Identication of Multi-Target Directed Inhibitors of SARS-CoV-2: Drug Repurposing

Background: COVID-19 remains a major global challenge while several drugs identied to ameliorate the syndromes are associated with incessant resistance and lack of prospective potentials to permanently curb the infection. This study is aimed at evaluating the potentials of the existing drug molecules to overcome the viral resistance through multi-target inhibition mechanisms. Methods: Molecular docking and molecular dynamics simulations have been extensively applied to virtually screen 2826 drugs from Selleckem.com library against some key bio-receptors implicated in the SARS-CoV-2 such as the viral nucleocapsid phosphoprotein, the viral spike glycoprotein and the human host ACE2. Results: Five drugs namely D-(+)-Ranose pentahydrate (1), (-)-Epicatechin gallate (5), 2797 (7), Rutin DAB10 (8) and Hyperoside (9) display higher inhibitory potentials against N-terminal NTD of SARS-CoV-2 (PDB 6M3M) with XP docking scores of -16.20, -11.98, -11.83, -11.81 and -11.41 than remdesivir and ribavirin with -10.27 and -9.06 respectively. Their estimated binding free energies against the receptor are -27.80, -27.91, -32.39, -27.91 and -29.50 kcal/mol compared to remdesivir and ribavirin with -24.27 and -15.37 kcal/mol respectively. Similar inhibition patterns were observed against the viral S-protein and the human ACE2 with high stability and bio-functionality. Conclusion: The identied compounds show promising potentials amenable for breakthrough against the drug-resistant COVID-19 upon further studies.


Introduction
Coronavirus disease 2019 (COVID-19) is a new viral infection of Severe acute respiratory syndrome coronavirus (SARS-CoV-2) which emerged in a suburb city of China towards the end of year 2019 and has spread across countries in all continents of the world. It was later termed COVID-19 pandemic disease by the World Health Organisation (WHO) [1], the situation of which threatens the global health, economy and population severely through daily loss of lives, leading to its further declaration as public health emergency of international concern by WHO in January, 2020 [1,2]. The death rate of COVID-19 as of 27 th August, 2020 stands at 5.0% with 24,331,524 of the global population con rmed infected out of which 16,872,542 patients have recovered and 829,664 [3][4][5]. The infants and aged population with weak immune systems as well as individuals suffering from co-existing complications of other ailments such as cardiovascular diseases, myocardial injury, hypertension, in uenza, diabetes, pneumonia etc are more vulnerable [6][7][8][9]. In fact, it's recently reported that not less than 20% and two-thirds mortality in China and Italy respectively had diabetes complication [10]. The age-dependent symptoms of the disease vary from mild such as fever, fatigue, nostril blockage, dry cough, u-like to Acute respiratory distress syndrome (ARDS), Acute lung injury (ALI), pneumonia, cardiac arrest, dyspnea, liver and kidney damage and death, although asymptomatic cases have been frequently reported [8,11,12].

