Numerous medicinal plants and their phytocompounds have demonstrated their antiviral properties against a large group of viruses. Consequently, the phytocompounds of Diplocyclos palmatus leaf extract were subjected to docking analysis in the current study. In previous studies, the tropical medicinal plant of D. palmatus has been reported for its anti-biofilm, anti-infection, and anti-photoaging activity using Caenorhabditis elegans model (Alexpandi et al., 2019). The list of quinoline-drugs (total 113) was retrieved from DrugBank database (https://www.drugbank.ca/categories/DBCAT000788). The canonical SMILES of the compounds was retrieved from the PubChem database. The canonical SMILES of quinoline,1,2,3,4-tetrahydro-1-[(2-phenylcyclopropyl)sulfonyl]-trans-(8CI) was retrieved from the Guidechem database (https://www.guidechem.com/reference/dic-395649.html). Then, the PDB-format 3D-structure of compounds was downloaded from the Openbabel online server http://www.cheminfo.org/Chemistry/Cheminformatics/FormatConverter/index.html.

The 3D crystal protein-structures of SARS-CoV-2 3CLpro (PDB ID: 6LU7) (Hall and Ji, 2020), SARS-CoV-2 spike protein-ACE-2 receptor-binding domain (RBD) (PDB ID: 6M17) (Wu et al., 2020), and human ACE2 (PDB ID: 1R4L) (Joshi et al., 2020) were obtained from the RCSB PDB database (http://www.rcsb.org/pdb). The 3D crystal structures of SARS-CoV-2 RdRp generated through homology modeling using ICM 3.7.3 modeling software was gifted by Prof. Hua Li, Hubei Key Laboratory of Natural Medicinal Chemistry and Resource Evaluation, School of Pharmacy, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China (Wu et al., 2020). The energy minimization of targeted protein structures was performed using the YASARA server. The protein preparation was done with AutoDock Tools Version 1.5.6.

The virtual screening of best scoring compounds was performed using the iGEMDOCK with blind-mode docking. The iGEMDOCK tool is a graphical-automatic drug design system mainly used for structure-based virtual screening of drug molecules (Hsu et al., 2011). After the selection of best binding compounds, the interaction on the active domains of therapeutic targets (3CLpro, RdRp, and Spike-ACE2 complex) of selected compounds were analyzed using the AutoDock Vina tool. It is an open-source docking software, which extensively improves the average accuracy of the binding mode predictions of compounds better than other docking tools (Trott and Olson, 2010). It implements a competent optimization algorithm for estimating the affinity of protein-ligand interactions and predicting the plausible binding modes of compounds (Goodsell et al., 1996). Then, the ligand-protein interactions were visualized by Maestro 10 (Schrödinger) (Balasubramaniam et al., 2019). In the present study, to compare the selected quinolines with already reported anti-SARS-CoV-2 drugs, lopinavir (Hall and Ji, 2020) and remdesivir triphosphate (Gordon et al., 2020) were selected as positive inhibitors of SARS-CoV-2 for in silico analysis in the present study. Further, the selected quinolines and remdesivir triphosphate were compared with the parental nucleotides (NTPs) of SARS-CoV-2 RdRp for understanding the inhibition mode of viral replication.

The drug likeliness and bioactivity of quinolines were analyzed using the Molinspiration server (http://www.molinspiration.com). Molinspiration tool is a cheminformatics software that provides molecular properties as well as bioactivity prediction of compounds (Mabkhot et al., 2016). In the Molinspiration-based drug-likeness analysis, there are two important factors, including the lipophilicity level (log P) and polar surface area (PSA) directly associated with the pharmacokinetic properties (PK) of the compounds (Beetge et al., 2000). In the Molinspiration-based bioactivity analysis, the calculation of the bioactivity score of compounds toward GPCR ligands, ion channel modulators, kinase inhibitors, nuclear receptor ligands, protease inhibitors, and other enzyme targets were analyzed by sophisticated Bayesian statistics (Mabkhot et al., 2016). This was done as the protein families, such as G protein-coupled receptors (GPCR), ion channels, kinases, nuclear hormone receptors, proteases, and other enzymes (RdRp), are the major drug targets of most of the drugs (Hauser et al., 2017).

