Repurposing FDA-approved drugs to fight COVID-19 using in silico methods: Targeting SARS-CoV-2 RdRp enzyme and host cell receptors (ACE2, CD147) through virtual screening and molecular dynamic simulations

Background Different approaches have been proved effective for combating the COVID-19 pandemic. Accordingly, in silico drug repurposing strategy, has been highly regarded as an accurate computational tool to achieve fast and reliable results. Considering SARS-CoV-2's structural proteins and their interaction the host's cell-specific receptors, this study investigated a drug repurposing strategy aiming to screen compatible inhibitors of FDA-approved drugs against viral entry receptors (ACE2 and CD147) and integral enzyme of the viral polymerase (RdRp). Methods The study screened the FDA-approved drugs against ACE2, CD147, and RDRP by virtual screening and molecular dynamics (MD) simulation. Results The results of this study indicated that five drugs with ACE2, four drugs with RDRP, and seven drugs with CD147 achieved the most favorable free binding energy (ΔG < −10). This study selected these drugs for MD simulation investigation whose results demonstrated that ledipasvir with ACE2, estradiol benzoate with CD147, and vancomycin with RDRP represented the most favorable ΔG. Also, paritaprevir and vancomycin have good binding energy with both targets (ACE2 and RdRp). Conclusions Ledipasvir, estradiol benzoate, and vancomycin and paritaprevir are potentially suitable candidates for further investigation as possible treatments of COVID-19 and novel drug development.


Introduction
Since the outbreak of the COVID-19 pandemic in late 2019, many have attempted to find a reliable treatment. Based on the reports of WHO, more than 17 million global cases have been confirmed thus far [1]. Given the data on structural properties of SARS-CoV-2, the causative agent of the COVID-19, and its homology to beta-coronaviruses of familiar viruses, such as SARS and MERS, similar therapeutic strategies have thus far been applied [2]. However, no effective drug has yet been introduced to treat and combat specific structures of viral components, despite the prescription of various medications, including hydroxychloroquine [3], azithromycin [4], remdesivir [5], idasanutlin [6], and favipiravir [7]. Therefore, pharmacological studies concerning COVID-19 treatment are still in progress. Many research studies have addressed the COVID-19 treatment, focusing on the drug repurposing technique for implementing which recognizing pathogenic targets is essential. In this case, biological insights toward genomic and structural properties of SARS-CoV-2 have identified many features of viral pathogenic targets [8]. RNA-dependent RNA polymerase (RdRp) is integral for preserving viral life, and several reports indicate that positive-sense viruses have conserved RdRp enzymes [9]. In line with these findings, X-ray crystallography and structural properties of SARS-CoV-2 RdRp were conducted to introduce novel antiviral drug designs and drug repurposing approaches [10]. Spike glycoprotein, as a structural protein of SARS-CoV-2, plays a critical role in the initial steps of pathogenesis. Structural studies and biochemical experiments have confirmed the binding of SARS-CoV2 spike to human angiotensin-converting enzyme (ACE2) receptors [11]. According to these studies, receptor-binding domains (RBD) of the spike undergo transient movements that trigger further up and down conformations for receptor attachment [12]. Based on surface Plasmon resonance studies, ACE2 binds to the RBD of SARS-CoV-2 spike with nearly ten to twenty-fold higher affinity than SARS-CoV spike ectodomain [13]. ACE2 is expressed in the lower respiratory tract, skin keratinocytes, small intestine cells, and oral epithelial cells [14]. Furthermore, the expression of ACE2 and the infection severity were correlated in vitro [15]. It has been shown that ACE2 expression patterns increase during tumor propagation [16], and factors such as age, diabetes, and cardiovascular diseases also impact expression patterns [17]. Recent studies have revealed functional patterns of CD147 receptor involvement in infection dissemination [18]. CD147 participates in inflammation, nutrient, and drug transporter activity, as well as microbial pathology, and developmental processes. Also, it has efficacy in certain infectious diseases, such as malaria, neisseria meningitides, and HIV-1 [19]. It has been reported that COVID-19 creates a novel route for CD147-spike protein (SP) through which it invades host cells [20]. It is known as Basigin or extracellular matrix metalloproteinase inducer and also a cell receptor in erythrocytes for the parasite Plasmodium falciparum [21]. The expression levels of CD147 receptors increase in patients with asthma, making them a susceptible group to SARS-CoV-2 infection [22]. Considering all the above aspects, this study selected RdRp, ACE2, and CD147 and screened them against 2471 FDA-approved libraries by in silico methods. The result showed significant binding of many approved small molecules to the chosen targets. Then, the complexes with the top-docked results were simulated for 100 ns to analyze the stability. In such a case, it is possible to repurpose several predicted drugs to prevent and treat SARS-CoV-2.

