Identification of high-affinity inhibitors of SARS-CoV-2 main protease: Towards the development of effective COVID-19 therapy

Graphical abstract

Highlights

• •The main protease (M^pro, 3CL^pro) of SARS-CoV-2 is considered as a potential drug target for drug development.
• •Using advanced computational studies, we have identified two potent inhibitors of SARS-CoV-2 main protease.
• •Both 6-Deaminosinefungin and UNII-O9H5KY11SV bind strongly to the residues active site pocket.
• •The complexes of both compounds with the main protease were stable throughout the MD simulation.
• •Both the molecules may be further exploited as preclinical leads for therapeutic management of COVID-19.

Abstract

Coronavirus disease 2019 (COVID-19) is an infectious disease, caused by a newly emerged highly pathogenic virus called novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Targeting the main protease (M^pro, 3CL^pro) of SARS-CoV-2 is an appealing approach for drug development because this enzyme plays a significant role in the viral replication and transcription. The available crystal structures of SARS-CoV-2 M^pro determined in the presence of different ligands and inhibitor-like compounds provide a platform for the quick development of selective inhibitors of SARS-CoV-2 M^pro. In this study, we utilized the structural information of co-crystallized SARS-CoV-2 M^pro for the structure-guided drug discovery of high-affinity inhibitors from the PubChem database. The screened compounds were selected on the basis of their physicochemical properties, drug-likeliness, and strength of affinity to the SARS-CoV-2 M^pro. Finally, we have identified 6-Deaminosinefungin (PubChem ID: 10428963) and UNII-O9H5KY11SV (PubChem ID: 71481120) as potential inhibitors of SARS-CoV-2 M^pro which may be further exploited in drug development to address SARS-CoV-2 pathogenesis. Both compounds are structural analogs of known antivirals, having considerable protease inhibitory potential with improved pharmacological properties. All-atom molecular dynamics simulations suggested SARS-CoV-2 M^pro in complex with these compounds is stable during the simulation period with minimal structural changes. This work provides enough evidence for further implementation of the identified compounds in the development of effective therapeutics of COVID-19.

1. Introduction

Coronavirus disease 2019 (COVID-19) emerged from China and globally affected a larger population through human-to-human transmission (Guan et al., 2020). Due to the high reproduction rate of novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and unavailability of effective drugs and vaccines, the number of reported cases and death are increasing day by day (Huang et al., 2020). The disease is now pandemic across the globe claiming the death of more than 0.6 million lives worldwide (https://www.worldometers.info/coronavirus/). SARS-CoV-2 belongs to Nidovirus superfamily of Coronavirus, which is an enveloped, positive-sense, single-stranded RNA virus, and the seventh coronavirus having the potential to infect humans (Chary et al., 2020; Naqvi et al., 2020). The genome of SARS-CoV-2 is 29.9 kb in size and shows a high similarity to the severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East respiratory syndrome coronavirus (MERS-CoV) (Lu et al., 2015b).

The genome of SARS-CoV-2 is composed of 13–15 open reading frames (ORFs) flanked by the 5′-untranslated region (UTR) and 3′-UTR (Lu et al., 2020). All the ORFs are arranged as main replicase assembly encoding 27 structural and non-structural proteins (Liu et al., 2020; Minakshi et al., 2014). Two-third of its RNA situated in the first ORF (ORF1ab) that encodes a 7096 amino acids long polyprotein. The viral genome of SARS-CoV-2 translates into four major structural proteins viz. envelope (E), membrane (M), spike (S), and nucleocapsid (N) proteins. All these proteins play a vital role in the viral assembly (Zhang and Holmes, 2020).

The spike protein (S) is important for coronavirus transmission, as it mediates receptor binding and membrane fusion of the virus to the host angiotensin-converting enzyme 2 (ACE2), a cell receptor for SARS-CoV-2 (Lu et al., 2015a; Wang et al., 2016). The S1 domain of S protein helps in receptor binding as it recognizes and binds to the host receptor through the receptor-binding domain (RBD), and the subsequent conformational transition of the S2 domain take place to facilitate the fusion of cell membrane (He et al., 2004; Wong et al., 2004; Zhou et al., 2020). Transmembrane protease, serine 2 (TMPRSS2) plays a key role in the fusion of virus and host cell by priming S protein. The RBD binds to receptor ACE2 of the host which provides entry of SARS-CoV-2 into the respiratory tract (Zhou et al., 2020). The affinity with which S protein of SARS-CoV-2 binds to the ACE2 is more than 10 times than that of SARS-CoV. The increased affinity of human ACE2 for the viral protein facilitates its high contagious nature among the human population (Wrapp et al., 2020).

