Identification of lead small molecules for the design and development of potent severe acute respiratory syndrome coronavirus 2 main protease inhibitors

Abstract

The repercussions of the COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are catastrophic, and the world has yet to achieve full recovery. Several inhibitors targeting SARS-CoV-2 main protease are experiencing diminished efficacy owing to resistance-inducing mutations. The current situation implies that the quest to find potent and resilient SARS-CoV-2 main protease drugs to overcome resistance must be a continuous effort. Here, multiple receptor virtual screening and molecular dynamics (MD) simulation techniques were employed to identify novel binders from an integrated small-molecule database as leads for the discovery, design, and development of antivirals immune to resistance by SARS-CoV-2 main protease. The small-molecule database was initially screened separately against five SARS-CoV-2 main protease structures with different substrate-binding site conformations using the GOLD program, after which the fitness score of a control compound was used as the cutoff to create a shortlist of potential hits in each case. Then, 21 compounds at the intersection of all five shortlists were selected as virtual screening hits. The hits were subjected to MD simulations, identifying four novel compounds capable of remaining bound to SARS-CoV-2 main protease for up to 100 ns. Analysis of the mode of binding and interactions between each of the four compounds and SARS-CoV-2 main protease revealed that the compounds fit better into the conserved subpockets of the substrate-binding site than the control and interact with important amino acid residues. Conjointly, MD simulations, binding energy, and toxicity analysis results further demonstrated that the compounds are promising leads for the discovery, design, and development of potent drugs to augment the fight against SARS-CoV-2 main protease resistance.

Keywords

COVID-19 SARS-CoV-2 main protease multiple receptor virtual screening molecular dynamics

Introduction

The repercussions of the COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), a positive-sense single-stranded RNA virus, on health, education, economies, geopolitics, and sociocultural lifestyles are catastrophic, and the world has yet to achieve full recovery.^1
–5 The World Health Organization reports that COVID-19 remains a concern because new SARS-CoV-2 variants, infections, and related deaths are being recorded, and the global death toll from COVID-19 stood over seven million as of September 2024.⁶ Despite the identification of numerous SARS-CoV-2 drug targets^7
–9 and potential therapeutic agents,^10,11 the consistent emergence of new SARS-CoV-2 variants poses a threat to COVID-19 treatment owing to drug resistance.^12
–20 To tackle these challenges, the discovery of a variety of SARS-CoV-2 inhibitors or leads for the development of effective drugs is required.

The SARS-CoV-2 main protease (SARS-CoV-2-M^pro), referred to as 3-chymotrypsin-like protease, is a cysteine protease that plays an important role in the viral life cycle because it is associated with the processing of polyproteins to produce functional proteins and enzymes essential for viral replication and assembly. SARS-CoV-2-M^pro is functional in the dimeric form,^21,22 and its monomeric structure consists of three domains (I, II, and III) with the catalytic H41-C145 dyad situated within the cleft of domains I and II (see Figure 1). Surrounding the catalytic dyad is a binding pocket (defined by residues from all three domains) consisting of subpockets S1, S2, S3, S4, and S5 to contain the substrate (see Figure 1).^23,24 Upon binding of the polyprotein substrate to M^pro, H41 abstracts a proton from the thiol group of C145 to give rise to a thiol nucleophile. The nucleophile attacks the scissile amide bond, resulting in bond breaking, the release of the first product, and deprotonation of H41 to restore its original state. Subsequently, the generated thioester is attacked by a nucleophilic hydroxide ion (OH⁻) formed by the deprotonation of a water molecule by H41, leading to product release and restoration of the catalytic dyad.^25,26 Considering the critical role of M^pro and the fact that it is conserved among coronaviruses and lacks a human analog, it has been recognized as an attractive target for SARS-CoV-2 drug development.^27,28 Compounds that strongly bind to the substrate-binding site of M^pro can act as inhibitors by competing with or dislodging the polyprotein substrate. Although many antiSARS-CoV-2-M^pro inhibitors have been identified,^29
–31 the majority experience diminished efficacy owing to mutations in M^pro that confer drug resistance.^32
–37 For instance, a decreased potency of nirmatrelvir, the first SARS-CoV-2-M^pro inhibitor approved for clinical use, has been observed following T21I, L50F, N142L, E166M/V, Q189E/I, Q192T, P252L, and T304I mutations in M^pro.^35,38 The situation implies that the quest to find potent and resilient SARS-CoV-2-M^pro drugs to overcome resistance must be a continuous effort.

Figure 1.

Structure of the main protease (M^pro) of SARS-CoV-2. The domains 1, 2, and 3 are colored cyan, gray, and green, respectively. The substrate-binding site and subpockets (S1, S2, S3, and S4) are shown.

In this study, multiple receptor virtual screening (MRVS) and molecular dynamics (MD) simulation techniques were employed to identify novel binders from a small-molecule database generated by merging peptidomimetic and M^pro-targeted libraries as leads for the discovery, design, and development of antiviral agents immune to resistance by SARS-CoV-2-M^pro. The small-molecule database was originally screened independently against five structures of M^pro with varying substrate-binding site conformations using the GOLD program,^39
–41 after which the fitness score of a control compound was used as the cutoff to create a shortlist of potential hits in each instance. Subsequently, 21 compounds at the intersection of all five shortlists were selected as virtual screening hits. The hits were subjected to MD simulations, identifying four novel compounds capable of remaining bound to M^pro for up to 100 ns. Analysis of the mode of binding and interactions between each of the four compounds and M^pro revealed that the compounds fit better into the conserved subpockets of the substrate-binding site than the control and interact with important amino acid residues. Collectively, MD simulations, binding energy, and toxicity analysis results further demonstrated that the compounds are promising leads for the discovery, design, and development of potent drugs to augment the fight against SARS-CoV-2-M^pro resistance. This study covers a wider chemical space and identifies novel compounds as drug leads.

Methods

Selection and preparation of SARS-CoV-2-M^pro receptor models

The three-dimensional coordinates of SARS-CoV-2-M^pro in both apo and holo forms were obtained from the Protein Data Bank (PDB). Three apo crystal structures with PDB IDs 7BAJ,⁴² 7DAV,⁴³ and 7JP1⁴⁴ were downloaded. For the holo forms of M^pro, crystal structures with PDB IDs 6LU7 (M^pro complexed with covalent inhibitor N3),⁴⁵ 6LZE (M^pro complexed with covalent inhibitor 11a),⁴⁶ 6WTT (M^procomplexed with covalent inhibitor GC376),⁴⁷ 6Y2F (M^pro complexed with covalent inhibitor 13b),⁴⁸ 7D3I (M^pro complexed with covalent inhibitor MI-23),⁴⁹ 7CA8 (M^pro complexed with noncovalent inhibitor shikonin),⁵⁰ and 7P2G (M^pro complexed with noncovalent inhibitor Z222979552)⁵¹ were retrieved. These holo structures were selected based on the presence of well-characterized inhibitors.⁵² All retrieved structures were examined for suitability as receptors for structure-based virtual screening (SBVS). The final selection of receptor models for MRVS was based on the configuration, orientation of amino acid side chains, and uniqueness of the substrate-binding site. Receptors with substrate-binding sites blocked by the side chains of amino acid residues were rejected. Ultimately, five structures, including 7BAJ (apo M^pro), 6LU7 (N3-bound M^pro), 6Y2F (13b-bound M^pro), 7CA8 (shikonin-bound M^pro), and 7P2G (Z222979552-bound M^pro), were selected for MRVS exercises. All water molecules were deleted from each of the selected structures, followed by replacement of all missing atoms and amino acid residues using the SWISS-MODEL server.^53,54 Each structure was fully protonated, energy-minimized, and saved as a mol2 file using the protein preparation tool incorporated in the BIOVIA Discovery Studio 2016.

