3.1 Biological interpretation

The inhibitory biological activity of unsymmetrical aromatic disulfides against the SARS-CoV Mpro has been previously evaluated in vitro (Wang et al., 2017). Due to the close similarity between the SARS-CoV and SARS-CoV-2 viruses, we have opted for investigating in silico the behavior of disulfides regarding SARS-CoV-2 Mpro. Choosing this enzyme as a target gives important leads in antiviral treatment because it participates in the proteolytic processing of replicase polyproteins; therefore, it plays a major role in viral gene expression and replication. Thus, the inhibition of this enzyme could hamper the replication of the viral genome and then the multiplication of SARS-CoV-2. In this study, we suggest that molecules that could inhibit 3CLproof SARS-CoV may result in the same interest to inhibit ofSARS-CoV-2 Mpro, because of their high sequence similarity. This hypothesis was investigated using MD analysis.

Based on the MD results, the two critical amino acids in the active site of the target are Cys145 and His41. The inhibition of SARS-CoV 3CLpro has been reported to mainly occur via an irreversible covalent bond with Cys145 and, in other several cases, by reversible interaction with Cys145 and His41, which form a catalytic dyad, where Cys145 acts a nucleophile (Ramajayam et al., 2010; Shao et al., 2008; Jacobs et al., 2013; Thanigaimalai et al., 2013a; Liu et al., 2014; Thanigaimalai et al., 2013b) and His41 acts as a base that polarizes and deprotonates the nucleophile (Cys145) to increase its reactivity; this dyad is then attacked by the inhibitor to form an intermediate complex before releasing the enzyme. In our study, compounds 6, 7, 10, and 11 display interactions with Cys145 and His41, suggesting that these molecules could have an inhibitory activity against SARS-CoV-2 Mpro. For compounds 8 and 9, even if they form a stable complex, they will have no pharmacological effect because there is no stabilization of Cys145 or His41. SARS-CoV-2 Mpro is a cysteine protease, in which the unsymmetrical aromatic disulfides may play the role of an inhibitor, which competes with the substrate (competitive inhibitor). We believe that compounds 6, 7, 10, and 11 could be potential inhibitors for this enzyme and, consequently, may act as drug candidates against COVID-19. To confirm this assumption, an in vitro study against SARS-CoV-2 Mpro is recommended.

3.2 Pharmacokinetics and drug-likeness prediction

Pharmacokinetics and drug-likeness prediction for the six compounds were performed using SwissADME (Daina et al., 2017) to evaluate individual ADME behaviors of target compounds.

Despite the fact that chemical compounds have activity and selectivity, they may not be viable candidates for drug development. There are a number of rules and models to follow in the classification of such compound as potential drug candidate. The most widely accepted one is Lipinski’s rule of five (Lipinski et al., 1997). If two parameters are out of range, “poor absorption or permeability is possible”.

The compound may be absorbed in gastrointestinal tract if any one of the parameters is not met. Hence, Lipinski’s rule suggests that active molecules with good bioavailability and oral absorption have molecular weights <500, Log P < 5, less than five hydrogen bond donor (HBD) groups, and less than 10 hydrogen bond acceptor (HBA) groups (Duchowicz and Castro, 2009). A candidate that violates more than one of these criteria is less likely to be developed as a prospective oral drug.

The studied compounds are confronted to the Lipinski criteria, the results show that only four compounds listed in Table 4 meet perfectly these criteria. Thus, there would be no problem with oral bioavailability for all proposed compounds.

The Veber rule defines drug-likeness constraints as rotatable bond (RB) count ≤10 and polar surface area (PSA) ≤140 (Veber et al., 2002). PSA is strongly related to the absorption properties of active compounds. The obtained PSA for compound 8 is greater than 140, suggesting that this compound has strong polarity and is not readily absorbed by membranes. In addition, the obtained PSAs for compounds 6, 7, 9, 10, and 11 are all less than 140; which indicates that these active compounds have good oral absorption; namely good solubility and better penetration through protein membranes. The number of RBs for the six tested compounds are less than 10; as the number of RBs increases, the molecules become more flexible and more adaptable for efficient interaction with a particular binding pocket. The screening of potential candidates using Veber’s rules, allowed us to select only five compounds (6, 7, 9, 10, and 11) that meet the criteria of drug-likeness assessment, whereas, compound 8 is rejected for reason of violation of one Veber criterion (Table 3).

In drug discovery schemes, oral administration and high solubility will favor complete absorption, while poor solubility limits the absorption of the drug in the gastrointestinal tract. Data listed in Table 4 suggests that compounds 6, 7, 9, 10, and 11 have good to moderate solubility, the Log S value is between −6 and −4. Thus, the afore mentioned compounds could exhibit good oral adsorption. The lower value of Log S obtained for compound 6 indicates that is hydrophilic and is more soluble (−4 < LogS < −2), and that absorption takes place in the gastrointestinal tract. The surface area and lipophilicity of this organ influences the rate of absorption of compounds. Compounds 6, 7, and 11 display high predicted gastrointestinal absorption (Table S3).

