Quinoline and Quinazoline Alkaloids against COVID-19: An In Silico Multitarget Approach

,e recent outbreak of the highly contagious coronavirus disease 2019 (COVID-19) caused by the novel coronavirus SARS-CoV-2 has created a global health crisis with socioeconomic impacts. Although, recently, vaccines have been approved for the prevention of COVID-19, there is still an urgent need for the discovery of more efficacious and safer drugs especially from natural sources. In this study, a number of quinoline and quinazoline alkaloids with antiviral and/or antimalarial activity were virtually screened against three potential targets for the development of drugs against COVID-19. Among seventy-one tested compounds, twentythree were selected for molecular docking based on their pharmacokinetic and toxicity profiles. ,e results identified a number of potential inhibitors. ,ree of them, namely, norquinadoline A, deoxytryptoquivaline, and deoxynortryptoquivaline, showed strong binding to the three targets, SARS-CoV-2main protease, spike glycoprotein, and human angiotensin-converting enzyme 2. ,ese alkaloids therefore have promise for being further investigated as possible multitarget drugs against COVID-19.


Introduction
In December 2019, a global health concern had been raised by the outbreak of coronavirus disease 2019 (COVID- 19) which is caused by SARS-CoV-2. It belongs to the singlestranded positive-sense RNA coronavirus family [1]. Its genome consists of different regions including a spike protein (S) gene, an envelope protein (E) gene, a membrane protein (M) gene, and a nucleocapsid protein (N) gene [2]. e sequence of SARS-CoV-2 showed more than 50% identity to SARS-CoV and MERS-CoV and closer relation to bat-SL-CoVZC45 [3,4]. SARS-CoV produces several functional proteins while its main protease is emerging as a promising therapeutic target as it is responsible for the processing of translated polyprotein. us, inhibition of the main protease was confirmed to affect the viral replication [5]. Its high sequence conservation with SARS-CoV main protease suggests the effectiveness of HIV-1 protease inhibitors such as nelfinavir against it [6]. SARS-CoV-2 shares the mode of transmission with SARS-CoV and MERS-CoV, after which it binds to ACE2 on the surface of host cells via the receptor-binding domain (RBD) in its spike proteins [7,8]. Blocking the ACE2 and RBD interaction by antibodies and inhibitors would be an effective way to stop the virus infection [9].
Symptoms of COVID-19 greatly resembled viral pneumonia ranging from mild to more severe eventually ending in several organ malfunction [6]. Discovery of efficacious drugs for this deadly disease could be achieved by one of the three options: testing the existing antiviral drugs which are already used to treat viral infections, secondly, screening of different existing drugs, and finally, discovery of new specific drugs based on the individual coronavirus genome [10]. Chloroquine, HIV protease inhibitors, ACE-2 inhibitors, and many other drugs were predicted to be COVID-19 drug candidates [11,12].
Medicinal plants have pulled in noteworthy consideration since they incorporate bioactive components that could be utilized to design medications against a few ailments with insignificant side effects [13]. e medicinal active compounds of plants have been widely used to treat microbial diseases being antifungal [14], antibacterial [15], and antiviral [16]. Quinoline and quinazoline alkaloids are N-based heterocyclic compounds with a wide range of activities, and many of them have been reported to have antiviral effects [17,18]. It is therefore imperative to test their effectiveness in COVID-19.

2.
1. Data Sources. Literature was surveyed for various phytochemicals with known antiviral and/or antimalarial activities. 71 bioactive alkaloids (quinoline and quinazoline) from different natural sources were selected to be investigated for their activity against COVID-19 virus [17,18].

Target Selection and Preparation.
COVID-19 main protease domain (PDB ID: 6LU7), COVID-19 spike glycoprotein (PDB ID: 6LZG), and human angiotensin-converting enzyme 2 (ACE2; PDB ID: IR42) were selected as the target proteins. eir 3D structures were retrieved from the RCSB PDB (protein data bank) database. Swiss PDB viewer V.4.1.0 software [39] was used for structure optimization, and the active sites were verified from the UniProt databases.

Ligands' Preparation.
e 3D SDF structures of some compounds were downloaded from the PubChem database, and 2D structures of the rest were illustrated using ChemSketch. e files were then converted into the PDB format with the aid of Open Babel, energy minimized, and converted into the PDBQT format using the graphical user interface version of PyRx virtual screening tool-python prescription 0.8.

