Structure-based inhibitor screening of natural products against NSP15 of SARS- CoV-2 revealed Thymopentin and Oleuropein as potent inhibitors
SStructure-based inhibitor screening of natural products against NSP15 of SARS-CoV-2 revealed Thymopentin and Oleuropein as potent inhibitors Ramachandran Vijayan , Samudrala Gourinath*School of Life Sciences, Jawaharlal Nehru University, New Delhi, India.*To whom correspondence should be addressed: Dr. Samudrala Gourinath, School ofLife Sciences, Jawaharlal Nehru University, Delhi-110067, India. Email: [email protected], Tel. No: +91-11-26704513; Fax: +91 -11-26742916/2558 tructure-based inhibitor screening of natural products against NSP15 of SARS-CoV-2 revealed Thymopentin and Oleuropein as potent inhibitorsAbstract:
Coronaviruses are enveloped, non-segmented positive-sense RNA viruses thathave the largest genome among RNA viruses. The genome contains a large replicaseORF encodes nonstructural proteins (NSPs), structural and accessory genes. NSP15 is anidoviral RNA uridylate-specific endoribonuclease (NendoU) has C-terminal catalyticdomain. The endoribonuclease activity of NSP15 interferes with the innate immuneresponse of the host. Here, we screened Selleckchem Natural product database ofcompounds against the NSP15, Thymopentin and Oleuropein showed highest bindingenergies. The binding of these molecules was further validated by Molecular dynamicsimulation and found very stable complexes. These drugs might serve as effectivecounter molecules in the reduction of virulence of this virus. Future validation of boththese inhibitors are worth consideration for patients being treated for COVID -19.
Aims:
NSP15 of SARS-CoV-2 is vital for its life cycle and replication; hence, it is anattractive target for structure-based drug design of anti-SARS drugs. Worldwide researchis underway, yet, FDA has not approved any vaccine for its treatment. Infection causedby this virus spreads through human-to-human contact, and has taken a toll of severalhuman lives worldwide. As per the WHO guidelines, lock-down was implemented as aprecaution to control the spread of the disease. The outbreak of COVID -19 raised globalconcerns due to its virulence, and initiated emphasize on the urgent need to find effectivedrugs for the treatment.
Methods:
The structure-based drug designing approach wasutilized to find anti-SARS drugs.
Key findings:
We performed virtual screening of theSelleckchem Natural product database of compounds against the NSP15 of SARS-CoV-2. We have identified Thymopentin (FDA approved drug), Ginsenoside and Oleuropeinwith high clinical potential against SARS-CoV-2.
Significance:
Further validations arenecessary to test the efficacy and safety of these drugs to be used for efficienttherapeutics against the COVID -19.
Keywords:
Drug designing, Virtual screening, Severe Acute Respiratory SyndromeCoronavirus 2 (SARS-CoV-2), Natural products ntroduction: in-silico
Materials and methods:
The crystal structure of the Non-structural protein 15 (NSP15) was retrieved from theRCSB Protein Data Bank (entry code 6W01) and, was used as target for our modelingstudies. Starting structure of NSP15 for docking studies was prepared with ProteinPreparation Wizard [28]. The process adds hydrogen, neutralizes appropriate amino acidchains, and relieves steric clashes. Also, it performs a series of restrained, partialminimizations on the co-crystallized structure, each of which employs a limited numberof minimization steps. It is not intended to minimize the system completely. In our study,the minimization (OPLS 2003 force field) [28] was stopped when RMSD of the non-hydrogen atoms reached 0.30 Å, which is the default specified limit.Selleckchem Natural product libraries with about 900,000 unique molecules and1,350,000 drug-like and lead-like screening compounds were chosen for virtualscreening. Libraries were prepared by energy minimization of 100 steps with the Ligprepmodule of Schrodinger [29]. Experimental studies on NSP15 have revealed that His235, His250, and Lys290 [10] arethe critical residues for catalysis that are located in the C-terminal domain of the enzyme. Docking studies [30-34] were performed using GOLD [35], and the protein was kept as arigid molecule, and the number of Genetic Algorithm (GA) runs was set to 10 runs perligand with the default search algorithm parameters. GOLD score was then used as thefinal scoring method.In order to estimate the accuracy of binding affinity of GLIDE [36], MOE docking [31]and Auto-dock [37] were used for cross-docking analysis. In the Glide-docking [36], the prepared structure was used to generate the receptor gridand no scaling was done for the Van der Waals radii of nonpolar receptor atoms. Anenclosing box was used as the docking space, centered on the His235, His250, andLys290, catalytic triad using the crystallographic position as their reference; the liganddiameter midpoint box was set to the default value (10 Å). Docking experiments wereerformed using 