Conformational perturbation of SARS-CoV-2 spike protein using N-acetyl cysteine: an exploration of probable mechanism of action to combat COVID-19

Abstract The infection caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) resulted in a pandemic with huge death toll and economic consequences. The virus attaches itself to the human epithelial cells through noncovalent bonding of its spike protein with the angiotensin-converting enzyme-2 (ACE2) receptor on the host cell. Based on in silico studies we hypothesized that perturbing the functionally active conformation of spike protein through the reduction of its solvent accessible disulfide bonds, thereby disintegrating its structural architecture, may be a feasible strategy to prevent infection by reducing the binding affinity towards ACE2 enzyme. Proteomics data showed that N-acetyl cysteine (NAC), an antioxidant and mucolytic agent been widely in use in clinical medicine, forms covalent conjugates with solvent accessible cysteine residues of spike protein that were disulfide bonded in the native state. Further, in silico analysis indicated that the presence of the selective covalent conjugation of NAC with Cys525 perturbed the stereo specific orientations of the interacting key residues of spike protein that resulted in threefold weakening in the binding affinity of spike protein with ACE2 receptor. Interestingly, almost all SARS-CoV-2 variants conserved cystine residues in the spike protein. Our finding results possibly provides a molecular basis for identifying NAC and/or its analogues for targeting Cys-525 of the viral spike protein as fusion inhibitor and exploring in vivo pharmaco-preventive and its therapeutic potential activity for COVID-19 disease. However, in-vitro assay and animal model-based experiment are required to validate the probable mechanism of action. Communicated by Ramaswamy H. Sarma Spike protein is used to target human ACE2 receptor for viral fusion into the human cell. N-acetyl cysteine can interestingly inhibit this viral fusion process by increasing the probability of structural deformation of this viral spike protein. This process may indirectly reduce the intermolecular binding affinity between two enzymes


Introduction
COVID-19 is caused by the highly pathogenic RNA virus, SARS-CoV-2 (Renhong et al., 2020).Unlike SARS-CoV-1 and Middle East Respiratory Syndrome Coronavirus (MERS-CoV), the rapid rate of infection caused by SARS-CoV-2 with a mortality rate of 1-2% has resulted in a pandemic across the globe (Lan et al., 2020).Phylogenetic analyses of the coronavirus genomes showed that SARS-CoV-2 belongs to the Beta coronavirus genus.The genome of SARS-CoV-2 is a single-stranded RNA consisting of about 30 kb nucleotides (Junwen et al., 2020).SARS-CoV-2 encodes for four major structural proteins, namely spike protein, membrane protein, envelope protein, and nucleocapsid protein.
The entry of the SARS-CoV-2 into human cells initiates through the transmembrane spike protein that forms trimers protruding from the virus cell surface (Junwen et al., 2020;Wrapp et al., 2020).The spike protein protrudes from the envelope of the virion and consists of two subunits -(1) a receptor-binding domain (RBD) that interacts with the receptor proteins of host cells (2) a second subunit that facilitates the fusion of the viral membrane into the host cell membrane.The receptor-binding domain (RBD) of the spike protein binds strongly with the host ACE2 receptor to enter into host cells with a dissociation equilibrium constant (K d ) of 15 nM (Renhong et al., 2020;Wrapp et al., 2020).ACE2 is a type I membrane protein expressed in multiple organs such as lungs, heart, kidneys, testis intestine, and also endothelial cells of arteries, with a primary physiological role in the maturation of angiotensin, a peptide hormone that controls vasoconstriction and blood pressure (Renhong et al., 2020).
An examination of the crystal structure of the spike protein and ACE2 complex showed that there are 13 hydrogen bonds and 2 salt bridges at the interface between RBD of the SARS-CoV-2 spike protein and the ACE2 receptor (Lan et al., 2020).Stereo specific orientations of the interacting amino acid residues of the spike protein are provided by its structural architecture, where four intra-molecular disulfide bonds between cysteine residues contribute.Among these four disulfide linkages of the spike protein, three disulfide bonds (cys336-cys361, cys379-cys432 and cys391-cys525) stabilize the b sheet structure and the fourth disulfide bond (cys480-cys488) joins the loops at the distal end of the receptor binding motif that belongs to RBD of the spike protein (Figure 1) (Lan et al., 2020).
Earlier it is reported that the binding affinity between spike protein and ACE2 enzyme was significantly impaired when all of the disulfide bonds of both ACE2 and SARS-CoV/CoV-2 spike proteins were reduced to thiol groups (Hati & Bhattacharyya, 2020).It is also reported that several reducing agents and thiol-reactive compounds are able to inhibit viral entry by disruption of the disulfide bonds (Man� cek- Keber et al., 2021).Based on this theory, Poe et al. hypothesized that NAC might be a potential therapeutic molecule to treat COVID-19 (Poe & Corn, 2020).Thus, in the present study we proposed to explore the mechanism of action of N-acetyl cysteine (NAC) against SARS-CoV-2.The reduction of the solvent accessible disulfide bond followed by the conjugation of NAC leads to the perturbation of the functionally active structure of spike protein in SARS-CoV-2 and thereby reducing the infectivity of the virus.In the present study, we have successfully reduced the most accessible disulfide bond (cys391-cys525) of purified S1 subunit of SARS-CoV-2 spike protein through thiol/disulfide exchange mechanism, executed by N-acetyl cysteine (NAC), which is a commonly prescribed antioxidant and a mucolytic agent and it is now under the investigation for anti-COVID-19 therapy like other repurposing FDA approved drug molecules (de Lizarrondo et al., 2017;Flora et al., 2020).Upon covalent conjugation of NAC with cys525, a comparative in silico analysis showed significant perturbation in the stereo specific orientation of interacting residues of RBD of spike protein with ACE2 receptor.As a proof of concept, LC-MS study was carried out for understanding the conjugation process of NAC with important cysteine residues of spike protein for its structural perturbation and the potential dysfunction against COVID-19.