The uncontrolled situation, mass infection, and high death along with the lack of effective therapeutic measures are emergent issues related to COVID-19 globally. Several strategies have been employed to address COVID-19 by using existing broad-spectrum antivirals, including anti-HIV drugs such as ritonavir and lopinavir (Sheahan et al., 2020). Amongst the well-characterized drug targets of coronaviruses, main protease (M^pro, also called 3CL^pro) is mostly considered as it is essential for processing the polyproteins that are translated from the viral RNA (Anand et al., 2003). M^pro cuts polyproteins translated from viral RNA to produce functional viral proteins. This enzyme is primarily involved in the regulation of replication and transcription of virions, thus targeted to design and develop effective therapeutic molecules for COVID-19.

Computational methods can serve as a great asset in the present times due to their ability to work at a rapid and helping manner to find potential inhibitors for drug targets of SARS-CoV-2 (Shamsi et al., 2020). Structure-based drug design and discovery approach is used to find effective molecules with high affinity and target specificity (Khan et al., 2019). The identification of new leads with high specificity is achieved by virtual high-throughput screening (vHTS) (Mohammad et al., 2020). This method identifies considerable drug-like candidates from a large chemical library based on the binding of ligands to target proteins with high affinity (Naqvi et al., 2018).

Prompt development of effective drugs for COVID-19 therapy is a difficult task as the conventional drug development process usually takes a long time and cost in billions. The optimization and reconsideration of available antiviral compounds offers an alternate approach to rapidly identify potential leads for the quick development of effective and safe drugs. In this study, we have extensively analyzed the available structures of SARS-CoV-2 M^pro co-crystallized with different compounds and employed to structure-guided rational drug design approach to find potential inhibitors with improved pharmacological properties (Dai et al., 2020; Zhang et al., 2020). First, redocking of 121 reported compounds (co-crystallized with SARS-CoV-2 M^pro) was carried out to search high-affinity binding partners of SARS-CoV-2 M^pro which were served as scaffolds to screen the PubChem database with >90 % structural similarity. Then, we performed molecular docking-based virtual screening of the PubChem compounds and estimated their binding affinities with SARS-CoV-2 M^pro. Finally, after the interaction analysis and based on the biological properties, we have identified two compounds for molecular dynamics (MD) simulations. A systematic approach based on structure-based drug designing was used in this study as described in Fig. 1 . We have identified two compounds, 6-Deaminosinefungin (PubChem ID: 10428963) and UNII-O9H5KY11SV (PubChem ID: 71481120) as high-affinity inhibitors of SARS-CoV-2 M^pro.

Fig. 1.

The workflow demonstrates the process of virtual high-throughput screening used in this study. RO5, Lipinski's rule of five; ADMET, Absorption, Distribution, Metabolism, Excretion, and Toxicity.

2. Materials and methods

2.1. Structure-based virtual screening

This study was performed on the DELL® Workstation running on Ubuntu version 18.04.4 LTS. Computational tools, such as MGL Tools (Jacob et al., 2012), AutoDock Vina (Trott and Olson, 2010), and Discovery Studio visualizer (Biovia, 2015) were used for structure-guided vHTS. A total of 158 SARS-CoV-2 M^pro structures were retrieved from the PDB. Most of them were in complex with different inhibitors. We found a total of 121 unique ligands co-crystallized with SARS-CoV-2 M^pro and extensively analyzed to get insights into their mode of binding and inhibition to SARS-CoV-2 M^pro. We re-docked these 121 ligands with SARS-CoV-2 M^pro (PDB ID: 6M03, Resolution: 2.0 Å) to estimate their binding affinity while using the molecular docking approach and selected top high-affinity binding partners (hits) of SARS-CoV-2 M^pro.

