Human malignant embryonic rhabdomyoma (RD) cells, susceptible to EV71, were maintained in our laboratory (Li et al., 2013). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen, United States) supplemented with 10% fetal bovine serum (FBS, Invitrogen) (DMEM-10% FBS) at 37°C with 5% CO₂. The EV71 (GenBank accession No. KJ508817) was provided by the Chinese Center for Disease Control and Prevention. CVA16 (GenBank accession No. JF695003.1) was provided by the Henan Provincial Center for Disease Control and Prevention. The EV71-luciferase pseudovirus (EV71-luciferase) encoding a luciferase reporter gene in the viral genome was pre-packaged before the experiment, as previously described (Xu et al., 2014).

Female breeder specific-pathogen-free (SPF) grade BALB/c mice were purchased from the Liaoning Institute of Biological Products (China). Newborn BALB/c mice born within 24 h from breeder mice were selected as subjects for in vivo experiments. All experiments were performed in a BSL-2 facility with all experimental methods carried out in accordance with the regulations and guidelines set forth by the Animal Experiments Committee of Jilin University. The protocols for mouse experiments were approved by the Laboratory Animal Ethics Committee of the School of Life Science, Jilin University (Approval Number: 2018-nsfc010).

All botanical drugs in this study were collected in the Korqin area of Tongliao City, Inner Mongolia, China, in July 2018, and were identified by Prof. Buhebateer working at Inner Mongolia Minzu University. A voucher (no. 20180720) was deposited at the School of Mongolian Medicine and Pharmacy, Inner Mongolia Minzu University. Luteolin purchased from Meilun Biotech Co., Ltd. (China) was used as a positive drug control in vitro (Xu et al., 2014).

Water and alcohol extracts of botanical drugs were prepared using water or ethanol to heat and reflux and then recover the solvent. Essential oils from botanical drugs were prepared by steam distillation. The dried ethanol extract was suspended in water and extracted sequentially using petroleum ether, dichloromethane, ethyl acetate, and n-butanol. The extracts of the botanical drugs were obtained after the solvent was recovered (ethanol, petroleum ether, dichloromethane, ethyl acetate, n-butanol; all were 100% pure and purchased from Beijing Chemical Works) at room temperature. All extracts were dissolved in DMSO to 2 mg/ml stocks and diluted with DMEM-2% FBS to working concentrations. To obtain the extract of T. sibirica, also named Messerschmidia sibirica L., 2.0 kg whole plant was ground-dried, then. Was extracted with 95% ethanol (20 L) three times, for 4 h each time. Evaporation of the solvent under reduced pressure yielded the ethanol extract. The ethanol extracts (250.0 g) of T. sibirica. were mixed with water (mass–volume ratio of 1:2, g/mL), and petroleum ether was added each time (mass–volume ratio of 1:2, g/mL). The preferred number of extractions was between 25 and 30. All petroleum ether extracts were combined and recovered under reduced pressure to obtain a petroleum ether extract of T. sibirica (PE-TS) at a total amount of 47.2 g. The PE-TS was stored at −20°C until its use. The PE-TS was also dissolved in DMSO and DMEM-2% FBS, as described above.

Briefly, RD cells (5  × 10³ cells/well) were seeded in 96-well plates at 80% confluence. To determine the cytopathic effect (CPE) inhibition activity, the cells were treated with botanical drug extracts in DMEM-2% FBS solution at working concentrations of serial 2-fold dilutions (8,000–1.95 μg/ml). Meanwhile, the cells were infected with EV71 (MOI = 0.05) or CVA16 (MOI = 0.005) for 48 h at 37°C with 5% CO₂. To determine the cytotoxicity, the cells were treated with only the botanical drug extracts in DMEM-2% FBS at serially diluted concentrations for 48 h. Cell viability was measured using a cell counting kit (CCK)-8 (Boster Biological Technology Co. Ltd., United States), and the OD490 was determined using a microplate reader (CliniBio 128C, BIOMAD). Antiviral activity was measured using the following equation: (T - Vc)/(Cc - Vc), where T, Vc, and Cc are the absorbance values of botanical drug extract-treated cells, virus control, and cell control, respectively. The 50% cytotoxicity concentration (CC₅₀) and the 50% effective concentration (EC₅₀) were analyzed using GraphPad Prism 5.0, as reported and calculated using regression analysis (Rudi et al., 1988). Selective index value was calculated by the ratio of CC₅₀/EC₅₀.