Solvent accessibility study of disulfide bonded cysteine residues
Solvent accessibility of cysteine residues-based disulfide bonds was examined via in silico method.To do this experiment, viral spike protein was obtained from the crystal structure of spike protein-ACE2 receptor (human) complex available in protein databank (PDB ID: 6LZG) (Wang et al., 2020).After taking that extracted protein, VEGAZZ software was used by considering default parameters to check the solvent accessibility in their pockets near to eight cysteine residues that are disulfide bonded in the spike protein (Pedretti et al., 2004).

Binding pocket identification analysis and molecular docking study
To detect probable binding pockets inside ligand free spike protein, CASTp web server was utilized by taking solvent probe sphere 1.4 Å.The pocket identification assessment was created on recent theoretical and algorithmic results of computational geometry, which incorporates delaunay triangulation, alpha shape, and discrete flow (Binkowski et al., 2003).
A molecular docking study was performed in AUTODOCK 4.2 (Morris et al., 2009) after the selection of the probable binding pocket in spike protein.Before the docking study, the structure of NAC was prepared in Sybyl-X 1.3 (Sybyl-X 1.3, 2010).Afterwards it was subjected to geometrical optimization applying the Powell energy minimization algorithm, Gasteiger-Huckel charges, and 0.001 kcal/(mol Å) as convergence criteria (Debnath et al., 2017).
To investigate the binding mode of NAC, protein coordinates of spike protein extracted from the co-crystal of spike protein-ACE2 receptor (human) (PDB code: 6LZG) was considered.For docking simulation, the protein was prepared in Sybyl-X 1.3 by pondering the same procedure as adopted in case of study molecule.Here, cysteine residue based molecular docking was performed to dock the NAC molecule adjacent to cys480 and cys488 residues of the spike protein which is placed in the interface of human ACE2 receptor and viral spike protein.The prepared spike protein and the corresponding study molecule were taken in AUTODOCK 4.2 for the flexible docking analysis by previously reported methods.Finest docked conformers were collected from a population of 150 samples by observing minimum Gibbs free energy values.Pymol and MOE software were employed to visualize the interactions of protein and docked molecules (Seeliger & de Groot, 2010;MOE, n.d.).