Five isoenzymes of the CYP enzyme system are mainly involved in the metabolism of most drugs. All six compounds are predicted to be inhibitors of CYP2C19 and CYP1C2. This suggested that they may be metabolized in the liver. Moreover, they are predicted not to be inhibitors of isoenzyme CYP2D6. This is very useful, because these compounds are expected to not have CYP metabolism interactions with other drugs and would not present an increased hepatotoxicity risk.

Drug oral bioavailability is the fractional extent of the drug dosage that finally reaches the therapeutic active site; an acceptable probability score of 55% indicates that the drug candidate has met the rule of five. All candidate compounds have shown a score of 55%, indicating good bioavailability.

For the parameters of the distribution, the blood–brain barrier (BBB) protects the brain from harmful substances. The capacity of a medicine to cross into the brain is a critical criterion to consider when lowering side effects and toxicities or improving the efficacy of treatments having pharmacological action in the brain. It is believed that no compounds pass through the BBB; also, the results show that there is no P-gp substrate for all the screened compounds.

When a drug is bioavailable and administered at the proper dose rate, it reaches a steady-state level. After evaluating characteristics of compound permeability through various barriers, we predicted the excretion parameters total renal clearance (log mL/min/kg); OCT2 substrate.

The OCT2 is a protein transporter that has a vital contribution to renal absorption, disposition and clearance of drug compounds. Evaluation of the transfer of a candidate compound by OCT2 provides useful information as regards not only its clearance but also its potential contraindications. In silico analysis revealed that all compounds except 6 and 7 were not substrates of human OCT2, which is important for the excretion of cationic molecules. So, compounds 6 and 7 are predicted to undergo a renal uptake process and stay longer in the body unlike other compounds that will not have a problem with renal absorption.

The faster the excretion process, the greater the compound's CLTOT value, From Table S4 it can be seen that the log CLTOT value of the test compounds ranges from −0.151 to 0.045 ml/min/kg and from these values can be predicted that all compounds tested are relatively rapidly excreted from the body.

3.3 Molecular dynamics simulations

The MD simulation of Mpro and associated complexes was performed for 100 ns. The RMSDs of the proteins and associated complexes were calculated with respect to their initially prepared structures (Fig. 2A). The RMSD values of each system reached equilibrium after 40 ns and then leveled out. The average RMSD of Mpro alone is 0.232 ± 0.052 nm. The RMSDs of complexes of compounds 6, 7, and 11 are 0.206 ± 0.043, 0.265 ± 0.034, and 0.197 ± 0.028 nm, respectively. The RMSD curves of Mpro alone and all associated complexes were similar with only minor deviations. The RMSD of the entire system is less than 0.3 nm, indicting the good stability of the complexes in the aqueous environment.

The average RMSF values of backbones C_α atoms of Mpro alone and associated complexes are calculated to assess dynamic behavior of amino acids of the SARS-CoV-2 Mpro. The analysis of RMSF gives useful insights regarding the structural fluctuations of different regions of the protein. Greater fluctuations in residues signify a less stable nature of the protein. The RMSF of Mpro alone and associated complexes is shown in Fig. 5B and remained below 0.2 nm for the majority of the amino acids of Mpro, indicating the stability of the ligands with the test protein. However, several fluctuations are observed at the N- and C-termini that may be due to the hanging position of these residues. Several fluctuations are also found in the certain regions of Mpro in the presence of ligands, which may be due to the dynamic behavior of ligands within the bonding region. The average RMSF of individual atoms of ligands were also calculated to check the dynamics of ligand and showed some variation, denoting the dynamical shift from their initial position.

The R_g of Mpro in the absence and presence of ligands was also calculated (Fig. 3A) as this indicates the stability of protein during MD simulation. In general, a compact protein shows less variation in R_g values compared with those of an expanded protein. The average R_g values of Mpro alone and in complex with compounds 6, 7, or 11 were 2.188 ± 0.013, 2.198 ± 0.018, 2.207 ± 0.012, and 2.233 ± 0.021 nm, respectively. These values signify the stability of the ligands with Mpro. The stability was further verified by calculating the changes in SASA. The SASA of Mpro and associated complexes is shown in Fig. 3B. The SASA of all complexes are similar to Mpro alone, further validating the stability of the ligands with the protein.