We further screened the PubChem database to get compounds with >90 % structural similarity (Tanimoto threshold) with the high-affinity hits of SARS-CoV-2 M^pro. The identified compounds from the PubChem database were screened further based on their physicochemical properties following the Lipinski’s rule of five (RO5), and showing appreciable ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity), and other drug-like properties with no carcinogenic and PAINS patterns. To identify high-affinity inhibitors of SARS-CoV-2 M^pro, molecular docking-based virtual screening was performed. Before docking simulations, energy minimization was performed on the protein structure to remove possible steric clashes in finding its most stable and lowest energy conformation state. The protocol for docking has been described in our previous communication (Shamsi et al., 2020).

Values of inhibition constant (pKi), the negative decimal logarithm of inhibition constants were calculated from the ΔG, generated from the docking study using the following formula-

$$\Delta G=RT(\text{Ln}Ki{}_{\text{pred}})$$ (1)

$$K{i}_{\text{pred}}=e^{(\Delta G/RT)}$$ (2)

$$p{K}_i=-log{(Ki}_{\text{pred}})$$ (3)

where ΔG, binding affinity (kcal/mol); R (gas constant), 1.98 cal*(mol*K)⁻¹; T (temperature), 298.15 K; pred, predicted.

Ligand efficiency (LE) is a commonly applied parameter for lead selection by comparing the values of average binding energy per atom (Hopkins et al., 2004). The following formula was applied to calculate LE:

LE = -ΔG/N

where LE is the ligand efficiency (kcal mol⁻¹ non-H atom⁻¹), ΔG represents binding affinity (kcal mol⁻¹) and N is the number of non-hydrogen atoms in the ligand.

The docking result was marked off for high-affinity compounds. Subsequently, all possible docked conformers of the hits were splitted for their interaction analysis. All conformers were analyzed using PyMOL and Discovery Studio visualizer to explore their binding pattern and interactions with SARS-CoV-2 M^pro. The interaction analysis was performed to get highly selective compounds that preferentially bind to the SARS-CoV-2 M^pro substrate-binding pocket.

2.2. Biological activity predictions and structure-activity analysis

To investigate the biological properties of the selected compounds, we have predicted their possible biological functions through the PASS webserver (Lagunin et al., 2000). The PASS analysis allows for exploring the effects and properties of chemical compounds on the basis of their molecular formula. It uses multilevel neighbors of atoms (MNA) descriptors, suggesting the biological activity of a compound is the function of its chemical structure. It defines the prediction of biological properties of a compound based on the ratio of probability to be active (Pa) and probability to be inactive (Pi). Higher the Pa value for a prediction means the compound is having more probability to be active under that particular activity or biological property. Here, we selected only those compounds showing antiviral properties and protease inhibitory potential and subsequently discussed their analog properties with parent compounds.

2.3. MD simulations

MD simulations were performed on three systems, one, the apo- SARS-CoV-2 M^pro and the other two with the selected ligands, 10428963 and 71481120 for 50 ns at the molecular mechanics level using GROMOS 54A7 force-field in GROMACS 5.1.2 at 300 K. Compounds 10428963 and 71481120 were extracted out from the docked complexes; subsequently, their topology and force-field parameters were produced through the PRODRG webserver and then combined into the M^pro topology to make the Gromacs complexed systems. All three systems were soaked in the Simple Point Charge (spc216) model for solvation and energy minimized using steepest descent approach under 1500 steps. Final MD run was performed for 50,000 ps (50 ns) for each system and the generated trajectories were analyzed using the inbuilt tools of GROMACS as described in our preceding communications (Mohammad et al., 2019; Naqvi et al., 2018).

2.4. Principal component analysis

To study the conformational sampling and atomic motions of M^pro and its docked complexes, principal component (PC) and free energy landscape (FEL) analyses were performed using the essential dynamics approach employing the calculation of the covariance matrix (Altis et al., 2008). The covariance matrix was calculated while using the following formula:

C_ij = < (x_i - < x_i>) (x_j - <x_j>) >

where x_i/x_j is the coordinate of the i^th/j^th atom of the systems, and < - > in the ensemble average.

The FELs of a protein can be attained using the conformational sampling approach that allows exploring the protein conformations near the native state (Papaleo et al., 2009). FELs were generated to investigate the stability and native states of SARS-CoV-2 M^pro, before and after compounds binding. The FELs were generated as:

ΔG(X) = -K_BT ln P(X)

where K _B is the Boltzmann constant, T is the temperature of simulation, and P(X) is the probability distribution of the system along with the PCs.

