Identification of Potential SARS-CoV-2 Main Protease and Spike Protein Inhibitors from the Genus Aloe: An In Silico Study for Drug Development

Severe acute respiratory syndrome coronavirus (SARS-CoV-2) disease is a global rapidly spreading virus showing very high rates of complications and mortality. Till now, there is no effective specific treatment for the disease. Aloe is a rich source of isolated phytoconstituents that have an enormous range of biological activities. Since there are no available experimental techniques to examine these compounds for antiviral activity against SARS-CoV-2, we employed an in silico approach involving molecular docking, dynamics simulation, and binding free energy calculation using SARS-CoV-2 essential proteins as main protease and spike protein to identify lead compounds from Aloe that may help in novel drug discovery. Results retrieved from docking and molecular dynamics simulation suggested a number of promising inhibitors from Aloe. Root mean square deviation (RMSD) and root mean square fluctuation (RMSF) calculations indicated that compounds 132, 134, and 159 were the best scoring compounds against main protease, while compounds 115, 120, and 131 were the best scoring ones against spike glycoprotein. Compounds 120 and 131 were able to achieve significant stability and binding free energies during molecular dynamics simulation. In addition, the highest scoring compounds were investigated for their pharmacokinetic properties and drug-likeness. The Aloe compounds are promising active phytoconstituents for drug development for SARS-CoV-2.


Structure-Based Virtual Screening and Molecular Docking of Aloe Phytochemicals on SARS-CoV-2 Spike Glycoprotein and Main Protease
High-throughput virtual screening of compounds from Aloe, was followed by molecular docking and MD simulation. Since ligand binding to a protein of interest is the first step in drug discovery, molecular docking is widely used to predict and identify ligands that fit into the binding pocket of a protein of interest [28]. Our screening was performed against two major drug discovery and therapeutic targets of SARS-CoV-2, spike glycoprotein and M pro proteins [7,12]. SARS-CoV-2 main protease M pro is critical for the life cycle of the virus. Approximately, two thirds of the SARS-CoV-2 genome is translated into polyproteins pp1a and pp1ab, that are cleaved with M pro into nonstructural proteins that are involved in the production of viral membrane, spike and nucleocapsid proteins [29]. M pro is a dimer that has cysteine and histidine in the active site which form a catalytic dyad, conserved among coronaviruses making it an ideal therapeutic target [12]. In molecular docking studies, the ligand-receptor interaction with protein active site residues is established by formation of some interactions including hydrogen bonds, Van der Waal force  36.3%, chromones 27.4%, coumarin 0.8%, flavonoids 4.2%, simple phenolic compounds 8%, phenyl pyran and phenyl pyrone derivatives 7%, benzofurans 2%, naphthalene derivatives 5.9%, alkaloids 1.2%, fatty acid derivatives 1.2% and miscellaneous compounds 5.5%.

