theory (DFT) is a computational quantum mechanical modeling method used to examine the electronic structure and also to investigate the interactions involved between the receptors and the ligands. The electronic and structural properties of the five best hit compounds were calculated using the Becke3-Lee-Yang-Parr (B3LYP) method with the 6-31G(d,p) basis set aided by Gaussian 09. The calculated parameters used in this study include the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) energies, electron affinity, and electrophilicity index. The molecular electrostatic potential surfaces (MEPs) were obtained from the population analysis calculations and visualized using Gauss View. These parameters play an influential role in explaining the magnitude of ligands interaction in the binding pocket of COVID-19 main protease.

To validate the stability of the hit compound, we have performed molecular dynamics simulation using NAMD full setup through the M.Wmolecular weight; log S wat -aqueous solubility (-6.5-0.5); log K HSA -logarithm of predicted binding constant to human serum albumin (-1.5-1.5); QPlogPwwater/gas partition (4.0-45.0); QPlogPC16 -hexadecane/gas partition (4.0-18); log BB -logarithm of predicted blood/brain barrier partition coefficient (-3.0-1.2); Caco-2 -cell membrane permeability (<25 poor >500 good); HBA -number of hydrogen bond acceptors (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) ; HBD number of hydrogen bond donors (0-6); QP polrzpredicted polarizability (13-70); log HERG the predicted IC 50 value for the blockage of HERG K þ channels (concern below -5); QPPMDCKpredicted MDCK cell permeability in nm/sec (<25 poor >500 great); log K p -predicted skin permeability and 95% of drugs: (-8 to -1);, log K HSA -logarithm of predicted binding constant to human serum albumin (-1.5-1.5), Human Oral Absorption -1-low, 2-medium, 3-high. MDWeb interface. The simulation process includes cleaning of the protein structure; fixing of side chains; addition of hydrogen atoms; neutralization, the addition of a solvent box and heating solvent to 300 K; reducing the restraints to just the protein backbone and minimization and equilibration of the system to finally achieve the structure prepared by simulation. To achieve a dry trajectory for the simulated protein, water molecules and ions were removed from the system.

To validate our docking protocol, we redocked native ligand and the selected 27 compounds, into the binding pocket of COVID-19 main protease using AutoDock 4.2 module in PyRx tool with default parameters. Moreover, the binding affinity of the native ligand was selected as a benchmark. Five of these compounds exhibited superior binding affinity as well as good interaction compared to efavirenz and the native ligand (-6.5 and -7.4 kcal/mol respectively), thus selected for visual analysis. Refer to Table 1 for full data.

The five selected compounds include simvastatin and lovastatin, HMG-CoA reductase inhibitor; oxacillin, a penicillinase-resistant β-lactam; podophyllotoxin, which is an antimitotic; and Gefitinib an epidermal growth factor receptor (EGFR) inhibitor [21, 22, 23, 24, 25, 26] . The identification of the key contributing residues in the binding pocket of the C19MP was performed using the Discovery Studio Visualizer.

The docked results showed that inhibitors swing between the space of hydrophobic residues; Met165, His41, Met49, Cys145, His165 and Leu27 which gave more conformational freedom, on the one hand; residues Glu166, His163, Thr26, Phe140, Thr190, His164 and Gln189, Gln143 showed hydrogen bond interaction with the ligands. Also, proteinligands interaction showed that the presence of naphthalene, quinazoline, isoxazole, benzyl, and tetrahydropyran rings played an important role. The details of the structures with the best docking scores are provided in Figures 3 and 4 . Figure 6B ). The phenyl group attached to the isoxazole ring interacts hydrophobically with Pro168 also contributes to the stabilization of the complex. Other residues like Ala19, Gln192, Agr188, Gln189, His164, 163, Gly143, Ser144, and Leu141 make close contact with the ligand without interactions ( Figure 6A) .