Structural implications for infection and therapeutic targets
Page 3/31 The main pathology still remains elusive [12]. According to virologists, SARS-CoV-2 is a β-CoV family member of coronaviridae of the order Nidovrales, similar to SARS-CoV-1 and Middle East respiratory syndrome coronavirus (MERS-CoV) [13,14], with an enveloped single stranded positive ribose nucleic acid (RNA) and crown-like density of proteinous membrane (corona) for binding to targets [2]. The structure of SARS-CoV-2 consists of envelope (E) and membrane (M) proteins responsible for structural fusion, spike (S) glycoprotein responsible for cell-interaction with targets and nucleocapsid (N) protein, a viral suppressor protein which resists host defence mechanism and interacts the viral RNA to E, M and S proteins in genome encapsulation (of around 30 000 nucleotides), and the RNA dependent RNA nucleocapsid, a non-structural protein for the viral transcription, virion formation and RNA replication [2,13,15]. The highly conserved N-nucleocapsid proteins consist of three main domains, the N-terminal domain (NTD), the central linker rich in serine/arginine and the C-terminal domain (CTD), suggesting them as targets for regulating pathogenic activities [16]. Upon gaining entrance into the host respiratory track, SARS-CoV-2 infuses the host cell in similar way to SARS-CoV-1 and MERS-CoV via the membranebound angiotensin-converting enzyme (ACE2) and employs the transmembrane protease serine 2 (TMPRSS2) for S protein activation, the progression of which the viral fusion and replication occur [2,10,14]. Although, the possibility of interactions between the viral spike protein and some other human cell adhesion factors such as the dipeptidyl peptidase, cycophillins and ezrin cannot be overruled [17]. Also, sex-dependent epidemiology is possibly observed in the current COVID-19 due to the encoding of ACE2 gene expression on X chromosomes [2], supporting the fact that the signal pathway of the human ACE2 is imperative in the control of COVID-19 infection in addition to the viral S-protein and the nucleocapsid NTD [18][19][20]. These receptors, nucleocapsid NTD and spike glycoprotein of SARS-CoV-2 are important therapeutic targets and their inhibitors have been clinically experimented in curtailing the pandemic along other angiotensin receptor blockers (ARBs), although, the chance of ACE2 upregulation and altered angiotensin II expression resulting in endothelial dysfunction still a major concern [2, 21].

Current treatment
The ambiguity surrounding the underlying aetiology of COVID-19 contributes to the scarcity of speci c vaccine till moment [11]. In the scienti c exploration, WHO recommends the megatrial of various strategies including the repurposing of the existing drugs with different mechanisms such as the antibiotics (e.g. amoxicillin and azithromycin), anti-viral (e.g. ribavirin), anti-CoV (e.g. remdesivir), corticosteroids (methylprednisolone) and antimalarial drugs (chloroquine and its hydroxyl-derivative), using the in-depth relationship in genomic information and pathological features of CoVs [11,22]. The potential candidates under investigation are classi ed by inhibition of structural enzymes and proteins responsible for virus-host cells interaction; action on functional proteins and enzymes associated with viral RNA mutation and replication and restoration of human innate immunity (immunomodulation) [11,23].
The clinical experimental results so far reveal the inhibition of the S-protein and ACE2 by arbidol, the viral nucleocapsid NTD by remdesivir and ribavirin and ACE2 glycosylation by chloroquine analogues [24,25].
