In Silico Investigation on the Interaction of Chiral Phytochemicals from Opuntia ﬁcus-indica with SARS-CoV-2 M

: Opuntia ﬁcus-indica is a cactaceous plant native to America but, nowadays, widely found worldwide, having been the most common domesticated species of cactus grown as a crop plant in semiarid and arid parts of the globe, including several Mediterranean basin countries. Opuntia ﬁcus-indica can be regarded as a medicinal plant, being source of numerous bioactive phytochemicals such as vitamins, polyphenols, and amino acids. The urgent need for therapeutic treatments for the COronaVIrus Disease 19 (COVID-19), caused by the Severe Acute Respiratory Syndrome (SARS)-Coronavirus (CoV)-2, justiﬁes the great attention currently being paid not only to repurposed antiviral drugs, but also to natural products and herbal medications. In this context, the anti-COVID-19 utility of Opuntia ﬁcus-indica as source of potential antiviral drugs was investigated in this work on the basis of the activity of some of its phytochemical constituents. The antiviral potential was evaluated in silico in docking experiments with M pro , i.e., the main protease of SARS-CoV-2, that is one of the most investigated protein targets of therapeutic strategies for COVID-19. By using two web-based molecular docking programs (1-Click Mcule and COVID-19 Docking Server), we found, for several ﬂavonols and ﬂavonol glucosides isolated from Opuntia ﬁcus-indica, good binding afﬁnities for M pro , and in particular, binding energies lower than − 7.0 kcal/mol were predicted for astragalin, isorhamnetin, isorhamnetin 3- O -glucoside, 3- O -caffeoyl quinic acid, and quercetin 5,4 (cid:48) -dimethyl ether. Among these compounds, the chiral compound astragalin showed in our in silico studies the highest afﬁnity for M pro ( − 8.7 kcal/mol) and also a low toxicity proﬁle, emerging, thus, as an interesting protease inhibitor candidate for anti-COVID-19


Introduction
Opuntia ficus-indica, also known as Indian fig and nopal cactus [1], is a dicotyledonous angiosperm cactaceous plant widespread worldwide in tropical and subtropical regions, including those surrounding the Mediterranean sea, with semi-arid and arid climates [2]. Recently, Opuntia ficus-indica's nutritional and health benefits have been suggested due to its phytochemical composition particularly rich in polyphenols, which are the main compounds responsible for antioxidant properties of plant extracts [3], but also in vitamins and amino acids [3][4][5][6]. Opuntia ficus-indica is a fast-growing and productive plant [7,8] with potential utility in biogas [9] and biofuel [10] production, rehabilitation of degraded soils, and desertification/climate change mitigation, with important consequent environmental benefits [11][12][13]. Fruits and cladodes are widely used for human nutrition [14][15][16], with the latter also largely being used as livestock forage in South America and Africa [17,18]. [14][15][16], with the latter also largely being used as livestock forage in South America and Africa [17,18].
Three deadly RNA viruses of the family of Coronaviridae have emerged in the last eighteen years: SARS-CoV-1 (also known simply as SARS-CoV), MERS-CoV, and SARS-CoV-2. These coronaviruses (CoVs) can cause severe pneumonia in infected patients that can be lethal in several cases. SARS-CoV-2 is causing the current pandemic of COronaVIrus Disease 19  which has provoked enormous sanitary as well social and economic impacts on the globe [33]. Prophylactic [34] and therapeutic strategies are currently under investigation worldwide, and among the latter, old antiviral drug repurposing [35,36] is one of the main ways to respond to the urgency of an effective anti-COVID-19 therapy. Drug repurposing, in fact, involves drugs already known and in use for other pathologies and for which safety has already been assessed previously [35]. Similarly, plant extracts or isolated compounds already used as dietary supplements or ingredients of traditional medicine are attracting much interest in the therapy of SARS-CoV-2 [37,38].
In this context, considering the previously ascertained utility of Opuntia ficus-indica plant extracts or isolated compounds as antiviral and anti-pneumonia agents, the potential of some of the same identified molecules was tested in the present in silico study as inhibitors of M pro (Figure 1). In fact, the main protease of SARS-CoV-2 represents one of the most attractive targets for COVID-19 drug development due to the absence of this type of proteins in humans and its essential role in viral replication [39].