Structure-Based Virtual Screening and Molecular Docking of Aloe Phytochemicals on SARS-CoV-2 Spike Glycoprotein and Main Protease
High-throughput virtual screening of compounds from Aloe, was followed by molecular docking and MD simulation. Since ligand binding to a protein of interest is the first step in drug discovery, molecular docking is widely used to predict and identify ligands that fit into the binding pocket of a protein of interest [28]. Our screening was performed against two major drug discovery and therapeutic targets of SARS-CoV-2, spike glycoprotein and M pro proteins [7,12]. SARS-CoV-2 main protease M pro is critical for the life cycle of the virus. Approximately, two thirds of the SARS-CoV-2 genome is translated into polyproteins pp1a and pp1ab, that are cleaved with M pro into nonstructural proteins that are involved in the production of viral membrane, spike and nucleocapsid proteins [29]. M pro is a dimer that has cysteine and histidine in the active site which form a catalytic dyad, conserved among coronaviruses making it an ideal therapeutic target [12]. In molecular docking studies, the ligand-receptor interaction with protein active site residues is established by formation of some interactions including hydrogen bonds, Van der Waal force interaction, π-sigma bond, π-π interaction, electrostatic interaction, and many other hydrophobic interactions. Hydrogen bonds are essential for interaction, lowering the binding energy and stabilizing the ligand-receptor docked complex. Pharmacologically, it is well-known that blockade of a receptor active site by a ligand terminates its functional activity [30]. Our molecular docking approach was validated by docking of hydroxychloroquine, a potent inhibitor of SARS-CoV-2 M pro . Hydroxychloroquine acts as a lysomotropic agent that inhibits viral entry and viral endocytosis. Viral entry and replication are highly dependent on the acidic pH of lysosomes and endosomes, and some host proteases which are also active in acidic pH (pH 5-5.5) [31]. Chloroquine and its analogues are diprotic weak bases that in their unprotonated forms, readily diffuse through cellular and organelle membranes such as lysosomes, endosomes and Golgi vesicles increasing pH from 6.3 to 6.7 [32][33][34]. In addition to disruption of endocytic pathway pH, chloroquine and hydroxychloroquine have been recently found to be potent inhibitors of SARS-CoV-2 M pro but not viruses that belong to Rhabdoviridae [35]. In our study, the compounds previously isolated from Aloe plants were virtually screened against SARS-CoV-2 main protease M pro (PDB ID: 6LU7) ( Figure 2) and spike glycoprotein (PDB ID: 6M0J) ( Figure 2) to find potential inhibitors for SARS-CoV-2. Using our docking approach, hydroxychloroquine interacted with SARS-CoV-2 protein M pro and docked hydroxychloroquine bound to the active site with and RMSD of 1.2 Å. Molecular docking data were filtered to remove compounds with scores > −6.5 for both SARS-CoV-2 main protease M pro ( Figure 3 and Table A1) and spike glycoprotein ( Figure 4 and Table A1). Molecular docking was performed by examining the interactions of these compounds with the active site residues of these proteins and analysis of results.
interaction, π-sigma bond, π-π interaction, electrostatic interaction, and many other hydrophobic interactions. Hydrogen bonds are essential for interaction, lowering the binding energy and stabilizing the ligand-receptor docked complex. Pharmacologically, it is well-known that blockade of a receptor active site by a ligand terminates its functional activity [30]. Our molecular docking approach was validated by docking of hydroxychloroquine, a potent inhibitor of SARS-CoV-2 M pro . Hydroxychloroquine acts as a lysomotropic agent that inhibits viral entry and viral endocytosis. Viral entry and replication are highly dependent on the acidic pH of lysosomes and endosomes, and some host proteases which are also active in acidic pH (pH 5-5.5) [31]. Chloroquine and its analogues are diprotic weak bases that in their unprotonated forms, readily diffuse through cellular and organelle membranes such as lysosomes, endosomes and Golgi vesicles increasing pH from 6.3 to 6.7 [32][33][34]. In addition to disruption of endocytic pathway pH, chloroquine and hydroxychloroquine have been recently found to be potent inhibitors of SARS-CoV-2 M pro but not viruses that belong to Rhabdoviridae [35]. In our study, the compounds previously isolated from Aloe plants were virtually screened against SARS-CoV-2 main protease M pro (PDB ID: 6LU7) ( Figure 2) and spike glycoprotein (PDB ID: 6M0J) ( Figure 2) to find potential inhibitors for SARS-CoV-2. Using our docking approach, hydroxychloroquine interacted with SARS-CoV-2 protein M pro and docked hydroxychloroquine bound to the active site with and RMSD of 1.2 Å . Molecular docking data were filtered to remove compounds with scores > −6.5 for both SARS-CoV-2 main protease M pro (Figure 3 and Table A1) and spike glycoprotein ( Figure 4 and Table A1). Molecular docking was performed by examining the interactions of these compounds with the active site residues of these proteins and analysis of results. Compounds scoring lower than −5.00 kcal/mol are expected to be active. These compounds were then filtered by RMSD value [30], to evaluate experimental stability of the docked ligand conformers. RMSD values around 1.5 Å , are considered successful and stable while those beyond 2 Å indicate instability of ligand conformation and docking parameters [36]. For SARS-CoV-2 protein M pro , the binding energy observed for these compounds ranged from−7.950 to −0.339 kcal/mol while for spike glycoprotein, binding energy ranged from −8.088 to −5.437 kcal/mol. The top three scoring compounds for SARS-CoV-2 protein M pro were compound 132 (2′-oxo-2′-O-(3,4-dihydroxy-E-cinnamoyl)-(2′R) aloesinol-7-methyl ether), compound 134 (2′-oxo-2′-O-(4-hydroxy-3-methoxy-(E)-cinnamoyl)-(2′R)-aloesinol-7-methyl ether) and compound 159 (rutin), (Table 1 docking scores and Figure 5, top panel). These three compounds showed the strongest interaction with the active site of SARS-CoV-2 protein M pro . Molecular 2D and 3D interactions complexes of compounds 132, 134 and 159 with SARS-SARS-CoV-2 protein M pro are shown in Figure 6. Compounds scoring lower than −5.00 kcal/mol are expected to be active. These compounds were then filtered by RMSD value [30], to evaluate experimental stability of the docked ligand conformers. RMSD values around 1.5 Å, are considered successful and stable while those beyond 2 Å indicate instability of ligand conformation and docking parameters [36]. For SARS-CoV-2 protein M pro , the binding energy observed for these compounds ranged from−7.950 to −0.339 kcal/mol while for spike glycoprotein, binding energy ranged from −8.088 to −5.437 kcal/mol. The top three scoring compounds for SARS-CoV-2 protein M pro were compound 132 (2 -oxo-2 -O-(3,4-dihydroxy-E-cinnamoyl)-(2 R) aloesinol-7-methyl ether), compound 134 (2 -oxo-2 -O-(4-hydroxy-3-methoxy-(E)cinnamoyl)-(2 R)-aloesinol-7-methyl ether) and compound 159 (rutin), (Table 1 docking scores and Figure 5, top panel). These three compounds showed the strongest interaction with the active site of SARS-CoV-2 protein M pro . Molecular 2D and 3D interactions complexes of compounds 132, 134 and 159 with SARS-SARS-CoV-2 protein M pro are shown in Figure 6.

