Facing COVID-19 via anti-inammatory mechanism of action: Molecular docking and Pharmacokinetic studies of six anti-inammatory compounds derived from Passiora edulis

SARS-CoV-2 is the causative agent of the COVID-19 disease. Pathophysiologically, high levels of proinammatory cytokines in the serum of SARS-CoV-2 patients are reported, which is so-called the cytokine storm. In this study, molecular docking calculations of six bioactive compounds from Passiora edulis with anti-inammatory activity in interaction with the main protease of SARS-CoV-2 were performed, and their pharmacokinetic properties were predicted. The results of their molecular simulations and the ADME-T proles of each ligand (Absorption, Distribution, Metabolism, Excretion and Toxicity) suggest their use as potential treatment for SARS-CoV-2. Among the six investigated compounds in which four avonoids and two alkaloids, the best docked ligands are quercetin (-8.2 kcal/mol), chrysin (-8.0 kcal/mol), kaempferol (-7.9 kcal/mol) and luteolin (-7.7 kcal/mol), both avonoids compounds. Their pharmacokinetic studies using SwissADME, preADMET and pkCSM Web servers establish the good ADMET prole for each ligand.


Introduction
Other than vaccine development, people around the world are waiting for the famous news from researchers: a molecule against the Severe Acute Respiratory Syndrome 2 (SARS-CoV-2) has been found.
The SARS-CoV-2 is the causative agent of the novel ß-Coronavirus (2019-nCoV) or the Corona Virus Disease (COVID-19) that is a pneumonia infection characterized by the hyperproduction of mainly proin ammatory cytokines (IL-1, IL-6, TNF-α, etc) [1]. In the active research of nding molecule that can treat COVID-19, two approaches are currently being used. One is to nd molecules that can be used as potential treatment against COVID-19 among several FDA-approved drugs [2][3], while the other is to identify from plants biodiversity potential inhibitors (phytochemicals) of SARS-CoV-2's main protease using molecular modeling approaches [4][5].
During molecular modeling approaches of nding potentials inhibitors, particular emphasis is placed on the signi cance of binding a nity of ligand-protein complexes and on their drug-likeness properties [4,6].
However, it should be mentioned that the biological activities of these molecules are as well very important [7]. Further, the most common trend is that an anti-COVID-19 molecule might be derived from a plant endowed with antiviral properties [8][9][10][11]. Nevertheless, what is abundantly clear in this moment is that the most important cause of COVID-19 related deaths is respiratory failure which is due to pneumonia (an acute in ammatory lung injury), which itself varies depending on the disease severity level, but also alveolar damage that can precipitate acute respiratory distress syndrome (ARDS) [12]. The innate immune response is then to produce pro-in ammatory cytokines and chemokines to contain and stop the infection. Pathophysiologically, previous studies have reported high levels of various cytokines (the socalled cytokine storm) and chemokines in the serum of SARS-CoV-2 patients [13][14]. In addition, Fidan and Aydoğdu recently reported that various pro-in ammatory cytokines such as IL-6, IL-1, the tumor necrosis factor (TNF-α) induce a migration of leukocytes into lungs, that then secrete the reactive oxygen species and proteases that damage capillary endothelium and alveolar epithelium [15].
Based on the relevant clinical characteristics, phytocompounds derived from Passi ora edulis whose isolated molecules have several therapeutic properties such as anti-in ammatory, antioxidant, antimicrobial, anti-cancer… can be used for the treatment of COVID-19 as supported in the following lines: Passi ora edulis (P. edulis), also known as passion fruit ( 1), exhibits potential effects for the treatment of in ammation. Several mechanisms, including the inhibition of proin ammatory cytokines: TNF-α and IL-1ß levels, enzyme: myeloperoxidase (MPO) and mediators: bradykinin, histamine, substance P, ni tric oxide (NO) release and/or action, appear to account for Passi ora edulis's actions. Interestingly, in a comparative study, Montanher et al. found that Passi ora edulis was more effective than dexamethasone (0.5 in inhibiting both MPO and NO levels) [16]. This latter, which is considered as an important steroidal anti-in ammatory drug, might hold the promise for the treatment of COVID-19 as recently reported by Ledford [17]. Cazarin and co-authors reported in 2015 the antiin ammatory activity of P. edulis leaves [18]. In a dextran sodium phosphate caused mice colitis model, P. edulis peel our was found to reduce TNF-α, IL-1ß, IL-6, IL-12, and IL-17 [19]. Molecules responsible for this effect could be compounds like C-glycosyl avonoids vicenin, orientin, chrysin, vitexin and kaempferol [20]. Finally, Harmol and harmine, two uorescent harmala alkaloids showed anti-in ammatory activity by signi cantly inhibiting the NF-kB signaling pathway [21][22].
With regards to the reactive oxygen species that are secreted by leucocytes, several studies highlighted the antioxidant activity of edulis fruit and leaf which can eliminate free radicals or inhibit the activity of free radicals [23][24].
Aqueous and ethanolic leaves extracts have shown in vitro effect on some viruses species including Herpes Simplex Virus Type 1 and 2, Varicella-Zoster Virus, etc. [25]. Bibliographical references were made using a bibliographical software "Mendeley".