System input setup and initial structures
Proteins: It was the ACE2-SARS-CoV-2 RBD complex, as the crystal structure of ACE2 (PDB id: 6M0J), that justified its use in this study [23]. The crystal structure of CD147 (PDB id: 3B5H) at the 2.8 Å resolution provides a suitable structural explanation by homo/hetero-oligomerizations and represents a general structure of other CD147 family members [24]. The predicted structure of RdRp coordinate files was also obtained from the I-TASSER server. The structural PDB files were investigated to detect various problems, such as undesirable HETATOMs, attached ligands, missing atoms, and possible chain breaks. Then, all crystallographic waters were removed from the structures, and molecular hydrogens were added to optimize hydrogen bonds and minimized using the GROMACS 5.1.7 package before docking [25].
Small molecules: The study used the section of small-molecules in the DrugBank database (https://www.drugbank.ca/about) to obtain all FDA-approved drugs [26]. Non-unique structures removed during the process included compounds containing rare atoms and organometallic compounds. Eventually, 2471 compounds were selected.

Virtual screening
The study used the Pyrx tool for virtual screening [27,28]. The unit-docking cell was defined with Pockdrug server [29] and Uniprot databases. Input pdbqt files for Autodock Vina and minimization steps were generated by the Pyrx tool. Finally, to screen libraries of compounds against targets, the study implemented AutoDock Vina for drug discovery. Docking was done using a 92 Å × 84 Å × 98 Å binding site grid box for ACE2, a 120 Å × 110 Å × 117 Å binding site grid box for CD147 and a 140 Å × 122 Å × 111 Å binding site grid box for RdRp. A total of the top eight poses were retained from the docking run. The interaction established in two dimensions was illustrated in Biovia Discovery Studio, and for 3D visualization of drug/target complexes, Molegro Virtual Docker was used [30].

Molecular dynamics simulation
The results of virtual screening indicated that five drugs with ACE2, four drugs with RdRp, and seven drugs with CD147 achieved the most favorable free binding energy (ΔG < − 10 kcal/mol). For further evaluations, these drugs were chosen for MD simulation to calculate the number of H-bonds and free energy of interaction. MD simulations were directed using the GROMACS 5.1.4 package [31]. The GROMOS 54a7 force field was utilized for the complexes [32]. The study used the ATB server for the preparation of the coordinates and topology of ligands. The study applied appropriate amounts of chloride ions and sodium to all simulation boxes to neutralize the system. Periodic Boundary Condition (PBC) was applied along every simulation box axis, and the SP3 water model was also utilized for system solvation [33] in each simulation system. The LINCS algorithms constrained all covalent bonds. MD simulations were done through a short-range electrostatic interaction as well as a 1.2 nm distance cutoff for the van der Waals interaction. The Particle Mesh Ewald (PME) algorithm calculated the long-range electrostatic interaction. The steepest descent algorithm fulfilled the energy minimization of all systems, and then the NVT ensemble for 500 ps equilibrated all the systems. Then, the NPT ensemble progressively directed the equilibration of each system and the Nose-Hoover algorithm temperature [34,35] was preserved at a temperature of 310 K. During the NPT equilibration, the Parrinello-Rahman barostat [36] maintained the pressures at 1 bar. The MD simulation was completed for the complexes in 100 ns.

Analyses
The nonpolar and polar interactions between CD147, ACE2, and RdRp with drugs are explainable by binding free-energy calculation. By exercising the MM-PBSA method, the binding free-energy was calculated using the g_mmpbsa tool [37]. The total amount of binding free-energy (ΔG) is realized by adding up the nonpolar interaction free-energy (ΔG nonpolar ) and the polar interaction free-energy (ΔG polar ) that can be explained as follows: Where ΔG elec , ΔG ps , ΔG vdW , ΔG nps are respectively the electrostatic energy, polar solvation energy, van der Waals energy, nonpolar solvation energy.

Results
This study applied virtual screening FDA-approved drugs against RdRp, ACE2, and CD147.

Virtual screening
According to the results, five drugs with ACE2, four drugs with RdRp, and seven drugs with CD147 achieved the most favorable free binding energy (Docking score < − 10 kcal/mol) ( Table 1). The hydrogen bonding of docked molecules was calculated using Molegro Virtual Docker and Biovia Discovery Studio v.4.5. 2D (Supplementary data).