Molecular Dynamics Simulation
Conventional docking approaches do not account for the inherent protein binding site flexibility and the many protein conformational rearrangements [37]. Computational tools for drug discovery such as molecular dynamics take into account structural flexibility and entropic effects which produce accurate predictions of small molecule-protein binding thermodynamics and kinetics [38]. Hence dynamical docking considers flexibility of drug-protein binding and conformational changes, solvation of drug-protein complex and temperature [38,39]. Unbiased millisecond-long can predict spontaneous drug-protein entire binding [40]. In addition, recent developments in dynamical docking such as

Molecular Dynamics Simulation
Conventional docking approaches do not account for the inherent protein binding site flexibility and the many protein conformational rearrangements [37]. Computational tools for drug discovery such as molecular dynamics take into account structural flexibility and entropic effects which produce accurate predictions of small molecule-protein binding thermodynamics and kinetics [38]. Hence dynamical docking considers flexibility of drug-protein binding and conformational changes, solvation of drug-protein complex and temperature [38,39]. Unbiased millisecond-long can predict spontaneous drug-protein entire binding [40]. In addition, recent developments in dynamical docking such as enhanced sampling for dynamical docking, path-based and alchemical transformations have greatly impacted drug discovery [38]. To validate molecular docking results, we subjected the top scoring compounds to unbiased molecular dynamics simulation experiments. The three top scoring M pro inhibitor hits 132, 134, and 159 were able to achieve stable binding inside the active site with low deviations across the course of simulations (Average RMSD = 3.22, 3.32, and 3.86 Å, respectively) and convergent binding free energies (∆G = −6.9, −6.8, and −6.5 kcal/mol, respectively), ( Figure 8A).