Molecular dynamics (MD) simulations
Docked model was further analyzed through MD simulation study by using NAMD 2.7 to explore the stability of these systems and the dynamics of molecular interactions between spike protein and NAC (Phillips et al., 2005).To perform this study, the protonation state of histidine residues of the protein at physiological pH (7.4) was assigned using the Hþþ web server (Gordon et al., 2005).CHARMM22 force Field (including CMAP correction) was considered to parameterize the protein structure (Brooks et al., 1983).This CHARMM22 force field produced enhanced dynamical and structural properties of proteins in MD simulations.Ligand (NAC) parameters were generated through the ParamChem web server (https://www.paramchem.org/).The MD simulation study was accomplished under periodic boundary conditions applying the solvated (water box) spike protein.The prepared spike protein-NAC docked complex was run for 100 ns using multi-core and CUDA supported NAMD package.The NPT method (Pressure 1 atm; Temperature 310 K) with 12 Å as cut-off for non-bonded atom interactions, Particle Mesh Ewald algorithm for long-range electrostatic forces and Langevin's piston in 'on' condition was employed to run the MD simulation (Dewaker et al., 2020).The relative binding free energy (MMPBSA) computation method was implemented using CaFE plugin along with the standard parameter files (CHARMM22) of MD simulation, default arguments and an internal dielectric constant value 1.6 (Li et al., 2018).The time step for dynamics integration was set to 2 fs.The simulation trajectories were documented at 1000 steps (2 ps) gaps.The complex system was satisfactorily minimized for 70 ps followed by the application of equilibration run for 100 ns.Visualization and assessment of the trajectories were done in VMD 1.9.3 (Humphrey et al., 1996).Finally by using of standard tcl scripts the RMSD and Rg of the respective protein trajectories were investigated.

Protein-protein molecular docking study
Protein-protein docking study was performed via HDOCK web-server (Yan et al., 2017).This method was accomplished for the crystal structure of ligand free viral spike protein with ACE2 receptor (human) as well as MD simulated NAC docked viral spike protein with the same ACE2 receptor (human) to recognize the difference in their binding affinities towards that ACE2 receptor which is present in human body.This study was based on a hybrid algorithm of template-based modeling and ab-initio free docking methods.Prodigy server was utilized to define DG and K d values of the predictive protein-protein interaction contacts (Xue et al., 2016).Pymol software was used to envision the interactions between two proteins.Additionally, NAC molecule was manually conjugated with free thiol group of cys525 residue by reducing the most solvent accessible disulfide bond (cys391-cys525) to form a new cys525-S-S-NAC connection.This was implemented on the basis of distance matrix analysis in between NAC and cysteine residues after MD simulation study.The study was carried out to verify the change in their binding affinities with ACE2 receptor via additional protein-protein docking analysis.

In silico ADMET property analysis
To assess the theoretical ADME property of the NAC molecule, pkCSM online prediction platforms was used.Pharmaceutically relevant properties such as H-bond donor, H-bond acceptor, octanol-water partition coefficient (LogP), surface area, and number of rotatable bonds were calculated through this software.Furthermore, effect of ligands on ADMET parameters like water solubility, Caco2 permeability, human intestinal absorption, skin permeability, P-glycoprotein I and II inhibition, volume of distribution, fraction of unbound drug, Blood Brain Barrier and CNS permeability, cytochrome P450 (CYP3A4 and CYP2C9 inhibition) inhibition, total clearance, action as renal OCT2 (organic cation transporter 2) substrate were also predicted.

In vitro methods
The purified S1 subunit of the spike protein of SARS-CoV-2 (Uniprot Accession ID: QHD43416) was obtained from RayBiotech (Georgia, USA).The expressed region of the protein spanned from Val16 to Gln690 residues.N-ethyl maleimide (NEM), dithiothreitol (DTT), iodoacetamide (IAM) and Nacetyl cysteine (NAC) were procured from Sigma (St. Louis, MO).Rapigest detergent was obtained from Waters (Milford, MA, USA).All solvents used were of LC À MS grade and all other chemicals used were of analytical grade.

Differential modification of the S1 subunit of spike protein with NEM and IAM
The S1 subunit of the spike protein (64 lg) was dialyzed against 50 mM ammonium bicarbonate buffer, pH 7.4.The dialyzed protein was incubated with NEM using protein/NEM ratio 1/20 mol/mol at 37 � C for 1 hr.Subsequently, the NEM conjugated protein was passed through a G-10 spin column to remove all unreacted NEM from the solution.Following this, the NEM conjugated protein was divided in two parts.The first part was subjected to proteolytic digestion with trypsin maintaining an enzyme/substrate ratio 1/10 mol/mol for overnight and this served as control for the LC-MS analysis.The second part was treated with DTT for the reduction of disulfides by maintaining a 50-fold excess of DTT to the concentration of protein in the solution, for 30 min at 60 � C. The reduced disulfides were further subjected to alkylation by incubating the solution with IAM for 1 hr in the dark at room temperature maintaining a 100-fold excess concentration of IAM to the concentration of protein in the solution.The differentially modified protein solutions was subjected to proteolytic digestion overnight with trypsin by maintaining enzyme/substrate ratio 1/10 mol/mol and further used for LC-MS analysis.