Moreover, His41 and Cys145 interacted with methyl group substituent, which was accompanied by one conventional H-bonding between the carboxyl oxygen atom and Cys145 (b.d M.Wmolecular weight; log S wat -aqueous solubility (-6.5-0.5); log K HSA -logarithm of predicted binding constant to human serum albumin (-1.5-1.5); QPlogPwwater/gas partition (4.0-45.0); QPlogPC16 -hexadecane/gas partition (4.0-18); log BB -logarithm of predicted blood/brain barrier partition coefficient (-3.0-1.2); Caco-2 -cell membrane permeability (<25 poor >500 good); HBA -number of hydrogen bond acceptors (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) ; HBD number of hydrogen bond donors (0-6); QP polrzpredicted polarizability (13-70); log HERG the predicted IC 50 value for the blockage of HERG K þ channels (concern below -5); QPPMDCKpredicted MDCK cell permeability in nm/sec (<25 poor >500 great); log K p -predicted skin permeability and 95% of drugs: (-8-1); log K HSA -logarithm of predicted binding constant to human serum albumin (-1.5-1.5), Human Oral Absorption -1-low, 2-medium, 3-high. and the meta-OCH 3 substituent on the phenyl ring; this was accompanied by one hydrophobic interaction. These interesting interactions potentially deepen the binding strength of podophyllotoxin to the C19MP receptor and seem to be critical (Figure 8) .

A binding affinity score of À7.8 kcal/mol was obtained for the docking of simvastatin in the binding site of C19MP (Figure 9 ) refer to supporting information, which also indicates strong interactions between the ligand and the receptor. The carbonyl oxygen group and the O atom in the tetrahydropyran ring interact with His163 and Glu166 via H-bond (b.d ¼ 2.21, 2.67 Å; b.a ¼ 115.6, 125.1 o , respectively). The naphthalene group extends into the hydrophobic pocket potentiating the ligandreceptor interaction (His 41 and Met165); the ring stacks vertically with the imidazole ring of His41 forming a π-π T-shaped interaction.

Additional stability to this complex is afforded by the formation of a hydrogen bond interaction between the carbonyl oxygen atom of dimethylbutanoate and residue Cys145.

The presence of a dimethyl substituent also assists with strong binding inside the active site of C19MP by forming hydrophobic interactions with residues His41, Cys145, and Met49, respectively. Physicochemical and ADMET descriptors of the five best dock compounds are shown in Table 2 .

Furthermore, to validate the docked results of the hit compounds, we performed MD simulations on the COVID-19 main protease and redocked all the five compounds. The re-docked results revealed an increase in the binding affinity of simvastatin and lovastatin (8.2 and 7.9 kcal/mol), respectively. Surprisingly, there was a decrease in the binding affinity of podophyllotoxin (7.7 kcal/mol), and the binding affinity of oxacillin and gefitinib (7.7 kcal/mol) remained constant. Interestingly, the ligands were bound inside the active site of a similar pose and also formed interaction with more residues, mainly through H-bonding Figure 10 .

Frontier molecular orbitals of five-hit compounds specify a crucial role of charge-transfer interactions with the binding site of COVID-19 main protease. The higher HOMO value denotes a molecule with a good electron donor, whereas a lower value implies a weak electron acceptor. Furthermore, a smaller energy gap between the LUMO and HOMO energies has a considerable influence on the intermolecular charge transfer and bioactivity of molecules. Thus, a wide energy gap observed in the hit molecules negatively affect the electron to move from the HOMO to the LUMO, which subsequently led to a weak affinity of the inhibitor for COVID-19 main protease. The E gap value decreases according to the following: oxacillin (0.1021eV) > podophyllotoxin (0.0145eV) > gefitinib (0.0351eV) > simvastatin (0.0230eV) > lovastatin (0.0145eV). Hence, the reactivity order increases according to: oxacillin (0.1021eV) > podophyllotoxin (0.0410eV) > gefitinib (0.0351eV) > simvastatin (0.0230eV) > lovastatin (0.0145eV) where the most reactive is clearly lovastatin (0.0145eV). The order of reactivity increases conforms with the decreases in energy gap values. The chemical potential (μ) indicate negative values for all the compounds, which implies good stability, and the formation of a stable complex with the receptor. Also, lovastatin and simvastatin have the least hardness values (η), among the hit molecules and which correlated with the trend of molecular docking (Table 3 ). The graphic results from DFT calculations are presented in Figure 11 . The red and green parts represent the cloud density of frontier orbital at HOMO or LUMO states. Finally, by comparing the values of molecular orbital energies (eV), global reactivity descriptors, and binding affinities value of five-hit compounds, simvastatin, and lovastatin may be considered as potential COVID-19 main protease inhibitors.

