Drug Designing: Open Access
Open Access

Caribbean Plants as Source of Novel Inhibitors for Main Protease, NSP-15, and RNA-dependent RNA polymerase (RdRp) of SARS-Cov-2

Arya Duncan, Juchara Margetson, Jada Roberts, Taquanna Baron and Neelam Buxani^{* *}Correspondence: Neelam Buxani, Department of Chemical and Physical Sciences, College of Science and Mathematics, St. Thomas Campus, The University of the Virgin Islands, VI-00802, United States, Tel: +1(340)-693-1682, Email:

Author info »

Introduction

The novel coronavirus SARS CoV-2, better known as COVID-19, is a new strain that belongs to the SARS family. This family of viruses replicates by binding to ACE receptors of cells and entering them to replicate. The virus itself consists of 2 subunits, one which binds to the ACE receptor and the other than binds to the cell surface to anchor the virus. The virus is most commonly found in a mass population among animals however, the new strain, SARS CoV-2 has been able to thrive among humans. A study of the genomic structure of the different coronavirus strains found that the new strain shares 88% similarity with the bat strain of SARS [1]. Reports state that the main mode of transmission for the SARS CoV-2 virus is through person-to-person contact specifically by way of droplets from an infected person coming into contact with a healthy host.

The trend of infection and symptom severity is directly linked to immune health whereby immunocompromised persons have a higher risk of infection and severe symptoms. Currently, 14 days is the known incubation period of the virus however there have been cases where the virus showed symptoms earlier or later than expected [2]. The main target of the virus is the respiratory system. As is common among the coronavirus family, symptoms presented include a dry coughing, fever, and fatigue however this new strain presented symptoms particular to the virus [3]. In addition to the aforementioned, SARS CoV-2 also caused disruption as the lower airway such as sneezing and sore throat in addition to gastrointestinal symptoms which are rare for the coronavirus family [3].

Due to how infectious the virus is, social distancing and mask regulations have been implemented leading to the shutdown or heavy patron restriction of closed spaces. Because of this disruption in day to day life, treatment for or immunity to the virus is necessary. Common methods of drug development take an extensive amount of time which is what makes computer-aided drug design a favorable computational tool. Computer-aided drug design is a computation drug design method that involves identifying active sites of target proteins and simulating protein- compound interactions to find inhibitors. The two types of computational drug design are structure-based drug design and ligand-based drug design [4]. Ligand-based drug design is used when the ligand structure is known but the receptor structure is unknown. Structure-based drug design on the other hand involves analyzing the known receptor structure to identify potential active sites, if unknown, and simulate ligand interactions at those sites to identify potential inhibitors that can compete with the biological ligand [4].

There are different proteins that are associated with SARS CoV- 2 functionality that are potential drug targets. Inhibition of these proteins could render the virus ineffective as interactions and mechanisms would be disrupted. The main protease (Mpro) is an essential drug target which, along with papain-like proteases catalyzes the processing of polyproteins translated from viral RNA and recognizes specific cleavage sites [5]. The RNA dependent RNA polymerase (RdRp) has been evident to possess nsp (7,8,12); each RdRp acts as responsive cofactors when stimulating polymerase complex activities. The CoV nsp7/nsp8/nsp12 complex constitutes nucleotide polymerization for its configuration [6]. In addition, the nsp 15 (active hexamer) is responsible for providing structural attributes for the development of innovative therapeutic agents [7]. The SARS-CoV-2 polyprotein assists with transportation mechanisms functioning with maintenance, transcription, and replication of the virus genome.

In this study, four medicinal plants aloe vera, lemon grass, Moringa oleifera and Lignum vitae native to the Caribbean were chosen to determine the inhibitory effects of the antiviral compounds isolated from them on different SARS CoV-2 proteins. Based on the literature [8-11] a list of 125 compounds were compiled in Table 1 and were virtually docked in the active sites of Main protease, RNA dependent RNA Polymerase, and nsp-15 proteins of SARS- CoV-2 by using molecular docking program (GLIDE, Schrodinger). Docking scores and binding energies of protein-ligand complexes were compared with the native ligand of the chosen protein structures. ADME properties of selected compounds were also determined.

Table 1: Complete list of compounds and their origin.

Materials and Methods

The virtual screening and molecular docking were performed using the Schrodinger small molecule drug discovery suite, Schrödinger Release 2020-2: Schrödinger, LLC, New York, NY, 2020.