Construction and preparation of small-molecule library

An integrated small-molecule library was constructed by merging a peptidomimetic library obtained from Life Chemicals and M^pro-targeted libraries sourced from Life Chemicals, Enamine, Otava, and TargetMol databases. A peptidomimetic library is a collection of compounds with special features that allow them to mimic the natural peptide substrate/ligand of a receptor.⁵⁵ A targeted library, on the contrary, comprises a selection of compounds possessing unique characteristics that may enhance their interaction with a particular target.⁵⁶ Four known inhibitors of M^pro, including N3,⁴⁵ 13b,⁴⁸ shikonin,⁵⁰ and Z222979552,⁵¹ were incorporated into the library as control compounds. These inhibitors are ligands for the holo-M^pro receptors selected for MRVS. The automatic ligand preparation tool in BIOVIA Discovery Studio 2016 was used to prepare the library, which entailed protonation, energy optimization, deduplication, and fixing the bad valences of the constituent compounds. A total of 15,427 unique compounds were prepared and saved as a mol2 file. Table 1 displays a breakdown of the number of compounds in the databases that constitute the integrated small-molecule library.

Table 1.

Composition of the integrated small-molecule library.

Database	Type	Number of compounds
Life Chemicals	Peptidomimetics	5836
Life Chemicals	M^pro-targeted	2334
Enamine	M^pro-targeted, noncovalent	3440
Enamine	M^pro-targeted, covalent	2636
Otava	M^pro-targeted	1017
TargetMol	M^pro-targeted	159
Control	—	5
Total		15,427

M^pro: main protease.

MRVS and hits selection

The GOLD docking program^39
–41 was used for all screening exercises. The suitability of the GOLD program for this study was validated by its ability to reproduce the experimental binding mode of the control with a root mean square deviation (RMSD) of 1.99 Å (see Supplemental Figure S1). GOLD employs a genetic algorithm (GA) to search the conformational space and predict the possible modes of ligand binding to a receptor. The predicted binding modes can be ranked using one of the built-in scoring functions (CHEMPLP, GoldScore, ChemScore, and ASP) or a user-defined function. The GoldScore function used in this study comprises a GOLD fitness score, which is the sum of four terms: the protein–ligand hydrogen bond (H-bond) score (S_{hb_ext}), protein–ligand van der Waals score (S_{vdw_ext}), contribution due to ligand intramolecular H-bonds (S_{hb_int}), which is switched off by default, and contribution of the ligand intramolecular strain (S_{vdw_int}). The GOLD fitness score is given by equation (1)

\begin{matrix} G O L D F i t n e s s S c o r e = S_{h b_e x t} + S_{v d w_e x t} \\ + S_{h b_i n t} + S_{v d w_i n t} \end{matrix}

(1)

The MRVS model adopted in this study requires sequential screening of the small-molecule library against the five M^pro receptor models. In each instance, the prepared receptor was uploaded to the GOLD wizard. The substrate-binding site of each receptor for the VS exercise was defined by the following amino acid residues: T24, T25, T26, L27, N28, H41, C44, T45, S46, E47, D48, M49, L50, P52, Y54, Y118, N119, F140, L141, N142, G143, S144, C145, H163, H164, M165, E66, L167, P168, T169, G170, V171, H172, F185, D187, R188, Q189, T190, A191, and Q192. After uploading the small-molecule library to the GOLD wizard, selecting the GoldScore scoring function, and specifying the default VS settings as the GA search option, the compounds were docked onto M^pro.

The protocol for selecting screening hits requires GOLD-generated best-ranking files, which contain the highest scoring poses for all compounds, along with a control to determine the selection cutoff. The control compounds N3 and 13b emerged as the best; however, compound 13b was chosen for hit selection because N3 obtained a very high score in its receptor (N3-bound M^pro) and would exclude most promising candidates. First, the best-ranking file corresponding to each of the five receptors was sorted in decreasing order of GOLD fitness score. The fitness score of the control compound in each sorted best-ranking file was used to set the selection cutoff, resulting in the shortlisting of all compounds with scores higher than the control as potential hits and the rejection of those with lower scores than the control. Thus, five sets of shortlisted compounds consisting of 532, 2488, 3006, 1353, and 546 entries for the apo, N3-bound, 13b-bound, shikonin-bound, and Z222979552-bound receptors, respectively, were generated. Eventually, the compounds at the intersection of all five shortlists were selected as VS hits, resulting in the identification of 21 promising leads. The 21 compounds were redocked onto the control-bound receptor (13b-bound M^pro) using the default settings of GOLD, and complexes involving the best ligand poses were saved as PDB files for further analyses. Although each of the selected 21 compounds is a potential M^pro binder, the complex structure of each compound with M^pro was subjected to MD simulations to identify compounds and scaffolds that could remain stably bound to M^pro. The simulation of each compound was categorized into phases 1, 2, and 3, involving 20 ns, 50 ns, and 100 ns simulation times, respectively. The RMSD and final pose of each compound were analyzed in each phase to ascertain stability, which served as a yardstick to determine progression to the next phase. The focus was then placed on the compounds that remained stably bound to M^pro for up to 100 ns. Two additional independent simulations were conducted for each stable compound.

MD simulations

Conventional MD simulations were conducted using the GROMACS 2020.4 simulation package.^57,58 The topology of M^pro was constructed using the Amber 99SB force field⁵⁹ and the TIP3P water model.⁶⁰ In contrast, the topologies of all ligands (selected compounds) were generated and the corresponding atom types and restrained electrostatic potential (RESP) charges⁶¹ were assigned using the antechamber tool included in the AMBER 21 suite.^62,63 In each instance, the M^pro–compound complex was situated in the middle of a cubic box with a minimum separation distance of 10 Å between the protein surface and box edges. Subsequent to the solvation of the simulation box with an explicit TIP3P water model,⁶⁰ the system’s net charge was neutralized at an ionic concentration of 0.1 M by incorporating the requisite quantities of Na⁺ and Cl⁻ ions. The Particle Mesh Ewald method⁶⁴ was employed to address long-range charge–charge interactions. The LINCS (linear constraint solver) approach⁶⁵ was used to restrain all bonds that involved hydrogen atoms at constant equilibrium lengths. The system was energy-optimized using the steepest descent algorithm. Position restraints were applied to the non-hydrogen atoms of M^pro and the compound, and initial velocities were arbitrarily apportioned to all atoms of the system. A 5 ns equilibration was conducted at 1 bar and 300 K in the NPT ensemble, utilizing the Berendsen barostat for pressure regulation and the V-rescale thermostat for temperature coupling.⁶⁶ A production run in the NPT ensemble was performed for the requisite simulation duration following the removal of position restraints, with data saved every 1 ps.

