Screening and Druggability Analysis of Marine Active Metabolites against SARS-CoV-2: An Integrative Computational Approach

: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections have triggered a recent pandemic of respiratory disease and affected almost every country all over the world. A large amount of natural bioactive compounds are under clinical investigation for various diseases. In particular, marine natural compounds are gaining more attention in the new drug development process. The present study aimed to identify potential marine-derived inhibitors against the target proteins of COVID-19 using a computational approach. Currently, 16 marine clinical-level compounds were selected for computational screening against the 4 SARS-CoV-2 main proteases. Computational screening resulted from the best drug candidates for each target based on the binding afﬁnity scores and amino acid interactions. Among these, ﬁve marine-derived compounds, namely, chrysophaentin A ( − 6.6 kcal/mol), geodisterol sulfates ( − 6.6 kcal/mol), hymenidin ( − 6.4 kcal/mol), plinabulin ( − 6.4 kcal/mol), and tetrodotoxin ( − 6.3 kcal/mol) expressed minimized binding energy and molecular interactions, such as covalent and hydrophobic interactions, with the SARS CoV-2 main protease. Using molecular dynamic studies, the root-mean-square deviation (RMSD), root-mean-square ﬂuctu-ation (RMSF), radius of gyration (ROG), and hydrogen bond (H-Bond) values were calculated for the SARS-CoV-2 main protease with a hymenidin docked complex. Additionally, in silico drug-likeness and pharmacokinetic property assessments of the compounds demonstrated favorable druggability. These results suggest that marine natural compounds are capable of ﬁghting SARS-CoV-2. Further in vitro and in vivo studies need to be carried out to conﬁrm their inhibitory potential.


Introduction
COVID-19 is one of the important epidemic diseases caused by the coronavirus (SARS-CoV-2) in the current century. It has spread to more than 210 countries, and more

Software/Servers Used
The marine preclinical-and clinical-level bioactive compounds are listed using PubMed database. All marine bioactive compound structures were collected from PubChem database. Protein structures were obtained from RCSB protein data bank. DruLiTo 1.0.0 software was used to analyze the physiochemical properties of marine bioactive compounds. pkCSM, a pharmacokinetics online server, was used to predict the ADMET properties, Open Babel v.2.3 and PyRx 0.8 were used for molecular docking studies, and the post-docking studies were carried out using Discovery studio 2017R2.

Target Protein Preparation
The 3D crystal structures of SARS-CoV-2 main protease (6LU7; 2.16 Å) were collected from the protein data bank (https://www.rcsb.org/ accessed on 5 February 2020). The unnecessary molecules, such as ions, inhibitors, ligands, heteroatoms, and water molecules, were removed from the COVID-19 target protein structure using BIOVIA Discovery Studio. The target protein structures were loaded in PyRx version 0.8 and converted into PDBQT format [22].

Ligand Preparation
The marine active compounds (preclinical and clinical levels) were used for ligand preparation ( Table 1). The 3D structures of the selected marine bioactive compounds were collected from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/ accessed on 5 February 2020) in SDF file format. Initially, the ligand files were loaded, and energy was minimized using Open Babel (MMFF94 method). After the energy minimization process, the ligand files were converted into PDBQT format. These energy-minimized ligands were subjected to further docking analysis [23].

Molecular Docking Studies
Following ligand and protein preparation, molecular docking was performed based on the grid box approach (X = −26.28, Y = 12.60, Z = 58.97) using AutoDock Vina inbuild PyRx version 0. 8 (Dallakyan and Olson, 2015). The docking analysis of COVID-19 proteins with bioactive drug candidates was evaluated using the binding affinities (kcal/mol). After docking analysis, the docked complex files were subjected to interaction studies. The protein and the ligand complex were loaded in the BIOVIA Discovery Studio, and different types of interactions such as covalent, carbon-hydrogen (C-H), hydrophobic interactions, and van der Waal attractions were analyzed [26,27].

Molecular Dynamic Simulation
MD simulation for the target SARS-CoV-2 main protease and hymenidin was carried out for 50 ns using WebGRO online server (https://simlab.uams.edu/index.php accessed on 5 February 2020). The lowest binding energy (most negative) docking conformation generated by AutoDock Vina inbuild PyRx was taken as the initial conformation for MD simulation. Initially, the hymenidin topology file was prepared using PRODRG server (http://davapc1.bioch.dundee.ac.uk/cgi-bin/prodrg accessed on 5 February 2020). The GROMOS96 43a1 force field was used for this study, and the SARS-CoV-2 main protease and hymenidin files' energy was minimized using steepest descents for 50,000 steps [28]. The calibration of NVT/NPT was completed at 300 K and 1 bar pressure. From the MD simulation, the root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), radius of gyration (ROG), and hydrogen bond (H-Bond) values were examined for SARS-CoV-2 main protease and hymenidin docked complex [29].

