Identification of Small Molecule Inhibitors of the Deubiquitinating Activity of the SARS-CoV-2 Papain-Like Protease: in silico Molecular Docking Studies and in vitro Enzymatic Activity Assay

COVID-19 is an ongoing pandemic caused by the SARS-CoV-2 virus with important political, socio-economic, and public health consequences. Inhibiting replication represents an important antiviral approach, and in this context two viral proteases, the SARS-CoV-2 main and papain-like proteases (PLpro), which cleave pp1a and pp1ab polypeptides, are critical. Along with protease activity, the PLpro possesses deubiquitinating activity, which is important in immune regulation. Naphthalene-based inhibitors, such as the well-investigated GRL-0617 compound, have been shown to possess dual effects, inhibiting both protease and deubiquitinating activity of the PLpro. Rather than binding to the canonical catalytic triad, these type of non-covalent inhibitors target an adjacent pocket, the naphthalene-inhibitor binding site. Using a high-throughput screen, we have previously identified the dietary hypericin, rutin, and cyanidin-3-O-glucoside compounds as potential protease inhibitors targeting the naphthalene-inhibitor binding site. Here, our aim was to investigate the binding characteristics of these compounds to the PLpro, and to evaluate deubiquitinating activity, by analyzing seven different PLpro crystal structures. Molecular docking highlighted the relatively high affinity of GRL-0617 and dietary compounds. In contrast binding of the small molecules was abolished in the presence of ubiquitin in the palm subdomain of the PLpro. Further, docking the small molecules in the naphthalene-inhibitor binding site, followed by protein-protein docking revealed displacement of ubiquitin in a conformation inconsistent with functional activity. Finally, the deubiquitinating activity was validated in vitro using an enzymatic activity assay. The findings indicated that the dietary compounds inhibited deubiquitinase activity in the micromolar range with an order of activity of GRL-0167, hypericin >> rutin, cyanidin-3-O-glucoside > epigallocatechin gallate, epicatechin gallate, and cefotaxime. Our findings are in accordance with mechanisms and potential antiviral effects of the naphthalene-based, GRL-0617 inhibitor, which is currently progressing in preclinical trials. Further, our findings indicate that in particular hypericin, rutin, and cyanidin-3-O-glucoside, represent suitable candidates for subsequent evaluation as PLpro inhibitors.


INTRODUCTION
Coronavirus disease  was declared a pandemic on the 11th of March 2020 (World Health Organization, 2020). The first reported cases of this disease came from Wuhan, China in late 2019, and the infectious agent responsible for causing this disease was identified as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (Coronaviridae Study Group of the International Committee on Taxonomy of Viruses, 2020). Since the start of the year, the scientific literature on COVID-19 has increased and the findings from these studies have formed an integral part of the public health response.
In regards to treatment options, a number of vaccine trials have been established and there is also a focus on drug repositioning (Bar-Zeev and Moss, 2020;Folegatti et al., 2020). The U.S Food and Drug Administration (FDA) has recently approved remdesivir, which is an RNA-dependent RNA polymerase inhibitor, as a COVID-19 treatment for hospitalized patients (FDA, 2020). Interestingly, the WHO Solidarity trial has produced contradicting findings regarding the effectiveness of remdesivir . In addition to antiviral drugs, the efficacy of compounds that have immunomodulating properties are also being investigated (de la Rica et al., 2020).
Further research is required to establish the precise mechanisms of action of potential therapeutic drugs and determine their biological targets (Zhou Y. et al., 2020). The virus replication cycle is comprised of several stages and compounds that inhibit key proteins involved in these steps may have antiviral properties (Jeong et al., 2020;Pandey et al., 2020). Papain-like protease (PL pro ) is a cysteine protease enzyme that is encoded by the multi-domain non-structural protein 3 (nsp3) and is required for polypeptide processing (Báez-Santos et al., 2015;Folegatti et al., 2020). The role of the SARS-CoV-2 PL pro in viral replication and the regulation of the innate immune response is being explored (Shin et al., 2020).
