Design, synthesis, and molecular dynamics simulation studies of quinoline derivatives as protease inhibitors against SARS-CoV-2

Abstract A new series of quinoline derivatives has been designed and synthesized as probable protease inhibitors (PIs) against severe acute respiratory syndrome coronavirus 2. In silico studies using DS v20.1.0.19295 software have shown that these compounds behaved as PIs while interacting at the allosteric site of target Mpro enzyme (6LU7). The designed compounds have shown promising docking results, which revealed that all compounds formed hydrogen bonds with His41, His164, Glu166, Tyr54, Asp187, and showed π-interaction with His41, the highly conserved amino acids in the target protein. Toxicity Prediction by Komputer Assisted Technology results confirmed that the compounds were found to be less toxic than the reference drug. Further, molecular dynamics simulations were performed on compound 5 and remdesivir with protease enzyme. Analysis of conformational stability, residue flexibility, compactness, hydrogen bonding, solvent accessible surface area (SASA), and binding free energy revealed comparable stability of protease:5 complex to the protease: remdesivir complex. The result of hydrogen bonding showed a large number of intermolecular hydrogen bonds formed between protein residues (Glu166 and Gln189) and ligand 5, indicating strong interaction, which validated the docking result. Further, compactness analysis, SASA and interactions like hydrogen-bonding demonstrated inhibitory properties of compound 5 similar to the existing reference drug. Thus, the designed compound 5 might act as a potential inhibitor against the protease enzyme. Communicated by Ramaswamy H. Sarma Highlights Quinoline derivatives have been designed as protease inhibitors against SARS-CoV-2. The compounds were docked at the allosteric site of SARS-CoV-2-Mpro enzyme (PDB ID: 6LU7) to study the stability of protein-ligand complex. Docking studies indicated the stable ligand-protein complexes for all designed compounds. The Toxicity Prediction by Komputer Assisted Technology protocol in DS v20.1.0.19295 software was used to evaluate the toxicity of the designed quinoline derivatives. Molecular dynamics studies indicated the formation of stable ligand-Mpro complexes.


