Synthesis, characterization, computational analyses, in silico ADMET studies, and inhibitory action against SARS-CoV-2 main protease (Mpro) of a Schiff base

COVID-19 disease caused by the severe acute respiratory syndrome coronavirus (SARS-CoV-2) has struck the whole world and raised severe health, economic, and social problems. Many scientists struggled to find a vaccine or an antiviral drug. Eventually, both vaccines and recommended drugs, repurposed drugs, or drug combinations were found, but new strains of SARS-CoV-2 continue to threaten human life and health. As part of the fight against COVID-19 disease, this study involves an in silico molecular docking analysis on the main protease (Mpro) of SARS-CoV-2. To this aim, a Schiff base compound was synthesized and characterized using spectroscopic techniques, including X-ray, FTIR, and UV-Vis. Surface analysis and electronic properties of this molecule were investigated using the DFT method. The drug-likeness parameters of the title compound were studied according to the rules of Lipinski, Veber, Ghose, Egan, and Muegge and were found in agreement with these rules. In silico toxicity analyses revealed that the new compound is a potentially mutagenic and carcinogenic chemical. The title compound was predicted to be an inhibitor of cytochrome P450 enzymes (5 CYPs). This inhibitory effect indicates a weak metabolism of the molecule in the liver. In addition, this compound was displayed good intestinal absorption and blood-brain barrier penetration. The druggability properties of the title compound were investigated, and SwissTargetPrediction predicted it to be a protease inhibitor. In this context, the SARS-CoV-2 main protease was selected as a biological target in molecular docking studies. Docking results were compared with the known native ligand N3 inhibitor. The value of binding energy between the Schiff base compound and the binding pocket of the main protease is higher than that of the reference ligand N3. The calculated free energies of binding of the Schiff base compound and the reference ligand N3 are −8.10 and −7.11 kcal/mol, respectively.


Results and discussion
molecule, the energy values of the HOMO, the LUMO, and the energy gap were found to be -5.629, -2.053, and 3.576 eV, respectively. Other global reactivity parameters can be seen in Table 1. The molecule has the smallest value of the energy gap among the similar studies we have performed previously [52][53][54]. From the calculated results, we can draw these conclusions about our compound: (i) The lowest value of the frontier orbital gap of the title molecule among the similar compounds [52][53][54] indicates that the molecule has high chemical reactivity, low chemical and kinetic stability, easy charge transfer, high polarizability, and softness. (ii) A high electrophilicity index (4.125) indicates a high electrophilic nature and good biological activity, while a relatively high softness value (0.279) indicates an increased probability of toxicity [55]. (iii) The effect of HOMO, LUMO, and energy gap values of the compounds on biological activity has been reported [56]. Low values of energy gap and LUMO lead to increased biological activity. This is due to the low energy required for electronic excitation and the strong charge transfer interaction between donor and acceptor atoms [57][58][59].     Chemical potential (μ = -(I+A)/2) -3.841 Electrophilicity index (ω = μ 2 /2η) 4.125 Maximum charge transfer index (ΔN max = -μ/η) 2.148

Hirshfeld surfaces and fingerprint analysis
To learn about the intermolecular interaction species in the molecule and their quantitative contributions, we performed Hirshfeld surface and 2D fingerprint analyses. For the above purpose, Crystal Explorer 17.5 software requires the CIF file of the compound. The calculated Hirshfeld surfaces of the title compound are shown in Figure 5 in six different maps, including dnorm, di, de, fragment patch, curvedness, and shape index. In the dnorm mapped surface, red dots show the interaction distance shorter than the sum van der Waals (vdW) radii of two atoms. White regions show the close interactions to the vdW radii, and finally, the more extended contacts from the sum of vdW radii are represented by blue areas [60]. On the dnorm surface, the bright red spot is on the O2 atom, indicating the presence of hydrogen bonding at this atom. Fingerprint plots showing the percentage contributions of each interaction type are shown in Figure 6 (If the contribution is greater than or equal to 1%). The largest contribution to the percentage of secondary interactions is the hydrogen‧‧‧hydrogen (50.8%) interaction, a type of vdW force. The second largest contribution is the hydrogen bonding interactions between oxygen and hydrogen (20.2%). The other contributors are C‧‧‧H (14.7%), C‧‧‧C (6.1%), and N‧‧‧H (5.8%).