Drug like Properties, and Pharmacokinetic Prediction of the Ligands
Drug-like properties and pharmacokinetic properties are intrinsic characteristi drugs that may need to be optimized independently from pharmacodynamics prope during drug development. It is a balance among molecular properties affecting pha codynamics and pharmacokinetics of small molecules. These molecular properties as membrane permeability and bioavailability are always connected to some basic m ular descriptors such as lipophilicity log P, (Tendency of a compound to partition int aqueous matrix versus lipid matrix), molecular weight (MW), topological polar sur area (TPSA), or hydrogen bond acceptors and donors count in a molecule. Lipophil impacts drug's absorption, distribution, metabolism, elimination (ADME) and pla protein binding properties. In addition, the number of hydrogen bond donors and hy gen bond acceptors influence drug's pKa (−log Ka). The solubility of small molecules pacts their bioavailability and the need for frequent dosing, hence we investigated With respect to SARS-CoV-2 spike glycoprotein, both compounds 120 and 131 were stable inside the binding site during MD simulation, with scoring average RMSDs of 2.81 Å and 3.96 Å, respectively, and ∆G of −7.4 and −6.8 kcal/mol, respectively ( Figure 8B). On the other hand, compound 115 was significantly less stable (average RMSD = 6.2 Å) inside the SARS-CoV-2 spike glycoprotein binding site, and this instability was further translated into a low binding free energy (∆G = −4.5 kcal/mol) compared to compounds 120 and 131 ( Figure 8B). RMSF is an expression of the average residual mobility throughout simulation in a structure and a higher RMSF value indicates more flexibility during MD simulation. We calculated the RMSF value for the top scoring compounds from Aloe genus with SARS-CoV-2 M pro and SARS-CoV-2 spike glycoprotein and plotted RMSF value versus residue number ( Figure 8C,D). The results indicate that compounds 159 and 120 had high RMSF values compared to other compounds. The RMSD and RMSF values indicate that the top scoring compounds from Aloe genus were stable and had greater random motion during the simulation. The inhibitors identified in in our docking analysis that showed interaction with SARS CoV-2 spike protein and M Pro are in agreement with previously reported results [41]. Arokiyaraj et al. found that several polyphenolic compounds from Geranii Herba, including geraniin, kaempferitrin, quercitin, gallic acid, and kaempferol interacted with amino acid residues in the SARS-CoV-2 RBD active site inhibiting the interaction of SARS-CoV-2 RBD with ACE2. Arokiyaraj et al. also reported that these polyphenolic compounds interacted strongly with amino acids in the active site of SARS-CoV-2 M pro and its proximity leading to blockade of the nucleophilic attack toward His 41 and blockade of proteolytic activity. In agreement with this, we found that quercetin interacted with SARS CoV-2 RBD and M pro with binding energies of −5 Kcal/mol and −5.5 Kcal/mol respectively, similar to the results reported by Arokiyaraj et al. for quercetin interaction with RBD and M pro −5.71 Kcal/mol and −6.49 kcal/mol, respectively. In addition, we found that gallic acid interacted with SARS CoV-2 RBD and M pro with binding energies of −4.19 Kcal/mol and −3.56 Kcal/mol, respectively, similar to the binding energies reported by Arokiyaraj et al. for the gallic acid interaction with SARS CoV-2 RBD and M pro , −4.21 kcal/mol and −4.46 kcal/mol, respectively. These finding indicate that phenolic compounds from Aloe are potential inhibitors for SARS CoV-2 RBD and M pro [41].

Drug like Properties, and Pharmacokinetic Prediction of the Ligands
Drug-like properties and pharmacokinetic properties are intrinsic characteristics of drugs that may need to be optimized independently from pharmacodynamics properties during drug development. It is a balance among molecular properties affecting pharmacodynamics and pharmacokinetics of small molecules. These molecular properties such as membrane permeability and bioavailability are always connected to some basic molecular descriptors such as lipophilicity log P, (Tendency of a compound to partition into an aqueous matrix versus lipid matrix), molecular weight (MW), topological polar surface area (TPSA), or hydrogen bond acceptors and donors count in a molecule. Lipophilicity impacts drug's absorption, distribution, metabolism, elimination (ADME) and plasma protein binding properties. In addition, the number of hydrogen bond donors and hydrogen bond acceptors influence drug's pKa (−log Ka). The solubility of small molecules impacts their bioavailability and the need for frequent dosing, hence we investigated the ADME properties, inhibition of cytochrome P450 (CYP), modulation of P-glycoprotein (Pgp), solubility, plasma protein binding and permeability of the top scoring compounds in our analysis. The best scoring compounds for both SARS-CoV-2 M pro and spike glycoprotein were tested for obeying Lipinski's rule of five parameters, which states that drugs having log P ranging from 0 to 5, have high possibility of oral absorption [42]. Data (Table 2) showed that the compounds have log P values that ranged from −1.06 to 2.8 that does not exceed 5.0 indicating reasonable probability of their good absorption. The number of hydrogen bond donors was variable and ranged from 4 to10 that is more than 5 and also hydrogen bond acceptors were 11-16 that is more than 10. All compounds have number of atoms that ranged from 40 to 43 which is within 20-70. In addition, the topological polar surface area (TPSA) of the compounds as parameter for the prediction of drug transport properties showed TPSA value greater than 140 Å 2 tend to be poor at permeating cell membranes. Despite violation of some rules, approved anticancer and anti-infective drugs from natural products or their semisynthetic derivatives such as taxol and amphotericin B have also some violations but are biologically effective as drugs. Therefore, these results don't interfere with the development of these compounds as potential SARS-CoV-2 therapeutic agents [43].  [44]. All compounds showed medium Caco-2 predicted permeability and medium MDCK predicted cell permeability [45]. Moreover, all compounds showed moderate predicted plasma protein binding (PPB) (Table S1) except for compound 159, which showed weak predicted PPB, which indicates predicted decreased excretion and increased predicted half-life. It is important to consider drug's interaction with plasma proteins, transporters and CYP450s for the successful selection of drug candidate. CYP2C19 and CYP2C9 inhibition leads to increased drug plasma concentration, leading to potential side effects [46,47]. All top scoring compounds were predicted to inhibit CYP2C19 and CYP2C9. CYP2D6 metabolizes many drugs and toxins [48]. None of the top scoring compounds showed predicted inhibitory activity to CYP2D6. CYP3A4 is also involved in metabolism of xenobiotics and is highly expressed in the liver and intestine [49]. The six top scoring compounds were predicted to inhibit CYP3A4. Drug resistance is a major concern in drug development. Multidrug resistance is regulated by a network of ATP-binding cassette (ABC) proteins that detoxify xenobiotics and act as drug transporters and efflux pumps. P glycoprotein (Pgp; ABCB1) is the most popular and well-studied efflux pump [50,51]. Pgp has intrinsic ATPase activity to drive active transport and generate a concentration gradient leading to transport of drugs to the extracellular space and inhibition of drug activity [51]. Pgp is highly expressed in blood-brain barrier cell, liver, intestine and kidney. Thus, it is important to predict drug's binding to Pgp. Only compound 115 was predicted to inhibit Pgp and hence it may affect the activity or excretion of other Pgp substrates. Compound 159 had the highest water solubility (217.207 mg/L) while the other five compounds had low water solubility, hence this should be considered during drug development. In addition, skin permeability is an important factor to consider during drug development for the potential of dermal drug delivery and risk assessment of drugs that may contact skin [52]. Skin permeability also increases drug's plasma concentration and activity. It has been reported that logP between −3 to +6 predict drug's skin permeability [53]. SKlogD, SKlogP and SKlogS are related to drugs' skin permeability and lipophilicity. All the six top scoring compounds had skin permeability values ranging from −4.6 to −3.6, indicating that they may not be absorbed through skin and thus should not pose a dermal exposure risk. Finally, all compounds did not have predicted ability to pass the blood brain barrier (BBB) and are not expected to be neurotoxic.