Results and Discussion
MDPs have been used as both nutraceutical and medicinal agents for the treatment of various health illnesses. In particular, marine secondary metabolites play an important role in pharmacological research. Drug-likeness properties are important features for drug design [30]. Moreover, the rapid development of high throughput screening (computational approaches) to rational drug design and new bioactive molecules from natural sources was an activity of the past. In total, 16 clinical and preclinical bioactive compounds from the marine ecosystem were selected for the present study ( Figure 1).

Molecular Docking Studies
The SARS-CoV-2 main protease is an important enzyme mainly involved in replication within the host system. To inhibit this enzyme activity, viral replication should be prevented [33,34]. As no homolog of the SARS-CoV-2 main protease has been identified in humans, it is achievable to develop effective and specific SARS-CoV-2 main protease

Molecular Docking Studies
The SARS-CoV-2 main protease is an important enzyme mainly involved in replication within the host system. To inhibit this enzyme activity, viral replication should be prevented [33,34]. As no homolog of the SARS-CoV-2 main protease has been identified in humans, it is achievable to develop effective and specific SARS-CoV-2 main protease inhibitors with extremely weak inhibitory activities on human proteases, thereby reducing the side effects caused by the SARS-CoV-2 main protease inhibitors [33,35]. It contains three domains: Domain I (8-101), Domain II (102-184), and Domain III (201-303). Domain I and Domain II have an antiparallel β-barrel structure, and Domain III has five α-helices. The substrate-binding site of the SARS-CoV-2 main protease is located between Domain I and Domain II.
The marine-derived preclinical, clinical, and approved drug candidates were docked with the SARS-CoV-2 main protease using PyRx version 0.8, and the results were tabulated ( Table 2). The geodisterol sulfates and chrysophaentin A were tightly bound to the COVID-19 targets of the main protease complex with minimized binding energy (−6.6 kcal/mol). Plinabulin and the hymenidin bioactive docked complex expressed the second highest binding affinity (−6.4 kcal/mol) compared with the other drug molecules. The molecular docking results were compared to standard hydroxychloroquine (HCQ) and paracetamol ( Table 2).

Molecular Dynamics Simulation
MD simulation is a computer-based approach used to predict the stability of the protein-ligand complexes, conformational flexibilities, and the dependability of protein-ligand affinities [41]. Therefore, a marine active compound of hymenidin with a low SARS-CoV-2 main protease docking score was submitted for MD simulations followed by binding energy calculations.
Root-mean-square deviation (RMSD) is used to examine the conformational stability of the protein-ligand complex (SARS-CoV-2 main protease-hymenidin), and it is defined as the "square root of an average value of the square of coordinate values of the protein". High values of RMSD represent the conformational instability of the docked complex. In general, the RMSD value should be 2 to 3 Å . In the present study, the SARS-CoV-2 main protease-hymenidin docked complex showed less than 3 Å RMSD value (0.1 to 0.3 nm) until 40 ns in Figure 3. This RMSD value indicates that the alkaloid class of hymenidin was tightly bound to the SARS-CoV-2 main protease and is in an acceptable range. According to [34], the cyclic depsipeptide of plitidepsin (from ascidian) with the SARS-CoV-2 main protease showed similar RMSD values with fluctuations around 0.3 nm at 300 K until 50 ns.