Protein preparation and receptor grid generation

The COVID-19 proteins, main protease (PDB ID: 6LZE), RNA dependent RNA polymerase (PDB ID: 7BV2) and nsp-15 (PDB ID: 6WLC) were retrieved from the Protein Data Bank [12,13]. Using Schrodinger’s Protein Preparation Wizard the imported proteins were preprocessed, optimized, and minimized. During preprocessing of proteins missing hydrogens and residues were added with the help of Prime (version). As there was a bound ligand with all these proteins, so the receptor grids with the dimensions 10 Å × 10 Å × 10 Å were generated by selecting the atom of ligand molecule using Schrodinger receptor grid version 8.7.

Ligand preparation

A total of 4 medicinal plants native to the Caribbean were selected to compile a list of compounds to be virtually screened and identify potential antiviral agents for COVID-19. A literary analysis of numerous research papers and screening of databases generated compounds from each of the 4 plants for a total of 125 compounds (Table 1). The 2D chemical structures of all compounds discovered were imported from the PubChem database in SDF format into Maestro. Using Schrodinger’s LigPrep module, all the compounds were converted into 3D structures and optimized under Force field OPLS3e, different ionization states were generated at pH 7.0±2.0 using Epik module (version 5.2) Specific chiralities of 3D structures were retained, while around other chiral centers all possible stereoisomers were generated. Total 153 structures were taken to the docking analysis.

In present study, co-crystallized ligand of each protein was used as reference ligand; a peptide inhibitor {N}-[(2~{S})-3- cyclohexyl-1-oxidanylidene-1-[[(2~{S})-1-oxidanylidene-3-[(3~{S})- 2-oxidanylidenepyrrolidin-3-yl]propan-2-yl]amino]propan-2-yl]- 1~{H}-indole-2-carboxamide (FHR) (Figure 1a) of main protease, Ramdesivir (Figure 1b) as an inhibitor of RdRp, and Uridine-5’- monophosphate (Figure 1c) inhibitor of nsp-15.

Figure 1a: {N}-[(2~{S})-3-cyclohexyl-1-oxidanylidene-1-[[(2~{S})-1- oxidanylidene-3-[(3~{S})-2-oxidanylidenepyrrolidin-3-yl]propan-2-yl] amino]propan-2-yl]-1~{H}-indole-2-carboxamide (FHR).

Figure 1b: Ramdesivir.

Figure 1c: Uridine-5’-monophosphate.

Molecular docking

Using Glide Standard Precision (SP) the interaction between each protein and the 153 compounds and native ligand were analyzed. Based on the binding affinity scores generated (docking score), the compounds that had scores close to or better than the native ligand were selected for Extra Precision (XP) docking. Of the 153 docked compounds those with docking scores better than the native ligand for the protein were selected for further screening in XP mode along with the native ligand. If no compounds had a better docking score than the native ligand, compounds with glide scores-6.00 were chosen and further screened. The docking scores were again analyzed to determine which compounds, if any, have a higher affinity than the native ligand.

Further analysis of the compounds was necessary to determine the binding energy of protein-ligand complexes. With the knowledge of which compounds and poses best interacted with the protein, the compounds from XP mode were analyzed using Prime MM- GBSA. The SP and XP docking results showed that the various compounds interact with the protein and bind to the active site while Prime-MM-GBSA shows the free binding energy of the ligand-protein association.

ADME properties

The conclusion of the molecular docking study identified different compounds that interact well with the protein in comparison to the respective native ligand. Using QikProp (v6.4) of the Maestro (v12.4), the ADME properties of the top docking scorers were analyzed to determine if any of the compounds showed drug-like properties.

Results and Discussion

Structure-based virtual screening

Using the binding site of the native ligand, FHR, bound to the main protease 6LZE, a receptor grid of the active site was generated to dock the researched 125 compounds. An analysis of the active site showed the native ligand interaction via hydrogen bonds (H-Bond) with the GLY 143, PHE 140, HIE 163, and GLU 166 residues (Figures 2a-2c). Ligand preparation of the 125 researched compounds generated a total of 153 different poses to be filtered with all 153 meetings the filter parameters. Because of the small compound library, preliminary virtual screening was carried out in SP mode and a total of 38 compounds were filtered out based on their docking score. The SP mode docking results did not yield any compounds with a docking score better than the native (docking score=-9.734) so all compounds with scores equal to or greater than-6.00 were selected for XP virtual screening.

Figure 2a: RInteraction diagram between 6LZE protein in complex with FHR.

Figure 2b: Interaction diagram between 7BV2 protein in complex with Remdesivir.

Figure 2c: Interaction diagram between 6WLC protein in complex with Uridine-5'-Monophosphate (site 1 and 2 left to right).

Like the 6LZE protein, 7BV2 and 6WLC had their active site analyzed and the compounds docked to those sites in SP mode to determine top scorers. There were 40 compounds that had better docking scores than the native ligand when docked in SP mode to 7BV2. On the other hand, because 6WLC had 2 different sites, the compounds were docked in SP mode to both. Site 1 did not produce any compounds that had better docking scores than the native so all compounds with SP scores >6.00 were selected for XP docking. Site 2 on the other hand had 18 compounds with better docking scores than the native in SP mode. All compounds that scored better than the native or had an SP docking score >6.00 when no better interactions were produced were docked to their respective sites in XP mode.