Modification of the S1 subunit of spike protein with NAC
About 7 lM of the S1 subunit of the spike protein was incubated with NAC in 50 mM ammonium bicarbonate buffer pH 7.4 for 4 h at 37 � C by maintaining a protein/NAC ratio of 1/10 mol/mol.Following incubation, the NAC modified protein was cleaned up using a G-10 spin column to remove all unreacted NAC from the solution.The NAC conjugated protein was denatured using 0.2% Rapigest at 80 � C for 15 min followed by proteolytic digestion using trypsin maintaining an enzyme/substrate ratio of 1/10 mol/mol at 37 � C and incubated overnight.Subsequently, the samples were acidified in 0.5% (v/v) formic acid (FA) and incubated at 37 � C for 90 min to hydrolyze Rapigest detergent present in the solution.Samples were then centrifuged for 45 min at 4000�g and the supernatant was isolated and used for LC-MS analysis.

LC-MS analysis
LC-MS analysis was performed using Synapt G2-Si High-Definition ESI Mass Spectrometer (HDMS, Waters) coupled to Acquity I class UPLC (Waters)a MassLynx TM 4.1 informatics solution was used for data acquisition and UNIFI 1.9.4 version was used for data processing.We used Acquity UPLC peptide CSH C18 column (130 Å, 1.7 mm, 2.1 mm X 150 mm, Waters) for the liquid chromatographic separation (Das et al., 2013).The column oven temperature was maintained at 65 � C and the autosampler was maintained at 10 � C. 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B) were used as mobile phase solvents.The flow rate was 0.2 mL/min and injection volume of the sample was 10 lL.The gradient program used was as follows: 0-2 min, 95% solvent A; 2-40 min, 57% solvent A; 40-42 min, 20% solvent A; 42-45 min, 20% solvent A; 45-46 min, 95% solvent A, 46-60 min, 95% solvent A.
Electrospray ionization (ESI) source was used for data acquisition and the mass spectrometer was used in the positive polarity with 40.0 V of sample cone voltage, and 3.0 kV of capillary voltage.Desolvation temperature was set to 350 � C and ion source temperature was 120 � C. The desolvation gas flow and cone gas flow were maintained at 800 (L/Hr) and 50 (L/Hr), respectively.Data were acquired in MS E mode and mass was scanned between the range 50-1995 Da with trap ramp collision energy from 30 to 50 eV.For lockmass correction, leucine-enkephalin ([M þ H] þ ¼ 556.2771) (0.2 ng/mL) was infused into the mass spectrometer as a reference lockmass.

