Natural products may interfere with SARS-CoV-2 attachment to the host cell

Molecular docking

Docking experiments (AutoDock Vina software) are performed using the HSPA5 solved structure (PDB ID: 5E84) after 50 ns of classical molecular dynamics simulation (performed using NAMD software) (Humphrey et al., 1996; Phillips et al., 2005; Trott & Olson, 2009). Four different conformations of HSPA5 representing the main four clusters (Chimera software) are used to test the ligands binding (Pettersen et al., 2004).

Thirteen different natural products-derived compounds are tested against the four different conformations of the host cell chaperone HSPA5 SBDβ, including; daidzein, genistein, formononetin, biochanin A, palmitic acid, linolenic acid, chlorogenic acid, hydroxytyrosol, caffeic acid, caffeic acid phenethyl ester, p-Coumaric acid, cinnamaldehyde, and thymoquinone. Additionally, six different physiological compounds are also docked to the HSPA5 SBDβ for comparison, including; estriol, estradiol, hydrocortisone, cholesterol, progesterone, and testosterone. All the dockings are done using flexible ligand into flexible active site protocol, where both the ligands and the active site residues (I426, T428, V429, V432, T434, F451, S452, V457, and I459) are treated as flexible during the search for a possible docking conformation using the vina scoring function of AutoDock Vina software (Trott & Olson, 2009; Yang et al., 2015). The grid boxes for the docking experiments were chosen to be of size 48 × 46 × 56 Å centered at (42.3, 54.9, −29.2) Å (with little differences between the different conformations of the HSPA5).

HADDOCK 2.4 web server is utilized to dock the spike model for SARS-CoV-2 against HSPA5 and the complex of HSPA5 with its docked ligands (van Dijk & Bonvin, 2006). The HADDOCK 2.4 easy interface was utilized in the study since there are no restraints to be defined (de Vries et al., 2010). Again the HSPA5 active site (I426, T428, V429, V432, T434, F451, S452, V457, and I459) is treated as flexible. In contrast, the C480-C488 region of the SARS-CoV-2 spike is treated as the active residues (binding site) in HADDOCK 2.4, as reported in a previous study by the author (Ibrahim et al., 2020).

After docking, the complexes are examined using the Protein-Ligand Interaction Profiler (PLIP) web server (Technical University of Dresden) (Salentin et al., 2015).

Docking results

Figure 3A shows the average binding affinities of different natural compounds to the HSPA5 SBDβ four different conformations with the error bars representing the standard deviations (SD). Pep42 (red column) is a cyclic peptide that recognizes explicitly cell-surface HSPA5 in vivo (Ibrahim et al., 2019; Kim et al., 2006). The average binding affinity of Pep42 is -6.73 ± 1.13 kcal/mol, which is used here as a reference to judge other compounds’ binding affinities. Phytoestrogens (green columns) show excellent average binding energies to HSPA5 ranging from -6.98 ± 0.19 kcal/mol (biochanin A) up to -7.80 ± 0.91 kcal/mol (daidzein). Compared to Pep42, the phytoestrogens have at least the same binding affinity to HSPA5 SBDβ. This means that a dietary supplement of phytoestrogens (found in Cicer arietinum) may contradict the binding of the SARS-CoV-2 spike to the cell-exposed HSPA5 preventing its recognition by the virus.

Open in a separate window

Figure 3.

The average binding affinity (in kcal/mol) calculated using AutoDock Vina software for the docking of the natural products bioactive compounds (A) and physiological compounds (B) into the four different conformations of the HSPA5 SBDβ. The cyclic peptide Pep42 (red column) is used as a reference due to its specificity in binding HSPA5 in vivo. Estrogens and phytoestrogen are among the best binders to HSPA5 SBDβ.