The molecular electrostatic potential (MEP) surface provides details about charge distribution and also predicts reactive sites for electrophilic and nucleophilic attack in a compound. Meanwhile, the MEP surface of simvastatin and lovastatin were evaluated using the DFT/b3lyp method as shown in Figure 11 most negative regions are shown by red, most positive regions in blue, and zero potential regions are visualized in green. MEP mapped surface of simvastatin range from À0.0678 a.u (deepest red) to 0.0678 a.u (deepest blue) and lovastatin range from À0.0672 a.u (deepest red) to 0.0672 a.u (deepest blue) Figure 12 .

In this research paper, virtual screening was successfully used to identify five new FDA approved candidate molecules similar to the efavirenz scaffold, and their binding affinity was superior to the native ligand in the active pocket of the COVID-19 main protease, i.e.; podophyllotoxin, oxacillin, lovastatin, simvastatin, and gefitinib. The redocked results after MD simulation revealed an increase in the binding affinity of simvastatin and lovastatin (8.2 and 7.9 kcal/mol), respectively. Notably, there was a decrease in the binding affinity of podophyllotoxin (7.7 kcal/mol), and the binding affinity of oxacillin and gefitinib (7.7 kcal/mol) remained constant. The docking results showed that H-bonds and hydrophobic interactions might play important roles in contributing to the molecular interactions between the active compounds and the COVID-19 main protease. The DFT calculations and molecular docking showed that lovastatin and simvastatin may be considered as potential hits as anti-coronavirus agents and can be selected for further studies like modification of the scaffold, characterization, and in vitro evaluation. The predicted physiochemical and ADMET parameters were within the acceptable optimal requirements for drug development.

Maryam Amra Jordaan: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.

Oluwakemi Ebenezer: Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

Nkululeko Damoyi, Michael Shapi: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. 

Virtual Screening of an FDA Approved Drugs Database on Two COVID-19 Coronavirus Proteins

Therapeutic options for the 2019 novel coronavirus (2019-nCoV)

Therapeutic drugs targeting 2019-nCoV main protease by high-throughput screening

The crystal structure of 2019-nCoV main protease in complex with an inhibitor N3

Role of lopinavir/ritonavir in the treatment of SARS: initial virological and clinical findings

through a drug-target interaction deep learning model

silico studies on therapeutic agents for COVID-19: drug repurposing approach

Self-attention based molecule representation for predicting drug-target interaction

PubChem Compound Summary for CID 5362440, Indinavir

Cobicistat (GS-9350): a potent and selective inhibitor of human CYP3A as a novel pharmacoenhancer

The use of caspofungin in HIV-infected individuals

Virtual screening

AutoDock Vina, Improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading

Mechanisms of action and efficacy of statins against influenza

The major resveratrol derivative in grape juices

Podophyllotoxin: a novel potential natural anticancer agent

A systematic review and metaanalysis on the therapeutic equivalence of statins

Detection of oxacillin resistance in Staphylococcus aureus isolated from the neonatal and pediatric units of a Brazilian teaching hospital

The authors declare no conflict of interest.

No additional information is available for this paper.