3. Results and discussion

3.1. Virtual screening

From the 158 PDB entries of SARS-CoV-2 M^pro, a total of 121 unique co-crystallized compounds were selected (Table S1). Redocking of these compounds resulted in the identification of many hits as high-affinity binding ligands of SARS-CoV-2 M^pro (Table 1 ). These compounds are showing considerable binding affinities to the SARS-CoV-2 M^pro. Here, we have selected the top 10 hits with their binding affinities within the range of -8.3 kcal/mol to -7.5 kcal/mol (Table 1).

Table 1.

List of selected hits from the co-crystalized ligands with M^pro showing notable binding affinity^*.

S. No.	Ligand ID (PDB)	Parent Protein ID (PDB)	Common name of the ligand	PubChem ID of ligand	Affinity (kcal/mol)	No. of hits*
	3WL	6M2N	Baicalein	5281605	−8.3	6930
	T7S	5RFO	1-[4-(piperidine-1-carbonyl)piperidin-1-yl]ethan-1-one	16394003	−8.3	1335
	T47	5RET	1-{4-[(3-chlorophenyl)methyl]piperazin-1-yl}ethan-1-one	19323586	−8.2	63
	A82	6YVF	2-[[(1R)-1-(7-methyl-2-morpholin-4-yl-4-oxidanylidene-pyrido[1,2-a]pyrimidin-9-yl)ethyl]amino]benzoic acid	44137675	−8.1	26
	K36	7BRR	(1S,2S)-2-({N-[(benzyloxy)carbonyl]-l-leucyl}amino)-1-hydroxy-3-[(3S)-2-oxopyrrolidin-3-yl]propane-1-sulfonic acid	118737648	−7.9	102
	T1J	5REC	2-{[(1H-benzimidazol-2-yl)amino]methyl}phenol	787400	−7.8	545
	T4V	5REX	1-{4-[(naphthalen-1-yl)methyl]piperazin-1-yl}ethan-1-one	8386889	−7.7	155
	SAM	6W61	S-Adenosylmethionine	34755	−7.6	1375
	SFG	6YZ1	Sinefungin	65482	−7.5	340
	T8A	5RFT	1-[(4S)-4-phenyl-3,4-dihydroisoquinolin-2(1 H)-yl]ethan-1-one	145998233	−7.5	52

^*PubChem hits, the number of the PubChem compounds selected with >90 % similarity (Tanimoto threshold).

3.2. Molecular docking

A total of 10,433 PubChem compounds carrying at least 90 % structural similarity with the top 10 hits were identified and screened for their drug-likeliness. After filtration based on physicochemical, ADMET, and other druglike properties, we got a set of 4802 compounds satisfying the threshold of the applied filters of drug-likeness and RO5. These compounds were further screened out based on their affinities with SARS-CoV-2 M^pro using molecular docking-based virtual screening and selected the top five compounds. These compounds show significant binding affinities toward SARS-CoV-2 M^pro (-8.1 kcal/mol to -8.7 kcal/mol, Table 2 ). The docking analysis suggested that the identified compounds have improved druglike properties compared to their parent compounds reported in the structures of co-crystallized SARS-CoV-2 M^pro.

Table 2.

List of selected hits from the PubChem compounds with their affinity and physicochemical properties.

S. No.	Compound ID (PubChem)	Common name	Affinity (kcal/mol)	LE*	pKi	MW (Da)	xLogP	#HBD	#HBA	Parent compound(PubChem ID)
	57789333	Azd-6482 (S)	−8.7	0.32	6.38	408.18	2.0	2	7	44137675
	5481231	Piscisoflavone B	−8.6	0.24	6.31	366.11	3.4	2	6	5281605
	71481120	UNII-O9H5KY11SV	−8.4	0.27	6.16	485.18	0.7	5	8	118737648
	54592323	2-Fluoro-5′-MethylthioAdo	−8.3	0.22	6.09	383.14	2.1	3	9	34755
	10428963	6-Deaminosinefungin	−8.1	0.23	5.94	366.37	−3.1	5	10	65482

Abbreviations: pKi negative decimal logarithm of inhibition constant; MW Molecular weight; #HBD Hydrogen Bond Donor Count; #HBA Hydrogen Bond Acceptor Count. *LE: Ligand Efficiency (kcal/mol/non-H atom).