0.80 to scale the VdW radii of the nonpolar ligand atoms with a chargecutoff of 0.15. Poses were discarded as duplicates if both, the rms deviations in the ligand(all atoms) was less than 0.5 Å and maximum atomic displacement was less than 1.3 Å.At most, 10 poses per ligand were retained. GlideScore XP [36] was used as the scoringmethod to finalize the screening. For MOE docking [31], the protein was kept as a rigid entity, and a maximum of 10conformations for each ligand was taken using the default parameters of MOE withTriangle Matcher placement. The top ranked conformations of NSP15 with the lead-likecompounds docked conformations was stored. On the basis of MOE scoring{Generalized-Born Volume Integral/Weighted Surface Area (GBVI/WSA)}, binding freeenergy calculation in the S field denotes the score. The GBVI/WSA is a scoring functionthat estimates the free energy of binding of the ligand for a given pose. For all scoringfunctions, lower scores indicate the more favorable poses. Top ranked dockedconformations of the lead-like compounds were selected for further evaluation. In the AutoDock 4.2 [37], empirical free energy function and the Lamarckian GeneticAlgorithm were used for all the docking calculations and the AutoDockTools (ADT)package [28] was used to generate input files and to analyze the results. The NSP15 wasset to be rigid and in small molecules all the torsional bonds were kept free during thedocking process. The solvent molecules were not considered during the docking process.A cubic grid box of 60 Å size (x62.520, y73.546, z29.278) with a spacing of 0.375 Å andgrid maps was created. The ten best solutions based on the docking scores were retainedfor further investigations. XSCORE [38] has been used to estimate the binding affinity ofthe molecules. Interactions between the protein and the compounds were calculated usingthe Discovery Studio program [39]. All Molecular dynamic simulation (MD) wereperformed with GROMACS [40], using the GROMOS 53A6 force field [40]. Furtherdetails of the simulation protocol have been used as described previously [25-27].
Results and Discussion:
Superposition of NSP15 with all other available coronavirus homologues revealsstructural conservation amongst these proteins. The structure of the SARS-CoV-2 yielded RMSD of 0.4 Å and 0.9Å with SARS-CoV, and Murine-CoV NSP15, respectively(Figure 1). Comparison of all the 3 structures showed some deviations in the active siteresidues as marked in Figure 2. These conformational changes at the active site of NSP15need to be considered carefully during the inhibitor design.
Figure 1: Superimposition of NSP15 at the active site from various organisms,SARS- CoV-2, SARS-CoV and Murine-CoV showed that the structures are verysimilar to each other. SARS- CoV-2 depicted in red, SARS-CoV depicted in green,Murine-CoV, depicted in cyanigure 2: Residues are shown in the drug-binding region. These conformationalchanges observed at the active site residues are highlighted. SARS- CoV-2 depictedin red, SARS-CoV depicted in green, Murine-CoV, depicted in cyan igure 3: Modes of binding and its interactions of the top five-inhibitor leads withNSP15Figure 4: Modes of binding and its interactions of the top inhibitor leads withNSP15o Protein Ligands GOLD score GLIDE scorekcal/mol Auto-dockkcal/mol MOEkcal/mol X-Scorekcal/mol -logKd
1. NSP15 Thymopentin 86.59 -11.030741 -9.7 -12.811 -9.63 8.492. NSP15 Oleuropein 84.69 -10.960358 -8.5 -17.1605 -8.70 7.503. NSP15 Ginsenoside 83.83 -9.856758 -8.0 -14.4559 -8.92 6.544. NSP15 Kaempferitrin 83.32 -9.17588 -7.0 -17.3375 -9.04 7.635. NSP15 Rhoifolin 82.14 -9.231593 -6.5 -12.1433 -8.43 7.186. NSP15 Demeclocyclinhydrochloride 81.71 -8.553255 -8.1 -18.0979 -8.15 6.987. NSP15 AkebiaSaponin D 80.17 -8.173799 -7.5 -12.3011 -8.89 7.258. NSP15 Liensinine 79.31 -7.752487 -7.7 -12.4259 -8.10 6.679. NSP15 Fidaxomicin 79.14 -10.166778 -7.6 -15.6436 -8.58 7.1010. NSP15 Amikacinhydrate 73.85 -8.754372 -7.0 -12.7334 -7.80 6.12
Table 1: Screening of NSP15 with the top-10 inhibitors and their estimated binding affinitiesFigure 5: The chemical structure of the top scoring compounds identified fromVirtual screening. olecular dynamic simulation [25-27] is considered to be a reliable approach withgreater insight into the dynamic behavior of proteins and that of the ligand conformations[25-27]. The simulations provide detailed information to better understand the motions ofthe individual atoms as a function of time and properties of the molecules. We carried outa 100 ns MD simulation to accurately predict the binding stability on the identifiedcompounds against NSP15 as explained in the Methods section. We have studied the rootmean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration(Rg), solvent accessible surface area (SASA), and hydrogen bonding interactions (NH)between the NSP15 alone as well as in complex with the top five inhibitors. A total of 6independent simulations were carried out, each with 100 ns simulation time.