Other conventional medication known for mitigation of ARDS, in uenza, ALI, in ammation, hypertension, immunomodulation have also been evaluated against the pandemic, including losartan, statins and ARBs [2,26]. Improved immune responses in SARS-CoV and MERS-CoV infected patients have also been investigated upon administration of chemokine and cytokine modulators such as tocilizumab and the human mesenchymal cells [27][28][29]. However, recurrence of the symptoms due to resistance by the virus, accessibility to all patients and unpleasant side effects are some major challenges with the few drugs in clinical trials of COVID-19 treatment, necessitating the search for new therapeutic candidates with improved pharmacological features and accessibility [21]. Whilst the discovery of novel molecules through the established structural information of the virus seems more promising for a long-term effect but could be costly and untimely, the repositioning of the existing FDA approved drugs is not only cheaper but more appropriate for the situational public-health emergency, even their extended application is very possible. From previous studies, the herbal respiratory detox shot and some other traditional Chinese medicines with known antiviral history were experimented in early prevention and control of COVID-19 through inhibition of SARS-CoV-2's 3CL pro , PL pro , spike protein etc by theoretical binding and in vitro study [1,30,31]. Melatonin being a protective agent against some pathogenesis associated with viral infections and ARDS has also been speculated as an adjuvant in the treatment of some chronic ailments induced by COVID-19 including ARDS, ALI and pneumonia [12], supporting the therapeutic potentials of natural products against the pandemic. Since the similarities in pathophysiologies of the existing and emerging CoV strains support the hypothesis of similar pathomechanisms of infection, similar approaches could be rationally employed for their prevention and treatment [14]. The existing drugs with multi-mechanistic inhibitory potentials from abundant natural resources are yet to be extensively explored, as such become viable for this scienti c quest.
More so, computational and bioinformatic tools have been extensively applied to understand the cellular factors such as the genome, mutagenesis and amino acid composition of the new Wuhan Seafood market SARS-CoV-2 isolate and predict its transmission and effective therapeutic targets for drug development [32]. Among them, molecular docking and molecular dynamic simulations are important techniques in screening protocols and approximated mimicry of molecular interactions within biocellular environments. Through these applications, some previous studies resulted in the identi cation of some COVID-19 RdRp and ACE2 potent plant-based inhibitors including hesperidin, sofosbuvir and thea avin [11,33,34], however there exist limited reports on multi-target directed inhibitors of SARS-CoV-2 among the existing FDA-approved drugs. Interestingly, the idea of "one-drug-multiple-targets" are reliable therapeutic strategies for tackling drug resistance and had been successfully employed in previous studies [35][36][37][38].
Therefore, this study is aimed at identifying multi-target directed inhibitors of SARS-CoV-2 among the existing drugs through molecular docking and molecular dynamics simulations. The computational tools will be employed to screen 2826 molecules from the drug repurposing library of Selleckchem.com Bioactive compounds expert (https://www.selleckchem.com/screening/drug-repurposing-library.html) against some selected receptor targets implicated in COVID-19 infection and treatment. The crystal structures of the viral nucleocapsid RNA-binding phosphoprotein for RNA mutation and replication of SARS-CoV-2 (PDB 6M3M), the densely glycosylated S protein for infusion into the human cells (PDB 6VSB) and the human ACE2 for viral reception (PDB 6M17) are carefully selected from RCSB PDB as virtual therapeutic targets for preventing viral translation, mutation, replication and infusion [18,19,39].
Theoretical inhibitory potentials against the targets, thermodynamic stability and biological applicability were assessed through molecular docking XP score and molecular dynamic trajectories according to some reported protocols [40][41][42][43].