Molecular Dynamics Simulation
MD simulation is a computer-based approach used to predict the stability of the protein-ligand complexes, conformational flexibilities, and the dependability of proteinligand affinities [41]. Therefore, a marine active compound of hymenidin with a low SARS-CoV-2 main protease docking score was submitted for MD simulations followed by binding energy calculations.
Root-mean-square deviation (RMSD) is used to examine the conformational stability of the protein-ligand complex (SARS-CoV-2 main protease-hymenidin), and it is defined as the "square root of an average value of the square of coordinate values of the protein". High values of RMSD represent the conformational instability of the docked complex. In general, the RMSD value should be 2 to 3 Å. In the present study, the SARS-CoV-2 main protease-hymenidin docked complex showed less than 3 Å RMSD value (0.1 to 0.3 nm) until 40 ns in Figure 3. This RMSD value indicates that the alkaloid class of hymenidin was tightly bound to the SARS-CoV-2 main protease and is in an acceptable range. According to [34], the cyclic depsipeptide of plitidepsin (from ascidian) with the SARS-CoV-2 main protease showed similar RMSD values with fluctuations around 0.3 nm at 300 K until 50 ns.
Root-mean-square fluctuation (RMSF) is similar to RMSD, and it is an important parameter to define the flexible areas of a protein-ligand system. It mainly involves individual amino acid residue flexibility. It is used to explore the conformation stability due to the individual amino acids of the SARS-CoV-2 main protease in the complex form with hymenidin. Fewer fluctuation coordinates represent more stability. The RMSF value of the SARS-CoV-2 main protease-hymenidin docked complex was calculated to be around 0.1 to 0.4 nm at 300 K temperature and 1 bar pressure in optimized conditions (Figure 3). The RMSD and RMSF for the SARS-CoV-2 main protease-hymenidin complex showed stable binding throughout 50 ns. The stable RMSD and RMSF showed that the hymenidin had a strong binding affinity to the SARS-CoV-2 main protease and may be reasonable to act as a good inhibitor against the SARS-CoV-2 main protease.
The radius of gyration (ROG) is a physical parameter used to calculate the distance between the center of mass of the protein (the SARS-CoV-2 main protease) taken with its rotational axis. The ROG analysis for the SARS-CoV-2 main protease with hymenidin was examined for 50 ns at 300 K temperature and 1 bar pressure in optimized conditions. The average value of ROG for hymenidin with the SARS-CoV-2 main protease was found to be around 2.1 to 2.2 nm (Figure 4). It represents the conformational stability of the formed protein-ligand complex between the SARS-CoV-2 main protease and hymenidin. According to [42], commercial drugs (such as 5-fluorouracil, methotrexate, and paclitaxel) found similar ROG values around 2.0 to 2.2 nm. The H-bond interaction is important for docking studies. It can be classified into two types: conventional and non-conventional Hbonding. Non-conventional H-bonding plays a vital role in molecular docking studies [43] as the stability of the small molecule in the active binding region of the protein is calculated in terms of the average number of non-conventional H-bonds. In the present study, the docked complex formed a maximum of four H-bond interactions (Figure 4). act as a good inhibitor against the SARS-CoV-2 main protease.
The radius of gyration (ROG) is a physical parameter used to calculate the distance between the center of mass of the protein (the SARS-CoV-2 main protease) taken with its rotational axis. The ROG analysis for the SARS-CoV-2 main protease with hymenidin was examined for 50 ns at 300 K temperature and 1 bar pressure in optimized conditions. The average value of ROG for hymenidin with the SARS-CoV-2 main protease was found to be around 2.1 to 2.2 nm (Figure 4). It represents the conformational stability of the formed protein-ligand complex between the SARS-CoV-2 main protease and hymenidin. According to [42], commercial drugs (such as 5-fluorouracil, methotrexate, and paclitaxel) found similar ROG values around 2.0 to 2.2 nm. The H-bond interaction is important for docking studies. It can be classified into two types: conventional and non-conventional H-bonding. Non-conventional H-bonding plays a vital role in molecular docking studies [43] as the stability of the small molecule in the active binding region of the protein is calculated in terms of the average number of non-conventional H-bonds. In the present study, the docked complex formed a maximum of four H-bond interactions (Figure 4).

Pharmacokinetic Properties Analysis
In silico ADMET screening has been widely used in drug development and the drug discovery process. It is used to minimize failure rates and reduce the time of drug discovery. Aqueous solubility, intestinal absorption, and membrane permeability are important criteria for the drug development process [44,45]. Aqueous solubility is an important criterion to study the ratio of drug uptake, transfer, and clearance. Gastrointestinal absorption (GIB) and drug distribution are major obstacles in oral drug delivery. A higher intes-

