Madin-Darby canine kidney (MDCK) cells were purchased from the American Type Culture Collection (NBL-2) and cultured in Eagle's minimal essential medium with 10% fetal calf serum, as described in our previous report (8). Human colon adenocarcinoma Caco-2 cells were previously purchased from the Riken BioResource Center (RCB0988) (9). Tryptanthrin (SML0310; Sigma-Aldrich; Merck KGaA) was dissolved in DMSO at a concentration of 1 mg/ml and then diluted 10-fold in ethanol (final concentration of the stock solution, 100 µg/ml).

Total RNA was extracted from Caco-2 cells using Trizol reagent (Invitrogen; Thermo Fisher Scientific, Inc.), and first strand cDNA was reverse-transcribed using total RNA as a template with Superscript IV (Invitrogen; Thermo Fisher Scientific, Inc.). Human ACE2 full-length cDNA (NM_001371415.1) was obtained by polymerase chain reaction (PCR) where Caco-2 cDNA was used as a template with the primer set: forward, 5'-gtggatgtgatcttggctca-3' and reverse, 5'-caaaatcacctcaagaggaaaaa-3'. The PCR product was inserted into the pTA2 TA-cloning vector (Toyobo). After amplification, the insert was excised at the NotI and HincII sites, and then inserted into the pCX4pur vector (10) at the NotI and HpaI sites (pCX4pur-hACE2). The absence of mutations was verified by sequencing.

MDCK cells (8x10⁴) were grown in a 6-cm dish to 60-70% confluence and were transfected with either pCX4pur-hACE2 or the empty pCX4pur vector (5 µg each) using the Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's instructions. Cells were selected by resistance to puromycin for two weeks.

Leaves of Polygonum tinctorium were collected in September 2020 in Aomori Prefecture, Japan. A voucher specimen was deposited in the herbarium of the medicinal herbal garden of Tohoku Medical and Pharmaceutical University and identified by KS.

We used our original extract, named ‘AOMORI-BLUE extract’, which has been patented in Japan (Japanese Patent no. 6389492). Powdered, air-dried leaves (100 g) were extracted with 1,200 ml of d-limonene (Wako Pure Chemical Industries, Ltd.) at room temperature for 48 h. After filtration, a pale-yellow extract was obtained, and the extract was subjected to HPLC. To analyze the composition, the extract was dissolved in ethanol (5.0 mg/ml) and passed over a COSMOSIL 5PE-MS column (i.d. 4.6x250 mm; Nacalai Tesque) (mobile phase 40% CH₃CN; flow rate, 1.0 ml/min; detection 254 nm; temperature 25˚C). Data were collected with a SIC Chromatocorder12 (System Instruments Co., Ltd.). A tryptanthrin standard was purchased from Sigma-Aldrich Japan. All other reagents were purchased from Fujifilm-Wako Pure Chemicals, Co. In the present study, indigo extract was diluted 10-fold in ethanol (stock solution). d-limonene was also diluted 10-fold in ethanol as a control stock solution. For reference, indigo leaves were extracted with ethanol (99.5%) according to the same procedures.

Recombinant protein SARS-CoV-2 S1 subunit tagged with mouse IgG2a Fc portion (S1N-C5257) and normal mouse IgG (mIgG; 140-09511) were purchased from ACROBiosystems and Fujifilm-Wako Pure Chemicals, Co., respectively. A Fluorescein Labeling Kit-NH₂ (Dojindo) was used to conjugate fluorescein to the S1 protein and mouse IgG, according to the manufacturer's instructions.

3x10³ of transfected or untransfected MDCK cells were suspended in 10 µl of culture medium, and were transferred separately into the bottom of the micro-Insert 4-Well (gasket) on a µ-Dish (35 mm, high, no. 80406; ibidi GmbH). After 3 h of incubation, 150 µl of medium were poured into the insert to fill all wells with identical medium. The next day, fluorescein-labeled S1 or mIgG was added to the culture medium at a concentration of 3 µg/ml. At the same time, indigo extract, d-limonene, or tryptanthrin was added at the indicated concentrations, depending upon the experiment. After cell cultures were incubated 24 h, the micro-insert gasket was gently removed, and cells were washed 3x with culture medium. Then, the µ-Dish was filled with 1 ml of culture medium, and was placed on the microscope stage of a C2+ confocal laser scanning system (Nikon, Tokyo, Japan). Fluorescein fluorescent images were captured with a 40x objective lens and analyzed on the Nikon C2⁺ computer system. Fluorescein intensity (arbitrary units per unit area) was measured at five randomly selected high-power fields for each well using Analysis Controls tools. Cell cultures in wells on µ-Dishes were prepared and measured in triplicate for each experimental group, and the mean and standard deviation of fluorescein intensities were calculated using ROI Statistics. Experiments were independently repeated three time with similar results.