Ligand preparation
The structure data le (SDF) formats of the 2826 drug molecules retrieved from drug repurposing library were imported into the workspace and prepared by Epik energy minimization, followed by the addition of hydrogens, ionization state at pH 7.0 and conversion of 2D to 3D structures. The geometry and partial atomic charges were calculated using Optimized potentials for liquid simulations (OPLS-3e) force eld (Schrodinger, 2019-Version LigPrep) [44], the prepared ligands were saved in LigPrep output le.

Protein preparation
Each PDB le of the selected x-ray structures of the protein targets was pre-processed, optimized by assigning charges and protonation states, re ned by deleting water and covalently linked molecules, and moderating the bond order, and regenerating het states using the default Epik pH 7 ± 2. Energy minimization was simulated using OPLS-3e force eld in Maestro 12.2 (Schrodinger, 2019-Version Protein Preparation Wizard) [44] followed by removal of non-focused chains.

Analysis of receptor active cavity and grid box generation
One of the main steps involved in a successful docking simulation is the identi cation of the active binding site of the receptor target. The active binding sites of the selected receptors were analysed using Biologics site map and sequence alignment modules available in Maestro 12.2 Schrodinger, 2019-version [45], information from the previously reported procedures [18] and available data in the RCSB PDB (https://www.rcsb.org/). The recently released crystal structures of SARS-CoV-2 nucleocapsid phosphoprotein consisting of 499 residues within active chains A, B, C & D, X-ray diffraction resolution of 2.7Å and free R-value of 0.293 (PDB 6M3M), the densely glycosylated S protein consisting of 2904 residues within active chains A, B & C, with electron microscopy resolution of 3.46Å (PDB 6VSB) and the RBD/ACE2-B0AT1 protein complex consisting of 3072 residues within chains A-D with an electron microscopy resolution of 2.90Å were retrieved from the RCSB and submitted to the workspace of Maestro suites one at a time. Each crystal structure was aligned with similarly resolved structures of other CoV strains for binding site comparison using protocols [14,18] and the Biologics site map function in Maestro [45] predicted ve active binding sites (hot-spots) for each with the contributions from the pocket volume, enclosure and hydrophilicity. The binding (druggable) sites were assessed basically with site score and Dscore analysis (eqn. 1 and 2) [46] embedded within the algorithms of Maestro.
Site score = 0.0733n ½ + 0.6688e -0.20p eqn. (1) Dscore = 0.094n ½ + 0.60e -0.324p eqn. (2) Where n = number of site points; e = enclosure; p = hydrophilic score The site with highest DScore and consistent con gurations with similar reported structures were selected for grid box generation according to the protocols [16,46]. The active residues within the selected active sites as well as the X, Y and Z coordinates were used to con gure the receptor binding grid centroid for docking simulation at 20 Å ligand length. Other parameters are kept at default. In order to validate the predicted binding sites of each receptor and establish the possibility of similar inhibition mechanisms by the drug conformers across different coronavirus strains, the residues of the N-terminal NTD of SARS-CoV-2 nucleocapsid receptor (6M3M) was aligned in sequence with residues of SARS-CoV-1 (PDB 1SSK) and MERS-CoV (PDB 4UD1) of similar targets, while the spike glycoprotein of SARS-CoV-2 (PDB 6VSB) was aligned with the crystal structures representing the same target in SARS-CoV-1 (PDB 6CRX) and MERS-CoV (5W9H).

Molecular docking
Molecular docking simulations were carried out to estimate the inhibition potential of each minimized conformers of the exible drug molecules via interactions with the active residues within the generated grid box of each protein target. The results were ranked in increasing order of glide extra precision (XP) scores and binding poses were used to analyse the bonded and non-bonded interactions of the drug molecules with protein residues. The higher the negative value, the stronger the binding.

Inhibition of SARS-CoV-2 RNA mutation and replication
The prepared conformers of the drugs were virtually screened through algorithms of molecular binding interaction and inhibition potentials against the N protein, NTD of SARS-CoV-2 according to the protocol of [40] incorporating remdesivir and ribavirin as references. The ligands were docked to the crystal structure PDB: 6M3M, within the grid box of the binding site-2 with x, y and z coordinates as 8.08, -7.11 and -30.57 respectively and a default grid space of 1 Å.

Inhibition of the viral S-protein
The potentials of the drug conformers to inhibit the viral S-protein priming to the human ACE2 were evaluated against the crystal structure of pre-fusion spike glycoprotein of SARS-CoV-2 using docking protocols [41], including remdesivir and ribavirin as references. The single RBD, PDB: 6VSB, within the binding site coordinates x, y and z as 212.56, 211.71 and 224.23 respectively represents the target.

Blocking of the human ACE2
The abilities of the drugs to block the human ACE2 (host cell) were theoretically evaluated using ligandreceptor docking protocols [42], incorporating remdesivir and ribavirin as references. The crystal structure PDB: 6M17, RBD/ACE2-B0AT1 protein complex within the binding site coordinates x, y and z as 141.53, 141.53 and 202.44 respectively represents the target. The evaluation was based on binding energy in terms of glide XP score, glide energy pro les and bonding and non-bonding interactions are shown by the binding poses. The cumulative results from the docking simulations were used to screen the drug molecules in order to retrieve those with optimum potentials. Since the theoretical biostability and biofunctionalities of a ligand-receptor system cannot be fully established by molecular docking, molecular dynamic simulations were employed for further studies of the selected drugs.

