Screening of honey bee pollen constituents against COVID-19: an emerging hot spot in targeting SARS-CoV-2-ACE-2 interaction

Abstract The attachment of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike to angiotensin-converting enzyme 2 (ACE-2) leads the cell fusion process, so spike blockade may be a promising therapy combating COVID-19. Bee pollen bioflavonoids with intrinsic bioactivities are of outmost importance to block SARS-CoV-2-ACE-2 interaction. Herein, we conducted a molecular docking assessment through natural phenolics/non-phenolics of pollen to investigate their affinity against SARS-CoV-2 spike. Finally, kaempferol 3-neohesperidoside 7-O-rhamnoside (compound a), quercetin 7-rhamnoside (compound b), delphinidin-3-O-(6-p-coumaroyl) glucoside (compound c), and luteolin-7-O-6″-malonylglucoside (compound d) showed the lowest binding affinity of −8.1, −7.7, −7.3 and −6.7 kcal/mol. The docking procedure was validated using protein-protein interactions between ACE-2 and SARS-CoV-2 RBD via HADDOCK webserver. MD simulations were fulfilled to investigate different ligands’ effects on protein movements. Collectively, compound a may possess the potency to disturb the binding of SARS-CoV-2 spike-ACE-2, which can be on the call for further in vitro and in vivo study to investigate its antiviral potential against SARS-CoV-2. Graphical Abstract


Introduction
The current deadly pandemic of COVID-19 infected millions of people globally (Ghosh et al. 2021), affecting all walks of life and left the world in lockdown (Lu et al. 2020, Zarei et al. 2020. So far, virtual screening techniques like molecular docking may improve information about the bioactive compounds' interaction with emerging viral diseases (Zhou et al. 2008). SARS-CoV-2 belongs to the beta coronavirus family (de Wit et al. 2016, Zarei et al. 2021, with >93% genetic sequence identity with bat coronavirus (RaTG13) (Lau et al. 2007), and its spike comprises S1 and S2 subunits. S1 attaches to ACE-2, and S2 facilitates the cell fusion process (Gui et al. 2017). S1 encompassing N-and C-terminal in which the residues from 319 to 529 make RBD (Receptor Binding Domain) of SARS-CoV-2 (Supplementary material Figures S1 and S2) (Walls et al. 2020). Among which, residues between 437 and 507 are the vital ones for attachment to ACE-2 (Supplementary material Figure S3) (Zhang et al. 2020), so SARS-CoV-2-ACE-2 interaction may trigger in silico investigation of bioactive compounds against SARS-CoV-2 (Wang et al. 2022). Honey bee pollen's natural compounds like delphinidin, kaempferol, luteolin, quercetin and their derivatives may earn therapeutic effects against SARS-CoV-2 due to their high bioactivities (Al Naggar et al. 2021). Kaempferol and its derivatives delineated possible effect on viral diseases e.g. SARS-CoV-2 (Adhikari et al. 2021), herpes simplex virus 1 (HSV-1) (Zhu et al. 2018), and japanese encephalitis virus (JEV) (Care et al. 2020). Delphinidin is a suitable suppressor of breast cancer (Han et al. 2019) and showed competent antiviral potency against zika virus (ZIKV), west nile virus (WNV), and Dengue virus (DENV) (V azquez-Calvo et al. 2017). Luteolin exhibited potent antiviral activity against JEV (Fan et al. 2016), EBV and HIV-1 (Wu et al. 2016). Quercetin and its derivatives represented potent biological activities such as antimicrobial, anti-inflammatory, antidiabetic, and antioxidant properties (Salehi et al. 2020). These scientific documents guaranteed high bioactivities of pollen natural compounds (Table S1 Supplementary material) and fuelled us to screen its bioactive phenolics/non-phenolics as inhibitors to check out their affinity against the active site of the SARS-CoV-2 spike RBD. Those natural inhibitors with the highest binding affinities are selected to investigate their effect on SARS-CoV-2 Spike dynamic using MD (Molecular Dynamics) simulations.

