We cultured 293T cells (ATCC, CRL-3216) and A549 cells (ATCC, CCL-185) in high-glucose Dulbecco’s Modified Eagle’s Medium (HyClone, South Logan, UT) supplemented with 10% FBS (Gibco, Carlsbad, CA), 1% penicillin–streptomycin solution (Gibco), and 2% 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (Gibco) in 5% CO₂ at 37°C. The cells were passaged every 2-3 days.

The HAdV5–fluc virus was purchased from Shandong Weizhen Company. HAdV3-GFP, HAdV4-GFP, andnHAdV7-GFP were stored in a P3 Laboratory of the School of Public Health, Southern Medical University, Guangzhou, China. HAdV14-GFP was a gift from the Institute of Respiratory Diseases, Guangzhou Medical University. HAdV55-GFP was a gift from the Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences.

The 1,813 approved drug compounds were purchased from TargetMol (Approved Drug Library L1000), and the 556 traditional Chinese medicine-sourced small molecule compounds were provided by the National Standard Chemical Control Library of NIFDC(110701-112025). The compounds were dissolved to 10 mM in dimethyl sulfoxide (Sigma-Aldrich, St. Louis, MO) and stored at −80°C until further use.

To test whether our system met the HTS conditions, Wells in columns 1 and 2 of the 96-well assay plates contained 293T cells as a control (0% response), whereas HAdV5 was added to all other wells (positive control, 100% response). The results were observed 24 h later. The signal-to-basal (S/B) ratio (S/B=mean signal/mean background), Z factor(Z=[1-3SD of sample +3SD of control]/|mean of sample - mean of control|) and CV were caculated. In the first round of HTS, 293T cells were seeded into 96-well plates at a density of 3 × 10⁴ cells/well (Corning, Inc., NY, USA). After overnight incubation, the cells were treated with the 10 μM compounds in duplicate, then incubated at 37°C, 5% CO₂ for 1 h. HAdV5–fluc (2.4×10⁴ plaque-forming units [PFU]/mL, MOI=0.67) was added to each well(50 μl), and luciferase activity was measured 23 h later. In short, 100 μl of culture medium was gently aspirated, and 100 μl of Bright-Glo luciferase reagent (Promega, Madison, WI) was added to each well to react with the cells for 2 min at room temperature. Luminescence was then measured using a GLOMAX 96 microplate luminometer (Promega). The percentage inhibition was calculated as 100 × [1 − (RLU in the presence of compound − RLU of negative control)/(RLU of positive control − RLU of negative control)](Negative control refers to the group of cells without drugs and viruses). The primary hits showed no apparent cytotoxicity and a >50% decrease in adenovirus replication in duplicate wells with 10 µM compounds. The 50% effective concentration (EC₅₀) was determined from the dose–response curve of HAdV5–fluc at eight serially diluted doses, 90–0.04 µM, in the second screening.

A549 cells were seeded at a density of 2 × 10⁴ cells/well into 96-well plates. The following day, the medium was replaced 2 h in advance of adding the indicated concentration of drug. The drug concentrations used were as described above. Four replicate wells were used for each drug concentration, and 50 μl of virus (4 × 10⁵ PFU/mL) was added to each well. The plates were incubated in a shaker for 30 min, and the medium was changed to the maintenance medium with the corresponding concentration of drug after 2 h. After 48 h, three fluorescence images (Nikon eclipse TE2000-U)were randomly recorded per well. After photographing, the GFP fluorescence intensity of HAdV-GFP in cells was analyzed by imagJ software. Five photos were taken here and three were analyzed after removing the highest value and the lowest value.

Cytotoxicity testing was performed using the CellTiter Glo luminescent cell viability assay kit (Promega) to determine the 50% cytotoxic concentration (CC₅₀) of each test compound in the absence of the virus. Serial drug dilutions, from 90 to 0.04 μM, were added to the 293T cells in 96-well plates, then 50 μL of complete medium was added instead of the virus. After incubation at 37°C and 5% CO₂ for 24 h, cell viability was measured using the GLOMAX 96 microplate luminometer.