Ubiquitin and ubiquitin-like proteins, such as interferonstimulated gene 15 (ISG15), are important effector molecules of the antiviral immune response (Jiang and Chen, 2011;Perng and Lenschow, 2018). Through binding to target proteins, various cellular pathways can be modulated (Jiang and Chen, 2011;Perng and Lenschow, 2018). Viruses have developed mechanisms to evade detection and destruction by the host's immune response, and these strategies continue to evolve (Nelemans and Kikkert, 2019). In terms of SARS-CoV-2, the deubiquitinating and deISGylating activities of the PL pro enzyme have been described (Bosken et al., 2020;Klemm et al., 2020;Rut et al., 2020;Shin et al., 2020). GRL-0617 is a naphthalene-based inhibitor that has been found to interfere with the protease, deubiquitinase and deISGylating activities of the SARS-CoV and SARS-CoV-2 PL pro enzymes (Ratia et al., 2008;Freitas et al., 2020;Gao et al., in press). The SARS-CoV-2 PL pro has consequently been identified as an attractive drug target and in this study, the deubiquitinase activity of this viral protein was of interest (McClain and Vabret, 2020).
The health-promoting properties of dietary compounds have been extensively explored over the years and in response to the COVID-19 pandemic, bioactive compounds are being investigated further Mani et al., 2020). There is a growing body of literature on the antiviral and immunomodulating properties of plant-based compounds, and their potential use as therapeutic agents against SARS-CoV-2 (Tiwari et al., 2018;Panyod et al., 2020;Tahir Ul Qamar et al., 2020). This includes traditional Chinese and traditional Indian medicinal compounds, vitamins, curcumin, glycyrrhizic acid, tea polyphenols and compounds derived from Allium sativum to name a few Divya et al., 2020;Donma and Donma, 2020;Tripathi et al., 2020). Enhancing the bioavailability of natural compounds continues to be a challenge however, their structures can be used as scaffolds for the development of novel drugs (Ngwa et al., 2020).
In silico methods were used to compare the binding mode of naphthalene-based inhibitors (GRL-0617 and 3k) to dietary compounds including hypericin, rutin, cyanidin-3-O-glucoside and (-)-epigallocatechin gallate. The antimicrobial, antiinflammatory and antioxidant activities of these phytochemicals have been reported and their structures may even be used as scaffolds in the drug development process (Mohammadi Pour et al., 2019). The aim was to determine whether the dietary compounds were able to bind in a similar manner as the positive control GRL-0617, and potentially interfere with the binding of ubiquitin. The results were validated further using in vitro assays.

Protein Structures and Ligands
Several crystal structures of the SARS-CoV-2 PL pro were obtained from the RCSB Protein Data Bank (PDB ID: 6xaa, 6w9c, 6wuu, 6wx4, and 7jrn) (Berman et al., 2000;Klemm et al., 2020;Osipiuk et al., 2020;Rut et al., 2020;Sacco et al., 2020). The SARS-CoV PL pro (PDB ID: 4mm3) and MERS-CoV PL pro (PDB ID: 4rf0) were also used for comparison in this study (Bailey-Elkin et al., 2014;Ratia et al., 2014). Crystallographic waters were removed and the native zinc ions were retained. A ubiquitin chain was present in the structures of 6xaa, 4mm3, and 4rf0, which was used to generate two sets of docking data for each protein: PL pro in complex with ubiquitin and, and apo PL pro in the absence of ubiquitin. The ubiquitin in each complex was also isolated for protein-protein docking. The ligands that were used in this in silico study were the naphthalene inhibitors GRL-0617 and 3k, and the dietary compounds (-)-epigallocatechin gallate, cyanidin-3-O-glucoside, rutin and hypericin. The structures of the dietary compounds were obtained from the National Center for Biotechnology Information PubChem (Kim et al., 2019). GRL-0617 and 3k were drawn using Chem3D 19.0 (Perkin Elmer, Massachusetts, USA).

Molecular Docking Using the Schrödinger Suite
Molecular docking was performed using the Schrödinger Suite (Schrödinger, 2020a). The protein structures were prepared using the Protein Preparation Wizard, while the compounds were prepared using the LigPrep tool (Madhavi Sastry et al., 2013;Schrödinger, 2020b). The default settings were used for both of these steps and the optimized potentials for liquid simulations 3e (OPLS3e) force field was selected (Jorgensen and Tirado-Rives, 1988;Jorgensen et al., 1996;Shivakumar et al., 2010;Harder et al., 2016). The top ranking ligand conformation was selected for the subsequent molecular docking stage.
The receptor grids were 20 × 20 × 20 Å in size and the OPLS3e force field was utilized. The compounds were then docked to each protein structure using the quantum-polarized ligand docking (QPLD) protocol, as previously described .

Blind Docking and the Prediction of Ligand-Binding Sites
The PL pro crystal structures and the compounds were prepared as macromolecules and ligands in PyRx, respectively (Dallakyan and Olson, 2015). The corresponding.pdbqt files were obtained and the receptor grid was generated around the entire surface of the protein. The exhaustiveness was set to 2048. AutoDock Vina was used for molecular docking and the jobs were run on the cloud-computing server Galileo (Hypernet Labs) (Trott and Olson, 2010;Hypernet Labs Galileo, 2020). In addition to blind docking, the PrankWeb server was used to identify potential binding pockets that were conserved in the 6xaa, 4mm3, and 4rf0 crystal structures (Jendele et al., 2019).