Docking studies
Docking experiments were performed on the active residues of the SARS-CoV-2 main protease to determine the inhibitory activity of the synthesized compound. The 3D crystal structure of the main protease with the native N3 inhibitor (PDB entry: 6LU7 [61]) was retrieved from the RSCB PDB website (https://www.rcsb.org/). Before the docking experiments, the protein and ligand structures were prepared by removing water, adding polar hydrogens, merging nonpolar hydrogen atoms, and adding charges using the AutoDock and Autodock tools. The grid box has centered on the active residues [62], and the grid dimensions are given in Figures 7 and S3. The docking experiments were performed using the Lamarckian genetic algorithm. In the docking experiments, we used a semiflexible docking method (rigid target/flexible ligand).      N3 is a peptidomimetic inhibitor of the main protease of SARS-CoV and SARS-CoV-2 [61,63]. It was used for comparison in this study. Both the native ligand N3 inhibitor and query compound were docked to the active sites of the target protein. Docking results of the title compound, including binding modes, interacting residues, and binding free energy, are given in Figure 7. The docking experiment was repeated ten times for the title compound. Docking scores were determined with a standard deviation of 0.057. The median value of the docking experiments was determined and accepted as the final docking score (-8.10 kcal/mol). The ligand efficiency was found to be -0.32. The calculated docking parameters for ten docking experiments are shown in Figure S4. The most stable conformations of the reference N3 inhibitor and query compound (Figure 8a), the top ten conformations produced in the active pocket of SARS-CoV-2 main protease (Figure 8b), and the whole protein surface, including the reference molecule and query compound (Figure 8c), were shown in Figure 8. According to the docking results, the query compound has higher binding energy than the reference N3 inhibitor. In our study, the binding energies of the native ligand N3 and the query compound were calculated to be -7.11 and -8.10 kcal/mol, respectively. Query compound is bound to the target protein via both hydrophobic interactions (LEU167: 3.91 Å, GLN192: 3.67 Å, GLN189: 3.33 Å) and hydrogen bond (TYR54: 2.11 Å) ( Figure 9). We compared the results of our study with those of the other studies ( Table 2). For Mpro (3CLpro) of SARS-CoV-2, the studies listed in Table 2 show the chemical class studied against Mpro, common functional groups with our compound, software, docking scores, and interaction status with the catalytic dyad of Mpro. Although our docking score is high, no interaction with the catalytic dyad of Mpro was detected in our complex interactions. Therefore, the title compound may not provide the desired inhibitory effect on the Mpro of SARS-CoV-2.

Druglike nature, medicinal chemistry and druggability
We examined some essential physical and biological parameters in medicinal chemistry using the SwissADME [40] web tool developed by the Swiss Institute of Bioinformatics. These parameters compose of six sections (Table 3), including physicochemical properties, lipophilicity, solubility, pharmacokinetics, drug-likeness, and medicinal chemistry. Druggability predictions ( Figure 10) of the title compound to determine the biological targets were performed using SwissTargetPrediction [64]. In Table 3, the pink-colored area of the polygon on the left shows the suitable physicochemical space for oral bioavailability, the white area shows the unsuitable space, and the red lines sign out the position of our compound in the whole space. This hexagon is related to lipophilicity (LIPO), size (SIZE), polarity (POLAR), solubility , and flexibility (FLEX) and is calculated by using appropriate domain borders of parameters, including XLOGP, molecular weight, topological polar surface area (TPSA), Log S, fraction of sp 3 , and number of rotatable bonds, respectively. As it can see from the red outline of the polygon, our compound has settled in the range suitable for oral bioavailability. When the BOILED-Egg model of the title molecule is examined from Figure  11, it can see that the molecule has a good intestinal absorption and can cross the blood-brain barrier. Cytochrome P450 enzyme families (CYPs) influence the pharmacokinetics of a drug. We investigated 5 CYPs, including CYP1A2, CYP2C19, CYP2C9, CYP2C6, and CYP3A4, which are related to the 80% of the metabolism of drugs in clinical use [65]. According to the predictions, the title compound is a potential CYPs inhibitor for five CYPs. This inhibitory effect means that our compound as a drug molecule candidate suppresses CYPs enzyme activity and decreases the metabolic rate in human liver, so the pharmacokinetic properties might not reach the desired efficiency. Drug-likeness was derived from the structures and properties of existing drugs and drug candidates. Before drug discovery, it is important to filter out unsuitable compounds [66]. This term was defined by Lipinski as meeting some proposed criteria for drug candidates [67]. Lipinski states that poor absorption and permeation are more likely in the following situations: i) Molecular weight is higher than 500. ii) LogP value is higher than 5. iii) Hydrogen bond acceptors          [72]. According to the mentioned rules, our molecule does not violate the above drug-likeness rules. We also investigated druggability predictions for our compound. The calculation results ( Figure 10) show that our compound can inhibit the following enzyme classes in the top 15 lists: protease, kinase, phosphodiesterase, family A G protein couplet-receptor, oxidoreductase, voltage-gated ion channel, and cytochrome P450.