To explore the specific interaction of these five compounds, detailed interaction analysis of their possible docked conformers (9 for each compound) was carried out to find their preferential interaction with the SARS-CoV-2 M^pro-binding pocket. We identified that all the selected compounds are showing many specific polar interactions with a set of critical residues, Thr25, Phe140, Leu141, Asn142, Gly143, Ser144, Cys145, His163, his164, Glu166, Gln189 of SARS-CoV-2 M^pro. All these compounds are mimicking the same binding pattern and sharing common interactions as co-crystallized SARS-CoV-2 M^pro inhibitor K36 (Fig. 2 ).

Fig. 2.

Binding pattern of the selected compounds and standard inhibitor K36 with SARS-CoV-2 M^pro. (A) Structural representation of the protein in-complexed with all the compounds making significant interactions with the functionally important residues of the SARS-CoV-2 M^pro binding pocket. (B) The potential surface of SARS-CoV-2 M^pro showing the binding pocket occupancy by the compounds.

All the compounds binding in the SARS-CoV-2 M^pro substrate-binding pocket were explored for their detailed interaction. All these compounds are showed to have many polar interactions with the critical residues of the SARS-CoV-2 M^pro substrate-binding site (Fig. 3 ). They are showing many hydrogen-bonds with Thr25, Phe140, Leu141, Asn142, Gly143, Ser144, Cys145, His163, his164, Glu166, and Gln189 (Fig. 3). Besides, all compounds occupied a deeper cavity of the SARS-CoV-2 M^pro substrate-binding pocket to hinder the substrate-accessibility thus possibly resulted in inhibition of protease activity. All compounds are mimicking of the same orientation and sharing a similar pattern of binding and showing common interactions as most of the co-crystallized inhibitors of SARS-CoV-2 M^pro binds (Jo et al., 2020).

Fig. 3.

2D plots of the SARS-CoV-2 M^pro substrate-binding pocket residues and their interactions with compound (A) 57789333 (B) 5481231 (C) 71481120 (D) 54592323 (E) 10428963 and (F) K36.

In the interaction analysis, we noticed that all the selected compounds are showing many hydrogen and halogen bonding with the SARS-CoV-2 M^pro substrate-binding site. These interactions help to bind the compounds within the substrate-binding pocket and consequently can inhibit the SARS-CoV-2 M^pro activity. The active site of SARS-CoV-2 M^pro, Cys145 (originally Cys3408) acts as a nucleophile, and responsible for protease activity is participating in direct interaction with the identified compounds (Zhang et al., 2020). Specific interactions and binding properties of these compounds make them potential leads to develop inhibitors of SARS-CoV-2 M^pro.

3.3. Biological activity and SAR analysis

To explore the possible biological activities and antiviral potential of selected ligands, the PASS server prediction was carried. The analysis suggested that the selected compounds are having similar classes of biological activities. We identified two compounds, 10428963 and 71481120 which show antiviral activity through the protease inhibitory potential, with Pa ranging from 0413 to 0,914 when Pa > Pi. The biological activities of the identified compounds with higher Pa value along with their ADMET properties are shown in Table 3 .

Table 3.

List of identified compounds and their biological activities along with ADMET properties.

S. No.	Compound ID (PubChem)	Pa	Pi	Biological Activity	A	D	M	E	T
	10428963	0914	0002	Antiviral (Picornavirus)	34.8	No	No	No	No
		0894	0002	Antiviral (Poxvirus)
		0544	0006	Antiviral (Herpes)
		0461	0007	Antiviral (Hepatitis B)
	71481120	0564	0002	Severe acute respiratory syndrome treatment	19.7	No	No	No	No
		0535	0002	SARS-CoV-2 Mpro inhibitor
		0479	0008	Antiviral
		0413	0004	Protease inhibitor

Pa = probability to be active; Pi = probability to be inactive. A: Absorption, GI absorption (%); D: Distribution, CNS & BBB permeability; M: Metabolism, CYP2D6 inhibitor/substrate; E: Excretion, OCT2 substrate; T: Toxicity, AMES & Hepatotoxicity.