Results and discussion
SARS-CoV-2 makes its way to lung cells via the ACE2 receptor.In this event, the spike protein of SARS-CoV-2 interacts with ACE2 receptor through its RBD where the main interacting residues are Leu455, Tyr473, Tyr489, Gln493, Asn501 andTyr505 (Figure 1).It is likely that the impairment in the binding of spike protein with human ACE2 receptor might be a feasible strategy to inhibit the SARS-CoV-2 infection in the host.In general, the tertiary/quaternary structure of a protein might be perturbed by reducing the disulfide bonds between cysteine residues that provide the structural architecture to the molecule.Crystal structure of spike protein (PDB ID: 6LZG) showed that out of nineteen cysteine residues, eight cysteines are covalently bonded by four disulfide bridges.Spike protein-ACE2 complex showed that the disulfide bond between cys480 and cys488 residues of spike protein is located closer to the interacting residues of ACE2 receptor in the complex, whereas other disulfide bonds are at a distant from the site of interaction (Figure 2).However, an analysis of the solvent accessible surface area (SASA) of different cysteine residues in the crystal structure (PDB ID: 6LZG) indicated that among the four disulfide bonds of spike protein, the bond between cys391 and cys525 residues is most accessible to the solvent (Table 1).Solvent accessibility analysis suggested that no binding site was available for NAC near to cys336-cys361 and cys379-cys432 regions.Reduction of surface exposed disulfide bonds in the protein can be achieved through thiol/disulfide exchange mechanism executed via free thiol group (-SH) of a reducing agent such as NAC, a drug that is commonly prescribed as an antioxidant and a mucolytic agent (Hati & Bhattacharyya, 2020).In the present study, spike protein was extracted from its complex with ACE2 receptor (PDB ID: 6LZG).Based on the SASA values of the disulfide bonds, CASTp web server based pocket identification analysis was carried out to find out the probable binding sites where disulfide groups are present inside the pocket or close to the pocket area.This study indicated the following possible sites for the docking of NAC in the spike protein: one is near to the vicinity of RBD site, pocket 1 (cys480: 17.29 Å and cys488: 7.91 Å) and the other is far apart from RBD site, pocket 3 (cys391: 57 Å and cys525: 44.74 Å). (Figure 3; Table 1).The solvent accessible area of pocket 1 near to the RBD site was smaller (74.09Å 2 ) than that of pocket 3 (89.68Å 2 ).However, during molecular docking, pocket 1 was considered for NAC binding as it was closer to the RBD site.It is used to check the initial binding interaction of NAC with the conserved domain of the viral protein which is available at the interface of human ACE2 and viral spike protein.The docking study showed that the H-bonding interaction occurred between NAC and following three residues of the spike protein, Pro479, cys480 and    Asn481 (Figure 4A,B).In addition, hydrophobic interactions were observed between NAC and nearby residues including Thr478, Pro479, Phe486, cys488.These interactions resulted in a considerable binding affinity (DG ¼ À 7.32 kcal mol À 1 ).(Table 2) Subsequently, to evaluate the stability of the NACspike protein complex, Molecular Dynamics (MD) simulation was performed for 100 ns.MD simulation study demonstrated that the molecular trajectories of the spike protein and NAC were not close to each other, due to a shift of NAC from the docking site to the N-terminal site of the spike protein.The trajectories of MD simulation were assessed by analyzing their Mean Square Division (RMSD), Radius of Gyration (RG), RMSF and through principal component analysis (PCA) (Grant et al., 2006).The RMSD, RG and RMSF of the system indicated that the trajectories followed a harmonious swirl throughout the dynamics (Figure 4E,G,H and Supplementary video S1 and S2). Figure 5A shows the results of PCA.The eigenvalues from the PCA indicated that first seven principal components collectively explained 62.5% the total mean square displacement (or variance) of atom positional fluctuations (Figure 5D).Among the components, PC1 to PC3 together explained 40.6% variance in the atom positional fluctuations.Projection of the distribution of trajectories onto the subspace defined by the largest principal components  (PC1, PC2 and PC3) are shown in Figure 5A-C.Figure 5E shows the residue-wise loadings for PC1 and PC2.During the course of MD simulation, the coordinates of the most important ACE2 interacting amino acid residues changed significantly.Furthermore, after 3 ns MD simulation, NAC unglued from the initial docked site (near C-terminal) and travelled across (about 9.9 Å) to be lodged/bound to a site (S1 domain) near the N-terminal.To verify this, we have repeated MD simulation for 20 ns with different initial velocities and the results obtained are almost same as explained above.The binding free energy calculation using MMPBSA approach clearly pointed out that the new binding site of NAC in the S1 domain of N-terminal of spike protein is more favorable (DG ¼ À 19 kcal/mol) to make a stabilized form than that of the initially docked complex (DG ¼ À 11 kcal/ mol) (.txt files are available in Supplementary information).The H-bond pattern of the trajectories from MD simulation are shown in Figure 4G.The figure indicates that number of H-bond formations increased after 35000 frames (trajectories) during the course of MD simulation.Various residues (Cys361, Ala522, Thr523, Cys525, etc.) were found to make Hbonds with thiol and carbonyl groups of NAC during the course of simulation.Among these residues, Thr523 in the S1 domain of N-terminal of spike protein showed maximum Hbond occupancy (2.79%) with thiol functional group of NAC.The distance between initial and final positions of NAC is 59.2 Å (Figure 4F).The new location of NAC is close to the disulfide bond between the cys391 and cys525 residues.The binding affinity of NAC with spike protein is improved by two strong H-bond interactions with the pocket residue Thr523 (Figure 4C,D).Docking study followed by MD simulation clearly showed that NAC can easily fit into the binding pocket 3 because of its better interaction with binding site and the availability of larger pocket volume, in comparison to other binding pockets of the spike crystal protein (Figure 3).In addition, NAC demonstrated hydrophobic interactions Spike glycoprotein-ACE2 receptor(human) interaction after NAC conjugation (in silico form) 1 DG (kcal mol À 1 ) À 7.   with other pocket residues such as cys361, Val362, and Ala522.From the last trajectory of the MD simulation study, the distance between centroid of NAC with cys391 and cys525 were observed to be 4.4 Å and 3.2 Å respectively (Figure 8).Thus, it improved the probability of thiol/disulfide exchange between the free thiol group (-SH) of NAC and the cys391-cys525 disulfide bond.Furthermore, in silico based ADMET (Absorption, Distribution, Metabolism, Excretion and Toxicity) analysis was carried out for NAC (Supplementary information, Table S1).It contains hydrogen bond acceptors (HBA, �10) and donors (HBD, �5) in accordance with the Lipinski's rule of five indicating good bioavailability of the candidate molecule.Additionally, a variety of key ADMET properties were also predicted through pkCSM server (Supplementary information, Table S1).Moderate Caco-2 permeability (0.486) was found for this NAC molecule.Fascinatingly, NAC showed 100% permeation potential across the intestinal membrane with greater than 77.6% Intestinal absorption (IA).Also, it showed good skin permeation (permeation > À 2.7).P-glycoprotein I and II were not predicted to inhibit the absorption of this molecule.In terms of toxicity prediction, NAC doesn't show any hepatotoxicity and skin sensitization issue as adverse effects.