Introduction
The novel coronavirus, leading to coronavirus disease 2019 , has so far spread from China to 219 countries around the world and the fatality rate has hitherto resulted in more deaths (>3%) than severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) combined Cheng & Shan, 2020;. Several independent research groups have identified that SARS coronavirus 2 (SARS-CoV-2) belongs to Betacoronavirus, with highly identical genome to bat coronavirus, pointing to bats as the natural host (Ghosh et al., 2020a;Pal et al., 2020). The novel coronavirus (SARS-CoV-2) uses the receptor angiotensin-converting enzyme 2 (ACE2) and mainly spreads through the respiratory tract (Bourgonje et al., 2020). Importantly, increasing evidence showed sustained human-to-human transmission, along with many exported cases across the globe. The clinical symptoms of COVID-19 patients include fever, cough, fatigue and also gastrointestinal infection symptoms Zhao et al., 2020) in a small population of patients. The elderly and people with co-morbidities are more susceptible to infection and serious outcomes, which may be associated with acute respiratory distress syndrome (ARDS)  and cytokine storm (Coperchini et al., 2020).
Currently, due to the exigency of the finding a suitable therapeutic treatment for COVID-19, several potent antiviral candidates as repurposed under urgent investigation (Ancy et al., 2020) to reduce the time and hassles of approval. Hence, an endeavour of prodigious momentum and magnitude has been set out to develop vaccines against SARS-CoV-2, with successive publication of its genome sequence. Its genome encodes several non-structural and structural proteins, comprised of spike (S), envelop (E), membrane (M) and nucleocapsid (N) proteins (Du et al., 2009;Ghosh et al., 2020b). During viral replication process of SARS-CoV-2 and MERS, the proteolytic cleavage of viral polyproteins into functional subunits is achieved by protease enzymes (Mpro, also known as 3C-like protease) Kumar et al., 2013;. Hence, proteases become a hot spot for drug target against corona viruses without any off-target toxicity Ghosh et al., 2020c;Ghosh et al., 2021;Sharma et al., 2021). Therefore, impeding the Mpro catalytic activity utilising protease inhibitors (PIs) leads to thwarting of viral replication, prompting the amended clinical upshots in case of COVID-19 and associated diseases (Bhardwaj & Purohit, 2020;Ghosh et al., 2020a).
The majority of the candidate vaccines under trial against COVID-19, intend to generate neutralizing antibodies against the viral spike proteins (S) and averting its uptake through human ACE2 receptor, and thus resulting in blockage of infections. The vaccine strategies traditionally employ the whole virus, either attenuated or inactivated, and aspire to procreate a broader, more heterologous polyclonal response against several viral antigens. Subsequently, an emergency use authorization (EUAs) has been issued for several vaccine candidates against SARS-CoV-2 by different national and international drug regulation agencies. SARS-CoV-2 vaccine should match the following parameters to prove to be an ideal vaccine: (i) it should not only be able to provide protection from severe disease but also thwart infection in the vaccinated population, (ii) it should be able to reduce the number of immunocompromised individuals and protect those with history of infection, (iii) it should provide long term memory immune response after a minimal number of immunizations or booster doses, and (iv) it should easily be assessable for worldwide vaccination at the cheapest cost. In this series, six vaccine developers, viz. Pfizer and BioNTech, Curevac, Sanofi-GSK, AstraZeneca and the University of Oxford, Johnson & Johnson and Moderna, have successfully introduced their vaccines, which are being used at present. Several other pharmaceutical giants are still engaged in different levels of clinical trials. In addition to traditional strategies of treatment, several platform strategies, viz. nucleic acid platforms, non-replicating viral vectored platforms, inactivated virus or recombinant subunit vaccines, have also come into existence as unprecedented expeditious vaccine development efforts (Kyriakidis et al., 2021). The human beings are lucky enough to have several vaccines (used under emergency conditions) against SARS-CoV-2 as a great respite. Nevertheless, vaccines, the prophylactic way of thwarting the disease, are not effective enough to guarantee full safety and save the precious lives of human beings, and hence the second line of therapeutic treatment must be there and this necessitates an urgent and undeniable need for safe and effective drug candidates.
The SARS-CoV-2 is a positive-strand virus and hence some broad-spectrum antiviral drugs such as remdesivir, favipiravir, Darunavir, X77 and lopinavir are currently being used for treatment. Recently, 2-deoxy-D-glucose has been approved by the Drug Controller General of India for treatment, however, its utility at the ground level is yet to be proved (Triggle et al., 2021).
Despite all these efforts, a number of aspects of anti-SARS-CoV-2 immunity are still unknown and the world is still facing the issues of morbidity and mortality caused by COVID-19. Thus, along with an ideal vaccine, an effective and safe drug that can abolish the virus from the human body, is privational prerequisite in order to efficiently ebb out the indisposition and mortality, and to establish a vivacity The N-heterocyclic scaffolds usually perform a vibrant role in the advanced biological activity and thus demonstrate promising therapeutic potential, probably due to charisma of nitrogen atoms (Badavath et al., 2020). Quinoline is a heterocyclic aromatic organic compound that contains an electronically rich benzene ring fused with an electronically deficient pyridine. Quinolines and substituted quinolines obtained from natural sources and microorganisms show remarkable biological activities. Many substituted quinoline derivatives have been synthesized and reported for their biological activities and are known to possess anti-microbial, anti-viral, anti-inflammatory, anti-tumor, anti-cancer, anti-dementia, anti-fungal, hypotensive, anti-HIV, and analgesics properties (Amer et al., 2018;Cocco et al., 2000;Mukherjee et al., 2001;Narender et al., 2005). Keeping in view the importance of quinoline nuclei in medicinal chemistry (Aldahham et al.,2020) and in the search of new anti-COVID agents, we have focused on design and development of some novel quinoline analogues and subjected them to in silico studies against novel coronavirus. The in silico studies were performed on a series of quinoline derivatives synthesized by our research group and out of a library of about one hundred molecules, eight promising molecules were selected for further studies. Furthermore, molecular dynamics (MD) simulations and other in silico pharmacokinetic assessments were also performed on these molecules. It is worth mentioning here that the selected quinoline derivatives (1-8) comprised of amide linkages, similar to well-known Michael acceptor inhibitors N3, B1 and B2, and thus provided a strong basis to investigate as plausible anti-COVID candidates (Aldahham et al., 2020;Gentile et al., 2020;. Design consideration of substituted quinoline derivatives, 1-8, is shown in Figure 1.

Physicochemical description and bioactivity score
Quinoline derivatives have been designed as possible antiviral agents using computational methods, i.e. in silico structure-based approach. Molinspiration and ChemDraw software were used to predict the physicochemical descriptors and pharmaceutically relevant properties of designed compounds. Assessment of properties was done using a thumb rule, i.e. Lipinski's rule of five in order to find out the dug-like characteristics of compounds and compounds violating more than one rule were rejected for further studies (Ghosh et al., 2020a(Ghosh et al., , 2020b(Ghosh et al., , 2020c(Ghosh et al., , 2021Naaz et al., 2018;Srivastava et al., 2018).