Protein-Protein Docking
The HDOCK server was used for ab initio template free proteinprotein docking in this study (Yan et al., 2017(Yan et al., , 2020. For the SARS-CoV-2, SARS-CoV, and MERS-CoV complexes, the main chain of PL pro was defined as the receptor molecule, whereas the ubiquitin chain was the ligand. Protein-protein docking was also performed with PL pro in the presence of compounds that were docked to the naphthalene-inhibitor site in the Schrödinger Suite. This was done for the 6xaa, 4mm3, and 4rf0 crystal structures in order to determine whether the presence of these ligands in the naphthalene-inhibitor region would affect the ability of ubiquitin to bind to PL pro .

Enzymatic Activity Assay
In vitro inhibition of the PL pro deubiquitinase activity was measured using a commercially available enzymatic activity assay (BP Bioscience, San Diego, CA, USA). The experiment was performed according to the manufacturer's instructions and all samples were assayed in triplicate. An excitation wavelength of 360 nm was used, and fluorescence was measured at an emission wavelength of 460 nm on the basis of the presence of a ubiquitinated fluoregenic substrate. The nonconvalent inhibitor GRL-0617 was provided as an internal positive control and was used at concentration of 100 µM in the assay. We tested the following compounds for potential inhibition of PL pro deubiquitinase activity: hypericin (89%, HWI pharma services GmbH, Germany), cyanidin-3-O-glucoside (reference standard, PhytoLab, Germany), and rutin (>94%), (-)epicatechin gallate (>98%), (-)-epigallocatechin gallate (>95%), and cefotaxime (European pharmacopeia reference standard) from Sigma-Aldrich (St Luis, MO, USA); 20 mM stock solutions of each compound were stored at −80 • C until use. Serial doubling dilutions were performed to yield a final concentration of 3.1 to 200 µM for assaying each compound. Readings (absolute fluorescence values at 1,200 gain), were made using a CLARIOstar R Plus fluorescence microplate reader (BMG Biotech, Oternberg, Germany). Where appropriate the % PL pro deubiquitinase activity was calculated as the ratio of activity in the presence of inhibitor and total activity, and taking into account background readings.

RESULTS
Molecular docking was performed to examine the potential inhibitor behavior of dietary compounds and naphthalene inhibitors GRL-0617 and 3k on the SARS-CoV-2 PL pro . Like SARS-CoV-2, SARS-CoV, and MERS-CoV are classified as betacoronaviruses and they predominantly affect the respiratory tract (Abdelrahman et al., 2020;Petersen et al., 2020). The PL pro from the novel SARS-CoV-2 was consequently compared to the SARS-CoV and MERS-CoV PL pro structures (Figure 1). In terms of the SARS-CoV and MERS-CoV PL pro , the pairwise alignment revealed they had a sequence similarity of 89 and 50% with the SARS-CoV-2 PL pro , FIGURE 1 | Crystal structures of the SARS-CoV-2 (6xaa), SARS-CoV (4mm3). and MERS-CoV PL pro (4rf0). The fingers, ubiquitin-like (Ubl), thumb and palm domains are labeled. The conserved naphthalene-inhibitor binding pockets that were identified through the PrankWeb server for 6xaa, 4mm3, and 4rf0 are shown. This region is colored ochre and the catalytic triad residues can be seen in red. GRL-0617 (colored violet) was the control and this compound was docked to the naphthalene-inhibitor site of each PL pro structure. The molecular docking results are depicted for 6xaa, 4mm3, and 4rf0. Ubiquitin is colored tan (ribbon representation), while PL pro is colored silver.
respectively. The pairwise alignment was used to identify the corresponding residues of the naphthalene-inhibitor binding site for the SARS-CoV and MERS-CoV crystal structures. The three crystal structures (6xaa, 4mm3, and 4rf0) were analyzed using PrankWeb, a binding site prediction tool, which identified the naphthalene-inhibitor binding site as a conserved ligand binding pocket. This region was ranked as pocket 2, pocket 3 and pocket 2 in the SARS-CoV-2, SARS-CoV and MERS-CoV PL pro structures, respectively (Figure 1 and Supplementary Table 1).