To investigate the effect of NAC on the disulfide connectivity of spike protein in vitro, the purified spike protein was subjected to differential alkylation using N-ethyl maleimide (NEM) in the first step followed by the reduction of disulfide bonds of the NEM-modified protein with dithiothreitol and finally alkylation of the reduced protein with iodoacetamide (IAM).Mass analysis of the trypsin digests of the differentially alkylated protein indicated that the cysteine residues within the proteins were in two different environments.When compared with the crystal structure of the spike protein in complexed with ACE2 receptor (PDB ID: 6LZG), the observed NEM versus IAM modification of cysteine residues in the mass spectra indicated that the following cysteine residues were disulfide connected in the spike protein: cys391, cys525, cys336, and cys432.Figure 6, panel A shows two peptide fragments with 679.7 m/z and 790.7 m/z, containing two cysteine residues, cys525 and cys538 modified with IAM and NEM respectively.The observed results correlated with the crystal structure of the protein complex where cys538 was in the free form and cys525 was disulfide bonded with cys391.Panel B, Figure 6 shows the proteolytic digest of the NAC treated spike protein, where the tryptic fragment consisting cys525 appeared with 660.7 m/z and the respective NAC conjugated peptide appeared with 714.4 m/z.The tandem mass spectra of the molecular ion with 714.4 m/z in panel C, Figure 6 indicates that the proteolytic fragment with 714.4 m/z in panel B originated from a stretch of amino acid residues 510-528 consisting of cys525 modified with NAC.All other modifications of disulfide bonded cysteine residues are shown in the supplementary figures S3-S5.In principle, NAC can form a covalent conjugate with a protein only through thiol/disulfide exchange process between its free sulfhydryl group and the disulfide bonded cysteine residues in the protein.The predicted dissociation equilibrium constant (K d ) values of protein-protein interaction clearly revealed that upon NAC conjugation, the binding affinity of spike protein with ACE2 was decreased three-fold, with a concomitant decrease in the release of free energy (DG) by 5 kcal mol À 1 (Table 1).The change in the number of polarpolar as well as polar-apolar/non polar interactions at the interface between two proteins provided the molecular insights into the decrease in binding affinity of NAC conjugate of spike protein with the ACE2 receptor (Table 2).In the in silico study, prior to MD simulation, the reduction of cys391-cys525 bond and subsequent conjugation between NAC with cys525 of the spike protein, the closest cysteine residue at 3.2 Å apart, was made manually by maintaining the protein coordinates (Supplementary information, Figure S2).Upon reduction of the disulfide bond via NAC, the spike protein was unfolded to a significant extent.However, no similar type of unfolding event was observed for other three disulfide bonds of spike protein when those were separately broken (Figure 7A,B,D).This clearly indicated that the cys391-cys525 disulfide bond significantly contributed to the structural integrity of spike protein.In Figure 7C, NAC is not conjugated with spike protein.Disulfide bond provides structural stability in protein.Thus, the cleavage in this SS bond perturbs the stability of protein structure.In the present study, we also observed that the protein conformation was significantly altered upon breakage of Cys391-Cys525 disulfide bond manually.Using our sybyl software, we performed manual disulfide bond cleavage to identify the most sensitive disulfide group, whose reduction might be responsible for significant perturbation in the structure of spike protein.
Here, MD simulation studies were not utilized as there is no such parameter available in MD simulation study to perform the cleavage of disulfide bond via thiol disulfide exchange mechanism.NAC has the ability to break a disulfide bond via thiol exchange mechanism and subsequently gets conjugated with one of the generated cysteine residue that is in the close proximity to it.In this study, the MD simulation was carried out only to explore the probable binding site of the NAC with more accuracy.During this experiment, interestingly we have observed that the binding stability of NAC with spike protein is significantly high to the region where most sensitive disulfide bond (Cys391-Cys525) exists in the spike protein.This might happen due to more solvent accessibility near to Cys391-Cys525 disulfide bond.
The foregoing information provides important insights into the cleavage of the disulfide bond between cys391 and cys525 residues of the spike protein by the action of NAC, which might impair the binding and consequently, inhibit the SARS-CoV-2 infection.After NAC conjugation with cys525, the energy minimization was carried out.The conformational change of the protein indicated that the unfolding of the spike protein led to an increase in the distance between C-and N-terminus of the S1 domain of spike (S) protein from 9.9 Å, as per the crystal structure of the complex, to 130.2 Å (Figure 8).This resulted in a significant perturbation of the stereo specific orientation of the interacting residues in the RBD of spike protein with those of ACE2 receptor which eventually might be reflected in a decreased binding affinity of the spike protein with the ACE2 receptor (DG ¼ À 7.48 kcal/mol) (Table 2) in comparison with NAC unbounded spike protein-ACE2 protein complex (DG ¼ À 12.57 kcal/mol) (Figure 8).Number of interaction contacts in between two proteins are also reduced (Supplementary information Table S2).
In recent years, several studies have reported the importance of the disulfide bond between Cys391-Cys525 along with the other disulfide bond between Cys379-Cys432 in the RBD domain of the spike protein for maintaining the overall structural integrity of the protein.Shi et al have performed in vitro assays with two thiol-based chemical probes that act as reducing agents, P2119 and P2165 to demonstrate that the disruption of the disulfides Cys391-525 and Cys379-Cys432 by these reducing agents decreases the binding of spike glycoprotein to its ACE2 receptor and thereby inhibits infection by SARS-CoV-2.Thiol-based chemical probes exhibit antiviral activity against SARS-CoV-2 via allosteric disulfide disruption in the spike glycoprotein.Another study reported the destabilisation of the structure of spike protein upon reduction of the Cys391-Cys525 disulfide bond in the RBD domain thereby leading to reduced host cell infectivity by the virus.Disruption of disulfides within RBD of SARS-CoV-2 spike protein prevents fusion and represents a target for viral entry inhibition by registered drugs.In addition, the anti-viral effect of NAC that helps to prevent infection in the host by SARS-CoV-2 has been thoroughly established in other independent studies as well.