Target prediction
Online available software SwissTargetPrediction was used to predict the macromolecular targets for designed molecules. The software is useful to predict off-targets and to evaluate the possibility of reprocessing therapeutically relevant compounds, based on "similarity principle," which normally states that two similar molecules are expected to have similar properties (Anbazhakan et al., 2019).

Molecular docking studies and density functional theory analysis
Selected ligands were optimized and prepared for molecular simulation by using BIOVIA/Discovery Studio 2020 Client (DS 20.1.0.19295 version) protocol default parameters (Lagos et al., 2008;Muegge & Martin, 1999;Singh et al., 2016). The 3D x-ray crystal structure of docking receptor i.e. active site of main protease enzyme (Mpro, 6LU7: PDB ID, www.rcsb.org), the catalytic core of SARS-CoV-2 was retrieved as an adduct from RCSB (www.rcsb.org) protein data bank (Rose et al., 2013). The ligand was prepared accordingly and docking was performed and the top-ranked key docking pose was selected for further analysis. Overall, the standard measures were employed for preparing, docking and scoring of ligands with protein (Srivastava et al., 2020). Predicted Ligand binding site for docking simulations is shown in Si- Figure 1.
The structure based density functional theory (DFT) analysis was also performed for the best screened compound using default parameters available in simulation tool panel of BIOVIA/Discovery Studio 2020 Client (DS 20.1.0.19295 version). Analysis of highest occupied molecular orbital and lowest unoccupied molecular orbital was done in solvent free condition with distance dependent dielectric constant parameter (Hoque et al., 2018;Mishra et al., 2021).

Preparation of receptor
Ligand docked with the target protein was extricated and all missing hydrogen atoms were added employing DS v20.1.0.19295. Positions of each atom was optimized by allatom CHARMm forcefield with Adopted Basis set Newton Raphson (ABNR) minimization algorithm till the root mean square (r.m.s.) gradient for potential energy become <0.05 kcal mol À1 Å À1 . Furthermore, target protein was minimized and defined as receptor employing the "Binding Site" 2.4.3.tool in DS 20.1.0.19295. The input site sphere having 5 Å radius on receptor, covered by ligand has been defined as binding site. The minimized receptor obtained this way from target protein was further employed for docking simulations (Srivastava et al., 2020).

Ligand setup
Built-and-edit unit of DS v20.1.0.19295 was employed for creation of 3D structure of each ligand and CHARMm forcefield using ABNR method was applied for minimization of ligands. Conformational selection of ligands was accomplished by MD approach. Moreover, the ligand was heated up to 700 K and then annealed up to 200 K for thirty times. On completion of thirty cycles, the conformation of ligand was obtained and subjected to local energy minimization employing ABNR method. All minimized conformations were then superimposed and the conformation with the lowest energy was selected as the most plausible conformation.

Docking and scoring
The DS v20.1.0.19295 Ligandfit protocol, combined with a shape comparison filter and a Monte Carlo transformational search, was used for docking of ligand with targeted protein.
Dreiding forcefield was employed to refine each docked pose by rigid body minimization of ligand with respect to the grid based calculated interaction energy. In docking simulations, the targeted protein receptor conformation was kept rigid. The minimization of docked poses was done employing all-atom CHARMm forcefield and smart minimization method until the r.m.s. gradient for potential energy was <0.05 kcal mol À1 Å À1 . While minimization, the binding sites of ligands and protein were concubine flexible employing simulation methods (MD, energy minimization and Monte Carlo simulation). The scoring (Interaction energy, Lig_Internal_Energy, Binding Energy and Dock Score) of ligands are explained in the result and discussion section. All conformations have been refined via LUDI scores. Ligand conformation exhibiting the highest LUDI scores was elected as the best conformation and was taken for further analysis. Negative binding energy of ligands defining the stability of ligand-protein complex, was evaluated by "Calculate Binding Energy protocol" in DS v20.1.0.19295 using the default settings .

Validation of docking protocol
In validation of docking protocols, the native crystallized ligand (remdesivir) and molecules were docked against the active site of Mpro protein receptor. The comparative analysis of docking simulation results alluded that the employed scoring function was appropriate as the root mean square deviation of molecules along with the native crystallized ligand and was under the justifiable limit, i.e., (root-mean square deviation [RMSD]) <2 Å. Therefore, the result supported the hypothesis that experimental binding modes could be reproduced with correctness employing pre-mentioned protocol.