Binding free energy calculations

The binding free energies (BFEs) and per-residue contributions (PRCs) by M^pro to the BFEs of M^pro–ligand complexes were estimated using the g_mmpbsa tool.⁶⁷ In each case, the calculations were performed on 1000 frames sampled from the final 50 ns simulation trajectory using a dielectric constant (pdie) of 6 and the default settings of all other parameters of the tool. The selection of the pdie of 6 was based on the observation that the M^pro–control complex obtained unfavorable BFEs at pdie values < 6. The BFE was calculated as the summation of the van der Waals energy (E_vdw), electrostatic energy (E_elec), electrostatic contribution to the solvation free energy (G_polar), and nonelectrostatic contribution to the solvation free energy (G_non-polar) terms, as summarized in equation (2).

B F E = E_{v d w} + E_{e l e c} + G_{p o l a r} + G_{n o n - p o l a r}

(2)

Pose, interaction, and MD simulation analyses

Ligand poses and interactions were analyzed using PyMol 1.7.4,⁶⁸ BIOVIA Discovery Studio 2016, and LigPlot 2.2 software.⁶⁹ Simulation trajectories were analyzed using the GROMACS 2020.4 tools,^57,58 including gmx rms for RMSD calculations, gmx gyrate for radius of gyration (Rg) calculations, gmx rmsf for root mean square fluctuation (RMSF) calculations, gmx hbond for H-bond analyses, and gmx covar and gmx anaeig tools for free energy landscape (FEL) analyses. For ligand RMSD and FEL analyses, the atoms of the receptor (M^pro) were least-squares fitted, and the calculations were performed on the ligand. For M^pro(receptor) backbone RMSD analysis, calculations were executed on the backbone atoms after the same backbone atoms were least-squares fitted. Regarding FEL analyses, the gmx covar utility was used to create eigenvectors, which were used as gmx anaeig input data to project principal components 1 (PC1) and 2 (PC2). The PC1–PC2 projections were used to generate FEL diagrams. For H-bond analyses, the default donor–acceptor angle of 30° and donor–acceptor separation distance of 3.5 Å of the gmx hbond tool were used.

Results

Selected compounds

Following the MRVS and ligand-selection criteria, 21 compounds emerged as hits. The chemical structures of the 21 compounds and the docking scores corresponding to each receptor model are shown in Figure 2 and Supplemental Table S1, respectively. Supplemental Table S1 shows that the compounds obtained better fitness scores than the control. For the 21 compounds, 18 entries (compounds with IDs starting with “F”) belong to the Life Chemical database, whereas two compounds (T2598 and T2727) and one compound (Z24395724) belong to the TargetMol and Enamine databases, respectively. The RMSD and pose analysis results of the 21 compounds obtained from the MD simulations and used to assess stability are presented in Supplemental Figures S2 and S3, respectively. The results show that four compounds, including T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control, remained stably bound to M^pro for up to 100 ns. Thus, the remainder of this study focused on compounds T2727, F1044-0016, F0514-3956, and F0514-4563 as representatives of the 21 VS hits.

Figure 2.

Chemical structures of the 21 compounds identified as virtual screening hits. The control and compounds that remained stably bound to SARS-CoV-2-M^pro for up to 100ns during MD simulations are indicated by green labels.

Default docking scores, poses, and interactions of compounds T2727, F1044-0016, F0514-3956, and F0514-4563

The default GOLD fitness, H-bond, and van der Waals scores obtained by compounds T2727, F1044-0016, F0514-3956, F0514-4563, and the control are presented in Supplemental Table S2. Compounds T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control, obtained fitness scores of 83.70, 84.12, 71.94, 69.35, and 68.39, respectively. Various chemical interactions and bonds occur between M^pro and compounds T2727, F1044-0016, F0514-3956, and F0514-4563. As illustrated in Figure 3(a), compound T2727 occupies subpockets S2, S3, S4, and S5 of the substrate-binding site of M^pro and forms H-bonds with amino acid residues T25, H41, F140, H164, E166, and Q192. Compound T2727 is also engaged in pi–pi interactions with H41 and P168. There are many other nonbonding contacts and interactions between T2727 and M^pro residues: M49, L141, N142, H163, M165, L167, H172, V186, D187, R188, Q189, T190, and A191. Compound F1044-0016 occupies subpockets S1, S3, S4, and S4 and makes H-bonds with E166 and Q192, as shown in Figure 3(b). A pi–pi interaction is observed between H41 of M^pro and compound F1044-0016, which makes numerous nonbonded contacts with T25, T26, L27, M49, L141, N142, G143, C145, H163, H164, M165, P168, D187, R188, and T190 of M^pro (see Figure 3(b)). Figure 3(c) shows that compound F0514-3956 occupies subpockets S1, S3, S4, and S5 of the substrate-binding site of M^pro. H-bonds are formed between G143 and E189 of M^proand compound F0514-3956. Residues T24, T25, T26, H41, M49, L141, N142, C145, H163, M165, E166, P168, D187, R188, T190, A191, and Q192 of M^pro establish nonbonded interactions with compound F0514-3956. Compound F0514-4563, which occupies subpockets S1, S3, S4, and S5, forms an H-bond with Glu166 of M^pro along with many nonbonded interactions with H41, M49, T54, F140, N142, G143, S144, C145, H163, H164, M165, P168, D187, R188, Q189, T190, and Q192, as illustrated in Figure 3(d). As shown in Supplemental Figure S4(a), the control occupies only subpockets S1, S3, and S4 and forms H-bonds with H41, G143, C145, H163, H164, and E166. A covalent bond is formed between the control compound and the thiol group of C145. Nonbonded interactions exist between the control compound and residues T26, M49, F140, L141, N142, M165, L167, P168, H172, D187, and Q189 of M^pro. The ability of compounds T2727, F1044-0016, F0514-3956, and F0514-4563 to occupy the subpockets of the substrate-binding site of M^pro better than the control explains why they obtained better docking fitness scores.

Figure 3.