Potential multitarget identification with fingerprint methods
To improve the information about the bioactivity properties of our compound, we used a web server, the polypharmacology browser (PPB), www.gdb.unibe.ch. This web server is used to identify potential targets of a compound based on six different fingerprints and some combinations. The results of PPB are given according to the various algorithms: atom pair fingerprint (APfp), extended atom pair fingerprint (Xfp), molecular quantum numbers (MQN), scalar fingerprint counting the occurrence of characters in SMILES (SMIfp), (SMIfp), substructure fingerprint (Sfp), and extended connectivity fingerprint (ECfp4). APfp works with molecular shape; Xfp perceives pharmacophores; MQN perceives atoms, bonds, polarity, ring features, constitution, topology, and molecular shape; SMIfp uses rings, aromaticity, and polarity; Sfp works with the detailed substructures; ECfp4 uses the combination of detailed substructures and pharmacophores [41]. We tabulated the top 20 targets selected by the six fingerprinting algorithms for our compound. The results were ordered by the calculated cumulative density (p-values) for each target in Table 4. The red hexagon in this table indicates that the specified fingerprint algorithm did not find a target; the green hexagon indicates a lower p-value (from 0.01 to 0) and a lower probability for the target; the blue hexagon indicates potential targets for which the p-value is greater than 0.01. Provided that the estimated p-value is greater than 0.01, we indicate the exact p-values. The ChEMBL-ID and common names of the targets, and the explanation of each target are listed in Table 4. The results show that our compound has a similar fingerprint to the molecules with the indicated number on the right side of  [88]. Our compound could interact with the listed targets as a potential ligand molecule and act as an inhibitor against target-related diseases mentioned above.

Toxicity analysis
Two web servers, ProTox-II and pkCSM were used to determine the toxicity parameters of the title compound. The calculated toxicity endpoints and models are shown in Table 5 and Figure 11 (ProTox-II) and Table 6 (pkCSM). From Table 5 and Figure 11, we can see that the title compound is classified as mutagenic and carcinogenic with a probability of 79% and 61%, respectively. From Table 6, we can see that the title compound has two alerts related to AMES mutagenicity and hepatoxicity. In summary, we can define the title compound as mutagenic, carcinogenic, and hepatotoxic. These toxic effects are generally reported as structural warnings for compounds with the nitro substituent [89][90][91][92]. Despite these   Target not found by fingerprint.  known facts for the toxicophoric nitro groups, many drugs containing the nitro group, such as flutamide and niclosamide, have been approved by the FDA, and the nitro group plays a direct role in the efficacy of a drug molecule [92]; therefore, we cannot exclude the compounds containing the nitro group, and can still consider them as drug candidates.

Gastrointestinal absorption and brain penetration
We examined the title compound to determine human intestinal absorption (HIA) and blood-brain barrier (BBB) penetration, two crucial pharmacokinetic properties in drug discovery. These properties were investigated using the Brain Or IntestinaL EstimateD permeation method (BOILED-Egg) developed by Daina and Zoete [93]. This model uses two physicochemical parameters, WLOGP and TPSA. It simultaneously predicts the intestinal absorption and brain access of the molecules. For our compound, the estimated model is shown in Figure 12, in which the yellow area (yolk) shows that the compounds can passively penetrate through the blood-brain barrier. The white region signifies a physicochemical space where the gastrointestinal system can absorb the molecules. In this graph, the white and yellow areas are not mutually exclusive. The small red cycle in the yolk shows that our compound can passively cross the blood-brain barrier and be absorbed by the human gastrointestinal tract. As a result, the molecule is active in the BBB, and the gastrointestinal tract can absorb it.

Conclusion
A Schiff base compound was synthesized via a condensation reaction between an aromatic aldehyde and amine molecule. The single crystal was analyzed using the X-ray diffraction method. The mentioned compound has space group P21/n and crystallized ii. Molecular orbital analysis provided information on the intramolecular charge transfer, molecular softness, stability, reactivity, and toxicity. The energies of the energy gap, HOMO, and LUMO orbitals were calculated to be -5.629, -2.053, and 3.576 eV, respectively. As a result, charge transfer between the HOMO and LUMO orbitals occurs easily; the title compound has high chemical reactivity, biological activity, polarizability, probably high toxicity, and low kinetic and chemical stability.
iii. Molecular stability is mainly established by H‧‧‧H interaction, followed by O‧‧‧H, C‧‧‧H, C‧‧‧C and N‧‧‧H interactions and others.     iv. Docking experiments were performed to determine the inhibitory effect of the candidate molecule. The title compound and reference inhibitor were docked to the COVID-19 main protease (Mpro). Our docking calculations showed that the binding energy of the complex of query compound/SARS-CoV-2 (-8.10 kcal/mol) is higher than that of the complex of N3/SARS-CoV-2 (-7.11 kcal/mol). Therefore, the title compound is a potent candidate for inhibition of the main protease. v. The title compound settled in the suit drug domain region according to the SwissADME algorithm and obeyed the known drug-likeness rules (Lipinski, Veber, Ghose, Egan, Muegge). pkCSM and ProTox-II tools uncovered mutagenic, carcinogenic, and hepatotoxic predictions on the title compound. Metabolism of the molecule in liver is likely to be low, as it was found to be an inhibitor of 5CYPs. There is no concern regarding human intestinal absorption and brain permeability.

Conflict of interest
The authors declare that there is no conflict of interest regarding the publication of this paper. Table S1. Single-crystal X-ray data of the title compound and the refinement parameters.