Molecular dynamics simulation
MD simulations were carried out for the target-ligand complexes (protease enzyme-compound 5 and protease enzymeremdesivir complex) for a period of 100 ns using GROMACS 2019.6 package (GROningenMAchine for Chemical Simulations) (Pronk et al., 2013;Purohit, 2014; with CHARMM36 force-field parameters (Huang & MacKerell, 2013). CgenFF web server was used to generate parameters for ligand molecules (Vanommeslaeghe et al., 2010). A unit cell defined as a cubical box, with a minimal distance of 15 Å from the protein surface to the edges of the box, was generated and solvated using the TIP3P water system. The systems were neutralized by adding appropriate number of sodium counter ions to them. Then the systems were energy minimized using steepest descent method to remove the bad contacts among the atoms, with 5000 steps and a force convergence of less than 1000 kcal mol À1 nm À1 . Subsequently, the minimized system was subjected to a twostep equilibration; 100 ps of NVT equilibration to maintain the system temperature at 300 K using the V-rescale thermostat (Bussi et al., 2007) and 100 ps of NPT equilibration to maintain a constant pressure of 1 bar for the system, using Parinello-Rahman barostat (Parrinello & Rahman, 1981). It also maintained the homogeneous density across the systems. The Linear Constraint Solver algorithm was used for constraining all the bonds (Hess et al., 1997). Further, in order to compute the long-range electrostatics forces, Particle Mesh Ewald (PME) method was employed with a cut off value of 1.0 nm (Darden et al., 1993). The MD simulations were run based on leap-frog algorithm (Van Gunsteren & Berendsen, 1988). A production run of 100 ns without any restraints was carried out for each system on NPT ensemble with step size 2 fs. The coordinates were saved every 10 ps during the production run. The resultant MD trajectories were analyzed using XMGRACE 2D plotting tool (Turner, 2005) and VMD 1.9.1 (Schuler et al., 2001).

Binding free energy calculation
The molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA), an efficient and reliable method, was used to investigate residual binding energies in molecular recognition processes (Genheden & Ryde, 2015;Homeyer & Gohlke, 2012;Wang et al., 2018;Sharma et al., 2020;Singh, Bhardwaj, Sharma, et al., 2021). The binding free energy (DG bind ) in a solvent medium was calculated as follows: In Equation (1), G complex is the total free energy of the protein-ligand complex, G protein and G ligand are the total energies of protein and ligand, respectively alone in a solvent. The free energies for each individual G complex , G protein and G ligand were estimated by: Equation (2), p can be protein, ligand, or complex. EMM is the average molecular mechanics potential energy in vacuum and G solv is the solvation free energy. The molecular mechanics potential energy was calculated in the vacuum as follows: Here E bonded is the total bonded interaction energy, like bond, angle, dihedral and improper interactions; E non-bonded is the total non-bonded interaction energy consisting of both van der Waals (E vdw ) and electrostatic (E elec ) interactions. E bonded is taken as zero. The solvation free energy (G solv ) is the sum of electrostatic solvation free energy (G polar ) and nonpolar solvation free energy (G non-polar ), which is determined using the Poisson-Boltzmann (PB) linear equation and the solvent-accessible surface area (SASA), respectively. The binding free energies of the complexes were calculated based on 500 snapshots taken at an equal interval of time during last 10 ns of MD simulation trajectories. The per-residue energy contribution was also computed to understand the contribution of individual amino acids to the total binding energy.

Chemistry
All chemicals were bought from Sigma Aldrich Chemical Company, USA and E. Merck India Ltd, India. All reactions were carried out in oven-dried apparatus and solvents used were dried and distilled. Column chromatography was carried out on silica gel (100-200 mess). Reactions were monitored on TLC, using silica gel 60F254 aluminium plates and visualized under ultraviolet light at 254 nm. Melting points recorded on electro thermal apparatus are uncorrected. NMR spectra were recorded on BRUKER-AV400 spectrometer (Bruker Co., Faellanden, Switzerland) in DMSO-d 6 ( 1 H at 400 MHz and 13 C at 100 MHz). Chemical shifts (d) are expressed in parts per million (ppm) and J (coupling constant) values in hertz. Multiplicities are indicated as s (singlet), d (doublet), dd (doublet of doublets), t (triplet), q (quartet), m (multiplet) and br (broad spectrum). Mass spectra were recorded on Micromass Q-Tof (ESI-HRMS). The elemental analyses were performed on a Perkin-Elmer 240-C equipment.

