A Computational Study on Selected Alkaloids as SARS-CoV-2 Inhibitors: PASS Prediction, Molecular Docking, ADMET Analysis, DFT, and Molecular Dynamics Simulations

Despite treatments and vaccinations, it remains difficult to develop naturally occurring COVID-19 inhibitors. Here, our main objective is to find potential lead compounds from the retrieved alkaloids with antiviral and other biological properties that selectively target the main SARS-CoV-2 protease (Mpro), which is required for viral replication. In this work, 252 alkaloids were aligned using Lipinski's rule of five and their antiviral activity was then assessed. The prediction of activity spectrum of substances (PASS) data was used to confirm the antiviral activities of 112 alkaloids. Finally, 50 alkaloids were docked with Mpro. Furthermore, assessments of molecular electrostatic potential surface (MEPS), density functional theory (DFT), and absorption, distribution, metabolism, excretion, and toxicity (ADMET) were performed, and a few of them appeared to have potential as candidates for oral administration. Molecular dynamics simulations (MDS) with a time step of up to 100 ns were used to confirm that the three docked complexes were more stable. It was found that the most prevalent and active binding sites that limit Mpro'sactivity are PHE294, ARG298, and GLN110. All retrieved data were compared to conventional antivirals, fumarostelline, strychnidin-10-one (L-1), 2,3-dimethoxy-brucin (L-7), and alkaloid ND-305B (L-16) and were proposed as enhanced SARS-CoV-2 inhibitors. Finally, with additional clinical or necessary study, it may be able to use these indicated natural alkaloids or their analogs as potential therapeutic candidates.


Introduction
SARS-CoV-2 afects the respiratory system, central nervous system, liver, heart, and kidneys, causing organ failure. Coughing, exhaustion, a loss of taste and smell, discomfort, and sore throat are typical symptoms. From these symptoms, approximately 10% of individuals can develop a severe illness requiring hospitalization and oxygen [1,2]. Till today, no specifc antiviral or SARS-CoV-2 inhibiting medication exists to treat this life-threatening illness. Antiviral medicines, such as lopinavir [3], danoprevir [4], ritonavir [5,6], remdesivir, umifenovir [7], chloroquine, and hydroxychloroquine [8], as well as antipyretics and mechanical respiratory support [9], are some therapeutic choices. Te current pharmacological recommendations for treating coronavirus disease in 2019 (COVID- 19) are based on the results of a number of studies that looked at severe acute respiratory problems. Western medicine works better when combined with herbal remedies and/or natural components from medicinal plants than when used alone [10][11][12]. Two proteases-a 3-C-like protease (M pro ) and a papain-like protease (PLpro)-that are involved in the processing and releasing of translated nonstructural proteins (NSPs) are encoded by the coronavirus polyprotein [13,14]. Main protease (M pro ), which is essential for COVID-19 replication and maturation, has been identifed as a therapeutic target among coronaviruses. SARS-CoV-2, a new coronavirus infection, is caused by an organism with a 30 kB singlestranded RNA genome [15]. Te weight of the full molecular structure of this protein (PDB ID: 6M03), which has 306 amino acid residues and deposited residues, 2454 atoms in all, is 33.83 kDa [16]. Te search for natural molecules to reduce COVID-19 remains a challenging problem despite the creation of various vaccines and supplementary therapies. To suppress SARS-CoV-2 M pro , a number of inhibitors have been proposed, including cilexitil, chloroquine, hydroxychloroquine, atazanavir, disulfram, dipyridamole, candesartan sulfacetamide, cimetidine, and maribavir [17]. Natural substances and alkaloid analogs have recently been the subject of intensive research for their robust antioxidant, antibacterial, anti-infammatory, and wide antiviral efects [18,19]. Natural substances may have antiviral capabilities, according to certain recent studies, and they may even be essential in the fght against COVID-19 [20][21][22][23][24]. Furthermore, research on SARS-CoV-2 inhibitory lead compounds derived from bioactive alkaloids is extremely scarce [20,25]. Alkaloids are typically found in plants, natural foods, fruits, and vegetables. Tey are nitrogen-containing molecules with at least one nitrogen acting as a heteroatom. Alkaloids are, therefore, likely to contain SARS-CoV-2 antiviral drugs due to their recognized bioactive and antiviral characteristics. Additionally, computational drug design is becoming increasingly important to quickly identify potential drug-like molecules [26,27]. Lipinski's rule of fve was used to pick numerous alkaloids from the PubChem database for this study's purposes, and PASS prediction studies were used to identify compounds with antiviral capabilities. So, the goal of this research is to do an in-depth computational study to fnd specifc lead drugs that can inhabit SARS-CoV-2. In this regard, we studied molecular docking, molecular electrostatic potentials surface (MEPS), molecular dynamics simulation (MDS), and ADMET characteristics of selected alkaloids to fnd their drug-liking properties as potential lead compounds against the M pro of SARS-CoV-2.