Mice were handled in accordance with institutional (SCXK(jing) 2017-0005)NIFDC, Beijing, China) guidelines for laboratory animal care and use, and the Animal Care and Use Committee at the NIFDC approved the study protocol. Four to five-week-old female BALB/c mice (n = 5 per group) were obtained from the Institute of Laboratory Animal Resources of NIFDC. To challenge the mice, 4.5 × 10⁶ PFU HAdV5 were administered via the tail vein to groups of five mice. Cardamomin (CDM) was dissolved in 0.5% methylcellulose (Sigma-Aldrich, St. Louis, MO) and aliquoted into three doses (150 mg/kg, 75 mg/kg, and 37.5 mg/kg), which were IP-injected into three groups of mice. The drug was administered 6 h before infection, then once daily until 5 dpi. Bioluminescence imaging (BLI) analysis was performed with the IVIS Lumina Series III Imaging System (PerkinElmer, Baltimore, MD) as previously described (Zhang et al., 2017). We used Living Image software (Caliper Life Sciences, Baltimore, MD) to analyze the regions of interest, which were represented by the total flux in photons per second.

Four to six-week-old male BALB/c mice (n 10 per group) were divided into three groups: A, drug control group; B, drug and virus experimental group; and C, virus control group. We inoculated 6.4 × 108 TCID50 of HAdV-5 to mice via the tail vein and CDM (125 mg/kg) was administered by IP injection 6 h before infection, then once a day until 5 dpi. The mice were weighed daily and the weights recorded and were euthanized at 1, 3, 5, 8 dpi.The livers, spleens, and lungs of mice that died of natural causes or were sacrificed daily were divided into three parts, which were used for pathological sectioning, immunohistochemical observation, and molecular analysis. For the molecular analysis, virus DNA was extracted by QIAamp DNA Mini Kit(Qiagen, Germany), and detected by Quantitative Real-time PCR (Zhang et al., 2020). For the pathological sectioning and immunohistochemical observation, tissue sections were embed in paraffin after washed with PBS three times. pathological sectioning by Hematoxylin-eosin (HE) staining and immunohistochemical by streptavidin-biotin peroxidase technique were performed respectively. A rabbit neutralizing antibody against HAdV5 was used for immunohistochemistry and the dilution ratio is 1:50.

GraphPad Prism 6 (San Diego, CA, USA) was used to analyze the EC₅₀ and CC₅₀ values of the compounds and the bioluminescence imaging data. The statistical comparisons were performed using the t-test and ANOVA in IBM SPSS Statistics 24.0. The values α = 0.05 and P < 0.05 were considered statistically significant.

Using HAdV5-fluc, we applied HTS to identify inhibitors of viral entry from 1,813 approved drug library and 556 traditional Chinese medicine-sourced small molecule compounds. The HTS conditions of cell-seeding density and HAdV5 dose were optimized at 30,000 cells/well and 1 × 10³ PFU/well, respectively. Under these conditions, the signal-to-basal (S/B) ratio and Z-factor were 209,321 and 0.76, respectively ( Figure 1A ). The above data indicates that the experiment met HTS conditions (Zhang et al., 1999; Zhang et al., 2006). The HTS schematic is depicted in Figure 1B . The primary screen was performed for a single dosage (10 µM) of each compound to identify those with activity against HAdV5. The results showed 42 compounds resulted in >50% inhibition, thus necessitating a second confirmation round. Each of the 42 compounds was diluted at a 1:3 ratio to generate four concentrations ranging from 90 µM to 0.04 µM to confirm their activity against HAdV5 and determine their EC₅₀ values. This evaluation of antiviral activity together with cytotoxicity was critical because compounds with cytotoxicity can appear to elicit antiviral activity, while the cellular inhibition assay is largely dependent on cell viability. Drugs with a dose-response relationship were further evaluated in parallel with cytotoxicity in uninfected 293T cells using eight-point dose response curves (starting from 90 µM) to generate EC_50HAdV5 and CC₅₀ values. The ratio of CC₅₀ and EC_50HAdV5 was calculated for use as a selectivity index (SI). After the final reconfirmation screening, three compounds: Tranilast, Cardamomin and Nuciferine were confirmed as possessing efficacious inhibition effects against HAdV5, low cytotoxicity, and SI values above 10 ( Table 1 and Figures 2A–C ).