Analysis of physicochemical properties of quinoline derivatives
The physicochemical data of all compounds were calculated online using Molinspiration, SwissADME and ChemDraw software. Physicochemical properties of compounds have been calculated on the basis of Lipinski's rule of five. Accordingly, the molecular weight of the compounds should be 500, hydrogen bond acceptor should be 10, hydrogen bond donor should be 5, log P value (lipophilicity) should be 5, and total polar surface area (TPSA) should be 140. Lower values of lipophilicity and TPSA referred to the good cell internalization of all compounds (except compound 3) like the reference drug remdesivir. The compounds having more than one violation were rejected for further studies because of difficulty in solubility and bioavailability. All physicochemical data of quinoline derivatives 1-8 are summarized in Table 1 and represented as radar graph in Si- Figure 2.
The bioactivity scores, viz. G-protein coupled receptor (GPCR) ligand, ion channel modulator, kinase inhibitor and nuclear receptor ligand were also evaluated to analyse some more drug-like properties. The compounds having bioactivity score more than zero are supposed to be pharmaceutically active, whereas the bioactivity score of compounds between 0.00 and À0.50, shows moderate activity and the score less than À0.50, shows inactivity. On the inspection of results, it was found that almost all compounds 1-8 were found within the permissible range of bioactivity score and thus showed good to moderate drug-like properties. Results are summarized in Table 2.

In silico ADMET prediction of quinoline derivatives (1-8)
ADMET properties, such as log Kp, log S, synthetic accessibility score and absorption (% ABS), blood brain barrier permeability (log BB), metabolism, excretion and toxicity parameters of all quinoline derivatives (1-8) were calculated by SwissADME online available software and are shown in Tables 3 and 4. The predicted intestinal absorption was more than 76% for all compounds and thus the compounds qualified the drug-like criteria of !70% absorption and hence good intestinal absorption was expected. Water solubility expressed as the decimal logarithms of the molar solubility, i.e. log S (mol/L) indicated the good drug bioavailability as the compounds showed water solubility value <(À4) (Ottaviani et al., 2010;Ritchie et al., 2013). The results thus demonstrated that all compounds were in the range of high to moderate solubility. Skin permeability, represented as log Kp (cm/s or cm/ h), plays a significant role in transdermal delivery of drugs and the molecule having higher À ve value of log Kp have lesser skin permeability. All molecules showed moderately permeability as value ranged between À2.74 and À2.80 (Potts & Guy, 1992;Singh & Singh, 1993). Compounds 3 and 8 were expected to cross blood brain barrier easily as their log BB values were higher than (À1).
Bioavailability of drug is also influenced by metabolism and cytochrome CYP450 enzymes are the most important class to study this effect. On analysis, most of the compounds were found to be inhibitors of CYP1A2, CYP2C19 and CYP2C9 and substrates for CYP3A4. All ADMET properties such as absorption, metabolism, distribution, excretion and toxicity were found to be favourable for all compounds. Results are summarized in Tables 3 and 4. Brain access and gastrointestinal absorption are two main drug-like behaviours critical to estimate at various stages of drug discovery processes. BOILED-Egg model presented in Si- Figure 3 (Daina & Zoete, 2016) revealed that compounds 7 and 8 lying in yellow zone could permeate through the blood-brain barrier (BBB). Gastrointestinal tract can easily absorb compounds present in white area.

Swiss target prediction
Swiss target predication of the quinoline derivatives 2-8 and the reference drugs are shown in Si- Figure 4.

