Pseudoephedrine and its derivatives antagonize wild and mutated severe acute respiratory syndrome‐CoV‐2 viruses through blocking virus invasion and antiinflammatory effect – Wiley

2.1 Compounds

All the ephedrine derivatives were purchased from commercial suppliers or from our laboratory compound library. All these molecules were at least 99.5% purity and were used without any purification. Chemical structures and compound ID of each molecule were shown in Figure 1.

2.2 Virtual screening and molecular docking

Protein preparation and binding-site definition were performed using protocols inside the program Discovery Studio 4.0 (v 4.0.0.13258). Small molecules were prepared under prepare ligands followed by full minimization protocols. CHARMm forcefield was used in both preparation procedures of proteins and small molecules. All compounds were tried to dock into the crystal structure of S-RBD and ACE2 complex (PDB id = 6M0J) (Lan et al., 2020). Lineage P.1 S-RBD and ACE2 complex (PDB id = 7NXC) were downloaded from PDB and were prepared using Discovery Studio 4.0. Libdock protocol was used in high throughput screening of small molecules with potential activity of disrupting the interaction between wild S-RBD and ACE2. The physical properties and libdock scores were used to conclude active ingredients. CDOCKER protocol was used in molecular docking experiments to judge whether active components could block the interaction between mutated S-RBD and ACE2. A combination of physical properties and scores of CDOCKER ENERGY and CDOCKER INTERACTION ENERGY were used to select the final pose.

2.3 Sequence alignment and homology modeling

The amino acid sequence of lineages B.1.1.7 and B.1.617 mutated S-RBDs was downloaded from PubMed website (https://www.ncbi.nlm.nih.gov). The S-RBDs of two x-ray structures of S-RBD–ACE2 complex (PDB id = 6M0J and 6ZGG; Lan et al., 2020; Wrobel et al., 2020) were used as templates for homology modeling for their high identity to mutated S-RBDs of SARS-CoV-2 lineages B.1.1.7 and B.1.617. Mutated S-RBD sequences downloaded in the previous step were aligned to the sequences corresponding to wild S-RBD crystal structures using align sequences protocol within the program Discovery Studio 4.0. For building of a protein amino acid sequence into its 3D structure, homology modeling is a more reliable protocol than build mutants procedure (Dubreuil et al., 2005). Thus, build homology models protocol was used to build mutated S-RBD models. The number of models generated was set as 20. Optimization level was set as high, and refine loops were set as true. A combination of probability density function (PDF) total energy, Ramachandran plot, and DOPE score was used to choose the final conformation.

2.4 Protein–protein docking and conformation refinement

The models of mutated S-RBD built in last step and the crystal structure of the spike protein of lineage B.1.351 (PDB id = 7LYN) downloaded from PDB as well as the ACE2 part of S-RBD–ACE2 complex (PDB id = 6M0J) were prepared and minimized using prepare protein protocol. We made an attempt to find out the strength of interaction of mutated S-RBDs and ACE2, so that their binding mode can be predicated based on their PPIs, which will also give a message about the small-molecule binding site between the two proteins.

ZDOCK protocol was used to perform protein–protein docking between the prepared proteins. The mutated S-RBDs were set as ligands, and the ACE2 was defined as receptor. Angular step size was set as 6°; 2,000 poses were generated for each docking process, and a distance cutoff of 8 Å was used to filter poses. A RMSD cutoff value of 6 Å and an interface cutoff value of 9 Å were used to cluster poses. The ranking of the poses was done, and electrostatic and desolvation energy were calculated during docking processes. The predicated docking conformations calculated by ZDOCK procedure were subjected to refinement and re-ranking using RDOCK protocol to minimize the complexes under CHARMm forcefield. Finally, the final poses were selected from a comprehensive consideration of ZDOCK scores, ZRank scores, and RDOCK scores among all refined conformations.