The interactions of ligands with Mpro were studied by calculating the hydrogen bond profiles. The number of hydrogen bonds with respect to time is shown in Fig. 4A. Hydrogen bonds are present between the ligands and Mpro over the entire simulation period. The residues of Mpro that are involved in hydrogen bond formation with compound 6areHis41, Asn142, Gly143, Ser144, Cys145, His163, and Gln189. Similarly, Thr24, Thr25, Thr26, His41, Ser46, Met49, Leu50, Asn142, Gly143, Ser144, Cys145, His163, Glu166, and Gln189 are involved in hydrogen bond formation with compound 7, and Ser144, His163, Arg188, Gln189, and Gln192 formed hydrogen bonds with compound 11; the lower number of hydrogen bonds formed with compound 11 may be due to the interaction of water molecules at the binding sites.

Close

Fig. 4

(A) Number of hydrogen bonds between SARS-CoV-2 Mpro and ligands (compounds 6, 7, and 11). (B) Percentage of secondary structure in Mpro and when in complex with ligands (compounds 6, 7, and 11).

Export to PPT

Close

Fig. 5

(A) Docked conformation poses of ligands 6, 7 and 11 inside the ACE2 receptor pocket and their two-dimensional visualization; (B) superimposed view of ligands 6, 7 and 11 inside the ACE2 receptor pocket.

Export to PPT

The effect of binding of compounds 6, 7, and 11 on the secondary structure of Mpro was also assessed. The percentages of secondary structures in Mpro and associated complexes with compound 6, 7, and 11 are shown in Fig. 4B. The percentages of coil, β-sheet, β-bridge, bend, turn, and α-helix were 25.44%, 26.35%, 2.02%, 7.05%, 16.44%, and 18.30%, respectively. The interaction of ligands with Mpro resulted in insignificant changes in the secondary structures of the protein. This data further validates the structural stability of Mpro in complex with compound 6, 7, or 11.

The binding energies and forces involved in the complexation of compounds 6, 7, and 11 with Mpro were studied using the MM-PBSA analysis. The binding energy was calculated for 100 snapshots extracted from the trajectory at equal intervals of the entire MD simulation.

The different energies observed in the principal interaction of the ligands with Mpro are presented in Table 4. In typical drug-protein interactions, the major forces observed are hydrogen bonds, electrostatic forces, hydrophobic interactions and Van der Waals forces, which positively or negatively contribute to the total binding energy (Qais et al., 2021). The energetic contribution of the Van der Waals forces is the major contributor and most favorable in the overall binding. Electrostatic energy and SASA energy also positively contributed to the binding of all ligands with Mpro. In contrary, altered solvation energy binding of all ligands to Mpro.

The total average binding energy for binding of compounds 6, 7, and 11 with Mprowas −69.489 ± 1.388, −81.212 ± 1.267, and −90.842 ± 1.079 kJ/mol.

3.4 Ligand-receptor interactions with human ACE2 receptor

To further support the results of molecular docking and molecular dynamics as well as ADMET properties related to the potential candidate molecules 6, 7 and 11 to inhibit SARS CoV-2 Mpro, we evaluate the ability of these molecules to interact with the active sites of the human ACE2 receptor (PDB 1R4L) (https://www.rcsb.org/structure/1R4L). This is to predict the most important interactions likely to occur in the inhibition of human ACE2 activity by the disulfide derivatives. We perform this procedure in order to predict the modes of interaction of molecules 6, 7 and 11 with human ACE2 receptors responsible for viral binding to host cells (Mohapatra et al.; Abdalla et al.) and to identify active reference sites likely to be targeted by disulfide derivatives to disrupt the interaction of SARS-CoV-2 virus with ACE2 receptors and thus inhibit entry of this virus. For this purpose, we prepared the ACE2 protein and its ligands 6, 7 and 11 with the same tools and methodology used to prepare the main protease of SARS-CoV-2 in this work. Coordinates (x = 40.605, y = 2.495, 22.919) and size (80x80x80) containing the entire ACE2 protein structure were identified as the insertion site for the 3 ligands. The summary of the obtained results by molecular docking is presented in Table 5 and the different conformations poses of the ligands in the ACE2 receptor pocket are presented in Fig. 5.

Through the results presented in Table 5, we notice that the binding energies of ligands 6, 7 and 11 with the predicted reference sites are high and close to each other.

Therefore, it is difficult to prefer one molecule among others, and this may be an appropriate reason to suggest the three-screened molecules as potential inhibitors of ACE2 as well as Mpro towards removing Coronavirus in the future.

Finally, we suggest the disulfides 6, 7 and 11 as SARS CoV-2 Mpro inhibitors because of their good interaction ability with tested protein. The compounds have also excellent bioavailability and do not pose toxicity concerns. The complexes of these compounds are also stable when simulated in aqueous medium. Therefore, based on these observations, the selected three compounds are potential candidates of SARS CoV-2 Mpro inhibitors.