For the saturated (palmitic) and unsaturated fatty acids (linoleic and chlorogenic acids) (yellow columns) the same conclusion can be drawn, with a better average binding affinity to HSPA5 for chlorogenic (−7.10 ± 0.96 kcal/mol) acid compared to other fatty acids (−6.05 ± 0.51 and −5.50 ± 0.46 kcal/mol for linoleic and chlorogenic acid, respectively). This pattern of HSPA5 binding affinities is in good agreement with the previous reports of the antagonistic effect of unsaturated fatty acids, chlorogenic and linoleic acids, against the saturated, palmitic, fatty acid which induces ER stress (Katsoulieris et al., 2009; Zhang et al., 2018; Zhang et al., 2012). Palmitic, linoleic, and chlorogenic acids may be used to counteract the SARS-CoV-2 recognition of the host cell-surface HSPA5 and hence may reduce the viral attachment. Additionally, the saturated fatty acid, palmitic acid, may be used to target stressed HSPA5-exposed cells (viral infected or cancer cell) and induce ER stress leading to cell apoptosis.

The bioactive component of olive leaf extract, hydroxytyrosol, (bink column) shows moderate average binding affinity (−5.20 ± 0.35 kcal/mol) to HSPA5 SBDβ. Hydroxytyrosol succeeded in a previous study as a prophylactic agent against myocardial infarction-mediated apoptosis (Wu et al., 2018). For the caffeic and p-Coumaric acids (light blue columns), that are found in grape skin, the average binding affinities to HSPA5 SBDβ are −6.3 ± 0.60 and −5.63 ± 0.57 kcal/mol, respectively. These values are slightly less than Pep42 (−6.73 ± 1.13 kcal/mol), but the differences are not significant. Caffeic and p-Coumaric acids may bind to cell-surface HSPA5 competing for its recognition by viral spike protein and contradict the attachment. The same effect can be concluded from the caffeic acid phenethyl ester (CAPE) (dark blue column) that can be found in honeybee hive propolis (average binding affinity to HSPA5 SBDβ is -7.13 ± 0.95 kcal/mol). This average binding energy value is better than the highly selective cyclic peptide, Pep42, which indicates the potential of CAPE as an HSPA5 SBDβ binder. Additionally, CAPE was reported to induce ER stress in an autophagy-dependent manner in human SH‐SY5Y neuroblastoma (Tanida et al., 2008; Tomiyama et al., 2018). Cinnamaldehyde (cyan column) and thymoquinone (violet column) show −6.25 ± 1.10 and −5.520 ± 0.12 kcal/mol average binding energies to HSPA5 SBDβ. These binding energies are comparable to the Pep42 cyclic peptide (−6.73 ± 1.13 kcal/mol) and hence the active components of cinnamon and the seeds of Nigella sativa may tightly bind to cell-surface HSPA5 and could be successful in contradicting SARS-CoV-2 spike recognition and attachment.

Not only the natural compounds can bind HSPA5 SBDβ with high affinity, but also other physiological molecules. Figure 3B shows the average binding affinities for the binding of estrogens (estriol and estradiol), cholesterol, progesterone, testosterone, and hydrocortisone (cortisol) to HSPA5 SBDβ. As implicated from the binding energy values, all the physiological compounds can tightly bond the HSPA5 SBDβ with values ranges from -7.20 ± 0.58 kcal/mol (Hydrocortisone) up to -8.40 ± 0.98 kcal/mol (estradiol). These values are lower (better) than that of the Pep42 cyclic peptide, which reported to target cell-surface HSPA5 (GRP78) in vivo selectively. It is important here to point out that HSPA5 SBDβ may act as a receptor for such hormones. The binding energies indicate that cell-surface HSPA5 may be critical for these hormones recognition and hence internalization which not only may downregulate the concentration of cell-surface HSPA5, and its associated chemotherapeutic resistance, but also may play an essential role in hormone internalization for cell-signaling (Niu et al., 2015; Zhang et al., 2015). As a consequence, these hormones may also be used as protective molecules during chemotherapy to revert the chemoresistance of the HSPA5 presenting cancer cells.