Moreover, the six compounds were docked to the 6wuu, 6w9c, 6wx4, and 7jrn crystal structures of the SARS-CoV-2 PL pro (Supplementary Figures 3-6). With the exception of the 7jrn PL pro structure, all of the dietary compounds had stronger GlideScores than the naphthalene-based inhibitors. Hypericin, rutin and (-)-epigallocatechin gallate were predicted to bind more strongly than GRL-0617 to the 7jrn PL pro structure, while cyanidin-3-O-glucoside had a similar GlideScore to the control compound. In the 6w9c, 6wuu, and 7jrn SARS-CoV-2 PLpro structures, 3k formed bonds with Y264 (6wuu: H-bond, 6w9c: H-bond, and salt bridge, 7jrn: salt bridge). In 7jrn, 3k formed a hydrogen bond with Y268. The intermolecular bonds that GRL-0617 formed with the protein residues varied amongst the crystal structures. GRL-0617 formed two hydrogen bonds with G266 and N267 in 6w9c, whereas a hydrogen bond was present with D164 in 6wuu. In 6wx4, there were two π-π interactions with Y264 and hydrogen bonds with G271 and Y268. There were π-π interactions with Y268 in 7jrn and a hydrogen bond with Q269.
Molecular docking was also performed on the SARS-CoV-2, SARS-CoV and MERS-CoV PL pro structures with the ubiquitin chain present (Figure 4). When GRL-0617, 3k, (-)-epigallocatechin gallate, rutin, cyanidin-3-O-glucoside, and hypericin were docked to PL pro , it was evident that they were binding distant from the naphthalene-inhibitor binding site. The position and orientation of these ligands, as well as their GlideScores, differed to when ubiquitin was absent (Figure 4 and Supplementary Table 4). GRL-0617 in the SARS-CoV-2 PL pro , for example, was found to form hydrogen bonds with P248 and G266. The control compound was no longer binding to the pocket that the C-terminal chain of ubiquitin extends into. Interestingly, hypericin was not able to bind in the presence of ubiquitin. Similarly, the blind docking results for PL pro in complex with ubiquitin revealed that the compounds that originally had poses within the naphthalene-inhibitor binding site were displaced from this region.
The PL pro -ubiquitin complexes were investigated further using the HDOCK server for protein-protein docking (Figure 5). The ubiquitin chain from each complex was isolated and was re-docked to the main PL pro structure. There was a clear overlap in the position between the top ranked model of the docked ubiquitin and the crystallographic ubiquitin. The rootmean square deviation (RMSD) values that were generated from protein-protein docking were 0.4, 0.6, and 0.5 Å for the SARS-CoV-2, SARS-CoV, and MERS-CoV complexes, respectively, and importantly, the C-terminal tail of ubiquitin extended into the catalytic and naphthalene binding region of PL pro . The C-terminus of ubiquitin in the SARS-CoV-2, SARS-CoV and MERS-CoV PL pro consists of residues R72, L73, R74, and G75. The C-terminal residue of ubiquitin is AYE76, GLZ76, and 3CN101 in the SARS-CoV-2, SARS-CoV and MERS-CoV structures, respectively. These three residues form a covalent bond with the catalytic cysteine residues and this is critical for deubiquitinase activity (C111 in SARS-CoV-2: 1.67 Å, C112 in SARS-CoV: 1.72 Å and C1592 in MERS-CoV: 1.45 Å). The 1.67, 1.72, and 1.45 Å correspond to the distances between the sulfur atom of the catalytic cysteine residue and the Cterminal ubiquitin residue in the crystal structures. Although non-covalent protein-protein docking was performed in the present study, the distance between AYE76 and C111 in SARS-CoV-2 was 1.94 Å. The distance between GLZ76 and residue C112 in SARS-CoV was 2.05 Å. Likewise, the distance between 3CN101 and C1592 in MERS-CoV was 1.48 Å.
In order to examine the ability of ubiquitin to bind to PL pro when compounds are present in the naphthalene-inhibitor binding site, the ligands that were docked using the Schrödinger Suite were retained in the protein structures (Figure 5). When GRL-0617, 3k, (-)-epigallocatechin gallate, rutin, cyanidin-3-Oglucoside, and hypericin were bound to this region, dramatic differences were observed in the binding mode of ubiquitin. The C-terminal chain of ubiquitin was no longer extending into the pocket that is located just above the catalytic triad. There were consequently changes in the position and orientation of ubiquitin for each ligand. When comparing the docked ubiquitin chain to the ubiquitin present in the original crystal structure, the RMSD values and docking scores were different (Supplementary Table 5). This was apparent for the SARS-CoV-2, SARS-CoV and MERS-CoV complexes.