RD cells at 90% confluence in 24-well plates were infected with EV71 (MOI = 0.2) in the presence of the PE-TS at a concentration of 250 μg/ml. Luteolin treatment (8.59 μg/ml) was used as a positive control. The supernatant was collected at 16 hpi for viral RNA extraction. EV71 RNA was extracted using the EasyPure^® Viral DNA/RNA Kit for Virus Detection (TRANSGEN Biotech Co., Ltd., Beijing, China). qRT-PCR was performed using the One Step SYBR^® PrimeScript™ RT-PCR Kit II (Takara Bio, Otsu, Japan) and the 5′ untranslated region (UTR) specific primers (sense 5′-TCC​TCC​GGC​CCC​TGA-3′ and antisense 5′-AAT​TGT​CAC​CAT​AAG​CAG​CCA-3′) as previously reported (Nijhuis et al., 2002) using the Bio-Rad CFX96 system (United States). Relative EV71 genomic RNA copies were normalized to the endogenous control (β-actin, sense 5′- CCA​CCA​TGT​ACC​CAG​GCA​TT-3′ and antisense 5′- CGG​ACT​CAT​CGT​ACT​CCT​GC-3′) using the 2^−ΔΔCt method (Livak and Schmittgen, 2001; Su et al., 2010).

RD cells (5 × 10³ cells/well) were seeded in 96-well plates, cultured to 90% confluence, treated with 250 μg/mL PE-TS or 8.59 μg/ml luteolin (positive control) in DMEM-2% FBS, and infected with EV71-luciferase (MOI = 0.5). Each concentration was set with three parallel wells and incubated at 37°C with 5% CO₂. After infection for 1 h, the inoculum was replaced with the drugs at corresponding concentrations in DMEM-2% FBS. The supernatant was discarded at 16 hpi, and 100 μl of Bright-Glo™ luciferase reagent (Promega, United States) was added to each well. Luciferase activity was determined with the VICTOR X2 Multilabel Plate Reader (PerkinElmer, United States).

RD cells (2 × 10⁴ cells/well) were seeded in 24-well plates, cultured to 90% confluency, infected with EV71-WT (MOI = 1) and treated with 250 μg/mL PE-TS. After infection for 1 h, the inoculum was replaced with the drugs at corresponding concentrations in DMEM-2% FBS. At 16 hpi, cell supernatants were collected. Thereafter, the virus titers in the cell supernatants were determined using endpoint dilution assays, and the 50% tissue culture infectious dose (TCID₅₀) was calculated using the Reed and Muench method (Lu et al., 2011).

Newborn BALB/c mice (born within 24 h) (n = 9–11 per group) were intracerebrally challenged with EV71 at a lethal dose of 1.5 × 10⁶ TCID₅₀ per mouse. Since the botanical extract PE-TS contained sticky insoluble matter after dissolution, PE-TS in the DMSO-water solution was filtered to exclude the influence of insoluble matter in the animal. The administration groups were set as follow: intraperitoneal injection of 0.4 mg/g or 0.13 mg/g filtered PE-TS, direct lavage of 0.4 mg/g or 0.13 mg/g filtered PE-TS, direct lavage of 0.8 mg/g or 0.26 mg/g unfiltered PE-TS. All extracts were added to sterile PBS supplemented with 10% DMSO, which was administrated once daily for seven consecutive days. Two groups were inoculated with sterile PBS supplemented with 10% DMSO as the uninfected control. The mice were monitored daily for 16 days to observe survival rate, clinical score, and body weight. Survival rate was estimated using the Kaplan-Meier method. Disease signs were evaluated to obtain a graded clinical score (0, healthy; 1, slow movement; 2, weakness in hind limbs; 3, paralysis in a single limb; 4, paralysis in two limbs; and 5, death) as described previously (Zhang et al., 2013).