2.5 NanoBiT-based SARS-CoV-2 S-RBD/ACE2 binding assay

The preliminary screening and 50% inhibitory concentration (IC₅₀) determination of the 17 compounds (Figure 1) to block SARS-CoV-2 S-RBD/ACE2 interaction were performed by NanoBiT-based assay according to our previous procedure (Yu et al., 2020). In brief, SARS-CoV-2 S-RBD-LgBiT and SmBiT-ACE2 fusion plasmids were transiently co-transfected into HEK293 cells in a six-well plate using FuGENE HD transfection reagent (Promega, Madison, WI) according to the manufacturer’s instructions. After transfection for 6 hr, cells were detached by gentle pipetting, suspended in fresh Opti-MEM medium (Gibco-BRL/Invitrogen, Carlsbad, CA) containing 2% FBS and reseeded into a 96-well plate. To measure if the test compounds could block the interaction between SARS-CoV-2 S-RBD and ACE2, the compounds were serially diluted to indicated concentrations and added to the wells of the 96-well plate and incubated for 3 hr at 37°C. Then, Nano-Glo live Cell Assay reagent was added, and luminescence was measured on an Envision plate reader (EnVision, Perkin Elmer, Waltham, MA). False positive results were considered by determining the cytotoxicity of the compounds on the HEK293 cells and the inhibitory effects of the compounds on NanoLuc (HEK293/Nanoluc stable cells), respectively. Data were analyzed and expressed using the inhibitory effects on SARS-CoV-2 S-RBD/ACE2 interaction (NanoBiT inh%), NanoLuc luciferase (NanoLuc inh%), and the cell proliferation (Cytotox inh%, CC₅₀) on HEK293 cells.

2.6 Cell viability assays

Raw264.7 cell viability impacted by MHJ-11 or MHJ-17 was detected with MTT method. Cells were treated with different concentrations of MHJ-11 or MHJ-17 for 48 hr; then, MTT solution (0.5 mg/ml) was added to each well. After incubating at 37°C for 5 hr, the formazan crystals in viable cells were dissolved with 100-μL DMSO. Solubilized formazan was spectrophotometrically quantified with Tecan M1000Pro Multiscan Spectrum (Tecan) at 490 nm.

2.7 qRT-PCR assays

LPS (0.1 μg/ml) was used to activate inflammatory response of RAW264.7 cells. Ctrl group represented normal cells without LPS or drugs treated. Model group represented cells treated with LPS. MHJ-11 groups represented cells treated with LPS and MHJ-11 (125, 250, and 500 μM). MHJ-17 groups represented cells treated with LPS and MHJ-17 (125, 250, and 500 μM). Raw264.7 cells were treated in a 37°C, 5% CO₂ incubator for 6 hr, and total RNA was extracted using TRIzol reagent (Sangon Biotech). Total RNA was then used for reverse transcription with PrimeScript™ RT reagent kit (TaKaRa, Tokyo, Japan), and TB Green™ Premix Ex Taq™ (TaKaRa) was used for qRT-PCR analysis. The mRNA levels were normalized to the β-actin mRNA. Relative quantification was performed as 2^−ΔΔCt. Three repeats were executed for each assay. Amplification primers were as follows: IL-6 (F: 5′-CTGCA AGAGA CTTCC ATCCA G-3′, R: 5′-AGTGG TATAG ACAGG TCTGT TGG-3′), TNF-α (F:5′-CTGTA GCCCA CGTCG TAGC-3′, R: 5′-TTGAG ATCCA TGCCG TTG-3′), and β-actin (F: 5′-GTCCC TCACC CTCCC AAAAG-3′, R: 5′-GCTGC CTCAA CACCT CAACC C-3′).

2.8 Enzyme-linked immune sorbent assays

Ctrl, Model, MHJ-11, and MHJ-17 groups were set as those in qRT-PCR analysis. Raw264.7 cells were treated for 24 hr, and the supernatants were collected for IL-6 and TNF-α detection. Protein levels were measured by enzyme-linked immune sorbent assay (ELISA) according to the manufacturer’s instructions (NeoBioscience, Shanghai, China). Three repeats were executed for each assay.