Interactions between M^pro and the four representative compounds. (a) Interactions between M^pro and compound T2727. (b) Interactions between M^pro and compound F1044-0016. (c) Interactions between M^pro and compound F0514-3956. (d) Interactions between M^pro and compound F0514-4563. Carbon, nitrogen, oxygen, sulfur, fluorine atoms are depicted by balls colored in black, blue, red, yellow, and cyan, respectively. Bonds linking the atoms of T2727, F1044-0016, F0514-3956, and F0514-4563 are colored magenta. Bonds connecting the atoms of amino acid residues involved in H-bonding are colored gray. The red “eyelashes” depict nonbonded contacts. H-bonds are illustrated as green dashes.

Ligand RMSD and convergence analyses

Three independent 100 ns simulations were conducted for each of the final selected M^pro–T2727, M^pro–F1044-0016, M^pro–F0514-3956, and M^pro–F0514-4563 complexes. Three independent simulations of the M^pro–control complex were also carried out. All trajectories in each case were used to compute the RMSDs of the compound during the simulations to ascertain the repeatability of the data and ligand stability. The RMSD results for compounds T2727, F1044-0016, F0514-3956, and F0514-4563, along with the control, are presented in Figure 4(a)–(e), respectively. The similarity of the curves in each case (see Figure 4) suggests reproducibility of the results. Furthermore, the stable RMSD curves in each case are indicative of the stability of the compound and its interactions with M^pro. The RMSD trajectories in Figure 4 show that the simulations converged by 50 ns. Therefore, to compare the RMSDs of the four compounds, the average RMSD of each compound was calculated using the last 50 ns of the three corresponding simulations. Compounds T2727, F1044-0016, F0514-3956, and F0514-4563, along with the control, obtained average RMSDs of 3.87 ± 0.25 Å, 2.48 ± 0.04 Å, 4.13 ± 0.22 Å, 2.01 ± 0.04 Å, and 3.15 ± 0.11 Å, respectively. Overall, the average RMSDs of the four compounds are less than 4.20 Å, indicating stability, and they compare well with that of the control compound. Given that the standard error is narrow in each case, it is noteworthy that the most stable trajectory associated with each compound was used for the remaining analyses.

Figure 4.

Ligand RMSD data. (a) RMSD of compound T2727. (b) RMSD of compound F1044-0016. (c) RMSD of compound F0514-3956. (d) RMSD of compound F0514-4563. (e) RMSD of the control. In each case, the red, green, and black curves show results for three independent simulations.

Receptor RMSD, RG, and RMSF analyses

RMSD, RG, and RMSF analyses were performed on the backbone of M^pro in the T2727-bound, F1044-0016-bound, F0514-3956-bound, F0514-4563-bound, and control-bound forms, as well as the apo form, to monitor receptor conformation and stability during the simulations. Receptor-backbone RMSD, RG, and RMSF analyses provide valuable insights into receptor stability, compactness, and local-atomic fluctuations, respectively, which can aid in understanding biological mechanisms and phenomena. However, large changes in RMSD, RG, and RMSF could indicate receptor instability. RMSD and RG are functions of simulation time and thus were calculated using the entire 100 ns of each trajectory. The RMSF, which requires sampling conformations from the stable region of a trajectory, was computed using the last 50 ns of each trajectory. The RMSD, RG, and RMSF results are presented in Figure 5(a)–(c), respectively. Compared to the apo form, no significant conformational and structural variations attributable to ligand binding can be observed in M^pro in the T2727-bound, F1044-0016-bound, F0514-3956-bound, F0514-4563-bound, and control-bound forms. The results suggest that M^pro was stable in all ligand-bound forms during the simulations.

Figure 5.

RMSD, RG, RMSF, and hydrogen bond data. (a) Backbone RMSDs of M^pro in T2727-bound, F1044-0016-bound, F0514-3956-bound, F0514-4563-bound, control-bound, and apo forms shown as green, cyan, blue, yellow, red, and black curves, respectively. (b) Backbone RGs of M^pro in the T2727-bound (green), F1044-0016-bound (cyan), F0514-3956-bound (blue), F0514-4563-bound (yellow), control-bound (red), and apo (black) forms. (c) Backbone RMSFs of M^pro in T2727-bound (green), F1044-0016-bound (cyan), F0514-3956-bound (blue), F0514-4563-bound (yellow), control-bound (red), and apo (black) forms. (d) The number of H-bonds between M^pro and T2727, F1044-0016, F0514-3956, F0514-4563, and control shown as green, cyan, blue, yellow, and red curves, respectively.

Hydrogen bond analyses

The number of H-bonds between M^pro and each of the compounds T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control, was calculated along the corresponding simulation trajectory. The results are shown in Figure 5(d), where the consistent H-bond counts over the simulation time, as illustrated by the curves, suggest that the interactions between M^pro and the compounds are stable. For each compound, the results from the last 50 ns were used to calculate the average H-bond count. Compounds T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control, obtained average H-bond counts of approximately 4, 2, 2, 1, and 4, respectively. Compound T2727 had the same H-bond count of 4 as the control; however, the overall results suggest that the other forces described in Figure 3 play significant roles in stabilizing the interactions between M^pro and compounds F1044-0016, F0514-3956, and F0514-4563 in particular.

FEL and conformational analyses

FEL analyses were performed on compounds T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control, to determine the energy minima and most stable binding orientations explored during the simulations. The two-dimensional (2D) FEL diagrams for compounds T2727, F1044-0016, F0514-3956, and F0514-4563 are presented in Figure 6(a)–(d), respectively. The FEL diagram for the control is shown in Supplemental Figure S4(b). The energy minima in the FEL plots are colored deep red and indicate the stable binding modes of the compounds to M^pro. Thus, a single energy minimum indicates one main ligand-binding mode, whereas many energy minima with distinct energy barriers imply several potential ligand-binding modes. As illustrated in Figure 6, the FEL diagrams for compounds T2727, F1044-0016, and F0514-3956 each show one main energy minimum, whereas the FEL plot for compound F0514-4563 shows two distinct energy minima. The FEL diagram of the control compound shows two energy minima, as illustrated in Supplemental Figure S4(b).

Figure 6.

FEL diagrams for the four representative compounds. (a–d) FEL diagrams for compounds T2727, F1044-0016, F0514-3956, and F0514-4563, respectively.

To examine the ligand poses and binding modes explored during the simulations, M^pro–compound complex structures were extracted from the main energy minimum/minima of each compound and compared with the corresponding initial docking complexes through all-ligand atom superimposition. The results of the superimposition are shown in Figure 7, where the carbon atoms of the compounds in the docking complexes are colored magenta, whereas the carbon atoms of T2727, F1044-0016, F0514-3956, and F0514-4563 in the corresponding simulation complexes are colored green, cyan, blue, and yellow, respectively. The results of the superimposition involving the control are shown in Supplemental Figure S4(c). The superimposition of the atoms of compounds T2727, F1044-0016, F0514-3956, and F0514-4563 in the simulation complexes on the same atoms in the corresponding docking complexes produced all-atom RMSDs of approximately 2.38 Å, 1.73 Å, 1.70 Å, and 0.88/0.11 Å, respectively. Figure 7(d) shows minor differences between the structures extracted from bins 1 and 2 of the FEL diagram for compound F0514-4563 and the docking structure. The superimposition of the simulation poses of the control extracted from bins 1 and 2 on the initial pose generated an all-atom RMSD of 2.00 Å in each case (see Supplemental Figure S4(c)), indicating that there is no major difference between the simulated poses. All compounds, including the control, maintained the ability to occupy the subpockets of the substrate-binding site of M^pro during the simulations. Generally, similar to the control, the pose and RMSD (< 2.5 Å) results of each compound indicate that the predicted binding mode and the corresponding interactions are stable.