Molecular Docking Studies
For the computational studies, we used the 3D structure of M pro with an unliganded active site (PDB ID: 6Y84), downloaded from the Protein Data Bank [40]. The structures for the natural compounds ( Figure 2) investigated as protease ligands were retrieved by us from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/, all links mentioned in this work were accessed on 11 December 2020), whilst those not present in PubChem were drawn with the molecular editor of 1-Click Mcule (Mcule Inc., Palo Alto, CA, USA) [41][42][43][44], an online docking platform making use of AutoDock Vina [45], by which the docking experiments herein described were performed. For the molecular dockings, we used, as atomic coordinates of the binding site, X: 9.204, Y: −4.557, and Z: 19.602, which were previously reported in the literature for the binding center [46], and we set the size of the binding site as 22 Angstrom. The docking poses with the most negative docking scores (kcal/mol), corresponding to the highest binding affinities, were selected for further analysis (Table 1). To validate the docking method, we applied it to other literature dockings having M pro as a protein target [47], and used the N3 inhibitor as a reference compound ( Figure 2), finding that the resulting scores were in line with those previously reported for ligands like the phenolic diterpene carnosol [46], bictegravir, and other molecules docked to the same protease [48]. We made use of the molecular graphics program embedded in 1-Clik Mcule for structural visualization of protein-ligand complexes and to obtain the snapshots of Figure 3A,B, while we obtained the protein-ligand interaction diagrams reported in the same figure ( Figure 3C,D) and in Figure 4 by ProteinsPlus (https://proteins.plus/) or PLIP (Protein-Ligand Interaction Profiler, https://plip-tool.biotec.tu-dresden.de/) [49,50]. For the N3 inhibitor/astragalin-bound M pro structural comparison ( Figure 3E), we made use of the matchmaker tool embedded in UCSF Chimera [51]. To this scope, we compared the structure with PDB ID: 6LU7 (containing the N3 inhibitor) and the best pose obtained from the docking of M pro with astragalin.

COVID-19 Docking Server Studies
The docking of astragalin with M pro was repeated with the COVID-19 Docking Server [52], a user-friendly platform realized by Kong et al. for docking peptides and antibodies, but also small organic compounds, against some of the main COVID-19-related proteins investigated by implementation of Autodock Vina and CoDockPP [45,[52][53][54][55]. The ligand file in mol2 format was obtained, starting from informative details on astragalin found in Pubchem (https://pubchem.ncbi.nlm.nih.gov, PubChem CID 5282102) and using the online tool available at the link: http://pasilla.health.unm.edu/tomcat/ biocomp/convert. We selected, in the COVID-19 Docking Server [52], (for ligand preparation, PDB IDs and description of protein targets, scoring and docking procedures see https://ncov.schanglab.org.cn/index.php) as 'Computational Type': 1-molecule Docking and as coronavirus protein target: 'Main protease'. After the docking run, the server provided the pose views for the different binding modes and the corresponding binding energy (score value, kcal/mol) and scoring function (RF-Score, pKd) values. The server used as binding center coordinates: x = −10.85; y = 12.58; z = 68.72 and as target the structure: 6LU7, edited by the developers of the COVID-19 Docking server (the inhibitor N3 [56] was removed). Astragalin-M pro complex was visualized in the structure viewer of the COVID-19 Docking Server and UCSF Chimera software [51]. Listed values of score values and RF score values corresponding to the top 10 poses for the complex in Table 2 were obtained as output of the docking experiment performed by the COVID-19 Docking Server (and they were found at https://ncov.schanglab.org.cn/dojmol.php?dir=202012310146538026).
To analyze the plant-derived compounds of the present work, we submitted to the PAINS Remover server their structure data as '.sdf' or '.mol' format. Toxicity properties of astragalin were predicted with ADMETlab (http://admet.scbdd.com/calcpre/calc_cf_ single_mol) using the isomeric SMILES format for the compound found in the PubChem database:

Results and Discussion
M pro (also called 3CL pro ) protease is considered an attractive protein target in COVID-19 drug discovery and, in particular, for the screening of compounds that could inhibit the replication of coronaviruses [39]. The SARS-CoV-2 protease shares more than 95% identity with SARS-CoV-1 M pro [59] and has a fundamental role in the life cycle of coronaviruses. No similar homologues are present in humans, rendering it a first-choice target in selective anti-COVID-19 therapy, and noteworthily, some of the repurposed drugs [35] used for COVID-19 therapy are protease inhibitors [60]. Based on the prominent role of M pro as a SARS-CoV-2 target, 10 phytochemicals ( Figure 2) identified in Opuntia ficus-indica were docked to the M pro structure with Mcule-1-Click [41,44], a docking server powered by Autodock Vina [45].
Five of the studied phytochemicals (from two phytochemical families: polyphenol acids and flavonoids) were endowed with a high affinity towards M pro , with binding energy values lower than −7.0 kcal/mol associated with the top ranked poses, as reported in Table 1.