Molecular Docking
The structure of the 3-Chymotrypsin-Like protease (3CLpro) or the COVID-19 virus main protease (M pro ) which is among the most studied SARS-CoV-2 proteases was obtained from PDB (Protein Data Bank) database (PDB ID: 2GTB) and imported into chimera for visualizing the binding domain of the complex and identifying the amino acids in the binding pocket as well. The hydrogen atoms were added to the protein in order to correct the ionization and tautomeric states of the amino acid residues. Furthermore, the water molecules and complexes bound to receptor molecule were removed before the docking.
Incomplete side chains were replaced using Drunbrack rotamer library. In addition, the protein was subjected to energy minimization by applying the AMBER 14SB force eld, and AM1-BCC was used for other residues with a maximum number of 200 steps at RMS gradient of 0.02. The optimized protein was saved in pdbqt format and imported to PyRx for molecular docking which was carried out by means of Autodock Vina virtual screening tool [26]. The validation of the docking study was performed by redocking the reference ligand into an appropriate protein cavity. Re-docking is accepted if the root mean square value (RMSD) < 2.0 Å. Figure 2 displays schematic structure of the SARS-CoV-2 M pro /3CLpro (a) and the complex formed between the SARS-CoV-2 M pro and 2GTB as a potential drug target for the new coronavirus-2 (b). According to Xu and co-workers, 2GTB is the main protease found in the coronavirus associated with the severe acute respiratory syndrome (SARS), and that the main protease in 2019-nCoV shares 96% similarity with that in SARS [27].

Generation of ligand dataset and pharmacokinetic pro les
The selected compounds derivatives from various literature resources [20,28] were drawn using Marvin JS. Figure 3 shows the 3D structures of the sketched compounds retrieved from PubChem/NLM. The 3D ligands were then saved in .sdf format. Ligands optimization was performed by using universal force eld (UFF) with conjugate gradients algorithm of 200 Steps, and then analyzed for pharmacokinetic properties. Bioinformatics resources have been employed in the prediction of ADME properties (Absorption, Distribution, Metabolism and Excretion) using the SwissADME database [29]. During the early stages of drug discovery, the ligand to be selected as a hit must be non-carcinogenic and non-hepatotoxic. The toxicity assessment (ADMET, T for Toxicity) that allows to predict the mutagenicity (Ames test) and carcinogenicity of the potential ligands was made using the preADMET server, Korea [30], while the hepatotoxicity and the oral rat acute toxicity were assessed using the pkCSM server [31].