Phytochemical Review of Genus Aloe
Intensive review of the literatures in ScienceDirect, PubMed, SciFinder and has been conducted to identify compounds from Aloe genus.

Preparation of Protein and Active Site Prediction
In this study, two SARS-CoV-2 proteins which facilitate viral-host interaction and replication were selected from the RCSB protein databank (https://www.rscb.org/pdb, accessed on 20 February 2021). The proteins are SARS-CoV-2 main protease (PDB ID: 6LU7, resolution = 2.16 Å) [54] and spike glycoprotein (PDB ID: 6M0J, resolution = 2.45 Å) [42]. The 3D protein structures were prepared using Molecular Operating Environment software (MOE 2014.0901) Ligx option. The site finder function used to calculate and predict possible active potential site of selected proteins for ligand binding in the receptor. PyMol 2.3 software was used for visualization.

Preparation of Ligand
Reviewing the available literatures identified 237 phytochemical compounds that were isolated from genus Aloe (Table A1 and Figures S1-S18). All these molecular structures were imported to MOE and subjected to 3D protonation and energy minimization using MMFF94s force field and ligand database was constructed. Ligand coordinate files were extracted from PDB files and used as reference structures for root mean square deviation (RMSD) calculations.

Docking Analysis
Flexible ligand-rigid receptor docking was performed with MOE-DOCK for molecular docking. The parameters of scoring were Triangle Matcher, scoring was set at London dG with 30 output poses and rescoring at GBVI/WSA dG retaining 10 refined poses. The docking score, root mean square deviation (RSMD), 2D and 3D interactions were recorded [55]. The results of docked ligands are chosen based on the most negative docking score. The docking score represents the best-bound ligand conformations and relative binding affinities. The best-docked conformations for comparison of the binding of the drugs and targets of SARS-CoV2 were selected based on number of hydrogen bonds, binding energy (kcal/mol), upper and lower bound RMSD, number of interacting residues, and forces which stabilized the receptor-ligand complex. RMSD and RMSF of the ligand interaction with the target protein were calculated using the following formulas: where N is the number of atoms, t ref is the reference time, r is the position of the selected atoms in frame x after superimposing on the reference frame, frame x recorded at time t x , T is the trajectory time over which the RMSF was calculated, r is the position of an atom. Poses of docked compounds are automatically calculated by docking function in Molecular Operating Environment software.