Molecular dynamics simulation
MD simulation for all complexes of the ACE2, RDRP, and CD147 with the top-selected drugs is performed for 100 ns. RMSD of alpha carbon atoms, RMSF of all amino acid residues, the number of hydrogen bonds, and free energy of interaction for the drug/protein complexes are investigated.

RMSD and RMSF
Figs. 1A, 2A and 3A depict the RMSD of ACE2, CD147, and RdRp Table 1 Docking results of FDA-approved drugs and ACE2, CD147 and RdRp with the best binding free energy (Docking score < -10) and the number of hydrogen bonds at 0, 20, 40, 60, 80 and 100 ns. The length of the Hydrogen bond ranges from 2.6 to 3.1 Å.

The number of hydrogen bonds and free energy of interaction
For further evaluations, the study utilized the MD simulation method to calculate the number of H-bonds and the amount of free energy of interaction. The number of H-bonds and free energy of interaction between ACE2, CD147, and RdRp and the drugs essential for stabilizing the complexes. The ACE2_vancomycin, ACE2_paritaprevir, CD147_estradiol, RdRp_vancomycin and RdRp-dactinomycin have the most hydrogen bonds over the 100 ns simulation time. Fig. 4 illustrated the number of H-bonds versus time at 310 K. The snapshots were generated at 0, 20, 40, 60, 80 and 100 ns, to investigate the stable hydrogen bonds between protein and ligands ( Table 1). The figures of these complexes were drawn using Molegro Virtual Docker (The figures for hydrogen bonds in these intervals are provided in supplementary data). Tables 2-4 illustrate the calculation of ΔG for the polar & nonpolar interactions between protein and drugs, indicating that ledipasvir with ACE2, estradiol benzoate with CD147, vancomycin with RDRP had the most favorable ΔG.