The PK properties, such as Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET), of quinolines were predicted using the admerSAR v2.0 server (http://lmmd.ecust.edu.cn/admetsar2/). The admerSAR server is an open-source computational tool for prediction of ADMET properties of compounds, which makes it a practical platform for drug discovery and other pharmacological research (Guan et al., 2019). In the ADMET analysis, the absorption (A) of good drugs depends on factors such as membrane permeability [designated by colon cancer cell line (Caco-2)], human intestinal absorption (HIA), and the status of either P-glycoprotein substrate or inhibitor. The distribution (D) of drugs mainly depends on the ability to cross the blood-brain barrier (BBB). The metabolism (M) of drugs is calculated by the CYP, MATE1, and OATP1B1-OATP1B3 models. Excretion (E) of the drugs is estimated based on the renal OCT substrate. Then, the toxicity (T) of the drugs is predicted on the Human Ether-a-go-go- related gene inhibition, carcinogenic status, mutagenic status, and acute oral toxicity (Shen et al., 2012).

In Table S1, the iGEMDOCK-based virtual screening result reveals that among the 17 phytocompounds, the novel quinoline (NQ) identified from D. palmatus, quinoline,1,2,3,4-tetrahydro-1-[(2-phenylcyclopropyl)sulfonyl]-trans-(8CI), has a lower binding energy (-6.8 Kcal/mol) with SARS-CoV-2 3CLpro, compared to other phytocompounds. As shown in Figures 1D,I, the NQ builds five hydrophobic interactions (Cys145, Met49, Met165, Phe140, and Leu141), 8 polar interactions (His41, His164, Gln192, Gln189, His63, His172, Ser144, and Asn142), two negative-charged interactions (Asp187 and Glu166), and one unspecified residue interaction (Arg188). As shown in Figure 1G, the NQ forms a hydrogen bonding interaction with Glu166 (2.044 Å distance). Importantly, the NQ has effectively interacted with the active site of catalytic-dyad (Cys145 and His41) of the SARS-CoV-2 3CLpro (Figures 1B,F), and hence we assumed that the NQ is able to hinder the protease activity of SARS-CoV-2 like lopinavir (−6.6 Kcal/mol) (Figure S1), which is a protease-inhibitor based anti-SARS-CoV-2 drug reported for COVID-19 (Choy et al., 2020).

Based on the virtual screening results of SARS-CoV-2 3CLpro, rilapladib, saquinavir, oxolinic acid, elvitegravir, batefenterol, sitafloxacin, CP-609754, GSK-256066, alatrofloxacin, and quarfloxin were predicted to be the best compounds from the commercial quinoline-based drugs (Figure 2 and Table S2). Out of these compounds, only saquinavir (−8.5 Kcal/mol) was found to interact with the active catalytic-domain (Cys145 and His41) of the SARS-CoV-2 2CLpro (Figures 1H,C). As shown in Figures 1E,J, saquinavir forms seven hydrophobic interactions (Cys145, Met165, Met49, Leu27, Leu167, Leu141, and Phe140), 12 polar interactions (His41, Asn142, His163, Ser144, Ser46, Thr25, Thr26, Thr24, Gln192, Gln189, Thr190, and His164), one negative-charged interaction (Glu166), and one glycine interaction (Gly143) with SARS-CoV-2 3CLpro. Therefore, for the development of strong interactions at the CysHis catalytic dyad, saquinavir was predicted to obstruct the 3CLpro activity.

As shown in Figure 3 and Table S3, amongst the tested-quinolines, elvitegravir, oxolinic acid, saquinavir, garenoxacin, rilapladib, pelitinib, difloxacin, batefenterol, danofloxacin, and LGD-2226 were shown to be the best-docked compounds against SARS-CoV-2 RdRp. Figures 4A,C reveals that elvitegravir and oxolinic acid have the same binding energy (−7.1 Kcal/mol) and same binding site at SARS-CoV-2 RdRp, which is almost similar to natural nucleotides such as ATP (−7.6 Kcal/mol), UTP (−7.1 Kcal/mol), GTP (−7.7 Kcal/mol), and CTP (−7.1 Kcal/mol) (Figure S2). As shown in Figures 4B,I, elvitegravir forms five hydrophobic interactions (Tyr455, Ala554, Tyr619, Pro620, and Cys622), seven polar interactions (Thr556, Thr680, Ser681, Ser682, Thr687, Asn691, and Ser759), four negative-charged interactions (Asp452, Asp618, Asp623, and Asp760), and five unspecified residue interactions (Lys545, Arg553, Arg555, Lys621, and Lys798) with SARS-CoV-2 RdRp. Notably, elvitegravir builds a hydrogen bond interaction with Lys621 residue (1.931 Å distance), as shown in Figure 4F. Similarly, it formed five hydrophobic interactions (Tyr456, Met542, Ala554, Val557, and Ala558), six polar interactions (Thr556, Thr680, Ser681, Ser682, Thr687, and Asn691), two negative-charged interactions (Asp452 and Asp623), and five unspecified residue interactions (Lys545, Arg553, Arg555, Arg624, and Lys676) (Figures 4D,J). Subsequently, oxolinic acid forms three hydrogen bonding interactions with Thy456 (1.224Å), Ser682 (1.815Å), and Arg624 (2.904) residues of the SARS-CoV-2 RdRp (Figure 4H). As shown in Figures 4E,G, both elvitegravir and oxolinic acid have the ability to bind with the NTP binding channel (a set of hydrophilic residues such as Lys545, Arg553, and Arg555) of the SARS-CoV-2 RdRp, similar to remdesivir triphosphate (−7.8 Kcal/mol) (Figure S3), which is a RdRp-inhibitor based anti-SARS-CoV-2 agent reported for COVID-19 (Gao et al., 2020; Gordon et al., 2020). As shown in Figure S4, due to the same binding sites, we expect these two quinolines (elvitegravir and oxolinic acid) can readily interact with the NTP binding channel of SARS-CoV-2 RdRp, more quickly than the parental-nucleotides (NTPs) such as ATP, UTP, GTP, and CTP. Hence, these two quinoline drugs are anticipated to arrest the viral replication of SARS-CoV-2, as is seen with remdesivir triphosphate.