Docking validation
The docking process was validated through the HADDOCK web service (Dominguez et al. 2003), using PDB code 6M0J, the chain E (SARS-CoV-2 Spike) and chain A (ACE-2) were extracted, re-docked with ACE-2 in HADDOCK and then simulated file of SARS-CoV2 S was aligned onto the initial screened SARS-CoV2 S using UCSF Chimera v 1.8.1 (Supplementary material Figure S4).

Molecular docking results
Molecular docking is a powerful approach to test the binding affinity of ligands with the active site of proteins (Meng et al. 2011). This study seeks to block SARS-CoV-2 spike: ACE-2 interaction to ruin cell entry process of the viral RNA using phenolics/ nonphenolics in pollen through Vina, and the ligands with the lowest binding energies are charted in Tables S2, S3 Supplementary material. Four natural compounds named Kaempferol 3-neohesperidoside 7-O-rhamnoside (compound a), and Quercetin 7-rhamnoside (compound b), Delphinidin-3-O-(6-p-coumaroyl) glucoside (compound c), and Luteolin-7-O-6 00 -malonylglucoside (compound d) have exhibited the lowest binding energies of À8.1. À7.7, À7.3, and À6.7 kcal/mol, respectively. Kaempferol and its analogous (compound a) shown a robust antiviral potential against bovine herpesvirus 1 (BoHV-1) (Zhu, Wang, Yuan and Zhu 2018) H1N1, H3N2, and H5N1 (Lee et al. 2016). Compound b also displayed dominant antiviral potency against porcine epidemic diarrhea virus (PEDV) replication with a 50% inhibitory concentration IC 50 of 0.014 mg/mL (Choi et al. 2009), and may be a promising candidate for liver injury treatment due to strong antioxidant properties (Huang et al. 2018). The former showed the lowest binding energy and ranked first among all tested naturally occurring compouds to block SARS-CoV-2-ACE-2 interaction, and also formed a p-alkyl bond with TYR 505 and relatively more H-bonds with the receptor (four H-bonding) with GLN 493, GLY 496, GLN 498, and ASN 501, whereas compound b has shown only three H-bonds. Moreover, compound b made a p-alkyl hydrophobic interaction with TYR 505 as well, which may cause stability in the conformation of the ligand. However, one unfavorable donordonor bond is illustrated between hydroxyl moieties of inhibitor and ASP 501. Given the favorable potential biological activities of compound a, which exhibits the lowest binding affinity of À8.1 Kcal/mol, forms a comparatively higher number of H-bonds, and last but not least, shows H-bonding with major amino acid residues of the SAES-CoV-2-spike-ACE-2 interaction (GLY 496, ASN 501, and GLN 498), it may dictate its high inhibitory potential against SARS-CoV-2-spike to ruin its interaction with ACE À2 (Supplementary material Figure S5a, b). Delphinidin has a punctual inhibitory effect against WNV, Zika and DENV (V azquez-Calvo, Jim enez de Oya, Mart ın-Acebes, Garcia-Moruno and Saiz 2017). Moreover, delphinidin can also ruin the cell entry process of the hepatitis C virus (HCV) (Calland et al. 2015). As is shown in Supplementary material Figure S5c, compound c formed five H-bonds with important residues TYR 449, GLY 496, GLY 502 and one with TYR 453, postulating its promising potential to block the SARS-CoV-2 cell entry. Compound c also made p-p T-shaped and p-p stacked between its aromatic rings and TYR 505, ranking first in terms of H-bonding among all and hydrophobic interactions may boost stability of the ligand. Furthermore, Luteolin exhibited high potent antiviral activities against JEV (Fan, Qian, Qian and Li 2016), EBV and HIV-1 (Wu, Fang, Hsu, Chen, Chou, Huang, Cheng, Lin, Chang and Tsai 2016), and showed significant anti-inflammatory and antioxidant activities (Boeing et al. 2020). Luteolin-7-O-6 00 -malonylglucoside (compound d) posseses convincable anti-inflammatory and antioxidant properties (Zhang et al. 2019), showing the lower binding mode of À6.7 kcal/mol. It bound with SARS-CoV-2-S via two H-bonds with GLN 493, GLY 496 and demonstrated p-p T-shaped, suggesting that these interactions might enhance stability of this natural flavonoid (Supplementary material Figure S5d). Briefly, the higher number of H-bonds and binding affinity, the stronger antiviral potency, in comparison to compound a, compounds c and d demonstrated lower binding energies and made lower H-bonds, so, compound a demonstrated supreme antiviral potential against SARS-CoV-2-S and might be probably nominated as a persuasive antiviral inhibitor to block SARS-CoV-2-S and ruin its binding with the host cell.