Figure 7.

Comparison of simulated poses of the four representative compounds with the corresponding docking poses. (a–d) Superimposed simulated poses of T2727 (green sticks), F1044-0016 (cyan sticks), F0514-3956 (blue sticks), and F0514-4563 (yellow sticks) on the corresponding docking poses (magenta sticks), respectively.

BFE and PRC analyses

The BFE analyses of the M^pro–T2727, M^pro–F1044-0016, M^pro–F0514-3956, M^pro–F0514-4563, and M^pro–control interactions produced BFEs of −112.79 ± 26.79 kJ mol⁻¹, −177.07 ± 16.70 kJ mol⁻¹, −177.83 ± 22.81 kJ mol⁻¹, −165.54 ± 12.95 kJ mol⁻¹, and −94.50 ± 15.98 kJ mol⁻¹, respectively. The PRC by M^pro to the BFEs of the M^pro–T2727, M^pro–F1044-0016, M^pro–F0514-3956, and M^pro–F0514-4563 complexes are illustrated in Figure 8(a)–(d), respectively, where only residues with favorable energy contributions ⩽ −2 kJ mol⁻¹ and unfavorable contributions ⩾ +2 kJ mol⁻¹ are labeled. As shown in Figure 8(a), the amino acid residues H41, N142, C145, M165, P168, Q189, T190, and A191 of M^pro made favorable contributions to the BFE of the M^pro–T2727 complex. Figure 8(b) shows that L27, H41, M49, C145, M165, C168, and Q189 of M^pro made favorable contributions to the BFE of the M^pro–F1044-0016 complex. The M^pro residues L27, H41, M49, N142, C145, M165, D187, and Q189 made favorable contributions to the BFE of the M^pro–F0514-3956 complex, as illustrated in Figure 8(c). For the M^pro–F0514-4563 complex, Figure 8(d) shows that L27, H41, M49, N142, C145, M165, D187, and Q189 made favorable contributions to the BFE of the M^pro–F0514-4563 complex. In each of the four complexes, E166 made the most unfavorable contribution to the BFE. For the M^pro–control complex, R4, H41, R60, R131, K137, L141, N142, C145, M165, and H172 are the main favorable energy contributors to the BFE, as shown in Supplemental Figure S4(d), where only residues with favorable energy contributions ⩽ −3 kJ mol⁻¹ and unfavorable contributions ⩾ +3 kJ mol⁻¹ are labeled. Supplemental Figure S4(d) also show that residues D48, E55, E166, D197, E240, and D289 are the main unfavorable contributors to the BFE of the M^pro–control complex.

Figure 8.

Per-residue contributions (PRCs) from M^pro to the binding free energies (BFEs) of complexes involving M^pro and the four representative compounds. (a–d) PRCs from M^pro to the BFEs of the M^pro–T2727, M^pro–F1044-0016, M^pro–F0514-3956, and M^pro–F0514-4563 complexes, respectively.

Toxicity assessment

The ProTox 3.0 web server,⁷⁰ which integrates molecular similarity, fragment propensities, pharmacophores, and machine-learning models for the prediction of various organ toxicity and toxicity endpoints, was used to assess the toxicity potentials of compounds T2727, F1044-0016, F0514-3956, and F0514-4563, as well as the control. For organ toxicity, the hepatotoxic, neurotoxic, nephrotoxic, and cardiotoxic potentials of each compound were assessed. Regarding toxicity endpoints, the carcinogenic, mutagenic, immunotoxic, and cytotoxic potentials were evaluated. Table 2 shows the results obtained where the prediction as “+” (toxic) or “−” (nontoxic) is made with the corresponding probability of the prediction indicated. The deep green, light green, light red, and deep red colors in Table 2 represent inactive, moderately inactive, moderately active, and active predictions, respectively. Overall, the compounds predominantly exhibit nontoxic or moderately toxic potentials, with compound T2727 flagged as a potential immunotoxin with high certainty. It is possible to reduce the toxic potentials of the compounds during lead optimization processes.

Table 2.

Toxicity assessment of T2727, F1044-0016, F0514-3956, F0514-4563, and control.

The deep green, light green, light red, and deep red colors represent inactive, moderately inactive, moderately active, and active predictions, respectively.

Discussion and future directions

Notwithstanding the computational cost, incorporating receptor flexibility into SBVS is considered a useful step toward identifying true binders and potential inhibitors.^71
–74 Although various strategies for integrating receptor flexibility into SBVS have been suggested, MRVS, which involves docking small molecules against different receptor conformations, enables extensive exploration of receptor flexibility.^71,75 Typically, MRVS involves docking small molecules onto superimposed structures of the receptor (ensemble docking)^72,76 or docking sequentially onto multiple conformations of the receptor.⁷⁷ Despite its speed, the ensemble docking approach only selects the best binding mode involving a compound (ligand) and one receptor conformation as the hit. Conversely, sequential MRVS offers a more robust approach to exploring the receptor conformational space by shortlisting receptor–ligand complexes involving all receptor conformations for further evaluation and selection of the best. The selection of compounds at the intersection of the top-ranked docking hits of all receptor models in sequential MRVS has been shown to be a reliable strategy for identifying true binders and reducing the associated challenge of false positives.⁷⁸

The sequential MRVS strategy and selection of intersection ligands, which is based on the proposition that a true binder can effectively bind to various binding site conformations,⁷⁸ identified 21 compounds (see Figure 2) as potential binders of SARS-CoV-2-M^pro. The selection of 21 compounds as hits underscores the usefulness of docking onto different receptor conformations and selecting the intersection ligands in the top-ranked solutions, considering that 532, 2488, 3006, 1353, and 546 compounds were shortlisted for the apo, N3-bound, 13b-bound, shikonin-bound, and Z222979552-bound receptor models, respectively. Inspection of the substrate-binding site in the five M^pro receptor models reveals backbone conformational variations in an unstructured loop (Supplemental Figure S5(a)) comprising residues T45, S46, E47, D48, and M49. Different side-chain orientations are also observed for some binding-site residues such as T24, H41, M49, N142, M165, and Q189, as illustrated in Supplemental Figure S5(b). These structural variations account for the differences in the number of compounds shortlisted for each receptor model, which is in agreement with the observation that each binding site conformation introduces a unique set of false-positive docking predictions.^77,78 Thus, the ligands at the intersection of the top predictions for all receptor models are true binders, immune to the effects of variations in backbone structure and side-chain orientations, owing to their ability to target the conserved conformation of the binding site. The receptor-binding potencies of such ligands may be immune to perturbations by side-chain mutations responsible for resistance.