The identified compound 10428963, commonly known as 6-Deaminosinefungin ((2S)-2-Amino-6-[(2R,3S,4R,5R)-5-(6-aminopurin-9-yl)-3,4-dihydroxyoxolan-2-yl]hexanoic acid) is a well-known antiviral compound exploited in cancer research (Peterli-Roth et al., 1994). Compound 10428963 is an active analog of S-adenosylmethionine, S-adenosylhomocysteine, and Sinefungin, an antiviral and inhibitor of SARS-CoV-2 M^pro (PDB IDs: SFG, 6YZ1) with improved pharmacological properties (Peterli-Roth et al., 1994).

The compound 71481120 is known as UNII-O9H5KY11SV ((2S)-1-Hydroxy-2-[[(2S)-4-methyl-2-(phenylmethoxycarbonylamino)pentanoyl]amino]-3-(2-oxopyrrolidin-3-yl)propane-1-sulfonic acid) is an antiviral in nature predicted to have possibilities for SARS treatment and appreciable inhibitory potential against SARS-CoV-2 M^pro. Compound 71481120 is an analog of K36 ((2S)-1-Hydroxy-2-[[(2S)-4-methyl-2-(phenylmethoxycarbonylamino)pentanoyl]amino]-3-[(3S)-2-oxopyrrolidin-3-yl]propane-1-sulfonic acid), co-crystallized with many viral proteases including SARS-CoV-2 M^pro (PDB IDs: K36, 7BRR).

Both the identified compounds are analogs of already known antiviral compounds with admirable protease inhibitory potential, improved pharmacological properties, and considerably high affinity towards SARS-CoV-2 M^pro. Both compounds are showing significant interaction with the amino acid residues of the catalytic site, Cys145, and His41, where Cys145 is the active site of SARS-CoV-2 M^pro, acts as a nucleophile, and responsible for the protease activity (Zhang et al., 2020). Binding of any compound to these residues will hinder the substrate accessibility and thus inhibition of protease activity of SARS-CoV-2 M^pro similar to the alpha-ketoamide/N3-ILP inhibitors reported (Zhang et al., 2020). All these properties of 10428963 and 71481120 make them potential leads to develop SARS-CoV-2 M^pro inhibitors that can be used in the development of safe and effective COVID-19 therapy.

3.4. MD simulations

Three systems, SARS-CoV-2 M^pro-apo, M^pro-10428963, and Mpro-71481^pro-71,481,120 complexes were subjected to MD simulation for 50 ns. Different systematic and structural properties of all three systems were explored to ascertain their stability and dynamics under the solvent condition during the simulation time.

Binding of any small compound can make large changes to a protein structure. Root-mean-square deviation (RMSD) is one such fundamental property that can be utilized to investigate structural deviation and compactness of a protein (Turab Naqvi et al., 2019). We explored the structural dynamics of SARS-CoV-2 M^pro-apo, and in complex with 10428963 and 71481120 by calculating their RMSD. The average RMSD of the M^pro in the free state, M^pro-10428962, and M^pro-71481120 complexes was found as 0.22 nm, 0.21 nm, and 0.23 nm, respectively. While comparing, the RMSD suggests that the M^pro is getting a little stabilized after 10428963 bindings as compared to the M^pro-71481120 complex. The binding of 10428963 and 71481120 cause an infrequently conformational change in M^pro from its native state (Fig. 4 A). The RMSD of M^pro-10428963 is showing a little decrease after 25 ns and equilibrated throughout the entire trajectory, suggesting the durable stability of the complex.

Fig. 4.

Structural dynamics of SARS-CoV-2 M^pro as a function of time. (A) Time evolution of the RMSD of M^pro before and after compounds binding. (B) Residual fluctuations plot of M^pro before and after 10428963 and 71481120 bindings. (C) Plot showing the radius of gyration of M^pro before and after 10428963 and 71481120 bindings. (D) SASA plot of M^pro of M^pro before and after 10428963 and 71481120 bindings.