Hypericin, Rutin, and Cyanidin-3-O-Glucoside Inhibit PL pro Deubiquitinase Activity in a Concentration-Dependent Manner
Inhibition of PL pro deubiquitinase activity by small molecules in vitro was investigated using an enzymatic assay. Overall, the findings indicate that hypericin, rutin, and cyanidin-3-Oglucoside resulted in a concentration-dependent inhibition of PL pro deubiquitinase activity, with hypericin clearly highlighting the most potent inhibition of the test ligands (Figure 6A). At the higher concentrations (>50 µM) hypericin inhibited PL pro deubiquitinase activity to a level akin to the internal FIGURE 3 | Blind docking results for the SARS-CoV-2 PL pro in the absence and presence of ubiquitin (6xaa). Blind docking was conducted on the apo and ubiquitin-bound crystal structure of the SARS-CoV-2 PL pro using the naphthalene-based inhibitors (GRL-0617: violet, 3k: orange) and the dietary compounds (cyanidin-3-O-glucoside: blue, rutin: green, (-)-epigallocatechin gallate: yellow, hypericin: red). The number of poses that were found to be in the naphthalene-inhibitor binding region (colored ochre) are shown. Ubiquitin is colored tan, while PL pro is colored silver.

DISCUSSION
The innate immune system is the first line of defense against foreign pathogens and various cellular and molecular components are involved in this process Huang et al., 2019). Post-translational modifications are also important regulators of immunity and this includes ubiquitination . Ubiquitin is a 76-amino acid polypeptide that can covalently interact with target proteins and ubiquitin itself can undergo ubiquitination at certain residues (Ciechanover et al., 2000;Pickart, 2001;Jiang and Chen, 2011). This results in the formation of lysine-linked polyubiquitin chains or linear polyubiquitin chains (Heaton et al., 2015). It is well-known that lysine 48 (K48)-linked polyubiquitylation promotes the proteasomal degradation of target proteins (Ciechanover et al., 2000). Conversely, lysine 63 (K63)-linked polyubiquitylation has been implicated in cellular processes such as the DNA damage response, inflammation and endocytosis (Panier and Durocher, 2009;Erpapazoglou et al., 2014;Zhou Z. et al., 2020). It is also important to note that other types of polyubiquitin chain linkages are being explored and that ubiquitin modifications can lead to different cellular outcomes (Komander and Rape, 2012;Ohtake et al., 2018).
Moreover, human deubiquitinases (DUBs) are enzymes that remove ubiquitin modifications and they contribute to homeostasis . Viruses are dependent on host cells for their survival and in order to complete their life cycle, they have developed strategies to evade the antiviral immune response (Nelemans and Kikkert, 2019). Interestingly, several viral proteins have been found to possess deubiquitinating activity and they can be used to antagonize or modulate the antiviral immune signaling pathway (Kumari and Kumar, 2018).
The deubiquitinating activity of the SARS-CoV-2 PL pro was the main focus of this study and the crystal structure of FIGURE 4 | Molecular docking results of the naphthalene-based inhibitors and the dietary compounds to the SARS-CoV-2 (6xaa), SARS-CoV (4mm3), and MERS-CoV (4rf0) PL pro in the absence and presence of ubiquitin. The compounds were docked to the naphthalene-inhibitor binding region (colored ochre) of the apo and ubiquitin-bound PL pro crystal structures. Ubiquitin is colored tan, while PL pro is colored silver. GRL-0617 is colored violet, 3k is colored orange, cyanidin-3-O-glucoside is colored blue, (-)-epigallocatechin gallate is colored yellow, rutin is colored green and hypericin is colored red.
PL pro in complex with ubiquitin propargylamide was utilized (Klemm et al., 2020). In addition to this, four other crystal structures of SARS-CoV-2 that were available on the RCSB PDB were evaluated. The crystal structures of the SARS-CoV PL proubiquitin aldehyde and MERS-CoV PL pro -ubiquitin complexes were used for comparison (Bailey-Elkin et al., 2014;Ratia et al., 2014). Molecular docking allowed for the binding properties of compounds to the known target site of naphthalene-based inhibitors to be predicted and examined.