Docking analysis of quinoline derivatives 1-8
Docking is the computational prediction of structures of ligand-protein complexes within a targeted binding site of protein receptor. To understand the binding mode of quinoline derivatives, 1-8, molecular docking was performed using DS (20.1.0.19295 version) software. Various anti-HIV drugs, viz. remdesivir, darunavir, lopinavir, atazanavir, etc. also displayed great binding affinity toward the active site of Mpro (Beck et al., 2020). Amongst them, remdesivir, darunavir and X77 ( Figure 2) have been employed recently by a number of researchers as standard substrates for comparative analysis of binding affinity vis-a-vis binding mode of several small molecules within active site of Mpro (Bhardwaj & Purohit, 2020;Ghosh et al., 2020b). Thus, the great binding affinity and other simulation results triggered us to subject these three anti-HIV drugs as standard Mpro inhibitors for further comparative analysis.
H-bonds, p-p (hydrophobic), and non-polar p-cation (noncovalent) interactions were used to explain the docking results. On inspections, results indicated that all compounds commonly formed H-bonds with five different amino acids such as His164, Glu166, Gln189, Tyr54 and Asp187 and p-bonds with His41 at the catalytic site of the PI, like reference drugs.
The existence of these hydrophilic and non-covalent interactions implied that the interatomic distances were within the range of 5.21 Å. The docking results of quinoline derivatives 1-8 with PIs are showed in Table 5.
Docking studies revealed that compound 1 having nitro group at para position of the sulphonyl group formed one hydrogen bond with amino acid His 164 at a distance of 2.68 Å, and compound 2 having chloro group at the para position to the sulphonyl group formed two hydrogen bonds with amino acids His164 and Glu166 at a distance of 3.18 and 2.82 Å, respectively.
Compound 3 having isopropyl group at 1, 2, and 3 positions of the aryl group formed two hydrogen bonds with amino acids His164 and Glu166 at a distance of 3.08 and 2.83 Å, respectively and two p-p bonds with amino acid  day were labelled as weak chronic toxicity and chemicals with LOAEL ranged from 10 to 50 mg per kg per day were labelled as medium chronic toxicity. T. Pyriformis toxicity ¼ Tetrahymena Pyriformis toxicity, Minnow toxicity ¼ Acute fathead minnow toxicity is basis of hazard and risk assessment for compounds in the aquatic environment. Structure-minnow toxicity relationship as follows: log LC 50 ¼ À0.94 log p þ 0.94 log (0.000068 p þ 1) À1.25 where p is the n-octanol/ water partition coefficient.    His41 at a distance from 3.81 to 4.69 Å. Compound 4 having trifluoro methyl group at the para position of the sulfonyl group formed three hydrogen bonds with the amino acid His164 at a distance of 2.72 to 3.01 Å, and two p-p bonds with the amino acid His 41 at a distance of 4.01-5.21 Å.
Compound 5 having methyl group at the para position of the sulphonyl group formed two hydrogen bonds with the amino acids Glu166 and Gln189 at a distance of 2.85-2.90 Å, and Compound 6 bearing thiazole moiety exhibited three hydrogen bonds with Tyr54 at a distance of 2.59 Å, His164 at 3.15 Å and Asp187 at 2.87 Å.
Compound 7 having benzoyl group formed one hydrogen bond with amino acid His41 at a distance of 2.72 Å and one p-p bond with amino acid His41 at a distance of 4.13 Å, and Compound 8 having propargyl group also formed two hydrogen bonds with amino acid Tyr54 at a distance of 2.68 Å to 2.96 Å.
Docking interactions of quinoline derivatives and reference drugs are shown in Figure 2. The binding of all compounds with protease enzyme confirmed their behaviour as PIs.
Analysis of result clearly revealed that all compounds exhibited robust interaction with amino acid residues of catalytic core of target enzyme hence all of them may behave as plausible lead candidates against SARS-CoV-2 and amongst all, compound 5 showed the highest extent of stability and safety profile.
Scoring descriptors, viz. Lib dock score (Lib DS), PLP1, PLP2, PMF, PMF04, DG and EC 50 values were calculated for all compounds to analyse the stability of protein-ligand  Lib-Dock score (Lib-DS), another factor generated from docking simulation again supported the interaction of the designed ligands with the Mpro protein in a better way. An extensive study revealed that the Lib-DS values for all molecules (ranging between 62.41 and 138.32) were found to be comparable to the reference drug remdesivir (Lib DS: 100.33), darunavir (136.25) and X77 (93.80). Ludi2 and Ludi3 are empirical scoring functions derived from the Ludi algorithm and used to select the exact conformations of proteinligand complexes. Free binding energy (DG) predicted via Ludi2 for all molecules (ranging between À6.41 and À4.48) was found in close proximity to the remdesivir (À6.00), darunavir (À6.21) and X77 (À7.80). EC 50 values, a measure of bearable toxicity and good pharmacological activity, were also evaluated using Ludi3 scoring function ranges between 4.27 Â 10 À6 and 7.94 Â 10 À8 , and concluded that the values were found to be comparable and even better in few instances. It can be inferred from the studies that higher DG (Àve) values indicated towards stable protein ligand complexes, which, in turn, supported lower EC 50 values. Results obtained here, clearly revealed that compounds 1-8 had either exhibited similar mode of binding to the Mpro catalytic site or had an excellent binding affinity with Mpro receptor protein, compared to the references employed in docking analysis. Results are presented in (Table 6).