Discussion
This study assessed the potential of FDA-approved small molecules to disrupt the interaction of virus-host cell components and interfere with the viral proliferation mechanism using drug repurposing approaches. The SARS-CoV-2 ACE2 receptor is integral in cardiovascular and renal diseases, diabetes, and lung injury [38]. Furthermore, the observations indicated that the expression of ACE2 correlated with the infection severity in vitro [15], which makes it a suitable candidate for pharmaceutical studies. Accordingly, many cases have been reported in which the administration of soluble ACE2 is observed as a competitive method for inhibiting virus binding to the host cell [39]. MD simulation results indicated that Ledipasvir with ACE2 had the most favorable ΔG (Table 2). Also, the results of molecular docking in a relevant study showed that Ledipasvir could interfere with the binding of S protein to ACE2 [40]. The stable binding energy during 100 ns MD simulations (− 399.338 kcal/mol), together with RMSD and RMSF, confirm Ledipasvir inhibitory effects. Additionally, paritaprevir has good binding energy with both ACE 2 and RdRp targets. In this study, paritaprevir-ACE2 in 0-100 ns, Asp 206, have a key part in the hydrogen interactions during MD simulation (Table 1). Based on structural data, Asp 206 is located in the interaction surface with SARS-CoV-2 spike proteins [40]. Also, the complexes of ACE2 are stabilized by H-bonded interactions offered by residues that some of which are in the binding site [41]. The recent studies aiming at screening the FDA against SARS-CoV-2 proteins indicated that ledipasvir, paritaprevir, and simeprevir are promising medication candidates for COVID-19 treatment [42,43]. The clinical trials to evaluate the efficacy of ledipasvir in the treatment of COVID-19 are underway [44,45]. Ledipasvir and paritaprevir are direct-acting antiviral (DAA) medications used as part of combination therapy to treat chronic Hepatitis C, which is an infectious liver disease due to HCV infection. The RdRp enzyme is involved in viral genome replication and transcription of structural and peripheral small guided mRNAs (sgmRNAs) [46]. Therefore, it has been considered a primary target for many antiviral drugs, such as Remdesivir [5]. RdRp has conserved structural motifs like other polymerases, such as thumb (residues 816-932), palm (residues 860-815), and finger (residues 366-581) [47]. Vancomycin has the highest hydrogen bond with ACE2 and RdRp (Fig. 4). Also, MD simulation results indicated that vancomycin with RdRp had the most favorable ΔG (Table 4). In such a case, most of the residues involved in the interaction sites of the Vancomycin-RdRp complex, are located in the preferred conserved motifs and it was observed that Asn 459, Asp 390, Thr 393 and Lys 391 have strong hydrogen bonds in 100 ns simulation. Relevant studies show that vancomycin has the main part in infection treatment in severe coronavirus disease patients. Secondary bacterial infections associated with COVID-19 are caused by gram-positive bacteria, including methicillin-resistant Staphylococcus aureus (MRSA), methicillin-resistant, coagulase-negative staphylococci (MRCNS), and Enterococci species. These common nosocomial infections can cause ventilator-associated complications, like pneumonia, and are commonly treated with an antibiotic known as vancomycin. However, it has a narrow treatment window and some patients react differently to the drug leading to sub-optimal vancomycin concentrations in patients [48]. As you can see in Tables 2 and 4, sirolimus and dactinomycin also have suitable binding energy with ACE2 and RdRp. A recent study used a network-based drug repurposing sirolimus plus dactinomycin as treatment candidates for COVID-19 [49]. Nowadays, CD147 is regarded as an integral target in treating inflammatory diseases [50]. In 2016, a study investigated the role of small molecules in inhibiting CD147, which was used a Pharmacophore model derived from the structure of CD147. The results confirmed that the small molecule targeting CD147 was able to disrupt CD147 dimerization specifically and inhibit the    motility and invasion of hepatocellular carcinoma (HCC) cells [51]. Therefore, inhibition of CD147, with the help of small molecules, can play an effective role in treating some cancers and viral infections, including COVID-19. However, no small-molecule inhibitors for CD147 have been developed to date as an FDA-approved drug. A new study assessing the efficacy and safety of Meplazumab, a humanized anti-CD147 antibody, examined patients with SARS-CoV-2 pneumonia. Meplazumab efficiently improved the recovery of patients with COVID-19 pneumonia with a favorable safety profile [20]. These results support the inhibition of CD147 as a treatment host-targeted strategy for COVID-19 pneumonia. Besides, MD simulation results indicated that estradiol benzoate with CD147 had the most favorable ΔG. Estradiol Benzoate is a pro-drug ester of Estradiol, a naturally-occurring hormone that endogenously circulates within the human body. Estradiol is the most potent form of all mammalian estrogenic steroids and acts as the major female sex hormone. Recent results show that the risk of severe complications of COVID-19 is lower in women than in men. One of the reasons suggested by researchers is the high levels of estradiol in women (52,53). In animal experiments, estrogen therapy suppressed inflammatory reactions and cured COVID-19 infection (54). Based on the results of this study, olaparib and irinotecan had a suitable free binding energy with CD147 (Table 3). A relevant study demonstrated that a combination of irinotecan (topoisomerase I inhibitor) and etoposide (a topoisomerase II inhibitor) can potentially inhibit cytokine storms in COVID-19 [52]. A type of clinical study using a drug repositioning strategy indicated the inhibitory effects of olaparib and mefuparib, as two PARP1 inhibitors, on COVID-19 [53]. Abemaciclib is an antitumor agent and dual inhibitor of cyclin-dependent kinases 4 (CDK4) and 6 (CDK6), that are involved in the cell cycle and promotion of cancer cell growth in case of unregulated activity.

Conclusion
The inhibition of CD147 and ACE2, as two main receptors of SARS-CoV-2, can prevent the entering of the virus into the host cells.  Besides, the inhibition of RdRp, as the main enzyme for viral replication, is effective in fighting the COVID-19. Accordingly, this study aimed to use drug repurposing by virtual screening to identify inhibitors for CD147, ACE2, and RdRp. The results of this study showed that five drugs with ACE2, four drugs with RdRp, and seven drugs with CD147 achieved the most favorable free binding energy (Docking score < − 10). These drugs were selected for MD simulation studies, the results of which indicated that ledipasvir with ACE2, estradiol benzoate with CD147, and vancomycin with RdRp had the most favorable ΔG. Drugs that can inhibit major virus receptors (ACE2 and CD147) were found to be effective in the early stages of viral infection, and inhibition of RdRp can also prevent disease progression. It was also shown for the first time in this study that, paritaprevir and vancomycin have good binding energy with both targets (ACE2 and RdRp). These drugs can be suitable candidates for further investigation as possible treatments of COVID-19 infection.

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

Author contributions
Soodeh Mahdian collected and interpreted the data and prepared the manuscript. Yunes Panahi revised the manuscript and contributed to its design. Mahboobeh Zarrabi contributed to the design and revision of the manuscript, and Somayyeh Dabbagh contributed to the authorship of the manuscript.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.