By comparing the docking fitness scores of the 21 selected compounds with the control (see Table S1), it is suggestive that each of these compounds can bind to SARS-CoV-2-M^pro and serve as lead for the development of potent inhibitors. However, four of the compounds, including T2727, F1044-0016, F0514-3956, and F0514-4563, demonstrated, through MD simulations, that they are capable of remaining stably bound to M^pro for up to100 ns. Close examination of the binding modes and interactions of compounds T2727, F1044-0016, F0514-3956, and F0514-4563 reveals that they interact with important M^pro residues, and their scaffolds fit into the subpockets of the substrate-binding site better than the control (see Figures 3 and 7, and Supplemental Figures S4(a) and S4(c)). The RMSD analysis results (see Figure 4) show that the four compounds are stable, and the average RMSD values compare well with the control compound, with compounds F0514-4563 (2.01 ± 0.04 Å) and F1044-0016 (2.48 ± 0.04 Å) projected as being relatively stable than the control (3.15 ± 0.11 Å). Furthermore, the estimated BFEs between the compounds and M^pro compare well with the control and confirm that the compounds have reasonable affinities for M^pro.

The results suggest that compounds T2727, F1044-0016, F0514-3956, and F0514-4563 strongly bind to SARS-CoV-2-M^pro, indicating the possibility of their optimization to produce novel inhibitors with improved pharmacological properties. By fitting perfectly into the conserved subpockets of the substrate-binding site of M^pro (see Figures 3 and 7), the scaffolds of these compounds could serve as starting materials for the design of potent noncovalent and covalent inhibitors with immunity to resistance conferred by mutations in SARS-CoV-2-M^pro. Given that structure is more conserved than sequence,⁷⁹ such inhibitors, like the substrate, could be immune to resistance induced by mutations in amino acid side chains. Structure–activity relationship (SAR) and chemical mutagenicity⁸⁰ studies on compounds T2727, F1044-0016, F0514-3956, and F0514-4563 could generate potent noncovalent inhibitors of SARS-CoV-2-M^pro. Covalent inhibitors of SARS-CoV-2-M^pro, such as the control,⁴⁸ form covalent bonds with the thiol group of C145. As illustrated in Figure 9, comparing the poses of T2727, F1044-0016, F0514-3956, and F0514-4563 to the control, along with the close proximity of some atoms to the thiol group of C145, indicates the possibility of covalent interactions. More importantly, Figure 9 also suggests the possibility of modifying the compounds by introducing the appropriate covalent warheads to achieve covalent interactions with C145. Many covalent warheads have been shown to readily form covalent interactions with cysteine amino acid residues.⁸¹

Figure 9.

Covalent bond-forming possibilities of the four representative compounds. (a–d) Poses of compounds T2727, F1044-0016, F0514-3956, and F0514-4563 compared with the control, respectively. Carbon atoms of compounds T2727, F1044-0016, F0514-3956, F0514-4563, and the control are colored green, cyan, blue, yellow, and magenta, respectively.

Conclusion

MRVS and MD simulation techniques were employed to identify novel compounds that could target the conserved subpockets of the substrate-binding site of SARS-CoV-2-M^pro and thus serve as starting materials for the discovery, design, and development of antiviral therapeutics to address the resistance challenge. An integrated database of small molecules was sequentially screened against five receptor models of SARS-CoV-2-M^pro, followed by the identification of compounds at the intersection of the top-ranked docking solutions from all five receptors as hits. A total of 21 VS hits were identified, out of which four compounds, including T2727, F1044-0016, F0514-3956, and F0514-4563, could remain stably bound to M^pro for up to 100 ns during MD simulations. The scaffolds of T2727, F1044-0016, F0514-3956, and F0514-4563 fit the conserved subpockets of the substrate-binding site of M^pro better than the control, which suggests that mutations and structural variations in amino acid side chains may not perturb their binding efficacies. Conjointly, the results suggest that the selected compounds are promising leads for the discovery, design, and development of potent drugs to buttress the fight against SARS-CoV-2-M^pro resistance. In vitro, in vivo, SAR, and mutagenicity studies are recommended to further evaluate the M^pro-binding and anti-SARS-CoV-2 characteristics of the compounds, as well as to optimize their general pharmacological features.

Supplemental Material

sj-docx-1-chl-10.1177_17475198251326083 – Supplemental material for Identification of lead small molecules for the design and development of potent severe acute respiratory syndrome coronavirus 2 main protease inhibitors

Supplemental material, sj-docx-1-chl-10.1177_17475198251326083 for Identification of lead small molecules for the design and development of potent severe acute respiratory syndrome coronavirus 2 main protease inhibitors by Elvis Awuni in Journal of Chemical Research

Footnotes

ORCID iD

Elvis Awuni

Statements and declarations

References

Brodeur

Gray

Islam

, et al. A literature review of the economics of COVID-19. J Econ Surv 2021; 35: 1007–1044.

Vindegaard

Benros

ME.

COVID-19 pandemic and mental health consequences: systematic review of the current evidence. Brain Behav Immun 2020; 89: 531–542.

Pollard

Morran

Nestor-Kalinoski

AL.

The COVID-19 pandemic: a global health crisis. Physiol Genomics 2020; 52: 549–557.

McKibbin

Fernando

The global macroeconomic impacts of COVID-19: seven scenarios. Asian Econ Pap 2021; 20: 1–30.

Naveed

Ahmad

Majeed

, et al. The global impact of COVID-19: a comprehensive analysis of its effects on various aspects of life. Toxicol Res 2024; 13: tfae045.

WHO. COVID-19 epidemiological update, https://www.who.int/publications/m/item/covid-19-epidemiological-update-edition-172 (2024, accessed 2 November 2024).

Dubey

SARS-CoV-2: potential drug targets and its virtual screening. In: Azar

Hassanien

(eds) Modeling, control and drug development for COVID-19 outbreak prevention. Cham: Springer International Publishing, 2022, pp. 203–244.

Jena

NR.

Drug targets, mechanisms of drug action, and therapeutics against SARS-CoV-2. Chem Phys Impact 2021; 2: 100011.

Gil

Ginex

Maestro

, et al. COVID-19: drug targets and potential treatments. J Med Chem 2020; 63: 12359–12386.

10.

Aboul-Fotouh

Mahmoud

Elnahas

, et al. What are the current anti-COVID-19 drugs? From traditional to smart molecular mechanisms. Virol J 2023; 20: 241.

11.

Choi

Kopp

, et al. Review of COVID-19 therapeutics by mechanism: from discovery to approval. J Korean Med Sci 2024; 39: e134.