In particular, astragalin showed, in silico, the highest affinity score (with a binding energy of −7.9 kcal/mol computed for the top scoring pose, Table 1) and, thus, the most stable complex for M pro . The predicted stability was also higher than that previously reported for the protease N3 inhibitor [61]. Interestingly, astragalin showed higher affinity than both commercial anti-COVID-19 drugs, like chloroquine, and natural molecules supposed to fight against the SARS-CoV-2 infection, like artemisinin and hesperidin, that showed binding energies ranging from −4.1 to −5.8 kcal/mol as emerged from previous computational studies on M pro inhibitors [62]. Other good binding energy scores were predicted for isorhamnetin, quercetin 5,4 -dimethyl ether, and isorhamnetin 3-O-glucoside, as well as for 3-O-caffeoyl quinic acid (−7.1 kcal/mol, for the top ranked pose).    We examined the astragalin-protease complex structure for the pose with the lowest binding energy by the ProteinPlus software and noticed that the plant compound was involved in multiple H-bonding, with residues Phe140 and Glu166 of the viral protein, two main amino acid residues also involved in the interaction of N3 inhibitor with the same protease target [56,65]. Interestingly, these two residues were also previously associated with the interaction of other molecules investigated as anti-COVID-19 drugs and endowed with M pro inhibitory activity [66,67].
In our prediction, isorhamnetin formed H-bonds with three protease residues, while its glucoside form interacted with six amino acids involving three additional H-bonds, as we expected for its more highly hydroxylated nature, and the hydrophobic interaction with Thr25 ( Figure 4C,D).
Comparing isorhamnetin and quercetin 5,4 -dimethyl ether, the latter was involved in a hydrophobic interaction with protease His41, while the former, with an additional phenolic OH, showed a higher propensity for H-bonding ( Figure 4B,D). Nevertheless, the overall binding energies (BE for the top-ranked poses and BE-average scores) that emerged from our simulation for the complexes formed by these two compounds with M pro were quite similar, as shown in Table 1.
two main amino acid residues also involved in the interaction of N3 inhibitor with the same protease target [56,65]. Interestingly, these two residues were also previously associated with the interaction of other molecules investigated as anti-COVID-19 drugs and endowed with M pro inhibitory activity [66,67].
However, the higher affinity for M pro predicted for astragalin with respect to the other natural ligands could not be fully explained by the only two H-bonds evidenced by ProteinPlus, and thus, we decided to investigate this aspect in more detail, using the program PLIP (Protein-Ligand Interaction Profiler, https://plip-tool.biotec.tu-dresden.de/), for better describing the interactions occurring between the protease binding site amino acids and the ligand.
Additional H-bondings with His163 and Gln189, as well as hydrophobic interactions with Thr25 and Glu166, were revealed for astragalin in a complex with M pro by PLIP ( Figure 3D). We superimposed and aligned the protein complex structures containing the N3 inhibitor (PDB ID: 6LU7) and astragalin (using the best ranked pose for the docking of the protease with the natural compound). By complex structures comparison, we observed that the sugar moiety of astragalin, whose H-bonding with Gln189 was evidenced by PLIP software, was found in the spatial region where the two Val3-involving amide bonds of N3 inhibitor were placed ( Figure 3E), whilst the aromatic flavonol moieties lay in the region occupied by the hydrophobic/aromatic 2-oxo-3-pyrrolidinyl and benzyl moieties of the N3 inhibitor ( Figure 3E), where they formed hydrophobic interactions with Thr25 and Glu166, as well as H-bonding with His163 ( Figure 3D).
The docking experiment with astragalin was repeated using COVID-19 Docking Server [52], a web platform for docking different classes of ligands to SARS-CoV-2-associated protein targets. The dockings on this web-based server were performed by us choosing the exhaustiveness level as 12, with a higher level of this parameter being considered to provide more precise docking results [52]. This web-based docking platform has been carefully validated by the developers to evaluate their docking protocols, by conducting several re-dock experiments for those targets whose experimental complex structures were available. Most of the complex structures, including that of the M pro with N3 inhibitor, were reproduced by using the docking procedure on the web server (Tables S1 and S2 in Supplementary Information of [52]).
The main results of this investigation are reported in Table 2. Interestingly, astragalin showed a high affinity for the selected COVID-19 protein target, with binding energy scores ranging from −6.90 to −8.70 kcal/mol and RF (pKd) scores varying between 5.40 and 6.03. The binding affinity predicted for the top ranked pose of the compound-M pro complex (−8.7 kcal/mol) was higher than that (−7.9 kcal/mol) indicated by 1-Click Mcule (Table 1) [44], whilst the structure of the complex ( Figure 5A,B) resembled only partially that shown in 1-Click Mcule structure viewer. In the complex structure predicted by COVID-19 Docking Server the sugar moiety replaced the astragalin phenol in the proximity of the region of the β-sheet made of His163-Glu166, belonging to the catalytic pocket ( Figure 5A,B).