After having detected the fluorescence from S1 proteins, we fixed the cells in µ-Dishes with methanol for 10 min at -20˚C, and blocked them with 2% bovine serum albumin for 30 min at room temperature. At this time, no fluorescein fluorescence was detectable on cells in any dishes. Then, cells were incubated with an antibody against ACE2 (E-11, sc-390851; Santa Cruz Biotechnology, Inc.) overnight at 4˚C, and were visualized with Alexa Flour 488-conjugated secondary antibody (anti-mouse IgG; Jackson ImmunoResearch). After washing with phosphate-buffered saline (PBS) three times, nuclei were labeled with DAPI (Molecular Probes) for 2 h at 4˚C. Fluorescent images were captured using a C2⁺ confocal scanning system equipped with 488-nm argon and 543-nm helium-neon lasers (Nikon).

Cell viability was assessed with a water-soluble tetrazolium-8 (WST-8)-based colorimetric assay using a Cell Counting Kit 8 (Dojindo, Kumamoto, Japan). MDCK cells (3x10⁴) were seeded in a 96-well plate in triplicate overnight, and then indigo extract was added to each well at indicated dilution rates. The next day, cells were incubated with WST-8 for 30 min, and the absorbance was measured at 450 nm using an automated microplate reader. Measurement of mitochondrial dehydrogenase cleavage of WST-8 to formazan dye provided an indication of cell viability.

Cells were washed in PBS and were lysed in a buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% Triton X-100 and 1 mM phenylmethylsulfonyl fluoride. After removal of impurities by centrifugation, lysates were subjected to Western blot analyses, as described in our previous report (11). Immunoreactive band intensities were quantified using ImageJ software (National Institutes of Health), as described previously (12).

We obtained the 3D structure of the of SARS-CoV-2 spike protein trimer from the Protein Data Bank (https://www.rcsb.org/) (PDB ID: 6Z97). We selected this structure because it satisfies the following conditions: i) the receptor-accessible ‘up’ conformation is available in the trimer, ii) the amino acid sequence is conserved without deletions, and, iii) the structure is available at a resolution <3.5 Å.

The 3D structure of tryptanthrin was downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov) (CID: 73549; no other conformers are known). Using these structures, we simulated 100 docking runs for the spike protein with tryptanthrin using AutoDock 4.2 in the grid box surrounding the Receptor Binding Motif (RBM). We then analyzed the number of dockings to the ACE2 binding site in the binding mode.

Comparative analyses were done on experiments consisting of more than two groups. We compared fluorescence intensities using one-way analysis of variance (ANOVA). We calculated the mean of each group, and compared mean values between two groups using the Bonferroni correction of one-way ANOVA. WST-8 data were analyzed with Dunnett's multiple comparison test. Comparisons between two groups were done with Student's t-test. P-values ≤0.05 were considered statistically significant. Correlations between cell densities and CADM1 protein levels using Spearman's rank test were considered significant if R²≥0.1 and P≤0.05.

An HPLC peak with a retention time of 26.328 min was identified as tryptanthrin, the concentration of which was estimated at 17.3 µg/ml (Fig. 1). This concentration was much higher than that in the extract prepared using ethanol (compare Figs. 1 and S1). There were other high peaks at retention times from 13-20, 35-40 and 45-50 min. These peaks could not be recognized as known compounds.

We transfected MDCK cells with the pCX4pur vector carrying the human ACE2 full-length cDNA and the empty vector, and we obtained cells that robustly expressed ACE2 (MDCK-ACE2) and the vector-control cells (MDCK-vector), respectively. Expression of ACE2 was confirmed by immunofluorescence and Western blotting analyses. An anti-ACE2 antibody clearly detected ACE2 expression on cell membranes of MDCK-ACE2 cells, and detected it as a ~150-kDa band in the cell lysate from MDCK-ACE2 cells, but not from MDCK-vectors or untransfected MDCK cells (Fig. 2A and B).