GRL-0617 was the control and this has previously been found to potently inhibit the SARS-CoV and SARS-CoV-2 PL pro in a non-covalent manner (Ratia et al., 2008;Freitas et al., 2020;Shin et al., 2020). The GRL-0617 inhibitor occupies the S3 and S4 pockets of the SARS-CoV and SARS-CoV-2 PL pro (Ratia et al., 2008;Gao et al., in press). Based on the molecular docking results from the current study, GRL-0617 was predominantly surrounded by hydrophobic residues in the SARS-CoV and SARS-CoV-2 crystal structures (Figure 1 and  Supplementary Figures 3-6) (Ratia et al., 2008). This ligand was predicted to form inter-atomic contacts with the protein residues and this included D164 in the 6wuu SARS-CoV-2 structure, as well as Q269 in the 7jrn SARS-CoV-2 structure. In the crystal structure determined by Gao et al.,, which was the cocrystallized ligand, was found to form hydrogen bonds with these FIGURE 5 | Protein-protein docking results of the SARS-CoV-2 (6xaa), SARS-CoV (4mm3), and MERS-CoV (4rf0) PL pro -ubiquitin complexes. The covalent bond that is formed between the C-terminal chain of the crystallized ubiquitin and the catalytic cysteine residue of the proteins can be seen. The HDOCK server was used to dock ubiquitin to the apo PL pro and to PL pro with ligands bound to the naphthalene-inhibitor region. Ubiquitin is colored tan, while PL pro is colored silver. GRL-0617 is colored violet, 3k is colored orange, cyanidin-3-O-glucoside is colored blue, (-)-epigallocatechin gallate is colored yellow, rutin is colored green and hypericin is colored red. critical residues (Gao et al., in press). In saying this, D164 and Q269 were found to surround GRL-0617 in the 6xaa, 6w9c, and 6wx4 structures of the SARS-CoV-2 PL pro .
GRL-0617 also interacted with Y268 in the SARS-CoV-2 PL pro and inter-atomic contacts were present with this residue in some structures. Interestingly, the naphthalenebased inhibitor 3k was found to consistently form inter-atomic contacts with Y264 in all of the SARS-CoV-2 crystal structures (Figure 2 and Supplementary Figures 3-6) (Bosken et al., 2020). Intermolecular bonds were also formed between 3k and D164 in three of the SARS-CoV-2 papain-like proteases, as well as Y268 in the 6wx4 structure (Bosken et al., 2020). Bosken et al. have identified these residues as playing an important role in the binding mode of 3k (Bosken et al., 2020).
The differences observed in the intermolecular bonds may be due to the conformations of the fingers domain and BL2 loop in the PL pro crystal structures (Figure 1) (Báez-Santos et al., 2015). In the SARS-CoV-2 and SARS-CoV PL pro structures, the BL2 loop corresponds to residues 267-272 (Lee et al., 2015;Gao et al., in press). In the MERS-CoV PL pro structure used in this study, the BL2 loop is comprised of residues 1,752-1,758 (Bailey-Elkin et al., 2014;Lee et al., 2015). The structural significance of the BL2 loop (blocking loop) has been discussed in a number of papers and its flexibility has been highlighted (Báez-Santos et al., 2015;Bosken et al., 2020;Klemm et al., 2020). Conformational changes have been observed in the BL2 loop and "open" or "closed" conformations have been reported in the literature (Báez-Santos et al., 2015). In regards to MERS-CoV, there are significant structural differences in the BL2 loop and it has been suggested that this affects inhibitor recognition specificity (Lee et al., 2015).
GRL-0617 is ineffective against MERS-CoV and in the study by Shin et al., it was discussed that this may be due to the presence of a threonine residue instead of tyrosine at a conserved position (Lee et al., 2015;Shin et al., 2020). In the SARS-CoV-2 and SARS-CoV PL pro sequences, the corresponding residues are Y268 and Y269, respectively (Shin et al., 2020). Y268 is required for the inhibitory effect of GRL-0617 and Shin et al. demonstrated that mutating this residue strongly reduces its potency (Shin et al., 2020). While GRL-0617 and 3k were predicted to bind to the MERS-CoV PL pro (4rf0) in this study, further docking to additional crystal structures may be required for comparison (Figures 1, 2). The naphthalene-based inhibitors were surrounded by the residues D165, Y269, and Q270 in the SARS-CoV PL pro . Ratia et al. and Báez-Santos et al., have also described the importance of these residues in the mechanisms of action of these ligands (Ratia et al., 2008;Báez-Santos et al., 2014).