Compound Screening Using PyRx Program.
Virtual screening was performed using PyRx software and Vina wizard as the engine for docking [40,41]. e amino acids' residues in the active site of the protein were selected. e results less than 1.0 Åin positional root-mean-square deviation (RMSD) were considered ideal and clustered together for finding the favorable binding. e highest binding energy (most negative) was considered as the ligand with maximum binding affinity.

In Silico ADME Properties.
e pharmacokinetic (ADME) properties of the selected compounds were predicted using the SwissADME web tool (http://www. swissadme.ch/) [42]. e compounds' structure was retrieved from databases using the import tool on the input zone of the SwissADME submission page and converted into the SMILES format, and then calculations were run. In some cases, the structures were converted into the SMILES format using the Online SMILES Translator (available at https:// cactus.nci.nih.gov/translate/).

In Silico Toxicity Risks' Assessment and Drug Likeliness.
OSIRIS Property Explorer open-source program (http:// www.organicchemistry.org/prog/peo/) [43] was used to evaluate the toxicity risks of the compounds retrieved from PubChem.

Molecular
Docking. Molecular docking was performed using AutoDock 4.0 software, based on the Lamarckian genetic algorithm, which combines energy evaluation through grids of affinity potential to find the suitable binding position for a ligand on a given protein [44,45]. Polar hydrogen atoms were added to the protein targets, and Kollman united atomic charges were computed. e targets' grid map was calculated and set to 60 × 60 × 60 points with the grid spacing of 0.375 Ǻ. e grid box was then allocated properly in the target to include the active residue in the center. e genetic algorithm and its run were set to 10 as the docking algorithms were set on default. Results were finally retrieved as binding energies.

Analysis and Visualization.
e resultant docking files with poses showing the lowest binding energies were visualized using DS Visualizer Client (Windows 64 bit) (267 MB).

Toxicity Risks and Drug Likeliness.
Twenty-seven compounds are predicted to have no risk of mutagenic, tumorigenic, irritant, or reproductive effects and have drug score >0.5 as shown in Table 1.
ese were selected for subsequent steps.

Physicochemical Parameters.
e physicochemical properties of the selected twenty-seven compounds were predicted by the SwissADME web tool. Ten compounds were found to be soluble, and seventeen were predicted to be moderately soluble as shown in Table 2.
e physicochemical properties influence the pharmacokinetic properties which in turn determines the ultimate biological effect.

In Silico ADME Properties.
e selected compounds were subjected to pharmacokinetic analysis using the SwissADME tool. e obtained results are summarized in Table 3.

Analysis and Visualization.
Molecular docking using AutoDock 4.0 was conducted for most promising compounds with DS values ≥0.5, no risk of toxicity, and considerable solubility ( Figure 1 and Table 4).
Visualization was then performed for five compounds, against each target, showing the best docking score (lowest binding energies). Figures 2-4 show the intermolecular interaction and hydrogen bonding between the ligands and proteins at the active sites. Important active site residues in each of the three targets are involved in ligands' binding (Table 5).

Discussion
SARS-CoV-2 is an RNA virus which tends to mutate more commonly than the DNA viruses. It has killed thousands of people around the globe with an increase in death rate every single day. In order to control and treat such pandemics, drug repurposing is highly recommended as it involves the use of derisked compounds with potentially lower development cost and shorter timelines. Moreover, discovery of multitarget drugs can be accelerated by in silico screening of potential ligands on a number of promising drug targets. In our present study, natural quinoline and quinazoline alkaloids were screened in silico against three macromolecules selected as potential therapeutic targets for treatment of COVID-19 infection.

Compounds' Screening.
e obtained results (Supplementary Table 1) revealed that most of the compounds have good affinity with binding energies ranging from −5.0 to −10.8 kcal/mol. erefore, in order to select the most promising and potential candidates, evaluation of toxicity risks and drug likeliness was performed.

4.2.
In Silico ADME Properties. Results predicted that twenty-three compounds have promise being drugs or drug leads since they showed no risk of mutagenic, tumorigenic,  TE  IE  RE  DL  DS  1 Uranidine    eir physicochemical properties indicated that they are less likely to have solubility problems which could lead to poor bioavailability as well as formulation problems [46]. Indeed, this acceptable profile of solubility has been translated in the ADME profile as solubility can also influence it. All the selected compounds were predicted to be highly absorbed from the gastrointestinal tract. Twelve compounds were predicted to permeate the BBB, of which four were also potential substrates for P-gp which could pump them out of the central nervous system (CNS) preventing any serious effects.
Many drugs are metabolized by cytochrome P450 (CYP) enzymes [47]; therefore, assessment of the effect of potential drug candidates on these enzymes is essential. Most of the selected compounds were predicted to inhibit one or two of the five CYP enzymes tested in this study. However, cuspareine, galipinine, acronydine, veprisine, 2-methoxyrutaecarpine, and 2-methoxy-13-methylrutaecarpine were predicted to inhibit four CYP enzymes indicating possible adverse effects and drug-drug interactions [48,49].