Molecular dynamics (MD) results
The RMSD experienced a smooth jump with a low slope around 80 ns, and then the system gained equilibration with the fluctuation about 0.25 nm for the rest of simulation. Although the protein has not experienced an intensive structural change to some point, 200 ns simulation is sufficient for such system. The RMSD for complexes comprising compounds a and c underwent severe fluctuations and showed the highest oscillations, which the former has risen to 0.4 nm in the middle of simulation before low oscillations till the end of the simulation. The later also gained the same point in 180 ns before stabilizing over the simulation. These changes reveal that protein experienced severe conformational alterations over the ligand binding (Supplementary material Figure S6).The protein interaction with the other two ligands were almost the same with a modest ups and downs at 0.25 nm RMSD value in average. According to RMSF analysis, except for complex comprising free protein and compound d, the amino acids of protein oscillated more in complex with ligands a, b, and c than it did in free form, and they passed higher oscillations in particular in the active site (residues between 485 and 505) (Supplementary material Figure S7). This can emphasize their good potency on the target structure. In comparison to the other ligands, the amount of RMSF deviation for complexes including protein and compounds a and b had a higher intensity, which may show their effective impact on the structure of SARS-CoV-2-S-RBD (Supplementary material Figure S7a).
The oscilation range of Rg in the complexes encompassing compounds a and b have undergone a rise in the final quarter of simulation, showing that protein may be unfolded in the vicinity of these two compounds. (Supplementary material Figure S8). Though analyses specify the movements in atoms of protein each by a few implies, they cannot interpret the most component of protein displacements. PCA is needed to explore changes in the main component of protein movements and to investigate different ligands' effects on protein movements. There is a clear contrast in the paradigm of 2 D PCA analysis protein in the vicinity of compounds a and b in comparison to those of other systems, showing that the protein structure might have been unfolded and the protein movements may be increased after binding to ligands. Such changes in protein movements can lead to significant effects on its biological activities (Supplementary material Figure S9b and c).
The interacted amino acid residues are altered after MD simulations e.g. though compound a bound to protein through four H-bonds with ARG 403, ARG 457, and SER 494, this compound showed four H-bonding, in docking analysis, but with different residues (GLN 493, GLY 496, GLN 498, ASN 501) (Supplementary material Figure S10a). However, compound b bound to the target through three H-bonds in docking analysis, it has made the lowest number of H-bonds (just 2) with ARG 403 and GLN 943 after MD simulations (Supplementary material Figure S10b). Compounds c and d have made four and three H-bonds after MD simulations, (Supplementary material Figure  S10c, d), but the former showed four and the latter made just two H-bonds, respectively, in docking analysis with absolutely different residues compared to those of interacted in MD. In comparison to the molecular docking results, changes in interacted amino acids with ligands after MD simulations notice the point that the binding site of the ligands is altered.

Conclusion
Herein, we performed an in silico study to test the affinity of 102 phenolics/nonphenolics extracted from honey bee pollen with high biological activities against SARS-CoV-2 spike to target its interaction with ACE-2. Then, compounds with the lowest binding energies were flexibly docked as the most active inhibitors. Then, fifty ns MD simulations were conducted to explore the target dynamics and its structural changes after binding with these 4 selected ligands, the attachment of spike to the inhibitors confirmed the interactions and protein stability. According to molecular docking results, 4 compounds revealed the lowest binding energies with viral receptor among which kaempferol 3-neohesperidoside 7-O-rhamnoside was ranked first and could be nominated as a strengthen blocker candidate against the interaction of SARS-CoV-2-ACE-2 to perturb the cell fusion process. Further investigations must be on the call to explicate the needed dose of these natural inhibitors to perturb the virus cell entry process, and more in vitro and in vivo studies are imminent to clarify their antiviral potency versus SARS-CoV-2 S.

Funding
The author(s) reported there is no funding associated with the work featured in this article.