Pharmacokinetic Properties Analysis
In silico ADMET screening has been widely used in drug development and the drug discovery process. It is used to minimize failure rates and reduce the time of drug discovery.
Aqueous solubility, intestinal absorption, and membrane permeability are important criteria for the drug development process [44,45]. Aqueous solubility is an important criterion to study the ratio of drug uptake, transfer, and clearance. Gastrointestinal absorption (GIB) and drug distribution are major obstacles in oral drug delivery. A higher intestinal absorption value indicates that the drug has good bioavailability in the system. The five lead MDPs (chrysophaentin A, geodisterol sulfates, hymenidin, plinabulin (NPI-2358), tetrodotoxin) against COVID-19 targets were analyzed for ADMET properties using the pkCSM online server. The results of six active compounds with high activity potentials are represented in Table 4. More than 30% of GIB values implies good absorbance. Chrysophaentin A showed the highest percentage of GIB (100%), followed by hymenidin (71.26%) and plinabulin (65.66%) which showed good absorption scores. Geodisterol sulfates (49.98%) and tetrodotoxin (36.93%) displayed a moderate absorption percentage. A skin permeability (SKP) value greater than −2.5 cm/h is considered low skin permeability, and all five drug candidates showed acceptable SKP values. Similarly, all five drug candidates displayed low Caco2 permeability (<0.9 cm/s). P-glycoprotein (PGP) is an important drug transporter, and it helps to determine the uptake and efflux of a range of drugs. The inhibition of PGP can result in the increased bioavailability of the susceptible drug, and the induction of PGP reduces the bioavailability [46,47]. All five drug candidates were shown to be a substrate for PGP. Chrysophaentin A and geodisterol sulfates were both observed to be inhibitors for PGP.
The VDss, CNS, and BBB membrane permeability were used to study the drug distribution [48]. A value greater than log 0.45 represents a relatively higher distribution volume. All drug candidates exhibited less than log 0.45 value, and plinabulin exhibited the better VDss (0.325) compared to the other four compounds. The BBB membrane permeability (range: log BB values > 0.3) and CNS permeability (range of log PS values > −2 to < −3) are important parameters in the distribution mechanism. All five marine drug compounds were predicted to be neither capable of crossing the CNS nor BBB membranes. CYP450 plays a vital role in all drug metabolism, and it has two important subtypes: CYP2D6 and CYP3A4. All five compounds were not substrates for CYP2D6 [49], and similarly, all compounds were not substrates for CYP3A4 except chrysophaentin A. Chrysophaentin A, geodisterol sulfates, hymenidin, and tetrodotoxin CYP1A2 inhibitors were not predicted as substrates for the CYP2C19, CYP2C9, CYP2D6, CYP3A4 inhibitors. Plinabulin was predicted to be an inhibitor for CYP2C19 and CYP3A4. This suggested that chrysophaentin A and plinabulin may be metabolized in the liver.
Drug excretion is related to the MW and hydrophilicity of marine active compounds. There are two important parameters involved in drug excretion: (i) total clearance (TCs) and (ii) the renal OCT2 substrate. TCs is measured using a combination of hepatic and renal clearance [50]. Hymenidin (1.027) showed the highest TCs score followed by tetrodotoxin (0.663), plinabulin (0.457), and geodisterol sulfates (0.27), and chrysophaentin A (−0.211) showed the least TCs score. None of the compounds were predicted as a substrate for renal OCT2. Toxicity is an important role in the selection of the most suitable drug compounds. AMES toxicity is used to predict the carcinogenic effect of drug compounds [51]. None of the compounds expressed AMES toxicity except plinabulin. hERG inhibition (I and II) is an important parameter that is mainly involved in cardiotoxicity [52]. None of the marine drug compounds expressed inhibitory actions of the hERG-I channel. Chrysophaentin A and plinabulin were involved in inhibitory actions of the hERG-II channel. All compounds were predicted as they may not have skin sensitization and hepatotoxicity (except hymenidin). The maximum tolerated dose (for humans), LD50, and LOAEL values were predicted and are tabulated in Table 4. The lead five drug candidates have some drawbacks due to their functional and structural properties. Further molecular modification is required to obtain a potential antiviral drug against SARS-CoV-2.

Conclusions
In summary, 16 marine-derived clinical-level compounds were investigated using in silico drug-likeness analysis, molecular docking, and ADMET properties. Among the 16 drug candidates, 5 compounds are proposed as potential hits against the SARS-CoV-2 main protease. In vitro, chrysophaentin inhibited MRSA, vancomycin-resistant Enterococcus faecium, and multidrug-resistant Staphylococcus aureus. The geodisterol sulfates also exhibited antibacterial activity against Candida albicans. Hymenidin inhibits HepG cytotoxicity activities. Tetrodotoxin was first reported as a molecular docking hit potential against SARS-CoV-2. The present study suggests that chrysophaentin A, hymenidin, and tetrodotoxin could be options to treat COVID-19-associated infections. Three compounds that were successful in binding to each of the targeted proteins were identified by the study and have shown stable behaviour, binding affinity, and molecular interactions. Our research discovered three compounds that were effective against each of the targeted proteins and showed stable behaviour, higher binding affinities, important residual molecular interactions, and good in silico pharmacokinetic properties. Overall, we suggest that these five compound hits might be promising pharmacological candidates for new COVID-19 treatments and suggest additional in vitro research on them.