The three drugs screened for HAdV5 inhibition using 293T cells were further tested for their effects on several other types of adenovirus: HAdV3, HAdV4, HAdV7, HAdV14, and HAdV55. Wildtype adenoviruses fused with GFP (AdV3-GFP, HAdV4-GFP, HAdV7-GFP, HAdV14-GFP, and HAdV55-GFP), which corresponding to HAd5-fluc were used for detection. The results show that CDM had an inhibitory effect on HAdV3-GFP but not on other types of adenoviruses at the cellular level. The result of CDM was showed in Figure 2D .

To test if the antiviral activities of the active compounds observed in vitro were translatable to in vivo HAdV5 infection, we evaluated the treatment effects of the three candidate compounds from the HTS. First, the three compounds (75 mg/kg/d) were delivered by IP injection into BALB/c mice 6 h before infection. The mice were infected with 4.5 × 10⁶ PFU HAdV5 (0 h) via the tail vein, and the drug was re-administered at 1,2,3,4 and 5 dpi. Based on the BLI analysis, we concluded that CDM treatment of mice resulted in decreased HAdV5 infection. The inhibitory effect was demonstrated by the reduced BLI signal compared with the control group. The other compounds did not show infection inhibition effects in vivo ( Figure 3 ). The experiment was subsequently repeated; CDM was administered in three doses, 150 mg/kg/d, 75 mg/kg/d, and 32.5 mg/kg/d, which exhibited obvious effects, with inhibitory rates of 80% (p < 0.05), 76% (p < 0.05), and 58% (p < 0.05), respectively ( Figure 4 ).

From the above results, CDM had a clear inhibitory effect on HAdV5-infected mice. A wildtype HAdV5 strain was used for further verification. The efficacy of the drug was evaluated by observing changes in virus DNA content in tissues, HE-staining, and immunohistochemical staining. There is no obvious change in DNA content in tissues.

Infected lung, liver, and spleen tissues were collected at 1-5 dpi for HE staining analysis. We compared the results for the control group and the experimental group at 1-4 dpi, and no significant differences were observed. On the fifth day, clear organizational differences were noted. The results are shown in Figure 5 .

At 5 dpi, the bronchial and alveolar walls of the lung tissue were not clearly abnormally structured in the three groups. Local alveolar wall thickening was observed in both the experimental and virus control groups. Alveolar wall granulocyte infiltration was more obvious and severe in the virus control group. The lung tissue damage in the virus control group(HAdV5) was more serious than that in the experimental group(CDM+HAdV5).

The liver lobules of mice in the drug control group were intact, the liver cells were arranged in a regular order, the liver cord was structured normally, and we observed no obvious inflammatory reaction. In the experimental group, spot-like necrosis of hepatocytes, balloon-like changes in some hepatocytes, swelling of the cell body, and loose staining of the cytoplasm (red arrow) were observed. Compared with the other two groups, the virus control mice showed more severe liver necrosis. Balloon-like hepatocytes were widespread, and there was abundant hepatocellular necrosis, cytoplasmic eosinophilic enhancement (yellow arrow), and a small amount of inflammatory cells infiltrate. Therefore, the liver tissue of the experimental group was less damaged than that of the virus control group.

In the control group, the red and white pith were clearly demarcated, and red pith lymphocytes were abundant. In the experimental group, the red and white pith were clearly demarcated, the germinal center was not obvious, and red pith lymphocytes were abundant. The red and white pith were less clearly demarcated in the spleen tissue of the virus control group. A small amount of red pulp caused congestion in the medulla sinuses, and a small number of extramedullary hematopoietic cells were seen (yellow arrow). Therefore, the degree of spleen tissue damage was slightly less in the experimental group than the virus control group. In summary, CDM had a protective effect on the tissues of mice infected with adenovirus.

We also performed immunohistochemical analyses using a rabbit neutralizing antibody dilution of 1:50 against HAdv5 a group of proteins expressed by adenoviruses—adenovirus capsid proteins. Immunochemical staining images of the liver at 1 and 5 dpi are shown in Figures 6 . The liver tissue of the experimental group was weakly positive for the virus on day 1 and was slightly more positive than the drug control group; however, the control group showed more positive staining at 5 dpi. According to the liver tissue immunohistochemical results on 5 dpi, CDM had a certain inhibitory effect on the virus.