To explore the local fluctuation in M^pro at the residue level, the vibrations in each residue of M^pro before and after compounds binding were explored as root-mean-square fluctuation (RMSF). A random distribution of residual fluctuations was observed in M^pro at regions near to N- to C-termini (Fig. 4B). The SARS-CoV-2 M^pro backbone fluctuations were compared with each residue after the binding of compound 10428963 and 71481120. These fluctuations were found to be minimized at several parts in the case of the M^pro-10428962 and M^pro-71481120 complexes. The local residual fluctuations at flap V20-T25, T45-D56, and T225-D245 were decreased after compounds binding. However, the RMSF of the M^pro-10428963 complex was increased at several parts including the flap spanning K137-G143, M165-H172, and D187-D197 with a sharp peak at G170. The average RMSF of the M^pro in the free state, M^pro-10428962 and M^pro-71481120 complexes was found as 0.12 nm, 0. 12 nm, and 0.11 nm, respectively. The RMSF suggested that most fluctuations are reduced in the region where the compounds bind. Decreased fluctuations were observed in M^pro upon 71481120 binding due to ligands adjustment in the M^pro binding pocket. The increased fluctuations observed in the ligand-binding region in the M^pro can be correlated with the docking results where a number close interactions are formed between the protein-ligand which results in local fluctuations as this region of M^pro is directly participating in the ligands interactions when comparing with the other regions in the M^pro.

The radius of gyration (R_g) is associated with the folding state and overall conformation of the proteins which can be used to get deeper insights into their compactness and folding mechanism (Naqvi et al., 2018). We assessed the conformational behavior of apo-M^pro, M^pro-10428962 and M^pro-71481120 systems by computing their average R_g values as 2.22 nm, 2.21 nm, and 2.20 nm, respectively. The analysis shows a minor decrease in the R_g values when in the bound states with the selected compounds. A little decrease in R_g is showing higher compactness of M^pro while its binding pocket is occupied by 10428963 and 71481120. However, initially up to 10 ns, the M^pro in presence of 71481120 was found with an increased R_g which suggesting initial adjustment of M^pro binding pocket occupied with the ligand. Here, no structural shift was observed in M^pro in the presence of the compounds where the R_g is attaining a stable equilibrium, suggests stability of protein-ligand complexes during the entire simulation (Fig. 4C).

The solvent-accessible surface area is calculated as an interface surrounded by a solvent (Ausaf Ali et al., 2014; Rodier et al., 2005). This solvent behaves differently with varying conditions and thus a useful parameter to study the conformational dynamics of a protein in the solvent environment. To investigate the conformational behavior of M^pro before and after the binding of 71481120 and 10428963, we have computed the SASA of all three systems. The average SASA values for apo M^pro, M^pro-10428962 and M^pro-71481120 were found as 148.47 nm², 149.75 nm^2, and 149.04 nm², respectively. A minor increase in the SASA of M^pro while binds with the compounds were observed possibly due to the exposure of some inner residues to the protein surface (Fig. 4D). The plot suggests that SASA attained an equilibrium without switching throughout the simulation signifying structural stability of M^pro before and after 10428963 and 71481120 bindings.

Intramolecular hydrogen bonding within protein molecules plays a fundamental role to stabilize their three-dimensional structure (Hubbard and Kamran Haider, 2001; Naz et al., 2018, 2017). To validate the stability of M^pro and its ligand-bound complexes, we have calculated the dynamics of intramolecular hydrogen bonds paired within 3.5 Å. The computed average number of intramolecular hydrogen bonds in M^pro apo, M^pro-10428962 and M^pro-71481120 was found to be 216 and 215, 215, respectively ( Fig. 5 A). A slight decrease in the number of average hydrogen bonds within M^pro itself is due to the occupancy of some intramolecular space of the binding pocket by compound 10428963 and 71481120. The Probability distribution function (PDF) analysis of H-bond dynamics shows that the complexes of M^pro-10428962 and M^pro-71481120 are quite stable during the entire simulation (Fig. 5B).

Fig. 5.

Structural compactness of M^pro before and after 10428963 and 71481120 binding. (A) Time evolution of hydrogen bonds formed intramolecular within M^pro (B) The PDF of intramolecular Hydrogen bonds.