2.9 Western blot

Ctrl and Model groups were set as those in qRT-PCR analysis. MHJ-11 group represented cells treated with LPS and 500-μM MHJ-11. MHJ-17 group represented cells treated with LPS and 500-μM MHJ-17. Raw264.7 cells were treated for 24 hr; then, cells were collected and lysed with NP40 buffer (Beyotime Biotechnology, Shanghai, China) according to manufacturer’s instructions. Western blots were performed as described using antibodies against phospho-NFκB p65 (Cell Signaling Technology, 3,033), phospho-p44/42 MAPK (Cell Signaling Technology, 4,376), phospho-SAPK/JNK (Cell Signaling Technology, 4,668), phospho-p38 (Cell Signaling Technology, 4,511), phospho-IκBα (Cell Signaling Technology, 2,859), and β-actin (Cell Signaling Technology, 3,700; Zhu et al., 2020). Detections were analyzed with Azure Biosystem (Azure c600, Azure Biosystem™, Dublin, CA). Three repeats were executed for each assay.

2.10 Data analysis

The raw data were analyzed using GraphPad Prism 8.0.1 software. Two-tailed Student’s t test was used to determine the significance of differences in the above-mentioned assays. All values are presented as mean ± SD. The P value was used to indicate statistical significance.

3.1 Virtual screening

The aim of our virtual screening was to search for active ingredients derived from TCM Ephedra sinica Stapf (ephedra), which could block the interaction between S-RBD of SARS-CoV-2 and ACE2. Therefore, a crystal structure of S-RBD and ACE2 complex (PDB id = 6M0J) was used as docking target. The long and narrow binding site was detected from receptor cavities by the function within Discovery Studio 4.0. Among the candidates successfully docked into the cavity, we found that ephedrine (MHJ-16) and pseudoephedrine (MHJ-17) were the most potent ingredients in disrupting the interaction between SARS-CoV-2 S-RBD and ACE2, and their CDOCKER interaction energies were lower than other candidates. Therefore, it could be predicted that ephedrine, pseudoephedrine, or their derivatives may have potent activity in blocking the interaction. The binding site of the crystal structure and docking pose of ephedrine were shown in Figure 2.

3.2 Blocking activities of three compounds on the SARS-CoV-2 S-RBD/ACE2 interaction

The inhibitory activities of ephedrine (MHJ-16), pseudoephedrine (MHJ-17), and 15 corresponding commercial derivatives against SARS-CoV-2 S-RBD/ACE2 interaction were investigated by NanoBiT assay. Fortunately, three compounds (MHJ-11, MHJ-16, and MHJ-17) were active according to preliminary screening results. Then, these three active compounds were further verified in a dose-dependent manner. As shown in Figure 3, the IC₅₀ values of these three compounds on S-RBD/ACE2 interaction (NanoBiT inh%), NanoLuc luciferase (NanoLuc inh%), and HEK293 cells proliferation (Cytotox inh%) were analyzed. Notably, none of the three compounds showed obvious inhibitory effects on Nanoluc luciferase (IC₅₀ > 100 μM), and they had low cytotoxicity to HEK293 cells (IC₅₀ > 100 μM; Figure 3). Among the three compounds, MHJ-17 (IC₅₀ = 13 μM) showed better blocking activity compared with MHJ-11 (IC₅₀ = 49 μM) and MHJ-16 (IC₅₀ = 73 μM). Therefore, these three compounds showed blocking activity against the interaction between SARS-CoV-2 S-RBD and ACE2, although MHJ-16 exhibited lower activity.

3.3 Homology modeling and protein–protein docking

Homology modeling is a powerful and widely used tool to build and predict 3D structures of proteins whose folded forms are unknown (Waterhouse et al., 2018). Although x-ray structure of lineage B.1.1.7 S-RBD complexed with antibody was reported (Supasa et al., 2021), the structure of the protein docked with natural receptor may be quite different. To reveal a more accurate result, homology modeling was used in this research. Two different S-RBD parts of x-ray structures of S-RBD–ACE2 complex (PDB id = 6M0J and 6ZGG) were used as templates. Both templates were combined with their endogenous receptor protein ACE2. SARS-CoV-2 lineage B.1.1.7 has a mutation called N501Y, and lineage B.1.617 is used to refer to two mutations in S protein at the positions E484Q and L452R. These varieties caused the infection rates of these viruses increased dramatically. Fortunately, the sequences of mutated viruses can be downloaded from PubMed website. As shown in Figure 4, sequence alignment revealed that these sequences are highly homologous, and mutations N501Y, E484Q, and L452R were highlighted in red. In the building procedure of S-RBD of lineage B.1.1.7, 20 molecular models were generated during initial procedure, and the best model (Figure 5a,b) with the lowest PDFs total energy and PDF physical energy of 8,820.1 and 591.4, respectively, was chosen for further CHARMm forcefield refinement and research. In another hand, S-RBD of lineage B.1.617 was built under the same procedure, and the PDF total energy and PDF physical energy of the best model were 9,075.5 and 764.0, respectively. Ramachandran plots of the two generated models were also shown in Figure 5c,d, and it was indicated that most of the amino acids in the best predicated model were considered favorable. Importantly, the highlighted mutated residues N501Y of B.1.1.7, and E484Q and L452R of B.1.617 were also in green, indicating that the built models are reliable.