Energetics and geometries
Noncovalent interactions, mainly H-bonds [32], van der Waals and π-π interactions (stacked/parallel and Tshaped/perpendicular conformations) [33] are forces that drive and determine the binding of ligand-protein interactions. The most common tool to evaluate the strength of binding between ligand-protein interactions is molecular docking. The docking results obtained using AutoDock Vina virtual screening tool between ligands 1-6, the native or reference ligand with the SARS-CoV-2's main protease (M pro or 3CLpro) are gathered in Table 1. Turning next to the types of noncovalent interactions established between ligands and the SARS-CoV-2 M pro , one can see in gure 4 below that the complexes are mainly stabilized by hydrogen bonding interactions, but also supported by van der walls and π/π interactions. At this stage, the stability of ligands 1-4 which are avonoids compounds over ligands 5 and 6 can be explained by the presence of multiple OH groups that can be act simultaneously as hydrogen bonds acceptors (HBA) and donors (HBD) [35].
In addition, the presence of three aromatic rings in avonoids compounds offer much possibilities to π-π interactions to take place. Such interactions are mainly stabilized by dispersion or van der Waals forces, important in the ligand-protein interactions [33].
H-bonds parameters (distances and angles) between the protein target and ligands 1-6 along with the involved groups (ligands) and the amino acids residues of the M pro engage in H-bonding interaction are summarized in Table 2.

Physicochemical properties and ADME-T pro les
Physicochemical property is an important parameter of a molecule that in uences e cacy, safety or metabolism which could be predicted by using Lipinski's rule of ve (RO5) that is: molecular mass < 500; Hydrogen-bond donors (HBD) ≤ 5; Hydrogen-bond acceptors (HBA) < 10; and Log P < 5 [36]. Prediction of in silico physicochemical parameters of the 6 ligands are grouped in Table 3. Table 3 shows that all ligands meet every single criterion of Lipinski's rule of ve and thus fully obey the rule. Consequently, all the investigated ligands are predicted to be easily absorbed and have good permeability and bioavailability. According to Ghose and co-workers, the molecular refractivity is a ubiquitous parameter for a drug molecule that cannot exceed 130 m 3 .mol -1 and not to be under 40 m 3 .mol -1 [37]. None violation is observed here for all the investigated ligands as can be seen in Table 3. Finally, as pointed out by Cerqueira and co-authors, for optimal drug absorption and distribution, the polar surface area (PSA) values cannot be higher than 140 Å [38]. Once again, none violation is observed here. Three potential candidates for the inhibition of the SARS-CoV-2 3CLpro have PSA values almost two to three times less than the recommended value (ligands 1, 5 and 6), while ligands 2, 3 and 4 have PSA value higher than 100 Å.