PAINS in screening libraries. [58] Out of the 10 evaluated compounds, only 3-O-caffeoyl quinic acid and caffeic acid were filtered out by the PAINS Remover and by SwissADME [57] (Figure 2), indicating no potential PAINS in the majority of the investigated Opuntia phytochemicals, including astragalin, the leading compound that emerged from our study ( Figure 2).
The last aspect examined in the present computational study was the potential hepa- Figure 5. Top ranked pose views (A) for the interaction between astragalin and M pro , as predicted by the COVID-19 Docking Server [52]. Comparison of the complex structures predicted for astragalin by 1-Click (B, left) [44] and the COVID-19 Docking Server (B, right) [52], as visualized with UCSF Chimera [51].
We identified pan-assay interference compounds (PAINS) in our collection of Opuntiaderived compounds to exclude the molecules with promiscuous behavior that, due to their chemical structures, could lead to false positive results in different biological screening assays. [69] To this scope, we used the PAINS Remover, a software able to identify PAINS in screening libraries. [58] Out of the 10 evaluated compounds, only 3-O-caffeoyl quinic acid and caffeic acid were filtered out by the PAINS Remover and by SwissADME [57] (Figure 2), indicating no potential PAINS in the majority of the investigated Opuntia phytochemicals, including astragalin, the leading compound that emerged from our study (Figure 2).
The last aspect examined in the present computational study was the potential hepatotoxicity of the most promising Opuntia-derived compound emerging from our work, together with other toxicity properties.
As can be seen in Table 3, there was no significant hepatotoxic effect predicted for astragalin. This is in accord with what has been reported previously on this flavonol glycoside, including the astragalin hepatoprotective effect [70], and suggests that this compound could be well-tolerated by the liver in a hypothetic therapeutic approach. Moreover, no drug-induced mutagenicity nor any important skin sensitization was predicted (Table 3). Table 3. Toxicity properties of astragalin predicted by ADMETlab. Notice the low toxic potential of the Opuntia-derived compound under investigation. The findings of this study are interesting, opening the way to further in silico and in vitro investigations on astragalin and, more generally, on the flavonol glycosides with similar structures extracted from Opuntia spp. for anti-COVID-19 therapeutic approaches.

Conclusions
We examined 10 compounds discovered in Opuntia ficus-indica, a typical cactus plant of common dietary and phytotherapeutic use, and found that five of them showed a good predicted affinity for the main protease of SARS-CoV-2, a first choice target in the COVID-19 drug discovery. Among these phytochemicals, the chiral astragalin emerged as the most promising potential inhibitor of M pro . This compound showed binding energy scores of −7.9 kcal/mol and −8.7 kcal/mol (corresponding to the top scoring poses obtained by 1-Click Mcule [44] and the COVID-19 Docking Server [52], respectively), higher than the M pro N3 inhibitor, revealing a strong interaction with the protease target. Overall, this chiral compound interacted with M pro amino acids like Phe140, His163, Glu166, and Gln189 by H-bonding and with Thr25 and Glu166 by hydrophobic interactions. Interestingly, these are also residues involved in the M pro binding of the N3 inhibitor [56]. Moreover, our computational studies suggested that Opuntia-derived flavonol glycosides are interesting derivatives with low toxic potential. These natural products show good binding affinities for M pro , and their direct use or chemical derivatization could lead to potentially effective treatments for COVID-19. Despite a predicted affinity for M pro lower than astragalin, quercetin 5,4 -dimethyl ether, a phytochemical isolated not only from Opuntia ficus-indica, but also from other plants like Rhododendron ellipticum [71], used as folk medicine to suppress cough [72], was an interesting candidate as an M pro inhibitor that was still endowed with a good binding affinity (−7.3 kcal/mol predicted for the top scoring pose). Since Autodock Vina software tends to underestimate the binding affinity of the ligands for the target [73], our complex could be endowed with an even lower binding energy and, thus, a higher affinity than that computed. In other terms, our study could inspire future clinical trials using some of the several commercially available Opuntia ficus-indica dietary supplements to ameliorate health conditions of COVID-19 patients, and from a molecular perspective, it could serve as a starting point for experimental investigations directed towards characterizing the above ligand-protease interactions and the in vitro/in vivo anti-SARS-CoV-2 effects of the different compounds extracted from Opuntia ficus-indica and from other Cactaceae family members.