In Figure S5 and Figure 6A, the crystal structure of spike protein-ACE2 (PDB: 6M17) revealed that the aminoacid residues of ACE2, including Gln24, Thr27, Asp30, Lys31, His34, Glu35, Glu37, Asp38, Tyr41, Gln42, Met82, Lys353, Gly354, Asp355, and Arg357, were recognized as the Spike protein receptor-binding domain (RBD) or entry receptor site to invade the target cells (Benítez-Cardoza and Vique-Sánchez, 2020). Based on the virtual screening results of ACE2 protein, CP-609754, saquinavir, rilapladib, quarfloxin, batefenterol, oxolinic acid, alatrofloxacin, dovitinib, GSK-256066, and rebamipide were predicted as the best compounds, exhibiting a high binding affinity to ACE2 with low energy (Figure 5 and Table S4). However, these quinolines were not predicted to interact with the RBD of the Spike-ACE2 complex (Figure 5). The only compound that could target the RBD interface between Spike and ACE2 was rilapladib, as shown in Figure 6B. Rilapladib was predicted to be positioned on the central shallow pit of the RBD of the Spike-ACE2 complex with strong interactions (Figure 6D). The aminoacid residues of the Spike-RBD-ACE2 complex that interact with rilapladib were His34, Glu35, Glu37, Asp38, and Leu39, as shown in Figure 6C. As a result of superimposing the Spike-RBD-ACE2 complex to the rilapladib-RBD-ACE2 complex, an individual overlap of rilapladib with the interface of ACE2 was observed, signifying that rilapladib may possibly interrupt the interaction of the Spike-RBD-ACE2 complex.

The adequacy of therapeutic drugs mainly depends on the molecular property and bioactivity of the compounds (Shen et al., 2012). To predict the drug-likeness and bioactivity of the selected quinolines, the in silico molecular property assessment was performed using the Molinspiration tool. This tool measures the _milogP value (Octanol-water partition coefficient logP) and TPSA (Topological polar surface area) values of the compounds using Bayesian statistics. The result shows that the _milogP value of quinoline,1,2,3,4-tetrahydro-1-[(2-phenylcyclopropyl)sulfonyl]-trans-(8CI) (3.63), saquinavir (4.26), elvitegravir (3.58), and oxolinic acid (0.68) were predicted as having ideal lipophilicity (logP < 5) (Han et al., 2019); rilapladib (7.33) was predicted as having poor lipophilicity (logP > 5) in the aspect of absorption and permeation (Figure S6). The TPSA of the quinoline,1,2,3,4-tetrahydro-1-[(2-phenylcyclopropyl)sulfonyl]-trans-(8CI) (37.38), elvitegravir (88.77), oxolinic acid (77.77), and rilapladib (54.79) were < 100, showing that these compounds had superior oral-absorption or membrane permeability than saquinavir (166.75), lopinavir (119.99), and remdesivir triphosphate (289.53) (Bakht et al., 2010). On the other hand, the majority of drug targets of existing drugs are in one of the following protein families: G protein-coupled receptors (GPCR), ion channels, kinases, nuclear hormone receptors, proteases, and other enzymes. In Figure S7, the in silico bioactivity prediction analysis also divulges that the NQ and saquinavir were predicted as protease inhibitors. Subsequently, elvitegravir and oxolinic acid were predicted as enzyme inhibitors, which means that these are able to inhibit other enzymes, including RdRp enyzme, except G protein-coupled receptors (GPCR), ion channels, kinases, nuclear hormone receptors, and proteases. This is because the protein families are the major drug targets of most of the drugs (Hauser et al., 2017). These data support the outcome of the predicted in-silico activity of selected quinolines against SARS-CoV-2.