12.

Uriu

Ito

Zahradnik

, et al. Enhanced transmissibility, infectivity, and immune resistance of the SARS-CoV-2 omicron XBB.1.5 variant. Lancet Infect Dis 2023; 23: 280–281.

13.

Moghadasi

Heilmann

Khalil

, et al. Transmissible SARS-CoV-2 variants with resistance to clinical protease inhibitors. Sci Adv 2023; 9: eade8778.

14.

Markov

Ghafari

Beer

, et al. The evolution of SARS-CoV-2. Nat Rev Microbiol 2023; 21: 361–379.

15.

Carabelli

Peacock

Thorne

, et al. SARS-CoV-2 variant biology: immune escape, transmission and fitness. Nat Rev Microbiol 2023; 21: 162–177.

16.

Faraone

Evans

, et al. Enhanced evasion of neutralizing antibody response by Omicron XBB.1.5, CH.1.1, and CA.3.1 variants. Cell Rep 2023; 42: 112443.

17.

Lewandowski

Tan

, et al. Naturally occurring mutations of SARS-CoV-2 main protease confer drug resistance to nirmatrelvir. ACS Cent Sci 2023; 9: 1658–1669.

18.

Nooruzzaman

Johnson

KEE

Rani

, et al. Emergence of transmissible SARS-CoV-2 variants with decreased sensitivity to antivirals in immunocompromised patients with persistent infections. Nat Commun 2024; 15: 7999.

19.

Igari

Sakao

Ishige

, et al. Dynamic diversity of SARS-CoV-2 genetic mutations in a lung transplantation patient with persistent COVID-19. Nat Commun 2024; 15: 3604.

20.

Iketani

DD.

SARS-CoV-2 resistance to monoclonal antibodies and small-molecule drugs. Cell Chem Biol 2024; 31: 632–657.

21.

Noske

Nakamura

Gawriljuk

, et al. A crystallographic snapshot of SARS-CoV-2 main protease maturation process. J Mol Biol 2021; 433: 167118.

22.

Pablos

Machado

de Jesus

HCR

, et al. Mechanistic insights into COVID-19 by global analysis of the SARS-CoV-2 3CL(pro) substrate degradome. Cell Rep 2021; 37: 109892.

23.

Xiong

Zhu

, et al. The SARS-CoV-2 main protease (Mpro): structure, function, and emerging therapies for COVID-19. MedComm 2022; 3: e151.

24.

Akbulut

Investigation of changes in protein stability and substrate affinity of 3CL-protease of SARS-CoV-2 caused by mutations. Genet Mol Biol 2022; 45: e20210404.

25.

Kneller

Zhang

Coates

, et al. Michaelis-like complex of SARS-CoV-2 main protease visualized by room-temperature X-ray crystallography. IUCrJ 2021; 8: 973–979.

26.

Xiong

Zhao

, et al. What coronavirus 3C-like protease tells us: from structure, substrate selectivity, to inhibitor design. Med Res Rev 2021; 41: 1965–1998.

27.

Ullrich

Nitsche

The SARS-CoV-2 main protease as drug target. Bioorg Med Chem Lett 2020; 30: 127377.

28.

Mengist

Dilnessa

Jin

Structural basis of potential inhibitors targeting SARS-CoV-2 main protease. Front Chem 2021; 9: 622898.

29.

Citarella

Dimasi

Moi

, et al. Recent advances in SARS-CoV-2 main protease inhibitors: from nirmatrelvir to future perspectives. Biomolecules 2023; 13: 1339.

30.

Janin

YL.

On the origins of SARS-CoV-2 main protease inhibitors. RSC Med Chem 2024; 15: 81–118.

31.

Pang

Liu

, et al. The research progress of SARS-CoV-2 main protease inhibitors from 2020 to 2022. Eur J Med Chem 2023; 257: 115491.

32.

Jiang

Zou

Zeng

, et al. Crystal structures of main protease (Mpro) mutants of SARS-CoV-2 variants bound to PF-07304814. Mol Biomed 2023; 4: 23.

33.

Chen

Arutyunova

, et al. SARS-CoV-2 Mpro protease variants of concern display altered viral substrate and cell host target galectin-8 processing but retain sensitivity toward antivirals. ACS Cent Sci 2023; 9: 696–708.

34.

Rauch

Costacurta

Schöppe

, et al. Highly specific SARS-CoV-2 main protease (Mpro) mutations against the clinical antiviral ensitrelvir selected in a safe, VSV-based system. Antiviral Res 2024; 231: 105969.

35.

Chatterjee

Bhattacharya

Dhama

, et al. Resistance to nirmatrelvir due to mutations in the Mpro in the subvariants of SARS-CoV-2 Omicron: another concern? Mol Ther Nucleic Acids 2023; 32: 263–266.

36.

Lewandowski

, et al. A yeast-based system to study SARS-CoV-2 Mpro structure and to identify nirmatrelvir resistant mutations. PLoS Pathog 2023; 19: e1011592.

37.

Costacurta

Dodaro

Bante

, et al. A comprehensive study of SARS-CoV-2 main protease (Mpro) inhibitor-resistant mutants selected in a VSV-based system. PLoS Pathog 2024; 20: e1012522.

38.

Sasi

Ullrich

Ton

, et al. Predicting antiviral resistance mutations in SARS-CoV-2 main protease with computational and experimental screening. Biochemistry 2022; 61: 2495–2505.

39.

Verdonk

Cole

Hartshorn

, et al. Improved protein-ligand docking using GOLD. Proteins 2003; 52: 609–623.

40.

Jones

Willett

Glen

, et al. Development and validation of a genetic algorithm for flexible docking. J Mol Biol 1997; 267: 727–748.

41.

Jones

Willett

Glen

RC.

Molecular recognition of receptor-sites using a genetic algorithm with a description of desolvation. J Mol Biol 1995; 245: 43–53.

42.

Amporndanai

Meng

Shang

, et al. Inhibition mechanism of SARS-CoV-2 main protease by ebselen and its derivatives. Nat Commun 2021; 12: 3061.

43.

Zhang

, et al. A comparison of Remdesivir versus gold cluster in COVID-19 animal model: a better therapeutic outcome of gold cluster. Nano Today 2022; 44: 101468.

44.

Lee

Worrall

Vuckovic

, et al. Crystallographic structure of wild-type SARS-CoV-2 main protease acyl-enzyme intermediate with physiological C-terminal autoprocessing site. Nat Commun 2020; 11: 5877.

45.

Jin

, et al. Structure of Mpro from SARS-CoV-2 and discovery of its inhibitors. Nature 2020; 582: 289–293.

46.

Dai

Zhang

Jiang

X-M

, et al. Structure-based design of antiviral drug candidates targeting the SARS-CoV-2 main protease. Science 2020; 368: 1331–1335.

47.