Conclusion and future perspective
In silico analysis indicated that the reduction of the disulfide bridge between cys391 and cys525 residues followed by the covalent conjugation of NAC with cys525 of spike protein resulted in a significant perturbation in the binding affinity that is critical in the interaction between spike protein and ACE2 receptor.Mass spectra analysis was carried out as a proof of the concept.These combination of data for NAC might be fruitfully explored through in vitro as well as in vivo analysis to pave a new way for identifying potent analogues of NAC against COVID 19.Additionally, this strategic approach might be explored in other viral infections as well where the disulfide bond (cystine residue) across surface accessible cysteine residues contributes significantly in the functionally active conformation of the viral protein.
Reduction of this crucial cystine residue might lead to structural impairment such that interaction between virus and host is weakened and subsequently the infection might be inhibited.

Figure 2 .
Figure 2. Distance between disulfide bonds (green color) of the viral spike protein with selected human ACE2 interacting residues (cyan color) present in the binding domain.Figure (A) represents the locations of disulfide bonds of the spike crystal protein and important interacting residues of human ACE2 protein.Figure (B) shows the distance (Å) between the centroid of the disulfide bond of cys480-cys488 and different nearby interacting residues of ACE2 protein present in the protein-protein interacting domain.
b A square angstrom (Å 2 ) is a non-SI (non-System International) measurement unit of area with sides equal to one angstrom (1Å).c Column D s G indicates the hydrophobic interaction between individual residues and solvent, in kcal/mol.