Inspection of
The next step to deal with is to establish the ADME/T pro les of each ligand. In fact, a major issue after identifying stables complexes, that is, lead or hit compounds, is to evaluate their ADME parameters and cardiotoxicity.
These pharmacokinetic properties are very important parameters in the computer-aided drug discovery since they allow one to retract some hit from early-stage trials. The ADME properties are evaluated by using SwissADME and pkCSM servers, but other parameters such as the Blood-Brain Barrier (BBB), the Human Intestine Absorption (HIA) and the skin permeability come from the preADMET server. The selected endpoints for toxicity are Ames test and Rodent Carcinogenicity (rat) in preADMET server, hepatotoxicity and oral rat acute toxicity (LD 50 ) in pkCSM server. These parameters are gathered in Table 4. According to the binding a nity values, it was derived the following decreasing order in the complexes formed between ligands and the SARS-CoV-2 3CLpro or Mpro: Ligand 4 > Ligand 1 > Ligand 2 > Ligand 3> Ligand 5 > Ligand 6. This order only re ects the thermodynamic stability of complexes. However, the stability over time of the ligand in a protein interaction site depends on other factors. For a ligand to be used for therapeutic purposes, its absorption, distribution, metabolism, excretion and toxicity are aspects to take into account. To pursue further, it is worthy to point out that rst of all, such a ligand must be nonhepatotoxic and non-carcinogenic [39].
Scrutiny of toxicity outcomes in Table 4 reveals that all potential ligands are non-hepatotoxic. With regards to the carcinogenicity, the results predict the carcinogenic activity only for the ligand 2. This encouraging result of toxicity assessment allows us to go back to ADME properties. The ability of a drug molecule to cross into the brain is an important propriety to improve the e cacy of drugs (reduce side cm/h. Finally, the bioavailability score which is evaluated to 0.55 con rms that ligands 1-6 have good absorption and distribution since all potential candidates may have more than 10% of bioavailability in rat [40].
Interestingly, the three best candidates according to their binding a nity are found to be non-inhibitors of CYP2D6 and CYP3A4 except hits 1 and 6 that affect the CYP3A4, and hit 6 which in addition affects the CYP2D6. This result rules the ligand 6 out from the list of potential candidates for the inhibition of the SARS-CoV-2 main protease, a result which is moreover in good agreement with its lower free enthalpy (-6.40 kcal/mol).
Turning next to excretion also called elimination, the total clearance is directly linked to the renal OCT2 activities. Passi ora edulis, also known as passion fruit, passion ower, purple granadilla, or "maracuja" (DR Congo or Brazil), is widely cultivated for its edible fruit. With numerous biological activities, the passion fruit also contains vitamins A, C, E, K, minerals such as Zn (0.10 g), Mg (29 mg), K (348 mg), Ca (12 mg), etc [41]. On one hand, vitamins A and K could help to ght the COVID-19 [42]. On the other hand, these chemical elements mainly Zn, although indispensable as enzymatic co-factors, a slight increase in their intracellular concentration inhibits the replication of retroviruses including SARS-CoV-1 [43] important in the management of COVID-19. Owing to its numerous therapeutic activities, P. edulis has traditional or ethno medicinal uses in many countries. Of complex phytochemistry, its secondary metabolites have numerous health bene ts and very recently, as stated above, Jabareen and co-workers reported in an experimental study the antiviral activity of P. edulis leaves on some viruses' species including Herpes Simplex Virus Type 1 and 2, Varicella-Zoster Virus, etc [25]. Since P. edulis exhibits anti-in ammatory capabilities essential to stem the cytokine storm, this study is conducted in order to identify potential inhibitors from a set of 6 phytochemicals endowed with anti-in ammatory activity. The 6 selected compounds (four avonoids and two alkaloids) reacted with the SARS-CoV-2 3CLpro or Mpro, and an order of thermodynamic stability was obtained.
Compared with Lopinavir and Nel navir that are protease inhibitors recommended for the treatment of SARS and MERS, the binding a nities of the four top compounds, both avonoids, are very close to those of two anti-HIV drugs, and even a bit higher to that of the ligand reference. Indeed, the ligand 1 or chrysin has anti-in ammatory, antibacterial and antioxidant activities [44][45], while ligand 2 or kaempferol exhibits antitumor, antioxidant and anti-in ammatory capabilities [46]. In fact, our previous studies showed that aloe vera represents potential treatment for COVID-19 [47], and three of its phytochemicals were identi ed as potential inhibitors of SARS-CoV-2 main protease, in which two compounds exhibit antiin ammatory effect [5]. The anti-in ammatory activity of luteolin in experimental animal models was reported by Ziyan and co-workers [48], and recently, a study by Lesjak and co-authors showed antioxidant and anti-in ammatory activities of quercetin and its derivatives [49].
One can remember that the hyperproduction of proin ammatory cytokines is the main reason that causes morbidity and mortality in SARS-CoV-2 patients. Thus, the application of anti-in ammatory molecules is a mechanistically-sound strategy for treatment development [50]. The established ADME-T pro le of each ligand suggests that the four avonoids (chrysin, kaempferol, luteolin and quercetin) might be used as potential treatment of SARS-CoV-2.
In conclusion, the strategy adopted in this work consisted in exploring the inhibitory power of six phytochemicals derived from P. edulis. Given the cytokine storm, six compounds exhibiting antiin ammatory activity, among which four avonoids and two alkaloids, were each paired with the SARS-CoV-2 main protease in order to evaluate rst their thermodynamic stability. The docking a nity scores showed that ligands 1-4 ( avonoids) are more stables than ligands 5 and 4 (alkaloids). Then, the Lipinski's rule of ve and the pharmacokinetic studies using SwissADME, preADMET and pkCSM showed that these phytochemicals have good ADME-T pro les, mainly the avonoids compounds. Consequently, chrysin, kaempferol, luteolin and quercetin as the four top compounds can be used via anti-in ammatory mechanism of action to ght the overproduction of proin ammatory cytokines in SARS-CoV-2 patients.

Declarations
Declaration of Competing Interest The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to in uence the work reported in this paper.