GC-MS analysis of PE-TS was performed using GC-MS (Model: GC MS-5977B, Agilent, United States) equipped with a 325/350C HP-5 MS fused silica capillary column of 30 m length, 0.25 mm diameter and 0.25 µm film thickness. For GC-MS analysis, electron ionization system with ionization energy of 70 eV was used. The carrier gas used was helium (99.9%), at a constant flow rate of 1.0 ml/min. Injector and mass transfer line temperature were set to 270°C and 260°C, respectively. The oven temperature was set from 70°C to 325°C at 15°C/min for 5 min and finally raised to 325°C for 10 min. One microliter of the sample was injected in a split mode with a scan range of 50–1,000 m/z. The total running time of GC-MS was 33 min. The GC-MS total ion chromatograms of the chemical constituents of PE-TS were obtained, and each chromatographic peak was searched using the NIST17. L mass spectral library. The chemical components with a matching degree higher than 90% were selected according to the mass spectral data and relative retention time.

The volatile organic compounds of the PE-TS were harvested using the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP, https://tcmspw.com/tcmsp.php). The compounds with oral bioavailability ≥30% and drug-likeness ≥ 0.18 were considered as potential active compounds. Targets of the active compounds of volatile organic compounds of the PE-TS were selected through both the similarity ensemble approach (SEA) (http://sea.bkslab.org/) and Swiss target prediction (STP) (http://www.swisstargetprediction.ch/) in “Homo Sapiens” mode. HFMD-related H. sapiens targets were searched in the GeneCards platform (https://www.genecards.org/) and DisGeNET (https://www.disgenet.org/search). Then, these collected targets were merged, and duplicates were removed. The overlapping targets between potential active compounds of the PE-TS and HFMD targets were identified. The network diagram between compounds and targets was established using Cytoscape 3.7.2 (http://www.cytoscape.org/). The overlapping targets were input into String 11.0 (https://string-db.org/) to obtain the interaction results among the targets, and the visual analysis results were obtained by Cytoscape 3.7.2.

The 3D structures of most EV71 viral proteins have been reported previously (Plevka et al., 2012). In this study, we used semi-flexible molecular docking program CDocker in Discovery Studio software package (Accelrys, San Diego, CA, United States) to evaluate the possible bindings between our compounds and EV71 viral proteins. CDocker was run with default parameters: random conformations and orientations were both set to “10”, the simulated annealing was set to “true”, “CHARMM forcefield” and “grid-based potential” were selected, the Top Hits was set to default value as “10”. After docking runs, the -CDocker Energy was interpreted as an approximated indicator of the binding affinity.

In vitro experiment values are expressed as mean ± standard deviation (SD) from n = 3 biological repeats. Graphing and analysis were performed using GraphPad Prism v7. Statistical analysis was performed using one-way ANOVA. Statistical analysis for survival rate was performed using the Log-Rank (Mantel-Cox) test. Statistical analysis for clinical score was performed using Ridit assay with SigmaPlot 12.0. Statistical significance represented by asterisks is marked correspondingly in the figures, *p < 0.05, **p < 0.01, and ***p < 0.001. All experiments were independently repeated twice.