Evaluation of Biological Activity.
Te prediction of activity spectra for substances (PASS) [28] was used to assess the biological activity of the chosen alkaloids. Pa and Pi values were used to forecast whether the ligands were possibly active or inactive. Te Pa and Pi values of the proposed active chemical should be near one and zero, respectively. In this work, 256 alkaloids' antiviral capabilities were examined and 50 projected alkaloids with potential antiviral characteristics were taken into consideration for additional research to determine the best SARS-CoV-2 inhibiting drugs.

Protein Preparation.
A single-chain mutation-free protein was chosen from UniPort (https://www.uniprot. org) [29], and the relevant protein was obtained from the RCSB Protein Data Bank (PDB) [30]. Using this, specifc proteins in the human body were located. Based on acceptable XRD data, chain number, amino acid residues, resolution, and the lowest value of RMSD, the crystal structure ( Figure 1) of the SARS-CoV-2 M pro protein (PDB ID: 6M03) was downloaded from the PDB. By eliminating water molecules from the modeled SARS-CoV-2 of M pro protein, we prepared it for docking analysis using PyMol version 1.1.0's protein preparation wizard [31] (no cocrystal ligand was attached). Protein energy was minimized using Swiss-Pdb Viewer 4.1.0 [32]. Te prepared fle was then translated using Open Babel into the PDBQT format [33].

Ligand Preparation.
Each alkaloid structure was downloaded in SDF format from the PubChem database site [34]. Ten, each alkaloid structure was optimized using the functional B3LYP and basis set 6-31G of the Gaussian 16 program [35].

Molecular
Docking. SARS-CoV-2 of M pro was the target molecule for docking with the selected ligands, and Auto-Dock Vina in PyRx, version 0.8, was used for the molecular docking investigation. Vina Wizard predicts the interaction between a protein and a ligand using its scoring function (binding energy in kcal/mol). Based on expected ligand binding sites, the grid box was modifed during molecular docking to include all of the binding sites for each protein, and the XYZ coordinates were recorded. Te grid center  Based on the binding energy as well as the probable hydrogen bonds (H-bonds) and hydrophobic contacts, the fnal protein-ligand interacting models were selected. Te calculations of H-bonds and nonbonded interactions were performed using a protein-ligand interaction profler. Te visualization of protein-ligand interaction was carried out using the Discovery Studio, followed by the creation of 3D stereo fgures using PyMOL version 1.1.0 [36].

Evaluation of Chemical Reactivity by Using Density
Functional Teory (DFT). By analyzing the electrical characteristics of the top 2 bioactive alkaloid ligands, the density functional theory was used to determine the chemical stability of our target molecules. Te highest occupied molecular orbital (HOMU) and lowest unoccupied molecular orbital (LUMO) energies are used to calculate global reactivity descriptors. With the help of the following equations, the global reactivity descriptors, such as softness (S), electron afnity (A), ionization potential 10 (I), electronegativity (χ), global hardness (ղ), global electrophilicity index (ω), and chemical potential (µ), were determined. (1)