Density functional theory analysis
DFT analysis has been done using the discovery studio v20.1.0.19295 software. DFT analyses of compound 5 and the reference drug remdesivir are shown in Figure 3. DFT analysis revealed that the band gap energy between the highest occupied molecular orbital and lowest unoccupied molecular orbital of compound 5 was less than that of remdesivir, which proved that the molecular charge transfer interaction is more feasible and prominent in compound 5 than remdesivir. Hence compound 5 might showed better binding affinity than remdesivir with target enzyme, which was required for good biological activities (Acar et al., 2017;Sulpizi et al., 2002;Tandon et al., 2019).

Molecular dynamics trajectory analysis
MD simulation presents an appropriate way to study atomic level information about binding of ligands to target molecules (proteins/DNA/RNA) (Hollingsworth & Dror, 2018;Kumari et al., 2021). On the basis of MD trajectories, the RMSD, root-mean square fluctuation (RMSF), radius of gyration (Rg), number of hydrogen bonds and binding free energy of the complexes were computed to analyse and get insight into their structural stabilities, binding modes and binding strengths of the designed molecules.

Conformational stability
The RMSD plots against simulation time, shown in Figure 4, elucidated smaller fluctuations for both reference complex remdesivir (black) and compound 5 (red) after 65 ns of the trajectory indicating attainment of a stable conformation. The average RMSD of the reference complex (0.33 nm) was relatively higher than that for compound 5 complex (0.28 nm). This observation suggested that compound 5 complex was relatively more stable than the reference complex.

Residue flexibility analysis
To evaluate the flexibility of the complexes, fluctuations of each amino acid residue presented by the RMSF plots as a   [compound 5 (red) and remdesivir (black)] over MD production run (b) Hydrogen bond numbers between Glu166, Gln189 residues and ligands [compound 5 (red) and remdesivir (black)] over MD production run. function of residue number were plotted ( Figure 5). High RMSF value (peaks) indicated the presence of loops, turns, terminal ends and loose bonding showing the flexibility of the protein structure while smaller value indicated the presence of secondary structures such as sheets and helices which rendered stability to structure. The graph illustrated that the complex with compound 5 showed high fluctuations than the complex with remdesivir. The catalytic pocket residues showed relatively more fluctuations than other residues (besides the last 100 residues at C terminus) showing a flexible catalytic pocket for compound 5 complex. According to the docking results, the residues Glu 166 and Gln 189 were important and these residues also showed a higher RMSF for compound 5 complex compared to reference complex, indicating higher flexibility and lower stability of compound 5 complex relative to reference complex. However, both complexes have similar average RMSF values (average RMSF for compound 5 complex is 0.14 nm and average RMSF for remdesivir complex is 0.11 nm) demonstrating the same inhibitory action and a balance between stability and flexibility, necessary for the activity.

Compactness analysis
The radius of gyration (Rg) accounts for the compactness or globular structures of the protein-ligand complexes. Generally, a higher Rg value indicates an expanded or open structural conformation of protein whereas a lower value indicates more compact structure. The plots of Rg versus time ( Figure 6) were quite similar for both complexes, having the mean values at 2.24 nm and 2.25 nm, for compound 5 (red) and remdesivir (black), respectively, with protease. Smaller and stable fluctuations in Rg value of compound 5 reiterated the previous finding that the complex was stable and compact resulting in stronger interaction between protease and ligand.  (Figure 7).