Most notably, the dietary compounds (-)-epigallocatechin gallate, hypericin, rutin and cyanidin-3-O-glucoside were predicted to bind more strongly to the naphthalene-inhibitor site of the SARS-CoV-2, SARS-CoV, and MERS-CoV papainlike proteases than the known inhibitors (Figure 2 and Supplementary Figures 3-6). They also formed multiple interactions with the key protein residues compared to GRL-0617 and 3k. Similarly, the blind docking results on the main PL pro chains showed that these natural ligands had multiple poses within this region (Figure 3 and Supplementary Figures 1, 2). (-)-Epigallocatechin gallate, rutin and cyanidin-3-O-glucoside are flavonoids, a biologically active class of the phenolic compounds (Bonvino et al., 2018). In a recent literature review conducted by Verma et al., the flavonoids were found to be the largest class of compounds with potential activity against coronaviruses (Verma et al., 2020).
In 2005, Li et al. published a study about the antiviral activities of natural compounds against SARS-CoV (Li et al., 2005). Lycorine was identified as a potent antiviral compound and potentially a candidate for the development of new medicines (Li et al., 2005). Natural compounds have been screened for their ability to target SARS-CoV-2 proteins. This includes extracts of medicinal herbs and several studies have focused on their inhibitory effects on key proteins, such as the spike glycoprotein and the main protease (M pro ) (Mani et al., 2020;Pitsillou et al., 2020;Russo et al., 2020;Smith and Smith, 2020). In a recent paper by Alamri et al. a structured-based computational approach was utilized to identify compounds that may act as pan-PL pro inhibitors and could be developed further as antiviral agents (Alamri et al., in press). Given the current situation, in silico methods have made it possible for large libraries of existing approved compounds to be screened in a relatively fast manner (Ojha et al., 2020). The structures of the hits identified from these computational studies could be optimized as part of the drug discovery process (Ojha et al., 2020). In addition to synthetic pharmacological compounds, herbal constituents can be screened in the same manner and this method has been described in many papers (Bhowmik et al., 2020;Chikhale et al., 2020;Ghosh et al., 2020;Gupta et al., 2020;Jena et al., 2020;Krupanidhi et al., 2020;Muhseen et al., 2020;Sinha et al., 2020;Subbaiyan et al., 2020). A number of studies that can be found on the World Health Organization's International Clinical Trials Registry Platform also involve plant-based compounds, particularly flavonoids. Ratia et al. determined the crystal structure of SARS-CoV in complex with ubiquitin aldehyde and they described how this polypeptide interacts with the palm and fingers regions of PL pro (Ratia et al., 2014). They emphasized that a significant amount of the binding energy of ubiquitin can be attributed to its C-terminal residues (R72-G76) and that this portion of ubiquitin forms an extensive number of intermolecular hydrogen bonds with PL pro (Ratia et al., 2014). The results from their study also indicated that the SARS-CoV PL pro had a preference for K48-linked ubiquitin and ISG15, over K63-polyubiquitin chains and mono-ubiquitin (Ratia et al., 2014). Most notably, two recognition sites on the surface of PL pro were characterized and were defined as either SUb1 or SUb2 (Ratia et al., 2014).
In the SARS-CoV-2 PL pro complex, ubiquitin propargylamide sits on the same subdomains as the SARS-CoV structure (palm and fingers regions), and the C-terminus extends into the active site (Klemm et al., 2020). Like SARS-CoV, the SARS-CoV-2 PL pro was also found to have a second ubiquitin binding site (SUb2) that is important for the binding of polyubiquitin (K48diubiquitin) and ISG15 (Klemm et al., 2020). The MERS-CoV (space group P6 5 22) PL pro -ubiquitin complex was solved by Bailey-Elkin et al. and in their paper, they refer to this structure as the closed conformation since the fingers domain is shifted toward the ubiquitin (Bailey-Elkin et al., 2014).
When ubiquitin was present in the SARS-CoV-2, SARS-CoV and MERS-CoV PL pro structures, the results from Schrödinger and blind docking showed that compounds were displaced from the naphthalene-inhibitor binding pocket (Figures 3, 4 and Supplementary Figures 1, 2). Compared to the apo PL pro , hypericin was unable to produce molecular docking poses for the PL pro -ubiquitin complexes. Likewise, the protein-protein docking results with the ligands already bound to this region in PL pro revealed that ubiquitin was binding in different conformations and that the position of the C-terminus was altered (Figure 5). This suggests that the dietary compounds may be able to interfere with the deubiquitinase activity of PL pro and in terms of in silico methods, this can be evaluated further using molecular dynamics (MD) simulations.