ZDOCK and RDOCK protocols were used to generate and refine protein–protein docking complex, respectively. As the crystal structure of spike protein of lineage B.1.351 (PDB id = 7LYN) was reported (Gobeil et al., 2021), the RBD part of it was directly used at this stage. As a result, eight poses for variety B.1.1.7, 12 poses for variety B.1.617, and 28 poses for variety B.1.351 complexed with ACE2 were generated, which were further refined and re-ranked using CHARMm forcefield. The best docking pose of each procedure was selected under a consideration of highest ZDOCK score and lowest E_RDOCK, which was considered as the conformation closest to the real. The best pose of lineage B.1.1.7 S-RBD–ACE2 complex was shown in Figure 6, and the contacting surface area was observed like the previous wild S-RBD–ACE2 complex. However, in the constructed model, the interacting amino acid residues of both parts are different from the previous one (Table 1 and Figure 6b,c,e,f,h,i). It was visualized that Arg403, Asn439, Tyr473, Gly476, Glu484, Cys488, Phe490, Pro499, Val503, and Gln506 in lineage B.1.1.7 S-RBD were interacted with ACE2 protein, which did not occur before. Residues Lys417 and Gly446 in the new model had no interaction with ACE2. Moreover, it could be observed that Ser19, Thr324, Gln325, Gly326, Glu329, Gly352, Phe356, and Ala386 were interacted with lineage B.1.1.7 S-RBD, which were not detected before. Also, Glu35, Glu37, Gln42, Arg357, and Arg393 were not interacted with mutated S-RBD as these interactions were collected using x-ray structures of wild S-RBD–ACE2 complex. On the other hand, S-RBD of B.1.617 variety complexed with ACE2 indicated that Arg403, Asn439, Tyr473, Gly476, Ser477, Gln484, and Pro499 were new interacted amino acid residues with ACE2. Furthermore, Ser19, Glu23, Thr324, Gln325, Gly326, Phe356, and Ala386 in ACE2 were interacted with S-RBD of B.1.617 variety, which has never happened before. At last, analysis of S-RBD of lineage B.1.351 complexed with ACE2 indicated that Arg403, Asp405, Lys458, Tyr473, Gln474, Gly476, Ser477, Lys484, Ser494, and Pro499 from S-RBD have new interactions with ACE2. Also, Ser19, Thr20, Glu23, Lys26, Thr324, Gln325, Gly326, Phe327, Phe356, Met383, Ala384, Ala386, and Ala387 were observed to have interactions with ACE2. As a result, all the mutated S-RBDs and ACE2 had more amino acid residues interact with each other. It may be explained that mutated viruses resulted in higher transmissibility than the previous viruses (Chan et al., 2020).