Figure 3 .
Figure 3. Predictive binding pockets of spike protein.Different colors indicate the position of binding pockets along with surface area for solvent accessibility.

Figure 4 .
Figure 4. Figures of different binding sites of NAC before and after MD simulation.(A) and (B) represent the docked poses of NAC before MD simulation.(A) shows 3D view of docked NAC along with their binding residues near to ACE2 receptor binding domain.Red dotted line (--) shows strong H-bond interaction of NAC with cys480.(B) shows 2D view of docked NAC along with surrounding residues.Green colour shows van der Waal interaction sites whereas violet colour shows the polar interaction sites.Proximity contour on each atom of NAC molecule indicates the ligand exposure sites.Similarly figures (C) and (D) display the binding site of NAC after MD simulation.Figure (E) is RMSD plot analysis of protein (blue) and ligand (brown).Figure (F) pointed out the positional distance (59.2Å) between centroids of NAC molecules before (cyan) and after (green) MD simulation.(G) Radius gyration (RG) of the spike protein after MD simulation (H) Root mean square fluctuation (RMSF) of different residues of spike protein-NAC complex after 100 ns MD simulation (I) H-bond formation in-between NAC and spike protein during MD simulation.
Figure 4. Figures of different binding sites of NAC before and after MD simulation.(A) and (B) represent the docked poses of NAC before MD simulation.(A) shows 3D view of docked NAC along with their binding residues near to ACE2 receptor binding domain.Red dotted line (--) shows strong H-bond interaction of NAC with cys480.(B) shows 2D view of docked NAC along with surrounding residues.Green colour shows van der Waal interaction sites whereas violet colour shows the polar interaction sites.Proximity contour on each atom of NAC molecule indicates the ligand exposure sites.Similarly figures (C) and (D) display the binding site of NAC after MD simulation.Figure (E) is RMSD plot analysis of protein (blue) and ligand (brown).Figure (F) pointed out the positional distance (59.2Å) between centroids of NAC molecules before (cyan) and after (green) MD simulation.(G) Radius gyration (RG) of the spike protein after MD simulation (H) Root mean square fluctuation (RMSF) of different residues of spike protein-NAC complex after 100 ns MD simulation (I) H-bond formation in-between NAC and spike protein during MD simulation.

Figure 5 .
Figure 5. PCA analysis of viral spike protein after 100ns MD simulation studies.(A) Protein conformation of the last frame of PC1 (blue) and PC2 (brown) trajectories respectively.Figure (B) indicates overall patterns of motion (PC1, PC2 & PC3) in the NAC-spike protein complex.(C) indicate the fluctuation of the residue positions in PC1 and PC2 trajectories.

Figure 6 .
Figure 6. Figure (A-C) represent the differential chemical modifications of two peptides containing cys525 and cys538 residues respectively.Panel A shows the peptide with 679.7 m/z (M þ3H) 3þ obtained upon conjugation of IAM to cys525, and peptide with 790.7 m/z (M þ2H) 2þ obtained upon modification of cys538 with NEM.Panel B shows the tryptic peptide with 714.4 m/z (M þ3H) 3þ obtained upon conjugation of NAC with cys525.Panel C represents the MS/MS spectra of the peptide with 714.4 m/z where the fragment ions, b and y, are labelled.

Figure 7 .
Figure 7. Figure (A-D) describe structural deformations after individual thiol formation at four different disulfide bond positions (across eight cysteine residues) of spike protein.

Figure 8 .
Figure 8. Difference in protein-protein interaction between ACE2 receptor and spike protein before and after NAC conjugation via thiol exchange mechanism: (A) shows interaction between the crystal forms of ACE2 receptor and NAC free spike protein whereas figure (B) shows in silico protein-protein interactions between ACE2 receptor and NAC conjugated spike protein.The red dotted lines (--) in (A) and (B) displays the distance between C terminal and N terminal of spike protein, before (9 Å) and after (130.2Å)NAC conjugation.

Table 1 .
Cysteine residues based solvent accessibility analysis of spike protein using a crystal structure (PDB ID: 6LZG).
a Column ASA indicates the solvent-accessible surface area, in Å 2 , for the respective residues.

Table 2 .
NAC binding comparative pocket analysis with the details of protein-protein interaction contacts (ICs) after MD simulation study.