First, we screened the inhibitory effects of 47 botanical drug extracts on EV71 infection by measuring the CPE on RD cells. The cells were infected with EV71 (MOI = 0.05) and treated with botanical drug extracts or luteolin as a positive control, and were measured for cell viability at 48 hpi. The cell viability of the virus control group decreased to 31.45 ± 0.29% of the non-infection control group. Most botanical drug extracts did not show obvious antiviral effects (inhibition of CPE less than 30%), whereas four botanical drug extracts displayed antiviral effects against EV71 (inhibition of CPE greater than 50%) (Figure 1). Among them, the petroleum ether extract of Tournefortia sibirica L. (PE-TS) displayed the highest CPE inhibition against EV71 infection (97%) (Figure 1). As a positive control, the natural compound luteolin (8.59 μg/ml) showed 71% CPE inhibition. T. sibirica, which belongs to the family Boraginaceae, is utilized in Inner Mongolia of China as a remedy for scrofula, eczema, sores, and ulcers (Wu, 2014; Ao and Na, 2015). To date, no study has shown that T. sibirica extract has an inhibitory effect on enteroviruses. Therefore, PE-TS was the focus of this study as a novel potential antiviral agent against EV71 infection.

Next, we determined the cytotoxicity of PE-TS in cells. RD cell viability decreased with increasing PE-TS concentration in a dose-dependent manner, with a CC₅₀ of 970 ± 55.6 μg/ml (Figure 2A). The inhibitory effect of PE-TS on EV71- or CVA16-induced CPE in RD cells was also investigated. As shown in Figure 2B, the optimal protection rate of PE-TS against EV71 infection was 97.25 ± 1.47% at a concentration of 250 μg/ml. The EC₅₀ against EV71 was 76.72 ± 8.03 μg/ml in RD cells. Selective index value was 12.6, suggesting the potent of development. The antiviral effect of PE-TS decreased at concentrations greater than 500 μg/ml, mainly due to its cytotoxicity. As shown in Figure 2C, the optimal protection rate of PE-TS against CVA16 infection was 94.75% ± 5.64% at a concentration of 125 μg/ml. The EC₅₀ against CVA16 was 30.57 ± 4.26 μg/ml, and selective index value was 31.7.

The effects of PE-TS on the biosynthesis of EV71 were examined based on genomic RNA replication and viral protein synthesis at 16 hpi without an obvious cytopathic effect (Lu et al., 2011). PE-TS (250 μg/ml) had a significant inhibitory effect on EV71 RNA replication in RD cells, measured using qRT-PCR, with an inhibitory rate of 90.45% (Figure 3A). To evaluate viral protein synthesis in cells, we used an EV71 reporter pseudovirus carrying a luciferase coding sequence in the viral genome. Luciferase activity, which can be measured by luminescence, acts as an indicator of viral protein synthesis when the reporter pseudovirus infects cells. PE-TS showed a significant inhibitory effect on viral protein synthesis at a concentration of 250 μg/ml, with an inhibition rate of 62.54% (Figure 3B). As a positive control, luteolin treatment also significantly decreased viral RNA content and luciferase activity (Xu et al., 2014). These results indicate that the antiviral effects of PE-TS against EV71 were due to the inhibition of viral RNA replication and protein synthesis.

To analyze the impact of PE-TS administration on the generation of EV71 progeny virus, the TCID₅₀ titer of EV71 in the supernatant was detected at 16 hpi. The results showed that PE-TS decreased the extracellular EV71 titer by 3.75 log at a concentration of 250 μg/ml (Figure 4), which proved that PE-TS inhibited the generation of EV71 progeny viruses.

We further investigated the protective effect of PE-TS against EV71 infection in a newborn mouse model (Dai et al., 2019). In this study, we aimed to simulate clinical oral administration and intravenous infusion by direct gavage or intraperitoneal injection, respectively. Considering the suspension character of PE-TS in a DMSO-water solution and its potential serious adverse outcome after peritoneal administration, PE-TS in the DMSO-water solution was filtered to remove insoluble matter. Subsequently, the in vitro antiviral activity and cytotoxicity of filtered PE-TS were determined. The results showed that the effective antiviral concentration of filtered PE-TS increased with an optimal antiviral concentration of 1,000 μg/ml (Figure 5A) compared with that of unfiltered PE-TS. The non-toxic concentration of filtered and unfiltered PE-TS were determined in 1-day-old BALB/c mice (Supplementary Figure S1).