Molecular Dynamics Simulation.
Te YASARA dynamics software program and the AMBER14 force feld were used to conduct the molecular dynamics simulation of the ligand-protein complexes [44][45][46]. Initial cleaning, optimization, and orientation of the hydrogen bond network of the docked complexes were performed. Te TIP3P solvation model was used to create a cubic simulation cell with   [47]. Beyond the complexes of ligand and protein, the simulation cell was elongated by 20Å each direction. Te simulation cell's physiological parameters included pH 7.4, 298 K, and 0.9% NaCl concentration.
For the preliminary minimization of energy, the steepest gradient algorithms (5000 cycles) were applied in the simulated annealing method. Te time step of the simulation system was adjusted to 1.25 fs. By using the Particle Mesh Ewald (PME) system and a cutof radius of 8.0Å, long-range electrostatic interactions were calculated [48]. Simulation trajectory data were saved every 100 ps. Simulations at a constant temperature, pressure, and Berendsen thermostat were conducted for 100 ns. Based on simulation trajectories, the root mean square deviation (RMSD), root mean square fuctuation (RMSF), solvent accessible surface area (SASA), the radius of gyration (Rg), and hydrogen bond were analyzed [44,[48][49][50].

Results and Discussion
3.1. Estimation of Biological Activities. Te PASS online tool was used to evaluate all substances' overall bioactivities based on their chemical structures [51]. Te Pa and Pi values of 256 compounds were sorted out with antiviral properties from this server. Only 102 compounds were available with Pa values of more than 0.3 (Table S1). From these data, the frst 50 predicted antivirals were used for molecular docking studies against M Pro (Table 1 and Table S1). Tis dataset of predicted antivirals may also be used in any other antiviral drug discovery.

Protein Ligand Interaction.
Hydrogen bonds and hydrophobic bonds (nonbond interactions) with residues were taken into consideration in the protein-ligand interaction analysis, whereas the others were omitted. Here, in order to determine the bond strength, the bond distance is also computed and taken into account. All relevant data (Table 3) and . Surprisingly, PHE294, ARG298, and GLN110 demonstrate hydrophobic and hydrogen bonds with the closest bond distances (GLN110: 1.95Å). Finally, it was found that the most frequent binding sites for a number of chemicals in this M Pro were PHE294, ARG298, and GLN110. It has been hypothesized that the formation of the residues PHE294, ARG298, and GLN110 by our suggested lead compounds may limit the function of viral protein M pro (6M03).

Molecular and Pharmacokinetic Properties.
Te chemical structures of the top ten ligands are shown in Figure 3. Based on molecular analysis of the top ten docking score compounds (Table 4), the number of hydrogen bond acceptors (H. Ac) ranges from 3 to 7 (Lipinski: 10) and the number of hydrogen bond donors (H. Do) ranges from 0 to 2 (Lipinski: 5). For a compound to be a good therapeutic candidate, its H-bond acceptors and donors may not be more than ten and fve, respectively [56]. Te fndings imply that all the shortlisted compounds have the potential to be therapeutic candidates. Topological polar surface area (TPSA) of possible therapeutic candidates ranges from 0 to 140 [57]. In this study, the TPSA values for our short-listed compounds lie between 21 and 88. However, the TPSA values in this instance are 204.0, 148.12, 272.27, 169.93, and 95.44 for the standard antivirals. It is also a matter of consideration or indepth research to identify o fnd a high standard TPSA value that will work well for approved drugs and potential drug candidates. Lipophilicity (XLOGP3) (Lipinski: XLOGP3 5) was discovered to be upto 3.39 [58].  intestines. Standard ESOL of 4.0 was determined to be within the range of −4.00 to −2.09 for water solubility [59]. More soluble L-7 has a higher oral bioavailability and permeability, which suggests that it is better absorbed in the gastrointestinal tract. All of the ten shortlisted lead compounds, which are interestingly all BBB-capable like remdesivir, are safer than ritonavir, lopinavir, oseltamivir, and ribavirin from that perspective. All of the shortlisted compounds, much like the other specifed standard medicines, have been found to have acute oral toxicity (AOT) of grade III. Similar to this, all anticipated chemicals have a bioavailability (BA) of around 0.55 and are noncarcinogenic (CAR). Tese parameters guarantee that the aforementioned lead compounds will be accepted in drug discovery methods.