The XP docking method was able to provide positive results as it showed there were compounds that interacted better than the native ligand. The XP docking yielded multiple compounds that had a better docking score than the native ligand for each protein (Tables 2-4).

Table 2: Top scoring compounds from 6LZE docking.

Table 3: Top Scoring Compounds from 7BV2 docking.

Table 4: Top scoring compounds from 6WLC docking.

The 6LZE protein had 13 compounds that docked better than the native ligand. These 13 interactions were analyzed using Prime MM-GBSA to determine the binding energy. Glide XP and Prime MM-GBSA scores for the top 13 compounds can be found in Table 2. The most prominent interaction between the protein- ligand complex was hydrogen bonding. In almost all complexes, hydrogen bonding occurred between the compound and the GLU 166 residue as is present in the native ligand-protein complex. This is the only common interaction when comparing the native ligand and the tested compounds, however when comparing the test compounds, hydrogen bonding with the CYS 145 residue and pi-pi stacking interaction between the HIE 41 residue and benzene derivatives are prominent (Figures 3a and 3b).

Figure 3a: 6LZE with FHR native ligand.

Figure 3b: 6LZE with Rutin (top XP scorer).

The Prime MM-GBSA analysis showed the free binding energy of each compound-protein complex. The top 13 compounds had a range of free binding energy from-18.06 kcal/mol to-70.54 kcal/ mol. The complex with the lowest energy, which is most favorable, belongs to Rutin which also has better binding energy than the native ligand which would suggest a potentially strong inhibitor of the main protease. Rutin also had the best docking score at-11.962. The ligand interaction diagram shows H-Bonds with GLY 143 and GLU 166 residues the same as the native ligand but also interacts via hydrogen bonding with ASN 142, CYS 145, HIE 41, and GLN 189.

A total of 36 compounds had better ligand interactions than the native ligand of 7BV2. Similar to 6LZE, the main interactions between the protein and the ligands is hydrogen bonding. The native ligand interacts with the protein by hydrogen bonds with ASN 691, U10, POP 1003 and ring interactions with U 20 and ARG 555. The binding energy for the top compounds range from- 9.1 to-48.84. The compound Orientin has the most favorable Peime-MM GBSA binding energy (-48.84) and has a better docking score than the 7BV2 native ligand. The ligand interaction diagram for Orientin shows that its interaction with the protein is strictly via hydrogen bonding with A 11, U 12, A 13, A 12, A 14, C 15, ASN 496 (Figures 4a and 4b).

Figure 4a: 7BV2 with Remdesivir native ligand.

Figure 4b: 7BV2 with Vicenin-2 (top XP scorer).

An analysis of the docking results for the 6WLC protein shows that one site has more favorable interactions with the compounds than the others. Only 1 of the 125 compounds interacts more favorably at site 1 (S1) than the native. Site 2 (S2) on the other hand had favorable interactions with 5 compounds (Table 4). The native ligand, Uridine-5'-Monophosphate, interacted with the protein via hydrogen bonding with LYS 65, ILE 64, SER 294 and LEU 346. Comparison of the native ligand interaction and the tested compounds shows that there are many residues in common between them. The tested compounds all form hydrogen bonds with at least one of the same residues as the native suggesting their importance in the active site and compound interactions (Figures 5a and 5d).

Figure 5a: 6WLC with Uridine-5'-Monophosphate native ligand.

Figure 5b: 6WLC with orienting (top XP scorer S1).

In addition to the docking score, Prime MMGBSA analysis shows that these compounds have the most favorable binding energy out of all the protein-ligand interactions. The binding energy for the native ligand is 0 kcal/mol for both sites. Of the 6 total ligands that interacted with the protein irrespective of site, half have equal binding energy to the native and the other half have better binding energy than the native at-12.04 kcal/mol. The common residue interaction among all compounds aside from Aloesin is the SER 294 residue, suggesting it is a key residue in ligand interaction in the active site. A complete list of H-bonding interactions for the top scoring compounds is given in Tables 2-5.

Table 5: H-bonding interactions of Top Scoring Compounds with different residues of 6WLC.

Among all the compounds, one showed promising interaction with all three proteins. Orientin, a compound native to Aloe Vera, not only showed docking scores that were better than the native ligand for all proteins but also had a better binding affinity to the protein than the native. It showed higher affinity with the 7BV2 and 6WLC proteins and had a similar score for 6LZE. Prime MMGBSA and XP docking scores for Orientin and each protein’s native ligands can be found in their respective Tables 2-4 for 6LZE, 7BV2 and 6WLC respectively Table 6.