According to the RMSD (Root-Mean-Square Deviation) calculation, shown in Figure 5,it can be seen that all the complexes tend to achieve equilibrium after 10 ns and lead to astable trajectory throughout the simulation. We found that Ginsenoside, Kaempgeritrinand Akebia saponin D complexes with NSP15 tend to reach a higher equilibriumcompared to the native, and remained distinguished throughout the simulation, resultingin the RMSD of 0.1 to 0.75 nm. The RMSD trajectories for the Thymopentin andOleuropein complexes with NSP15 observed a lower equilibrium compared to the nativewith minor difference in their trajectory that leads to a stable equilibrium through the endof the simulation, suggesting that the complexes stabilized themselves (Figure 6A). Thehigher RMSD obtained for all the complexes were limited to 0.75 nm that demonstratesthe stable trajectories and provided us with an appropriate basis for further investigation.(Figure 6A). igure 6: MD simulation results for NSP15 with ligands. (A) RMSD analysis for themodel system NSP15 protein alone and protein in complex with hits compounds (B)Radius of gyration plot for NSP15 and top5 compounds complexes. (C) RMSF plotfor NSP15 and top5 hit compounds complexes D) SASA plot for NSP15 and top5 hitcompounds complexes
In order to observe the ligand induced conformational changes of the protein, wecalculated the C-RMSF to observe the overall flexibility of the atomic positions in thetrajectory for the native and the protein-ligand complexes (Figure 6B). The flexibility ofthe residues in each of the protein-ligand was analysed by means of RMSF, where thehigher RMSF value describes the higher flexibility. According to Figure 5B, AkebiaSaponin D and Keampgertrin produced higher fluctuations in the residues positions 40,90, 110, 200, 250, 275, 290, 310, and 320 during the simulation, while low fluctuationswere observed for Ginsenoside, thymopentin, Oleuropein, except at the 150 th residueposition. Overall, the RMSF result showed that the Ginsenoside, Thymopentin, andOleuropein complexes were more stable than the Akebia Saponin D and Keampgertrincomplexes. (Fig. 6B).he Rg (radius of gyration) is used to calculate the compactness of a protein (Figure 5C).The radius of gyration is the root mean square distance of a particular atom or group ofatoms with its center of mass. The overall NSP15 structure at various time points duringthe trajectory can be analyzed for the competence, shape, and folding in the Rg plot ( .(Figure 5C). Akebia Saponin D-NSP15 complex showed a higher deviation with a Rgscore of 2.4 nm. All the other compounds, Ginsenoside, Thymopentin, Oleuropein, andKeampgertrin had an aggregate Rg score of 2.2 nm in a decreasing trend followed bystabilisation after 40 ns towards the end of the simulation (Figure 6C).We have measured the compactness of the hydrophobic core by analyzing thechanges in SASA. As shown in Figure 5D, an increase of the SASA values wereobserved for Akebia Saponin D and Keampgertrin after 10 ns of simulation timemaintained till the end, while the Ginsenoside, Thymopentin, and Oleuropein complexesshowed stable conformation throughout the simulations. As higher SASA value leads toan increased exposure, there was loss of hydrophobic contact between the protein-ligandcomplexes (Figure 6D). Finally, the number of hydrogen bonds for each of the protein complex wascalculated during the simulation. An increase of hydrogen bonds, Van der Waals, andelectrostatic interactions were observed for Ginsenoside, Thymopentin, Oleuropeincompounds, while Akebia Saponin D and Keampgertrin formed lesser interactions (Fig. 3& Fig. 4). The top-ranked compound obtained from the virtual screening is the FDAapproved drug, Thymopentin (Table 1). Thymopentin binds to the active site of NSP15endoribonuclease and may potentially block the replication of virus. Thymopentin (TP5)is a synthetic pentapeptide (Arg-Lys-AspVal-Tyr) belongs