Molecular dynamics simulations
In order to estimate a more accurate approximations of real biological system mimicry, ligand-receptor complexes of the nine drugs with cumulative highest glide XP gscores and good binding poses were selected for molecular dynamics simulations. The binding a nities, biostabilities, exibility trajectories and potential systemic functionalities were evaluated in comparison with drugs currently in clinical trial of the COVID-19 treatment, remdesivir and ribavirin. These were achieved through molecular mechanics (MM) force eld in Ligand and Receptor Molecular Dynamics server integrated with AMBER 16 force eld, CAVER 3.0, Charts.js, ChemAxon, JSmol, MDTraj, MolScript and R (http://chemyang.ccnu.edu.cn/ccb/server/LARMD) [47]. Each ligand-receptor complex was submitted for 'Int-mod' simulation, setting the all-atom MD time at 3000 ps, water explicit and other parameters as default. The average period for each MD simulation ranges between 10 -25 hours for each drug-receptor system after which the results were retrieved. The average distances between the atoms of the docked ligands as deviated from their original positions were represented by RMSD plots while the radius of gyration plots reveal the root mean square distance between each atom of the protein to their control. The conformational dynamics of each molecular complex system and energetics of the binding ligands were explored by capturing the transition states of the protein receptor with a folding free energy barrier represented by the fractions of native contacts. The uctuations and isotropic displacements that Where GE = total gas-phase energy; GBS = GB non-polar contribution to the solvation-accessible surface area; TS = TΔS the entropy contribution by the inhibitor; EE = Electrostatic energy; VW = van der Waals contribution; GBT = GB total; ΔG bind = nal binding free energy MM/GB(SA) The drug molecules with virtually higher inhibitory potentials, biostability and biofunctionalities than the reference drugs, remdesivir and ribavirin were identi ed and subjected to some biochemical antiviral activity search.

Antiviral activities of the selected drugs
The selected ve (5) drugs from molecular dynamics simulation were traced to some documented biochemical data to nd correlation between the theoretical ndings in this study and experimental inhibitory actions where possible. Interestingly, all the ve have been documented with some pre-clinical and clinical experimental assays in vitro and in vivo against some virulent viral strains with some similarity in genomic encodings. These include the assay to study inactivation, occulation and removal of Sindbis virus heat resistant (SVHR) strain in osmolyte solution of drug 1 [50], the inhibition assay of SARS-CoV-1 using quantum dots-conjugated oligonucleotide system for drug 5 [51], the clinical trial on patients infected with hepatitis delta virus (HDV) by administration of 200/400 mg in two doses of drug 7 for 28 days [52], the in vitro inhibitory assay on vero cells infected with infectious bronchitis virus (IBV) and the in vitro inhibitory assay on murine CoV for drug 8 [51,53], and the in vitro inhibitory assay on human hepatoma Hep G2.2.15 cells, HBeAg and HBsAg, and in vivo inhibitory assay on DHBV-DNAinfected duckling model at doses 0.05 g/Kg/day and 0.10 g/Kg/day for drug 9 [54].
3.0 Identi cation of the selected drugs to their current applications and natural sources The current applications of the selected drugs as well as their natural product sources were identi ed using Drug bank and natural plant database (Table 6).
Results And Discussion

Ligand and protein preparation
A total number of 6834 minimised conformers of the drug molecules were regenerated in a 'ready-todock' ligprep.out le, while the minimized receptors with binding pocket con gurations were prepared in glide_grid.zip les for docking.