Table 6: ADME Properties.

ADME properties

The 13 best compounds from molecular docking went through ADME analysis using the QikProp module on Maestro. A total of 11 properties were the main focus which include the molecular weight, the number of hydrogen donors, the number of hydrogen acceptors, the number of violations of the rule of five, predicted IC50 value for blockage of HERG K+ channels (QPlogHERG), percent human oral absorption, predicted aqueous solubility (QPlogS), prediction of binding to human serum albumin (QPlogKhsa), predicted octanol/water partition coefficient (QPlogPo/w), predicted blood/brain partition coefficient (QPlogBB), number of likely metabolic reactions (#metab), and predicted apparent Caco-2 cell permeability (QPPCaco-).

ADME analysis showed that each compound has its pros and cons with no one compound falling within the acceptable range for all the properties. Most of the compounds had at least 1 violation of the rule of five with a few that had no violations. Lipinski Rule of Five states that drug-like compounds should have molecular weight lower than 500, lipophilicity (logP) lower than 5, less than five hydrogen bond donors, and less than 10 hydrogen bond acceptors, but many of natural products drugs do not comply with the “Rule of Five” [14]. it is recommended to not apply overly rigid cut-off points, as it increases the risk of losing some valuable compounds in earlier stages of screening [14], especially in case of natural products. A comprehensive list of the compounds and their ADME properties can be found in Table 6.

Conclusion

The researched plants are all used for medicinal purposes in the Caribbean from centuries. Based on the present in-silco study, multiple compounds showed promising inhibitory potential against different SARS CoV-2 proteins. Some of the compounds like rutin, vicenin-2, and Orientin exhibited better protein- compound association in compare to respective native ligands and found to have more negative binding energy. Further research on the antiviral properties of these plants against the various proteins of the SARS CoV-2 virus could prove useful for the discovery of naturally occurring antiviral compounds for future drugs and may provide natural remedies against this fatal disease.

References

1. Chen Y, Guo Y, Pan Y, Zhao ZJ. Structure Analysis of the Receptor Binding of 2019-NCoV. Biochem Biophys Res Commun. 2020; 525(1):135-140.
2. Rothan HA, Byrareddy SN. The Epidemiology and Pathogenesis of Coronavirus Disease (COVID-19) Outbreak. J Autoimmun. 2020; 109:102433.
3. Yu W, Mackerell AD. Computer-Aided Drug Design Methods. Methods in Molecular Biology Antibiotics. 2016; 85-106.
4. Gurung AB, Ali MA, Lee JK, Farah MA, Al-Anazi KM. Unravelling lead antiviral phytochemicals for the inhibition of SARS-CoV-2 Mpro enzyme through in silico approach. Life Sci. 2020; 255:117831.
5. Kirchdoerfer RN, Ward AB. Structure of the SARS-CoV nsp12 polymerase bound to nsp7 and nsp8 co-factors. Nat Commun. 2019; 10(1):1-9.
6. Pillon MC, Frazier MN, Dillard L, Williams JG, Kocaman S, Krahn JM, et al. Cryo-EM Structures of the SARS-CoV-2 Endoribonuclease Nsp15. BioRxiv. 2020.
7. Quispe C, Villalobos M, Bórquez J, Simirgiotis M. Chemical Composition and Antioxidant Activity of Aloe Vera from the Pica Oasis (Tarapacá, Chile) by UHPLC-Q/Orbitrap/MS/MS. J Chem. 2018; 2018:1–12.
8. Abd Rani NZ, Husain K, Kumolosasi E. Moringa genus: A review of phytochemistry and pharmacology. Front Pharmacol. 2018; 9:108.
9. Nasr-Eldin MA, Abdelhamid A, Baraka D. Antibiofilm and antiviral potential of leaf extracts from Moringa oleifera and rosemary (Rosmarinus officinalis lam.). Egypt J Microbiol. 2017; 52(1):129-139.
10. Zhang B, Zhang Y, Jing Z, Liu X, Yang H, Liu H, et al. 6LZE: The crystal structure of COVID-19 main protease in complex with an inhibitor 11a. PDB. 2020.
11. Kim Y, Maltseva N, Jedrzejczak R, Endres M, Chang C, Godzik A et al. 6WLC: Crystal Structure of NSP15 Endoribonuclease from SARS CoV-2 in the Complex with Uridine-5'-Monophosphate. RCSB PDB. 2020.
12. Lionta E, Spyrou G, K Vassilatis D, Cournia Z. Structure-based virtual screening for drug discovery: principles, applications and recent advances. Curr Top Med Chem. 2014; 14(16):1923-1938.