Analysis of hydrogen bonding
Hydrogen bond formation between the ligand and target shows the binding stability of a target-inhibitor complex. The simulation results showed a large number of intermolecular hydrogen bonds formed between protein residues and both the ligands (remdesivir and compound 5) indicating a strong interaction (Figure 8a). Figure 8a gives the number of hydrogen bonds formed during the simulations with respect to time. The number of hydrogen bonds for remdesivir complex (average hydrogen bond number 1.65) is higher compared to compound 5 complex (average hydrogen bond number 0.27) as remdesivir is a bulkier molecule (77 atoms) than compound 5 (39 atoms). However, the hydrogen bonds were intact during the entire dynamics for compound 5 complex, indicating stronger binding between compound 5 and protease and leading to a very stable conformation. The docking studies showed that both remdesivir and compound 5 ligands formed two hydrogen bonds with protease residues Glu166 and Gln189. The hydrogen bond analysis of MD simulation also illustrated the bond formation between Glu166, Gln189 residues and both ligands (Figure 8b). The number of hydrogen bonds formed during the simulations is more in case of compound 5 (red) than remdesivir (black). From VMD analysis it was observed that for the last 35 ns, the bond between Glu166 and remdesivir is more stable while for compound 5-protease complex the bond between Gln189 and compound 5 is more stable. This observation illustrates that residue Gln166 is stable Figure 9. Energy contributions of individual amino acid residues to the binding free energy shown for Compound 5-protease complex (red) and remdesivir-protease complex.  in remdesivir-protease complex and residue Gln189 is stable for compound 5-protease complex.
3.6.6. Binding free energy and residue interaction energy  Table 6 for both complexes. The compound 5-protease complex showed lower negative binding energy (À64.30 kJ mol À1 or À15.37 kcal mol À1 ) than remdesivir-protease (À70.62 kJ mol À1 or À16.88 kcal mol À1 ) . The binding energy results showed that the reference complex exhibited more stability than compound 5-protease complex. The difference in the values of binding energies was also reflected in the docking binding energies studies. Decomposition into separate energy terms revealed that the polar solvation energy decreases the binding strength of inhibitors to the protease significantly, and thereby reduced the total binding energy in both complexes due to the positive energy contributions (Table 7). Among the various interactions, van der Waal energy (DE vdW ) showed the most favourable contributions towards the negative binding free energy of both complexes. In addition to the overall energies of the complex (Table  7), the contributions of individual amino acids to the binding free energy (DG bind ) of both complexes were also computed using the MM/PBSA method and presented in Figure 9. The per-residue interaction energy profiles revealed that compound 5-protease complex differed from reference complex at residue position 49, 61, 165, 167, 168 and 187, indicating that Met49, Met165, Lue167, Pro168 and Asp187 actively participated in interaction to give rise to a stronger binding and stability. There were also energy values that contributed negatively to the stability of compound 5 complex like Arg40, Glu47, Asp48, Tyr54, Arg105 and Glu166. The most unstable residue was Arg40 showing a high positive value þ14.19 kJ/mol. The most favourably contributing residue was Met49, having a binding energy of À8.41 kJ/mol, while that for reference complex was also Met49 with À6.32 kJ/mol binding energy.

Structural analysis during MD simulations
To analyse the conformational changes in compound 5 and remdesivir in complexed form with Mpro, the structural snapshots at the interval of every 10 ns for the entire 100 ns MD simulation were taken. In case of compound 5: Mpro complex exhibited more fluctuation in first 40 ns and later on due to more compactness in the binding pocket of receptor enzyme (reflected by radius of gyration and SASA analyses) the complex displayed less conformational changes and high stability from 70 ns to 100 ns, as presented in Figure 10.
Similarly, the snapshots were also extracted for Remdesivir: Mpro complex. It was observed that the large size of remdesivir facilitated its additional interactions with main protease atoms/residues as compared to compound 5. Here, remdesivir demonstrated conformational changes only for initial 30 ns and further remains almost constant for entire 100 ns period, also supported by the RMSD assessment. Thus the remdesivir: Mpro complex was also found stable for later 70 ns, as presented in Figure 11.
Overall MD analyses manifested that compound 5-PI complex showed properties comparable to remdesivir-PI complex and both were quite stable complexes.

Chemistry
The synthetic strategy adopted for synthesis of 8-amino-4methyl-1H-quinoline-2-one from ethyl acetoacetate is shown in Scheme 1. The quinoline derivatives, 1-8, have been synthesized by the reaction of 8-amino-4-methyl-1H-quinoline-2-one with different types of sulphonyl chloride/benzoyl chloride/propargylic bromide as outlined in Scheme 2. Purification and characterization data of these compounds have been provided in supporting information.

Conclusion
A new series of quinoline derivatives 1-8 have been developed as effective PIs of SARS-CoV-2 using in silico structurebased approach and synthesized via multiple steps. From docking analysis, it was revealed that some of the designed compounds showed high effectiveness and even better results as compared to the reference drugs. Docking results also indicated the good relationship between the predicted EC 50 , various energy terms and descriptors for all newly reported compounds against SARS-CoV-2. Subsequently, MD simulations positively supported the result of docking analysis by exhibiting comparable results of compound 5 and reference drug remdesivir in terms of radius of gyration, RMSD, RMSF, SASA and binding free energy, that established a robust relation between docking and MD results. In conclusive remark, a comparable stability as well as binding ability of compound 5 and reference drug remdesivir has been found against Mpro of SARS-CoV-2. Therefore, the present study provided a valuable advancement in the development of novel PIs and can be considered as a starting point for lead optimization.