The half-life (t_1/2) of drugs is a valuable pharmacokinetic factor as it provides an exact indication of the duration of time that the effect of the drug continues in an individual. The period of action of a drug is called its half-life. The distribution half-life (t_1/2a) is the time required to divide the plasma concentration by two after reaching pseudo-equilibrium, and not the time needed to eliminate half the administered dose. The elimination half-life (t_1/2b) of drugs is defined as the time required for the concentration of the drug to reach half of its original value in the plasma or the total amount in the body. As mentioned in Table 1, the elimination half-life (t_1/2b) of saquinavir (670.84 g/mol) is 7–12 h and their elimination half-life is 9–15 h (Taylor et al., 2001). The distribution half-life (t_1/2a) of elvitegravir is calculated at 8.7 h, and the elimination half-life (t_1/2b) as 12.9 h (Cada et al., 2013). Oxolinic acid (261.23 g/mol) has 1.3 and 84 h as its distribution half-life and elimination half-life, respectively (Samuelsen et al., 2003). Unfortunately, the half-life of rilapladib (735.81 g/mol) as well as quinoline,1,2,3,4-tetrahydro-1-[(2-phenylcyclopropyl)sulfonyl]-, trans-(8CI) (313.41 g/mol) are not available in the database. On the other hand, the elimination half-life of SARS-CoV-2 protease inhibitor lopinavir (628.8 g/mol) measured at 6.9 ± 2.2 h. Similarly, the reported anti-COVID-19 drug remdesivir has a very short half-life (0.39 h) and hence the human esterases hastily converted the remdesivir into nucleoside triphosphate metabolite (remdesivir triphosphate). However, the produced remdesivir triphosphate has a longer half-life of ~20 h in humans, in which the metabolite acts as an NTP analog and slows down the viral replication of SARS-CoV-2 (Eastman et al., 2020). Due to the long half-life of the NTP analog of remdesivir triphosphate, only one dose is required for daily administration for treating COVID-19. The data obtained revealed that the selected quinolines are significantly long in terms of drug half-life, in which the quinolines treatment will be successful in the drug repurposing against COVID-19.

In Table 2, the absorption (A) analysis reveals that the NQ and oxolinic acid were predicted to have high Caco-2 permeability. A HIA of <30% is classified to be poorly absorbed. The result reveals that all the quinolines were predicted to be highly absorbed by the human intestine. P-glycoprotein, a member of ATP-binding trans-membrane glycoprotein, excretes incoming drugs or other chemicals from the cells (Äänismaa and Seelig, 2007). The results revealed that the NQ and oxolinic acid are non-substrates or non-inhibitors for P-glycoprotein and other compounds are substrates/inhibitors for P-glycoprotein. In the distribution (D) analysis, the BBB permeability, logBBB >0.3, is thought to cross the BBB easily (Han et al., 2019). Here, all the quinolines were predicted to be BBB+ and they can easily cross the BBB. In the metabolism (M) analysis, the two main sub-types of cytochrome P450s (CYP) are CYP2D6 and CYP3A4, which are essential enzyme-systems for drug metabolism in the liver (Sams et al., 2004). The results showed that all the quinolines were non-inhibitors to CYP2D6; only NQ and saquinavir were substrates for CYP2D6 and CYP3A4. This data suggested that these quinolines may possibly metabolize in the liver. OATP1B1 and OATP1B3 are transporters expressed on the sinusoidal-membrane of hepatocytes, which interact with therapeutic-drugs as their substrates or inhibitors, and cause clinically relevant drug-drug interactions (Shitara, 2010). The result found that all the selected quinolines were predicted as inhibitors for OATP1B1-OATP1B3. The excretion (E) of drugs is associated with their hydrophilicity and molecular weight. In the kidney, organic cation transporters (OCTs) and multidrug and toxin extrusion proteins (MATEs) are the foremost transporters for the clearance of cationic drugs into the urine (Motohashi and Inui, 2013). The results showed that the quinolines were predicted as non-inhibitors to MATE-1, OCT-1, and OCT-2, which indicates the safety elimination profile of the quinolines. In the toxicity (T) analysis, the result showed that all quinolines were non-carcinogen and non-eye corrosive. In conclusion, the predicted results indicate that the ADMET properties of the quinolines are almost similar to the reported anti-SARS-CoV-2 drugs (lopinavir and remdesivir triphosphate), which promotes repurposing of these quinoline drugs for the treatment of COVID-19.