3.4 Molecular docking

After explaining the reasons for increased infectious capacity of the new viruses, it was urgent to evaluate whether pseudoephedrine (MHJ-17) and its derivative (MHJ-11) can block the interaction between mutated viruses and ACE2 protein. Molecular docking experiment using CDOCKER protocol within Discovery Studio 4.0 software was performed to solve this problem. It is hard to define binding sites between simulated mutated S-RBDs and ACE2, because the interactions between them are stronger than that of the wild one. Fortunately, both molecules (MHJ-11 and MHJ-17) could dock into the narrow cavities, indicating that these two compounds could also disrupt the interaction between mutated S-RBDs and ACE2. These results showed a prospect of two compounds. In the S-RBD of B.1.1.7 variety and ACE2 complex model, as shown in Figure 7a, MHJ-11 has several π relevant interactions with Gln388, Lys417, and Pro389 and has many strong hydrogen bond interactions with surrounding residues, such as Ala386, Ala387, Gln388, Asp405, and Tyr505. These amino acid residues are dispersed across the lineage B.1.1.7 S-RBD and ACE2. Moreover, MHJ-17 can also dock into the complex (Figure 7b). The amino group of MHJ-17 has a salt bridge with His34 and Glu37, and the compound also has several hydrogen bond interactions with Asn33 and Lys417. In addition, the interaction between lineage B.1.617 complexed with ACE2 could also be disrupted by MHJ-11 and MHJ-17 according to the results by molecular docking. As shown in Figure 7c, MHJ-11 performed π relevant interactions with Gln388, Pro389, and Lys417. Also, the compound has several hydrogen bond and salt bridge interactions with Ala387, Gln388, Ala386, and Asp405. Furthermore, MHJ-17 can form salt bridge interactions with the complex (Figure 7d). Also, it could be observed that the compound has two hydrogen bonds with Asn33 and Glu37. Under the same strategy, docking experiments were also performed to verify the two compounds on blocking activities for disrupting the interaction between ACE2 and S-RBDs of lineage B.1.351 and P.1. As a result, both MHJ-11 and MHJ-17 could dock into the complexes. As shown in Figure 7e,f, both MHJ-11 and MHJ-17 have several hydrogen bond interactions with the complexes, and they also have π-anion interactions with Glu37. In another hand, both active compounds have many strong hydrogen bond interactions with variety P.1 S-RBD and ACE2 complex (Figure 7g,h) and have π-cation interactions with surrounding amino acid residues. Moreover, according to CDOCKER interaction energy scores calculated in docking experiments, both MHJ-11 and MHJ-17 have stronger interactions with mutated S-RBDs complexed with ACE2 than that of the wild one. In addition, mutated amino acid residues such as N501Y, E484Q, L452R, K417N, and E484K seldom perform interactions with the two active compounds (only mutated amino acid Tyr501 of variety B.1.351 has interactions with MHJ-11 and MHJ-17, Figure 7e,f). Therefore, the mutations did not cause a decrease in activity in blocking PPIs between mutated S-RBDs and ACE2, and MHJ-11 and MHJ-17 can be used as reliable lead compounds for further modification.

3.5 Antiinflammation activities

As systemic inflammatory response is a major symptom for coronavirus infections, macrophages play critical role in this process. Therefore, we detected the antiinflammation activity of MHJ-11 and MHJ-17 in Raw264.7 cells. As shown in Figure 8, the cytotoxicity of MHJ-11 (Figure 8a) and MHJ-17 (Figure 8b) was not observed in Raw264.7 cells until the concentration reached 400 μM. After that, the expression and transcriptional levels of two representative inflammatory factors, IL-6 and TNF-α, were measured after treatment with different concentrations of MHJ-11 and MHJ-17. The results showed that MHJ-11 could significantly reduce the mRNA and protein levels of IL-6 at 125, 250, and 500 μM in LPS-activated Raw264.7 cells (Figure 8c,d). Similarly, MHJ-17 could also greatly downregulate IL-6 mRNA and protein expression of at 125, 250, and 500 μM in LPS-activated Raw264.7 cells (Figure 8c,d). Furthermore, MHJ-11 could significantly reduce the mRNA and protein levels of TNF-α at 500 μM (Figure 8e,f), and MHJ-17 could significantly reduce TNF-α mRNA and protein expression at 250 and 500 μM (Figure 8e,f). Thus, the antiinflammation effects were observed for MHJ-11 and MHJ-17.

We then explored the inflammation-related pathways regulated by MHJ-11 and MHJ-17. Phosphorylation of NFκB p65, p44/42 MAPK, SAPK/JNK, p38, and IκBα proteins was detected by Western blot with LPS-activated Raw264.7 cells treated with 500-μM MHJ-11 or MHJ-17. As shown in Figure 9, MHJ-11 and MHJ-17 could significantly inhibit protein phosphorylation of NFκB p65 (Figure 9b), p44/42 MAPK (Figure 9c), SAPK/JNK (Figure 9d), p38 (Figure 9e), and IκBα (Figure 9f), which indicated that the two active compounds could regulate NFκB and MAPK signaling pathway to inhibit inflammatory response.