Sacco

Hurst

, et al. Boceprevir, GC-376, and calpain inhibitors II, XII inhibit SARS-CoV-2 viral replication by targeting the viral main protease. Cell Res 2020; 30: 678–692.

48.

Zhang

Lin

Sun

, et al. Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved α-ketoamide inhibitors. Science 2020; 368: 409–412.

49.

Qiao

Zeng

, et al. SARS-CoV-2 M(pro) inhibitors with antiviral activity in a transgenic mouse model. Science 2021; 371: 1374–1378.

50.

Zhou

Zhang

, et al. Crystal structure of SARS-CoV-2 main protease in complex with the natural product inhibitor shikonin illuminates a unique binding mode. Sci Bull 2021; 66: 661–663.

51.

Rossetti

Ossorio

Rempel

, et al. Non-covalent SARS-CoV-2 Mpro inhibitors developed from in silico screen hits. Sci Rep 2022; 12: 2505.

52.

Agost-Beltrán

de la Hoz-Rodríguez

Bou-Iserte

, et al. Advances in the development of SARS-CoV-2 Mpro inhibitors. Molecules 2022; 27: 2523.

53.

Waterhouse

Bertoni

Bienert

, et al. SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res 2018; 46: W296–W303.

54.

Guex

Peitsch

Schwede

Automated comparative protein structure modeling with SWISS-MODEL and Swiss-PdbViewer: a historical perspective. Electrophoresis 2009; 30: S162–S173.

55.

Vagner

Hruby

VJ.

Peptidomimetics, a synthetic tool of drug discovery. Curr Opin Chem Biol 2008; 12: 292–296.

56.

Harris

Hill

Sheppard

, et al. The design and application of target-focused compound libraries. Comb Chem High Throughput Screen 2011; 14: 521–531.

57.

Lindahl

Hess

van der Spoel

GROMACS 2020.4 manual. Geneva: Zenodo, 2024.

58.

Van Der Spoel

Lindahl

Hess

, et al. GROMACS: fast, flexible, and free. J Comput Chem 2005; 26: 1701–1718.

59.

Cornell

Cieplak

Bayly

, et al. A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J Am Chem Soc 1995; 117: 5179–5197.

60.

Jorgensen

Chandrasekhar

Madura

, et al. Comparison of simple potential functions for simulating liquid water. J Chem Phys 1983; 79: 926–935.

61.

Bayly

Cieplak

Cornell

, et al. A well-behaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. J Phys Chem 1993; 97: 10269–10280.

62.

Case

Cheatham

Darden

, et al. The AMBER biomolecular simulation programs. J Comput Chem 2005; 26: 1668–1688.

63.

Pearlman

Case

Caldwell

, et al. AMBER, a package of computer programs for applying molecular mechanics, normal mode analysis, molecular dynamics and free energy calculations to simulate the structural and energetic properties of molecules. Comput Phys Commun 1995; 91: 1–41.

64.

Darden

York

Pedersen

Particle mesh Ewald—An N.log(N) method for Ewald sums in large systems. J Chem Phys 1993; 98: 10089–10092.

65.

Hess

Bekker

Berendsen

HJC

, et al. LINCS: a linear constraint solver for molecular simulations. J Comput Chem 1997; 18: 1463–1472.

66.

Bussi

Donadio

Parrinello

Canonical sampling through velocity rescaling. J Chem Phys 2007; 126: 1–7.

67.

Kumari

Kumar

Lynn

g_mmpbsa-A GROMACS tool for high-throughput MM-PBSA calculations. J Chem Inf Model 2014; 54: 1951–1962.

68.

Schrödinger

The PyMOL molecular graphics system, version 1.7.4, 2014. https://sourceforge.net/p/pymol/mailman/message/33108131/

69.

Wallace

Laskowski

Thornton

JM.

LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein Eng Des Sel 1995; 8: 127–134.

70.

Banerjee

Kemmler

Dunkel

, et al. ProTox 3.0: a webserver for the prediction of toxicity of chemicals. Nucleic Acids Res 2024; 52: W513–W520.

71.

Carlson

McCammon

JA.

Accommodating protein flexibility in computational drug design. Mol Pharmacol 2000; 57: 213–218.

72.

Huang

Zou

XQ.

Ensemble docking of multiple protein structures: considering protein structural variations in molecular docking. Proteins 2007; 66: 399–421.

73.

Barril

Morley

SD.

Unveiling the full potential of flexible receptor docking using multiple crystallographic structures. J Med Chem 2005; 48: 4432–4443.

74.

Cavasotto

Orry

Abagyan

RA.

The challenge of considering receptor flexibility in ligand docking and virtual screening. Curr Comput-Aided Drug Des 2005; 1: 423–440.

75.

Feixas

Lindert

Sinko

, et al. Exploring the role of receptor flexibility in structure-based drug discovery. Biophys Chem 2014; 186: 31–45.

76.

Erickson

Jalaie

Robertson

, et al. Lessons in molecular recognition: the effects of ligand and protein flexibility on molecular docking accuracy. J Med Chem 2004; 47: 45–55.

77.

Totrov

Abagyan

Flexible ligand docking to multiple receptor conformations: a practical alternative. Curr Opin Struct Biol 2008; 18: 178–184.

78.

Awuni

Reduction of false positives in structure-based virtual screening when receptor plasticity is considered. Molecules 2015; 20: 5152–5164.

79.

Illergård

Ardell

Elofsson

Structure is three to ten times more conserved than sequence-a study of structural response in protein cores. Proteins 2009; 77: 499–508.

80.

Lou

Yang

Deng

, et al. Chemical rules for optimization of chemical mutagenicity via matched molecular pairs analysis and machine learning methods. J Cheminform 2023; 15: 35.

81.

Huang

Han

Xiao

, et al. Covalent warheads targeting cysteine residue: the promising approach in drug development. Molecules 2022; 27: 7728.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

3.53 MB

Identification of lead small molecules for the design and development of potent severe acute respiratory syndrome coronavirus 2 main protease inhibitors

Abstract

Keywords

Introduction

Methods

Selection and preparation of SARS-CoV-2-Mpro receptor models

Construction and preparation of small-molecule library

MRVS and hits selection

MD simulations

Binding free energy calculations

Pose, interaction, and MD simulation analyses

Results

Selected compounds

Default docking scores, poses, and interactions of compounds T2727, F1044-0016, F0514-3956, and F0514-4563

Ligand RMSD and convergence analyses

Receptor RMSD, RG, and RMSF analyses

Hydrogen bond analyses

FEL and conformational analyses

BFE and PRC analyses

Toxicity assessment

Discussion and future directions

Conclusion

Supplemental Material

sj-docx-1-chl-10.1177_17475198251326083 – Supplemental material for Identification of lead small molecules for the design and development of potent severe acute respiratory syndrome coronavirus 2 main protease inhibitors

Footnotes

ORCID iD

Statements and declarations

References

Supplementary Material

Selection and preparation of SARS-CoV-2-M^pro receptor models