to the native thymic hormonethymopoietin [43]. Thymopentin has been suggested for the treatment of autoimmunediseases, and including chronic lymphocytic leukemia [44], rheumatoid arthritis [45],cancer immunodeficiency [46], acquired immunodeficiency syndrome (AIDS) [47], andchronic heart failure [48]. Thymopentin [45] is also known to be an effectiveimmunomodulatory agent and helps to improve immunological condition for patients.he second compound, Oleuropein showed strong binding with NSP15endoribonuclease. Oleuropein is a phenylethanoid, mainly found in the olive leaves.Oleuropein suppresses cancer cells by activating the gerosuppressor AMPK by reductionof growth in human primary cells leads to several transcriptomic signatures [49].Oleuropein helps reducing colonic microflora by Hydroxytyrosol (HT) and is mainlypresent in the olive leaf and olive oil. Oleuropein is known to be a potent antioxidantsobserved in nature to date [50]. The third rank compound, Ginsenoside also binds tightly to the catalytic center of NSP15structure. Ginsenosides are the natural products of steroid glycosides and triterpenesaponins. Ginsenosides family consists of the oleanane family and is pentacylic in nature,composed of a five-ring carbon skeleton [51]. Ginsenosides have shown diversepharmacological and biological properties, such as antitumorogenic, anti-inflammatory,antioxidant, and inhibitor of cell apoptosis.The top ranking compound Thymopentin [43-48], FDA approved drug is currentlyavailable in the market. The other two compounds, Oleuropein [49, 50] and Ginsensoside[51] are also promising ones. Hence, these drugs might help to reduce the virulence of thevirus and can be considered for patients being treated for COVID -19.
Conclusion:
In this study, structure based virtual screening followed by the validation throughMolecular dynamic simulation approaches were carried out to find antiviral leads againstNSP15 of SARS-CoV-2. Thymopentin (FDA-approved), Ginsenoside, and Oleuropeinwere identified as potential inhibitors of NSP15. Molecular docking and MD simulationsinvestigated the binding affinities, mode of binding, stability of binding and theirpotential interactions. The drug leads identified in this study sheds light on the pandemicinfectious disease that currently lacks specific drugs and vaccines. Further investigationswill be carried out for these drugs to check their efficiency in vitro and in vivo . eferences: [1] Cavanagh D. Nidovirales: a new order comprising Coronaviridae and Arteriviridae. Arch Virol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . upplementary dataNo Protein Ligands Reference MOE scorekcal/mol Table S1: Docking of NSP15 with the reported inhibitor and their estimated bindingaffinitiesNo Ligands Tumori-genicity Irritant DrugLikeness CLogP CLogS LE
1. Thymopentin None None -2.0657 -8.3593 -2.129 0.176282. Oleuropein None None -7.1766 -0.3124 -1.768 0.226263. Ginsenoside None None -11.64 -0.1028 -4.833 0.109144. Kaempferitrin None None 1.9289 -0.1371 -3.387 0.208725. Rhoifolin None None 1.1329 -2.142 -2.949 0.208746. Demeclocyclinhydrochloride None None 2.8917 -4.6648 -2136 0.271497. AkebiaSaponin D None None -12.698 0.2283 -4.85 0.127318. Liensinine None None 4.2826 4.0339 -5.513 0.189429. Fidaxomicin None None -2.7125 8.3037 -7.667 0.1138610. Amikacinhydrate None None 2.5019 -21.186 0.233 0.213651. Thymopentin None None -2.0657 -8.3593 -2.129 0.176282. Oleuropein None None -7.1766 -0.3124 -1.768 0.226263. Ginsenoside None None -11.64 -0.1028 -4.833 0.109144. Kaempferitrin None None 1.9289 -0.1371 -3.387 0.208725. Rhoifolin None None 1.1329 -2.142 -2.949 0.208746. Demeclocyclinhydrochloride None None 2.8917 -4.6648 -2136 0.271497. AkebiaSaponin D None None -12.698 0.2283 -4.85 0.127318. Liensinine None None 4.2826 4.0339 -5.513 0.189429. Fidaxomicin None None -2.7125 8.3037 -7.667 0.1138610. Amikacinhydrate None None 2.5019 -21.186 0.233 0.21365