A commercially available PL pro enzymatic assay was used to measure deubiquitinase activity using GRL-0617 as an internal positive control; in this specific assay GRL-0617 has been shown to have an IC 50 value of 1.7 µM for inhibition of PL pro deubiquitinase activity (BP Bioscience). Overall, our findings indicated inhibition with an order of potency of GRL-0617 and hypericin > rutin and cyanidin-3-O-glucoside > epigallocatechin gallate and epicatechin gallate >> cefotaxime. GRL-0617 and hypericin > rutin > cyanidin-3-O-glucoside and epicatechin gallate > cefotaxime and epigallocatechin gallate (Figure 6). The potent inhibition of PL pro deubiquitinase activity by hypericin which, at higher concentrations, was analogous to GRL-0617, is particularly encouraging. Hypericin is an anthraquinone derivative that can be found in the flowering plant Hypericum perforatum, which is also commonly known as St. John's Wort (Napoli et al., 2018). It has been identified as a lead compound for the SARS-CoV-2 spike protein and its antiviral properties have been the subject of numerous papers in the past (Jacobson et al., 2001;Shih et al., 2018;Chen et al., 2019). In addition to its antiviral effects, St. John's Wort is also being investigated for its antidepressant properties and synthetic hypericin (SGX301) has gained attention for its use as a photodynamic agent in the treatment of cutaneous T-cell lymphoma (Rook et al., 2010;Montoya et al., 2015;Apaydin et al., 2016).
While specific compounds were selected for use in this study, it would be important to expand this in the future to incorporate a greater number of phytochemicals that are present in various plant extracts. Network pharmacology is also being increasingly used in drug discovery and this systematic approach can assist with identifying potential protein targets and lead compounds, as well as understanding their mechanisms of action Pan H. D. et al., 2020). Nonetheless, the antiviral, antioxidant and anti-inflammatory properties of the compounds used in this study had been previously reported in the literature and were consequently suitable candidates. Molecular docking was used for virtual screening and although the scoring functions produced from docking aren't absolute binding energies, it allowed for predictions to be made about the protein-ligand interactions. Docking was performed using the Glide (XP) protocol of the Schrödinger Suite and in a study conducted by Wang et al., this was found to have a 90% success rate in identifying the correct binding poses of ligands (Wang et al., 2016). In this study, the inhibitory activities of the compounds were subsequently measured using an enzymatic activity assay. In silico tools are currently being utilized for the early stages of the drug discovery pipeline however, it is important to note that the pipeline involves multiple steps and is a time consuming process (Agostino et al., 2019). Potential drugs must be explored further in pre-clinical trials using a combination of techniques and clinical trials (Agostino et al., 2019).

CONCLUSION
Overall, on the basis of our in silico and in vitro evaluations, hypericin, rutin, and cyanidin-3-O-glucoside can be considered potential lead compounds. In particular, further clarification of the molecular mechanisms and antiviral properties of hypericin, which displayed high potency in the in vitro assay and favorable binding properties in the in silico studies, is warranted.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
TK and AH conceptualized the aims and methodology and were involved in supervision. TK was involved in the production of the first draft of the manuscript. EP performed data analysis, data curation, and was involved in production of the first draft of the manuscript. JL was involved in data analysis and curation and was involved in production of the first draft of the manuscript. KV performed formal data analysis and was involved in data curation. KL performed formal data analysis and validation. All authors contributed to editing and reviewing the manuscript.

ACKNOWLEDGMENTS
We would like to acknowledge intellectual and financial support by McCord Research (Iowa, USA). JL was supported by an Australian Government Research Training Program Scholarship. We are indebted to Alfonso Perez Escudero and the team at Crowdfight COVID-19 for enabling access to supercomputing facilities, and to Matthew Gasperetti and the team at Hypernet Labs; Galileo, for enabling cloud computing for this project. We thank the National Computing Infrastructure (NCI), and the Pawsey Supercomputing Centre in Australia (funded by the Australian Government). Further, we thank the Spartan High Performance Computing service (University of Melbourne), and the Partnership for Advanced Computing in Europe (PRACE) for awarding the access to Piz Daint, hosted at the Swiss National Supercomputing Centre (CSCS), Switzerland.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fchem. 2020.623971/full#supplementary-material Supplementary Table 3 | Blind docking of the naphthalene-based inhibitors and dietary compounds to the SARS-CoV-2 6w9c, 6wuu, 6wx4, and 7jrn PL pro crystal structures. No poses were found in the naphthalene inhibitor site for GRL-0617 and hypericin in the 6wuu structure.
Supplementary Table 4 | The GlideScores (kcal/mol) of the naphthalene-based inhibitors and dietary compounds for the SARS-CoV-2, SARS-CoV, and MERS-CoV crystal structures with ubiquitin are listed.
Supplementary Table 5 | The docking scores and RMSD values for the top-ranked PL pro -ubiquitin complexes from the HDOCK server are provided.