Tables 1 and ​and22 summarize the interactions established between the small molecules and the HSPA5. Two types of interactions are dominant, the H-bonding and the hydrophobic interactions. Additionally, π-stacking (residues in bold in the tables) is reported between the residue F451 and the estrogens (estriol and estradiol), phytoestrogens (daidzein, genistein, and formononetin), caffeic acid phenethyl ester (CAPE), and cis-p-Coumaric acid. Also, salt bridges (underlined residues in Table 1) are formed between the residue K460 and both linolenic acid and cis-p-Coumaric acid. Hydrophobic interactions are more dominant compared to the H-bonding, as can be seen from almost all the natural and physiological compounds. This is in good agreement with previous reports defining the function of HSPA5 SBDβ in the lumen of ER as to recognize unfolded proteins in the lumen of the ER mediating its degradation or refolding using cellular machinery (Ibrahim et al., 2019; Pfaffenbach & Lee, 2011; Roller & Maddalo, 2013). On the other hand, hydrocortisone and chlorogenic acid have more H-bonds than hydrophobic interactions (5:3 and 7:5 for hydrocortisone and chlorogenic acid, respectively). Additionally, caffeic acid forms four H-bonds and four hydrophobic contacts with the HSPA5 SDBβ.

Figure 4 shows the interaction analysis made by the PLIP web server for the docked structures of HSPA5 to estrogen (estradiol) (A), phytoestrogens (daidzein) (B), and biochanin A (C) as an example. The HSPA5 is shown in colored surface representations with its domains labeled. The ligands are represented in yellow sticks, where it appears how it fit in the binding site groove of the SBDβ. Enlarged views of the binding sites show how the interactions established upon docking. Residues in the binding site of HSPA5 SBDβ are represented in blue sticks and labeled with its one-letter code. In Figure 4, the hydrophobic interactions are described in dashed-gray lines, while H-bonds and π-stacking are depicted in solid blue lines and dashed-green lines, respectively. Docking scores are listed to reflect a binding affinity for each complex. Noticeably, the interacting residues are mainly hydrophobic, while hydrophobic interactions are dominant all the docking complexes. For estradiol, only one H-bond is formed through T458, while none is reported in daidzein. On the other hand, the biochanin A-HSPA5 complex show 5 H-bonds. Estradiol and daidzein but not biochanin A form π-stacking with residue F451 of HSPA5.

Open in a separate window

Figure 4.

The structure of the docked complexes of HSPA5 and the small molecules (A) estradiol, (B) daidzein, and (C) biochanin A. HSPA5 is represented in the colored surface while the docked small molecules are in orange sticks. The NBD, SBDα, and SBDβ domains of the HSPA5 are labeled, while the enlarged panels show the interactions that established upon docking. The active site residues in the expanded panels are marked with its one-letter code and represented in blue sticks. H-bonds, hydrophobic contacts, and π-stacking interactions are shown by blue lines, dashed-gray lines, and dashed-green lines, respectively. The docking score (in kcal/mol) is shown for each complex.

I performed molecular docking experiments using the four different conformations of HSPA5 after the 50 ns MDS using HADDOCK 2.4. The average docking score is −66.3 ± 5.7 indicating high binding affinity between the interacting proteins. Additionally, I tried to dock the small molecules-HSPA5 complexes to the SARS-CoV-2 spike protein model, but the spike doesn’t fit the HSPA5 binding site; this may be due to the presence of the small molecules in the SBDβ of HSPA5. The small molecules prevent the spike from binding to HSPA5 SBDβ in silico. The results support the effectiveness of natural products and physiological hormones to block HSPA5 SDBβ, preventing SARS-CoV-2 spike recognition (see the graphical abstract). The small molecules tested in this study may be used as prophylactic agents for high-risk personals like elders, medical staff in the front-line, or cancer patients.