Analysis of receptor active cavity and grid box generation
In line with the set parameters in Maestro, ve active sites (hot-spots for ligand binding) were predicted as sites 1-5 with site score, size, Dscore and volume for each receptor in decreasing order of site score ( Table  1). The site with the highest predicted Dscore was selected as main target of each receptor. Therefore, site 2, site 1 and site 3 were selected for PDBs 6M3M, 6VSB and 6M17 respectively. The x, y and z grid box coordinated of selected site of each receptor target as well as the potentially active residues for interaction with the drug conformers are presented ( Table 2). The site with the highest Dscore in the crystal structure of the N-terminal NTD nucleocapsid phosphoprotein of SARS-COV-2 is site 2 (blue and red cartoon region, Fig. 1A) with the active residue backbones (green labels, Fig. 1B) identi able within the unique drug targeting site of the NTD terminal (residue numbers 46-174) of the aligned structures of the nucleocapsid proteins of the three CoV strains (Fig. 1C). The residues of SARS-CoV-2 receptors smartly aligned in sequence with the residues of SARS-CoV-1 and MERS-CoV of the same NTD target. Furthermore, the N-terminal NTD SARS-CoV-2 (PDB 6M3M) has 100% homology identity with the resolved structure (PDB 6VYO) (supplementary Fig. 1) whose inhibitors have been demonstrated to be potent against similar targets of other CoV strains. Similarly, the active site of the crystal structure of the viral S glycoprotein of SARS-CoV-2 (PDB 6VSB) (red and blue cartoon, Fig. 1D) reveals the proposed active residues (green labels, Fig. 1E), whose residual sequence aligned smoothly with the resolved active sites for the same target in MERS-CoV (PDB 5W9H) and SARS-CoV-1 (PDB 6CRX) (Fig. 1F). These further validate the possibility of similar inhibition mechanisms by the drug conformers across different coronavirus strains as previously demonstrated [55,56], and also support the accuracy in the con guration of the binding sites of the receptor targets. The active RBD site as a target for blocking the host enzyme, ACE2 during infection is mapped (blue and red cartoon, Fig. 1G) with the most probable residues for interaction (green labels, Fig. 1H) traceable to the predictions in Table 2. The active binding pocket grid box of each minimized receptor was generated (in glide grid zip le) for ready-to-dock conformation.

Molecular docking
The glide XP glide scores, glide energy and binding poses of the nine most interacted drugs across the targets, N-terminal of the RBD of SARS-CoV-2 (PDB 6M3M), S-protein of SARS-CoV-2 (PDB 6VSB) and human ACE2 enzyme (PDB 6M17) are presented presented in Table 3-5. The most strongly interacted drug molecules are buried deep within the binding pocket cavity of each receptor shown by surface structures (Fig. 2). The binding poses to show the bonded and non-bonded residues with the drug molecules within the active site of each receptor as well as the nature of interactions are displayed in Fig.

3-5.
For convenience, the drugs are annotated with numeric identities as given below:  Considering the inhibition potentials against the viral N-terminal RNA-binding nucleocapsid (PDB: 6M3M) ( Table 3), all the drugs including the references interacted with residues within the therapeutic target, NTD region, residue number 46-174 (Fig.1C) of the receptor [18], an indication of inhibitory actions as proposed. Drug 1 has the least XP score followed by 2, while 9 has the least. In comparison to the reference drugs, R1 and R2, the selected drugs interacted more strongly as indicated by lower glide XP scores. The binding poses (Fig. 3)  compared to the reference drugs and this favours similar but stronger potential inhibition mechanisms against the mutation and replication of the SARS-CoV-2 RNA along the NTD region. The XP docking scores (Table 4) represent the potentials of the selected drugs to inhibit the priming of the SARS-CoV-2 S-protein (PDB 6VSB) to the human ACE2 in comparison with reference drugs in the clinical trial of COVID-19 treatment, remdesivir and ribavirin. Drug 1 shows the best potential in terms of XP docking score followed by 2, while the least is observed with 6. However, all the nine selected drugs displayed better interactions, an indication of a more promising potential than the reference drugs. They occupy almost same volume within the binding sub-pocket of the receptor as the references, mostly through H-bond interactions (Fig. 2) and this favours similarity in biological functions. From the binding poses (Fig. 4) [14,58]. The selected drugs mostly interacted with more active amino acid residues along various residual chains within the binding pocket of the receptor, an indication of a more promising potential to inhibit the viral S-protein along the therapeutic regions and prevent SARS-CoV-2 infusion to the host than the reference drugs. One of the major targets in the vaccination against SARS-CoV-2 infection is the human host cell protease, ACE2. The result (Table 5) shows XP scores as representation of the higher potentials of the selected drugs to block the RBD structure of human ACE2 (PDB 6M17) in comparison with R1 and R2, reference drugs. Drug 8 has the highest potential in terms of glide XP score followed by 1, while the least is observed in 9. Interestingly, all the selected nine drugs have lower glide XP scores than the references.
The surface and binding poses ( Fig. 2 and 5) indicate that all the drugs and the references occupied the same volume and interacted through H-bonding with amino acid residues within the active sub-cavity of the RBD as expected to indicate therapeutic potentials [14].