One-day-old BALB/c mice were intracranially challenged with a lethal dose of EV71 and administered PE-TS treatment for seven days via gavage or intraperitoneal injection, respectively. The survival rate, clinical score, and body weight of mice were monitored daily for 16 days after the virus change. Several dose gradients for unfiltered PE-TS (i.g.) and filtered PE-TS (i.g. and i.p., respectively) treatments were set in this experiment. Ribavirin was used as a control because it is a broad-spectrum antiviral drug (Pang et al., 2014). As a result, intraperitoneal injection of 0.4 mg/g filtered-PE-TS and direct lavage of 0.8 mg/g unfiltered-PE-TS significantly improved the survival rate (70% and 66.7%, respectively; Figure 5B). They also significantly relieved clinical symptoms and decreased weight loss in the infected mice (Figures 5C,D). Consistent with previous studies (Dai et al., 2019; Kang et al., 2021), ribavirin treated mice exhibited only slight improvement of mortality. This is probably due to ribavirin may exhibit more cytotoxicity in the EV71-infected suckling mice model. These results demonstrate that PE-TS effectively protected newborn mice from lethal EV71 infection.

The chemical constituents of PE-TS were further analysed by GC-MS of Single-Quadrupole systems. Such approach can preliminarily describe the profile of volatile organic compounds in PE-TS. The GC-MS total ion chromatograms of the chemical constituents of PE-TS were shown in Figure 6. The mass spectrum of each detected peak was retrieved in the reference standard spectral library NIST17.L. Sixty compounds were identified and analyzed from PE-TS (Supplementary Table S1), mainly including 15 esters, 13 hydrocarbons, 7 terpenoids, 5 ketones, 3 amides, 3 carboxylic acids, 2 nitrile compounds, and 12 compounds each belongs to unique type. None of these compounds has been reported anti-enterovirus activity from literature.

We firstly analyzed potential active compounds with oral bioavailability from volatile organic compounds of the PE-TS. Then, a total of 638 compound targets were identified from the potential active compounds. Further, a total of 5549 HFMD-related targets were acquired by screening the GeneCards and DisGeNET databases. After the intersection of 638 compound targets and 5549 HFMD-related targets, 346 overlapping targets were obtained. A protein-protein interaction network of 346 overlapping targets was established, containing 346 nodes and 1,694 edges. According to the twofold median degree value obtained by topological analysis, 8 core targets were further selected, including SRC, STAT3, HSP90AA1, AKT1, EGFR, MAPK3, PIK3R1, and MAPK1 (Figure 7).

To further predict the compound targets, molecular docking was performed. We focused on both EV71 proteins which are usually identified as inhibitor targets and host proteins which selected from compound-target interaction network analysis, as docking targets. Docking results showed potential targets, including EV71 3D polymerase, 3C protease and host protein SRC. EV71 3D protein is an RNA-dependent RNA polymerase which is responsible for viral RNA replication (McMinn, 2002). 3C protease is essential for cleavage of viral precursor polyproteins and also host proteins associated to antiviral responses (Li et al., 2002). SRC, namely steroid receptor coactivator, is a node of several signals involved in many biological processes such as proliferation, differentiation, motility and adhesion, and is also an important regulator of cell metabolism (Pelaz and Tabernero, 2022). As shown in Figure 8, hexadecamethyl cyclooctasiloxane (PubChem CID: 11170), 3-thiazolidinecarboxylic acid (PubChem CID: 560987), monobutyl phthalate (PubChem CID: 8575) efficiently binds to 3D polymerase with best -CDocker energy 74.283, 30.883 and 30.524 kcal/mol, respectively. Diheptyl phthalate (PubChem CID: 19284) efficiently binds to 3C protease with best -CDocker energy 41.453 kcal/mol. Furthermore, nona-2,3-dienoic acid, ethyl ester (PubChem CID: 533672) efficiently binds to SRC with best -CDocker energy 24.646 kcal/mol.