Drug-Likeliness Properties.
Drug research and development processes are sped up through the study of drug-like features. Te fve guidelines of Lipinski are employed as a criterion for locating possible drug candidates. However, if one rule is broken, it is acceptable to suggest a potential medicine [60]. Table 4 displays the outcomes after molecules are subjected to Lipinski's rule of fve, the Ghose flter rule, the Veber rule, the Egan rule, and the Muegge (LGVEM) rule. Tese criteria have their own rules for determining if a bioactive function is a potent drug. All of the alkaloids chosen for this study precisely followed the LGVEM guidelines without breaking any of them. Tis demonstrates the discussed chemicals' potential as drugs. Although they are recognized and accepted medications, remdesivir, ritonavir, lopinavir, and ribavirin have been found to have some violations of these LGVEM guidelines. In order to improve the efectiveness of drug-likeliness rules as a component of computational techniques in computer-aided drug design, this also creates a need for a separate, comprehensive study to build a deep association between drug-likeliness rules and parameters of approved medications (CADD).

ADMET Properties of the Selected Compounds.
Te pharmacokinetics of a drug (also known as absorption, distribution, metabolism, excretion, and toxicity, or ADMET) is the study of how pharmaceuticals enter, travel through, and leave the body [61]. Intestinal absorption in humans is greater than 94% for 9 out of 10 drugs but is only 71% and 65% for remdesivir and lopinavir, respectively. Finally, it showed that, in comparison to some of the traditional medications, the lead compounds indicated above had a higher capability for intestinal absorption by humans (Table 5). Te Caco-2 human colon epithelial cancer cell line permeability assay analyzes the rate of chemical fux through polarized Caco-2 cell monolayers to predict in vivo drug  (Table 5).
A drug's bioavailability may be increased by inhibiting P-glycoprotein or decreased by activating P-glycoprotein. When compared to other compounds, the L-7, L-16, L-21, L-30, and L-35 had a higher capability to block p-glycoprotein, indicating improved drug bioavailability compared to remdesivir, ritonavir, lopinavir, oseltamivir, and ribavirin (Table 5).

Frontier Molecular Orbital of HOMO and LUMO.
Due to their dominance in the coupling of the molecular bridge to electrodes, HOMO and LUMO spatial distributions are typically utilized to describe the transport properties [58]. Te chemical reactivity of the top two alkaloids and the HOMO and LUMO frontier molecular orbitals are shown in Table 6 and Figures 4(a)-4(f) in various hues to aid in comprehension. Additionally, a certain color map makes all other molecules accessible and available. Positive nodes are represented by the color green in HOMO, while negative nodes are represented by the deep radish hue. In order to create a chemical bond, the electrophilic attracting group can be linked to the HOMO portion of biologically active     Figures 5(a)-5(c) illustrate a 3D mapped electrostatic potential charge distribution. Tis diagram shows the attractive or repulsive force that a fxed charged particle (usually a proton with a point positive charge) experiences at different locations in space that are equally spaced apart from a molecular surface. Te negative electrostatic potential (in red) depicts the attraction of the proton by the area of high concentration of electrons in the molecule, whereas the positive electrostatic potential (in blue) corresponds to the repulsion of the proton by atomic nuclei [62,63].