Molecular dynamic simulations
The binding free energies and molecular dynamic trajectories in terms of statistics of H-bond, RMSD, RMSF, Rg and Q plots of the ligand-receptor complex systems of the nine selected drugs are presented in Table 3 Molecular dynamics simulation has become an indispensable structural and biophysical tool for an extensive study of biochemical ligand-receptor interactions [59]. From Table 3, the total binding free energy, MMGB(SA) is a sum of GBT and TS which comprehensively quantify the interactions that exist between the small drug molecules and the larger biological target through various conformations. The statistics of H-bond (Fig. 6A) indicates that drugs 8, 5, 1 and 9 possess the highest number of hydrogen bond interactions in descending order, then others along the simulation period. This accounts for their respective higher binding free energy than drugs 2, 4, 6 and R2 whose number of H-bond interaction occur at lower rates. The RMSD plot (Fig. 6B) shows that the receptor backbone initially experienced slight atomic deviation between 2.0 -3.5Å and become equilibrated around 1500 ps till the end of simulation with an insigni cant deviation of <1.0Å. This indicate that the 3000 ps selected for the simulation is adequate especially since the longer simulations do not necessarily in uence the binding free energy and dynamic conformations [60]. The selected drugs mostly entered the equilibrium conditions earlier and maintained it throughout the simulation periods with insigni cant deviations in the order 7<9<1<5<8<R1.
The dynamic thermal motion paths, transient channels and mean square isotropic displacements which allow the ligands to enter the internal cavities of the receptor are shown by B-factor or temperature factor and RMSF plots ( Supplementary Fig. 2C-D). Although all the selected drugs including the references undergo very insigni cant uctuations, however the least is observed in drug 1 and highest in R2. These indicate a good thermostability in the drug-receptor systems within the period of simulation. The compactness and stability of the drug-receptor complexes to further probe the structural activity were revealed by the radius of gyration (Rg) plot ( Supplementary Fig. 2E)  Table 4). The RMSD plot (Fig. 7B) indicates that the selected drugs, the references as well as the receptor backbone form stable systems with little or no deviation of atom/residue throughout the period of simulation. In spite of this, the least uctuation is observed in 1 and 2 while R1 and R2 produced the largest during the thermodynamic simulation as shown ( Supplementary Fig. 3C-D). The receptor enters into a folding state with drugs 1 and 8 the most during the simulation while the least is observed with R1 and R2 ( Supplementary Fig. 3E), although all the drugs including the references favour the folding of the receptor except 4 as indicated ( Supplementary Fig. 3F). On the average, the selected drugs display virtual inhibitory potentials, stability and bio-applicability in complex with the receptor representing the viral spike glycoprotein.
The plot of statistics of H-bond interactions throughout the simulation trajectory (Fig. 8A) indicates that drugs 7, 8 and 9 have the highest number while drug 4 show the least consistently with their binding free energy scores (Table 5). Others considerably compete with R1 and R2. The RMSD plot (Fig. 8B) reveals that the receptor backbones become stable around 1300 ps, and only undergoing an insigni cant uctuation of <1Å till the end of simulation while the drugs including the references form stable systems almost throughout the period of simulation ( Supplementary Fig. 4C-D). Although the compactness, protein folding state and stability ( Supplementary Fig. 4E-F) favour all the drugs including the references in complex with the receptor, however a little deviation is observed in R2. Summarily the drugs virtually interact more strongly and form stable systems with the receptor, an indication of potential abilities to actively block the human ACE2 against SARS-CoV-2 infection in better forms than the reference drugs. 4.6 Biochemical analysis of the antiviral activities of the selected drugs The selected drugs coincidentally have been experimented for e cacy against some virulent viral strains through anti-CoV/antiviral activities and the results ( Table 6) indicate potent inhibition. From the biochemical analysis, the drugs demonstrate strong inhibitory activity in vitro and in vivo experimental models including clinical trials against some coronaviruses and other virulent viral strains, corroborating the theoretical ndings presented above. Their mechanisms of antiviral actions include prevention of viral infusion, disruption of viral E proteins, inhibition of prenylation, disruption of the viral RNA for replication, inhibition of viral DNA and suppression of enteroviruses [50][51][52][53][54]. This phenomenon is reasonably feasible due to some degrees of genomic similarities that exist among active pathogenic sites of the viral strains [61]. Advantages of repurposing already approved drugs for this quest include the availability of physico-chemical and pharmacodynamics information as well as documented activity data relevant to the study, additionally supporting the promising inhibition in real biological system. HDV-RNA declined by 0.73 log and 1.54 log at low and high doses respectively. This is signi cantly higher than the effect of placebo. [52] 8 In vitro inhibitory assay on Vero cells infected with IBV; In vitro inhibitory assay on murine CoV Inhibits the viral replication in dose-dependents up to 4-6-folds titer reduction at 0.004 g/mL of crude extract rich in drug 8. The information presented in Table 7 indicate that the ve (5) selected drugs (Scheme 1) are either currently approved for clinical trials or in use for treatment of some ailments. More interestingly, they are commercially available and mostly traceable to vastly available natural products. This study thus, demonstrates their dual applicability potentials. Similar to other previously reported study [40][41][42][43], the application of computational tools such as molecular docking and molecular dynamics simulation have afforded fast and cheap demonstration of multi-target directed inhibition mechanisms against SARS-CoV-2 through the viral N-terminal NTD nucleocapsid phosphoprotein for translation and replication, the viral S protein along its various therapeutic targets for infusion and the human ACE2 to reception. The molecular docking protocols reveal that the drugs show stronger inhibition potentials against the targets than remdesivir and ribavirin, drugs in clinical trials. The molecular dynamics simulations to approximately mimic the real biological system also indicate better binding a nity, biostability and biofunctionalities in favour of the selected drugs than the references. Interestingly, the drugs have activity history which corroborate the ndings from the virtual studies. The selected drugs were also traced to abundant natural resources for global accessibility. The protocols reveal that the identi ed drugs possess stronger multi-target inhibition potentials against the SARS-CoV-2 pathogenic targets, similarly to some previous studies where multitarget inhibition mechanisms have been demonstrated in favour of some bioactive agents against cancer-related targets/pathways to overcome incessant resistance to drugs [62,63].