Molecular Dynamics Simulation
MD simulation experiments were performed as previously described [43]. Briefly, the Molecular Dynamics (NAMD) 2.6 software [45], employing the CHARMM27 force field [56] was used for simulations. Hydrogen atoms were added to initial coordinates of proteins using the psfgen plugin included in the Visual Molecular Dynamics (VMD) 1.9 software [57]. Subsequently, the protein systems were solvated using TIP3P water particles and 0.15 M NaCl. The equilibration procedure comprised 1500 minimization steps followed by 30 ps of MD simulation at 10 K with fixed protein atoms. Then, the entire systems were minimized over 1500 steps at 0 K, followed by gradual heating from 10 to 310 K using temperature reassignment during the initial 60 ps of the 100 ps equilibration MD simulation. The final step involved NTP simulation (30 ps) using the Nose-Hoover Langevin piston pressure control at 310 K and 1.013 bars for density (volume) fitting [58]. Thereafter, the MD simulation experiments were continued for 25 ns for the entire systems (20 ns for the enzyme-ligand complexes). The trajectory was stored every 0.1 ns and further analyzed with the VMD 1.9 software. The MD simulation output over 25 ns provided several structural conformers that were sampled every 0.1 ns (250 poses) to evaluate conformational changes of the entire protein structure to analyze the RMSD. All parameters and topologies of the compounds selected for MD simulation were prepared using the online software Ligand Reader & Modeler [59] and the VMD Force Field Toolkit (ffTK) [57]. Binding free energy calculations (∆G) were performed using the free energy perturbation (FEP) method through the web-based software Absolute Ligand Binder [60] together with MD simulation software NAMD 2.6 [45]. Hydrogen bonds and hydrophobic interactions between protein and ligand were also analyzed using Protein-Ligand Interaction Profiler [61].

Drug Like Properties, and ADME Prediction of the Ligands
The drug likeliness of best pose scoring compounds is specified by the Lipinski's rule and molecular properties prediction was calculated by the free access website https: //www.molinspiration.com/cgi-bin/properties, accessed on 20 February 2021. ADME Prediction were determined by PreADMET estimation server website [62].

Conclusions
In recent years, advances in computational resources and software tools led to emergence of molecular dynamics (docking and scoring tool), as the first phase in drug screening and discovery. In addition, absence of new cell culture models for working safely with highly pathogenic viruses makes virtual screening, docking and dynamics of great importance. Aloe genus is a rich source of phytochemicals with a wide range of therapeutic activity. Several natural products from Aloe have shown strong antiviral activity, inhibiting replication and entry and of HSV-1 and 2, human cytomegalovirus (HCMV), influenza A and polio. Aloin significantly reduces replication of oseltamivir-resistant (H1N1) influenza virus. In our study, we applied computational screening of 237 natural product compounds from Aloe genus and identification of six compounds as stable potential inhibitors of SARS-CoV-2 main protease and spike glycoprotein. Our molecular docking analysis showed that theses six compounds are stable and safe. Compounds 132, 134 and 159 were the top three scoring potential inhibitors of for SARS-CoV-2 main protease. These compounds interacted strongly with amino acids in the active site of SARS-CoV-2 main protease. Rutin (154) is known to have antiviral activity against influenza virus [63]. Compounds 115, 120 and compound 131 were the top scoring potential inhibitors of SARS-CoV-2 spike glycoprotein. The results highlighted chromone derivatives as potential inhibitors for SARS-CoV-2 according to their best scores of binding affinities to the mentioned target proteins among the examined compounds. The results of this in-silico investigation (docking and molecular dynamics simulation) should have a great impact for drug repurposing studies. In the future, in vitro, in vivo and clinical studies shall be conducted to further validate the effectiveness of these compounds as potential treatments for COVID-19 and to identify compounds with best pharmacokinetic profiles. In addition, it will be of great importance to apply newly-developed algorithms and utilize the development of steered molecular dynamics for evaluating the binding of the top scoring compounds to SARS-CoV-2 target proteins [64].

Supplementary Materials:
The following are available online. Figures S1-S18: Compounds isolated from genus Aloe, Table S1: Predicted pharmacokinetics of top scoring compounds.

Conflicts of Interest:
The authors declare that they have no competing financial interests or personal relationships that could influence the work reported in this paper.
Sample Availability: Samples of the compounds are not available from the authors.