Molecular Dynamics Simulation.
To investigate the structural rigidity of the top three protein-ligand complexes and validate the docking scenarios for these complexes, molecular dynamics simulations were performed. Te stability of protein-ligand complexes was investigated using the RMSD values of the C-alpha atoms. Te ligand-protein complexes containing fumarostelline (L-1), strychnidin-10-one, 2,3-dimethoxy (L-7), and alkaloid ND-305B (L-16) displayed preliminary RMSD increases owing to the instability of these complexes, as shown in Figure 6(a). Te fumarostelline-M pro complex had a higher increase in RMSD on average than the other two complexes. Contrariwise, the alkaloid ND-305B-M pro complex exhibited a lower RMSD value on average than the other two complexes. Te RMSD profle of the fumarostelline-M pro complex fell drastically at roughly 25 ns, then stabilized at around 45 ns and remained steady with just minor changes for the remaining 55 ns of the simulations. Te strychnidin-10-one, 2,3-dimethoxy-M pro complex showed a slightly greater RMSD value than the other complexes at 55-70 ns, which could elucidate their enhanced fexibility. Despite this, the RMSD trend of all the three complexes did not surpass 2.5Å, indicating that the complexes remained stable across the entire simulation time [64].
To measure the variations in the surface of the SARS-CoV-2 main protease (M pro ) in response to interaction with the ligand molecules, the SASA values of the top three docking score complexes were evaluated. Te surface area of the protein infates with the increased value of SASA, while protein truncations occur when SASA decreases [65]. In general, the SASA of the fumarostelline-M pro complex was higher on average than the other complexes, suggesting that the complex's surface area was extended than those of the other two ( Figure 6(b)). Te alkaloid ND-305B-M pro complex had lower SASA on average than the other two complexes before peaking at 60 ns of the simulation period. After 60 ns of the simulation time, the fumarostelline-M pro complex, strychnidin-10-one, 2,3-dimethoxy-M pro complex, and alkaloid ND-305B-M pro complex reached a steady-state and remained stable for the remaining 40 ns simulation period with only slight fuctuations. Using Rg values, it was determined whether the protein complexes were more compact or labile. A greater value implies a more labile protein complex, whereas a lower value designates that the simulated protein complex is stifer. All three complexes displayed an initial increase followed by a decrease of the Rg value. In contrast to the other two complexes, the strychnidin-10-one, 2,3-dimethoxy-M pro complex exhibited relatively low Rg at around 50-70 ns of the simulation period, representing a more stable structure of this complex (Figure 6(c)). Other than that, the other two complexes diverged only slightly but comparatively had lower Rg values.
Te docked complexes' hydrogen bonds were examined since hydrogen bonding is important for keeping protein integrity and stability. Te fumarostelline-M pro complex, strychnidin-10-one, 2,3-dimethoxy-M pro complex, and alkaloid ND-305B-M pro complex all formed a great deal of hydrogen bonds throughout the simulation trajectory,  (Figure 6(d)). To further understand M pro ' s fexibility across the amino acid area, the RMSF of the ligand and M pro complexes were investigated. Te RMSF profles of approximately all amino acid residues in the top three docked complexes were below 2.5Å except at the beginning. Te lower RMSF value of the top three docked complexes indicated the decreased fexibility of the complexes since lower RMSF values are connected with the higher stability of the complexes (Figure 6(e)).

Conclusion
On the basis of computer-aided drug design, which includes protein purifcation, molecular optimization, molecular docking and visualization, molecular characteristics, ADMET analysis, and molecular dynamic simulations, the frst 50 compounds were chosen out of 252 for detailed studies (MDS). Te highest possible docking score was −8.5 kcal/mol (L-7 and L-1). Te top 10 compounds are chosen for an additional indepth investigation to identify SARS-CoV-2 major protease inhibitory drugs based on the docking score. We ran molecular dynamic simulations of L-1, L-7, and L-16 to test their stability in our biological systems after examining all the fndings and comments. After comparing all of the outcomes to the FDAapproved antiviral drug remdesivir, the following compounds are suggested as improved SARS-CoV-2 inhibitory lead compounds: fumarostelline, strychnidin-10-one (L-1), 2,3dimethoxy-brucin (L-7), and the alkaloid ND-305B (L-16). Based on the study of molecular docking, protein ligand interactions of docked complexes, and surface molecular electrostatic potentials, our proposed mechanism is as follows: the formation of the PHE294, ARG298, and GLN110 residues by our proposed lead compounds may limit the function of the viral protein M pro (6M03) to combat SARS-CoV-2. On the other hand, suggested compounds can also be used for further in vitro and in vivo studies.

Data Availability
Te datasets used and analyzed during the current study are available from the corresponding authors on reasonable request.