We tried to quantitatively detect binding of S1 proteins to ACE2 in live cell cultures. For this purpose, we used S1 proteins fused with mouse IgG Fc, and labeled them with fluorescein. The resulting protein (S1-Fc-fluorescein) was added to untransfected MDCK, MDCK-vector, and MDCK-ACE2 cell culture in µ-Dishes at a concentration of 3 µg/ml. After 24 h, live cells were washed with PBS and observed through a confocal laser microscope. Fluorescein fluorescence was clearly detected on cell membranes of MDCK-ACE2 cells, but was not seen in untransfected MDCK or MDCK-vector cells (Figs. 3A and S2). For a negative control, mouse IgG was labeled with fluorescein, and added to the cultures. No fluorescence was detected in either type of cells (Fig. S3). Fluorescent intensities were calculated for each cell type and treatment setting. MDCK-ACE2 cells incubated with S1-Fc-fluorescein had much higher fluorescence intensity than the other types of cultures (P<2x10^-16; Figs. 3A and S3 and Table SI).

WST-8 assays revealed that indigo extract had no substantial effect on MDCK cell viability when diluted ≥8,650-fold (Fig. 2C). The extract did not change the medium pH in this range of dilution (4.39-7.44). Together with S1-Fc-fluorescein, the indigo extract stock solution was added to confluent MDCK-ACE2 cell cultures in µ-Dishes at a dilution of 1,730-fold (17,300-fold for the original extract), and after 24 h, cells were observed alive. Fluorescent signals were weakly detectable (Fig. 3A). The fluorescence intensity was much lower than that in MDCK-ACE2 cell cultures without indigo extract (P<2x10^-16; Fig. 3A and Table SI).

Since indigo extract contained tryptanthrin at a concentration of 17.3 µg/ml (Fig. 1), the tryptanthrin concentration was estimated to be 1.0 ng/ml in the above treatment. Instead of indigo extract, we added tryptanthrin (molecular weight 248.24) alone, together with S1-Fc-fluorescein, to MDCK-ACE2 cell cultures at a concentration of 1.0 ng/ml, or 4.0 nM. The fluorescent intensity of fluorescein decreased, but the degree of diminution was smaller than that caused by indigo extract (Fig. 3A).

Next, we used indigo extract with a further 5-fold dilution, i.e., an 86,500-fold dilution. The inhibitory effect on S1-ACE2 binding substantially disappeared, suggesting that the effect of the indigo extract is dose-dependent (Fig. 3B). Then, we changed the ratio of concentrations of indigo extract and S1-Fc-fluorescein added to the culture. There was a significant correlation between the ratio and S1 fluorescent intensity. That is, higher relative concentrations of indigo extract resulted in decreased S1 fluorescence intensity, while higher ratios of S1 increased S1 fluorescent intensity (Fig. 3B).

We conducted Western blotting analyses and immunofluorescence on MDCK-ACE2 cells treated with indigo extract or d-limonene at a 17,300-fold dilution and tryptanthrin at 1.0 ng/ml. Both experiments revealed that ACE2 expression was essentially unchanged with either treatment (Fig. 2A).

Molecular simulation methods were used to examine possible mechanisms by which tryptanthrin inhibits binding of S1 protein to ACE2 on MDCK cells. We searched the PDB, drew the 3D structure of the spike protein trimer bound to ACE2 using PDB ID 6M0J, and obtained the 3D structure of the spike protein trimer PDB ID 6Z97 (Tables I and II and Fig. 4). We conducted docking simulation analyses to examine how tryptanthrin binds to RBM. We found that tryptanthrin molecules preferentially bound to the ACE2 binding site reported in 6M0J (13) in more than half of the 100 docking runs (Tables I and II). Notably, most of this binding showed the same binding mode, in which the molecular plane of tryptanthrin was nearly perpendicular to the surface of RBM around Leu455-Phe456 and Cys488-Ser494 (Fig. 4 and Table III). In these two regions, S1 spike protein was simulated to have four amino acids that are involved in binding to ACE2 and tryptanthrin (Table III).