Molecular Docking Analysis and Visualization.
e most potential 23 compounds with DS score ≥0.5 and no risk of toxicity were redocked against the three targets using AutoDock 4.0. Table 4 shows their different binding energies with 6LU7 (ranging between −6.5 and −9.64 kcal/mol), IR42 (ranging between −6.2 and −11.01 kcal/mol), and 6LZG (ranging between −5.86 and −9.53 kcal/mol). Compared to nelfinavir, a known protease inhibitor, all docked compounds show lower binding energy to the main protease of SARS-CoV-2 indicating that they have the potential to inhibit viral replication. It is important to note that residues Glu166, His163, His164, Phe140, Cys145, Ser144, and Gln189 are involved in binding of nelfinavir and the selected alkaloids. ese residues were recently reported in an inhibitor-bound SARS-CoV-2 protease [50]. e receptor-binding domain of SARS-CoV-2 spike glycoprotein is involved in the virus entry to host cells by binding to ACE2. Our results predicted that the tested alkaloids bind to the receptor-binding domain of SARS-CoV-2 spike glycoprotein which could therefore inhibit its binding to its receptor. Notably, important amino acids (Gln493 and Glu484) are responsible for the higher binding of SARS-CoV-2 to human ACE2 compared to SARS-CoV as confirmed by mutagenesis studies. Moreover, they are predicted to bind to human ACE2 in its active site which is not the binding site for virus spike glycoprotein. However, conformational changes in the three-dimensional structure of the protein were reported which affect residues which are important for the binding of the spike protein [51]. In addition, a novel ACE2 inhibitor discovered by molecular docking has also shown effectiveness in blocking SARS-CoV spike protein-mediated cell fusion [9]. Given the high similarity of the SARS-CoV-2 and SARS-CoV binding to the host receptor (ACE2), the same could happen with SARS-CoV-2. Eighteen of the alkaloids docked within the active  Journal of Chemistry site of human ACE2 exhibit higher binding affinities than lisinopril, which is a prototype the ACE2 inhibitor, with binding energies ranging from 7.31 to 11.01 kcal/mol compared to −7.2 kcal/mol for lisinopril. Interestingly, three alkaloids, namely, norquinadoline A, deoxytryptoquivaline, and deoxynortryptoquivaline, are predicted to strongly bind the three proteins studied here. ese alkaloids are biologically active secondary metabolites isolated from the mangrove-derived fungus Cladosporium sp. PJX-41 and showed anti-influenza A (H1N1) activity [52]. A recent study in which a number of fungal secondary metabolites (including a number of quinoline and quinazoline alkaloids) were virtually screened against five targets of SARS-CoV-2 has also identified norquinadoline A and deoxynortryptoquivaline as potential inhibitors to main protease [38]. e reported binding energy of  erefore, earlier prediction is essential. e three alkaloids were predicted to have good pharmacokinetic and safety profiles indicating their promise of being taken forward to be tested in vitro and in vivo to prove their effectiveness against SARS-CoV-2.

Conclusion
Computer-based drug discovery usually involves the search for small-molecule leads with attractive pharmacokinetic and toxicity profiles. ese molecules are then tested in vitro and in vivo to confirm their therapeutic potential. Since natural products have played and continue to play a great role in efficacious drug discovery, this study was conducted in order to investigate the potential of selected natural alkaloids against the highly contagious virus SARS-CoV-2 which is responsible for the current pandemic. e study identifies twenty-three alkaloids as possible candidates with the potential to inhibit targets that prevent virus entry and/ or preventing their replication. Of these, norquinadoline A, deoxytryptoquivaline, and deoxynortryptoquivaline inhibit three protein targets and exhibit good pharmacokinetic and safety profiles, suggesting them as possible natural multitarget drugs against COVID-19. erefore, these three alkaloids could be starting points for future drug development

Data Availability
e data used to support these findings are available within the manuscript and its supplementary file.

Conflicts of Interest
e authors declare no conflicts of interest.

Authors' Contributions
Esraa conducted the pharmacokinetic and toxicity analysis. Esraa, Shaza, and Mona performed screening and docking and wrote the paper draft. Wadah and Hassan wrote the introduction. Esraa and Ramzi revised and edited the final draft.