*2.1. Cells, Reagents, and Antibodies*

Vero E6 and 293T cells were obtained from Cell Bank of Chinese Academy of Sciences and were cultured in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS, *v*/*v*) at 37 ◦C in a humidified incubator. The medium also contained 10 mM HEPES (pH 7.4), 2 mM GlutaMAX, and penicillin–streptomycin–fungizone (100 units/mL of penicillin, 100 μg/mL of streptomycin, and 250 ng/mL of amphotericin B). The DMEM and all the supplemented ingredients were purchased from Invitrogen (Shanghai, China). An insect cell-expressed S protein (40589-V08B1, S1 + S2 ectodomain) and a polyclonal antibody cross-reactive to SARS-CoV-2 S (40150-T62) were purchased from Sino Biological (Beijing, China). The Steady-Lumi firefly luciferase reporter gene assay reagent was purchased from Beyotime (Nantong, China). An HIV-1 NL4-3 luciferase reporter vector that contains defective Nef, Env, and Vpr was purchased from MiaoLing (P20782, Wuhan, China). pCMV3- SARS-CoV-2-S (VG40589-UT, accession number MN908947.3) with codon-optimized *S* gene for mammalian cell expression and pCMV3-ACE2-HA (HG10108-CY) were purchased from Sino Biological. The pCMV3-SARS-CoV-2-S was modified in the lab by deletion of the last 19 amino acid residues to generate pCMV3-S since the endoplasmic reticulum (ER)-retention signal from the cytoplasmic tail was reported to interfere with virus preparation [15,16]. Glycyrrhizin ammonium salt (#50531, purity ≥ 95%) and HRP-conjugated anti-HA antibody (H3663) were purchased from Sigma-Aldrich. A stock solution of 50 mM GA was prepared by dissolving the compound in distilled water with pH adjusted to 7.4 using NaOH.

#### *2.2. Spike Protein-Pseudotyped Virus (Lenti-S) Preparation*

The spike protein-S-pseudotyped lentivirus (Lenti-S) was generated using a 2-plasmid system as previously reported [17]. Stocks of single-round infection of S protein-pseudotyped virus were produced by cotransfection of 293T cells (1.0 × 107 cells per 10 cm dish) with 2 μg of pCMV-S plasmid and 8 μg of pNL4.3-Luc using PEI reagent. The supernatant was harvested 34 h following transfection. After clarification by centrifugation at 1500× *g* followed by 0.45 μm filtration, pseudovirus-containing medium was collected and aliquots were stored at −80 ◦C. The virus was validated by testing for its ability to deliver the luciferase gene to ACE2-expressing HEK293T cells [18].

#### *2.3. Infection Assay*

Pseudotyped viruses provide an efficient way to determine an antiviral effect by measuring reporter gene expression. For gene-transducing assay, the permissible Vero E6 cells in 96-well plates (#3599, Costar, Corning, NY, USA) were infected with varying amounts of Lenti-S. Luciferase expression was determined at approximately 24 h PI using Steady-Lumi reagent on a GloMax 96 luminometer (Promega, Madison, WI, USA).

For blocking assay, Vero E6 cells were detached using 3 mM ethylenediaminetetraacetic acid (EDTA). After washing twice with serum-free medium (SFM), the cells were resuspended in ice-cold SFM-containing 2% (*v*/*v*) FBS (107 cells/mL). The cells were then aliquoted (5 × 105 cells/sample), and duplicate samples were incubated on ice with Lenti-S in the absence or presence of GA or a blocking reagent as indicated. The mixtures were incubated on ice for 60 min with occasional mixing. At the end, the cells were then washed with ice-cold medium 3 times to remove non-bound virus then plated in 12-well plates (Corning™ Costar, #3512) without supplementation of GA or the blocking reagent. Luciferase expression in duplicate samples was determined at 24 h PI.

#### *2.4. Protein Biotinylation and Binding Assay*

An insect-cell-expressed S protein (2 μg), resuspended in 100 μL of 50 mM NaHCO3, was labeled with freshly prepared sulfo-NHS-biotin (21217, Pierce, Carlsbad, CA, USA, 0.5 mg in 50 μL NaHCO3). The reaction lasted for 10 min at room temperature. At the end, unreacted sulfo-NHS-biotin was quenched by reaction with 1 mg glycine dissolved in 20 μL NaHCO3. The labeled protein was dialyzed against phosphate-buffered saline (MW cutoff 12 kD) and was used for protein-binding assay.

To measure the effect of GA on S protein binding, detached Vero E6 cells were incubated on ice with the biotinylated S protein in the absence or presence of varying amounts of GA or excess amounts of unlabeled S protein. After incubation on ice for 60 min, the cells were washed 3 times with ice-cold medium. After the cell lysates were separated by a 6% SDS-PAGE gel, cell-attached S protein was detected by immunoblotting analysis using an HRP-conjugated anti-biotin antibody (A0185, Sigma-Aldrich, St Louis, MO, USA). Anti-actin (sc-8432, Santa Cruz, Dallas, TX, USA) was used for loading controls.

#### *2.5. Surface Plasmon Resonance (SPR) Studies*

SPR analysis was performed using the Biacore T200 system and a CM5 sensor chip. The chip consists of two channels that were loaded with S protein. Briefly, the carboxymethylated dextran matrix was activated via the passage of EDC-HCl and NHS (0.2 M and 0.05 M in water, respectively). Then, the S protein at 20 μg/mL (in 1 mM sodium acetate) was passed over the surface. Any remaining un-reacted active ester groups were quenched by the passage of ethanolamine solution. The sensor chip surface was regenerated after each experiment by passing sodium dodecyl sulfate (SDS, 100 mM) over the surface for 2 min. This simple procedure reliably returned the original signal response seen before the binding experiment started. For testing, GA at concentrations was passed over the chip, and SPR angle changes were recorded and reported as response units (RUs). Data fitting was performed using the 1:1 Langmuir model in the BIAevaluation software package (GE Healthcare, Waukesha, WI, USA).

#### *2.6. Autodocking*

Docking analysis was performed using AutoDock Vina 1.1.2 (Windows version) with default scoring function [19]. The molecular structures of SARS-CoV-2 S protein (PDB id: 6vsb) and ACE2 (PDB id: 6m18) were downloaded from the Research Collaboratory for Structural Bioinformatics-Protein Databank (RCSB-PDB) [20]. The 3D structure of GA was from ZINC (https://zinc.docking.org/, id: 960251743495; last accessed on 20 September 2021). The structures were prepared using AutoDock Tools (ADT) by addition of polar hydrogens and the Gasteiger charges. Structures were exported in the pdbqt format after assigning the AD4 (AutoDock 4) atom type [21]. In the AutoDock Vina configuration files, the parameter num\_modes was set to 1000 and exhaustiveness to 100. We chose all the rotatable bonds in ligands to be flexible during the docking procedure, and we kept all the residues of S protein inside the binding pockets rigid. Multiple rounds of definition of the grid coordinate (x-, y-, and z-coordinates) with a defined grid box size were conducted to screen through the entire extracellular part of the S protein. The resulting docking structures were further analyzed in PyMOL.