Conclusion
The rigorous molecular simulations adopted in this research work have aided the identi cation of potential inhibitors of various targeted human/viral enzymes and protein receptors implicated in SARS-CoV-2 infection such as the human ACE2, the viral spike glycoprotein and the viral nucleocapsid protein than the remdesivir and ribavirin previously reported. The ve drugs for various applications from Selleckem.com drug repurposing library (Table V), D-(+)-Ra nose pentahydrate (1), (-)-Epicatechin gallate (5), 2797 (7), Rutin DAB10 (8) and Hyperoside (9) have virtually demonstrated higher inhibitory potentials, stabilities and biological functionalities through different mechanisms against the selected therapeutic targets than the references. Their biological activities against some virulent viral strains including SARS-CoV-1 and murine CoV previously documented are consistent with the theoretical ndings in this study, a strong promise of bona de inhibition against SARS-CoV-2. Interestingly, their commercial availability and presence in natural products (in measurable amounts) could enhance the global research-based developmental studies in the discovery of effective therapeutic candidates against the pandemic. Although, the study requires computing resources and more e cient algorithms for accurate and comprehensive understanding of real biological processes, common to computational approaches in drug design, and more robust biochemical analysis targeting speci cally SARS-CoV-2 is required to further a rm the potentials of the drugs as bona de inhibitors, however, it presents a model amenable for further studies towards designing more effective and accessible therapeutic candidates, acting through multi-target mechanisms to overcome the incessant drug resistance associated with the pandemic COVID-19.

Declarations Con ict of interest
No con ict of interest was declared in the research.