3.5. Principal component and free energy landscape analysis

The dynamics of a protein structure can be illustrated through their phase space performance. We did PCA to study the conformational sampling of M^pro in the free state as well as in complexed states with 10428963 and 71481120 via studying their collective motions while using the essential dynamics approach (Amadei et al., 1993). The PCA analysis allows us to explore the dynamic motion and flexibility of a protein structure in conformational space. The conformational sampling of M^pro and its docked complexes in the essential subspace is illustrated in Fig. 6 A. This projection shows the structural conformations of M^pro along with the eigenvector (EV)-1 and EV-2 projected by C^α atoms. Here, we observed that the M^pro-10428962 and M^pro-71481120 complex occupied the same conformational subspace as M^pro in the free state. An increased dynamic was observed at EV1 in the case of the complexes, but no overall shift of the M^pro in complexed with 10428963 and 71481120 was observed at both EVs (Fig. 6B). Both complexes are overlapping the stable clusters with phase space of M^pro-apo in the free state. PCA analysis including the RMSF indicates that M^pro and its complexes are pretty stable during the entire simulation.

Fig. 6.

Principal component analysis. (A) 2D projections of trajectories on eigenvectors (EVs) showing conformational projections of M^pro (B) The projections of trajectories on both EVs with respect to time (C) Residual fluctuations of M^pro on eigenvector 1.

To study the conformational behavior and native states of M^pro and its complexes, the FELs were created using the first two EVs. The FELs of apo M^pro, M^pro-10428962, and M^pro-71481120 complex systems are shown in Fig. 7 . When studying the plots, a deeper blue color is signifying the conformational states with lower energy near to native states. We observed that M^pro is having only a single global minimum confined within a local basin. Similarly, M^pro in presence of 10428963 and 71481120 doesn’t acquire multiple minima and showing single minima but different conformational motion. Both complexes didn’t go to the development of multiple minima and showing a single global state. The plots suggest that the binding of 10428963 and 71481120 to M^pro affects the size and the location of the sampled essential subspace but with a stable single global minimum ( Fig. 7B-C).

Fig. 7.

The Gibbs energy landscapes for (A) free M^pro (B) M^pro-and 10428963 (C) M^pro-71481120.

Altogether, the drug-like properties including physicochemical and ADMET properties, higher and specific binding towards the SARS-CoV-2 M^pro substrate-binding site, and stability during MD simulation studies suggest that 6-Deaminosinefungin (PubChem ID: 10428963) and UNII-O9H5KY11SV (PubChem ID: 71481120) can act as potential leads in drug development against SARS-CoV-2 infection. Both compounds are analogs of known antivirals and showing admirable protease inhibitory potential with improved pharmacological properties and considerably high affinity and stability with SARS-CoV-2 M^pro. These findings may be further implemented in the development of effective therapeutics to Covid-19 after required experimental validation.

4. Concluding remarks

In this study, we extensively analyzed the available SARS-CoV-2 M^pro structures, co-crystallized with different molecules and inhibitor-like compounds investigate their binding pattern and mechanism of inhibition. We utilized the structural information of co-crystallized M^pro structures to find high-affinity SARS-CoV-2 M^pro inhibitors from the PubChem database. The compounds were further screened for their physicochemical properties, drug-likeliness, and molecular interactions with SARS-CoV-2 M^pro. Finally, we identified two compounds, 6-Deaminosinefungin (PubChem ID: 10428963) and UNII-O9H5KY11SV (PubChem ID: 71481120) showing the high binding affinity and interact to the conserved residues of the substrate-binding pocket of SARS-CoV-2 M^pro. Both compounds are analogs of known antiviral compounds with protease inhibitory potential and improved pharmacological properties. All-atom MD simulation studies suggested complex stability of M^pro in the presence of both compounds with minimal structural changes.

Funding

This work is funded by the Indian Council of Medical Research.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.

CRediT authorship contribution statement

Taj Mohammad: Conceptualization, Methodology, Software, Investigation, Data curation, Writing - original draft, Visualization. Anas Shamsi: Conceptualization, Methodology, Investigation, Data curation, Writing - original draft, Writing - review & editing. Saleha Anwar: Methodology, Investigation, Writing - original draft, Visualization, Software, Validation, Investigation. Mohd. Umair: Methodology, Investigation, Writing - original draft, Software, Validation. Afzal Hussain: Methodology. Md. Tabish Rehman: Validation, Visualization. Mohamed F. AlAjmi: Software, Writing - review & editing. Asimul Islam: Validation, Data curation, Writing - review & editing, Supervision, Funding acquisition. Md. Imtaiyaz Hassan: Conceptualization, Validation, Data curation, Writing - original draft, Writing - review & editing, Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no conflict of interest.

Appendix A. Supplementary data

The following is Supplementary data to this article: