Anti-enterovirus 71 activities of Melissa officinalis extract and its biologically active constituent rosmarinic acid

Enterovirus 71 (EV71) infection is endemic in the Asia-Pacific region. No specific antiviral drug has been available to treat EV71 infection. Melissa officinalis (MO) is a medicinal plant with long history of usage in the European and Middle East. We investigated whether an aqueous solution of concentrated methanolic extract (MOM) possesses antiviral activity. MOM inhibited plaque formation, cytopathic effect, and viral protein synthesis in EV71-infected cells. Using spectral techniques, we identified rosmarinic acid (RA) as a biologically active constituent of MOM. RA reduced viral attachment and entry; cleavage of eukaryotic translation initiation factor 4 G (eIF4G); reactive oxygen species (ROS) generation; and translocation of heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) from nucleus to cytoplasm. It alleviated EV71-induced hyperphosphorylation of p38 kinase and EPS15. RA is likely to suppress ROS-mediated p38 kinase activation, and such downstream molecular events as hnRNP A1 translocation and EPS15-regulated membrane trafficking in EV71-infected cells. These findings suggest that MO and its constituent RA possess anti-EV71 activities, and may serve as a candidate drug for therapeutic and prophylactic uses against EV71 infection.


Results
Anti-EV71 activity of methanolic extract of MO (MOM). The syrup of MOM was diluted in water, and tested for its antiviral activity. Using the plaque reduction assay, we examined the anti-EV71 activity of MOM. RD or Vero cell monolayers were infected with 100 PFU/well or 200 PFU/well of EV71 strains, BrCr, 1743, or 4643, for 1 h, and were overlaid with semisolid media containing 0.3% agarose and 0, 78, or 156 μg/ ml of MOM. MOM diminished plaque formed by these virus strains in RD and Vero cells at concentrations up to 156 μg/ml (Fig. 1a). At these concentrations, MOM was non-cytotoxic to RD cells (CC 50 = 370.9 ± 1.07 μg/ ml, Supplementary Fig. S1a) or Vero cells (CC 50 = 555.4 ± 1.05 μg/ml, Supplementary Fig. S1b). Additionally, MOM inhibited virus-induced cytopathic effect (CPE) in RD and Vero cells ( Supplementary Fig. S1c). Infected cells exhibited CPE characterized by cells rounding, and the presence of crescent-shaped nuclei with condensed chromatin due to CPE at 16 h post infection. The percentage of cells with crescent-shaped nuclei decreased after MOM treatment in a dose-dependent manner ( Supplementary Fig. S1d,e). It was accompanied by diminished expression of EV71 proteins, including viral caspid (VP2), RNA dependent RNA polymerase (3D p°l ), and protease (3CD pr°) , in MOM-treated infected cells. Protein levels of 3D p°l were reduced by 98% and 99%, respectively, in infected RD cells treated with 156 and 312 μg/ml of MOM (Fig. 1b). Expression levels of VP2 were lowered by more than 99% in infected cells treated with 156 and 312 μg/ml (Fig. 1b). The 50% inhibitory concentration (IC 50 ) of MOM for inhibitory effect on EV71 in RD cells was 45.92 ± 1.05 µg/ml (Fig. 1c). These findings indicate that MOM possesses an anti-EV71 activity.

Identification of antiviral constituents in MOM.
To identify the antiviral constituents in MOM, we diluted the dark green syrup of MOM in water and partitioned with equal volume of ethyl acetate (EtOAc) (Fig. 2a). The aqueous and organic fractions were evaporated, and the resulting fractions MOMW and MOME were obtained. These fractions were tested for their antiviral activity. Both MOMW and MOME fractions reduced plaque formation for all viral strains tested (Fig. 2b). At these concentrations, MOMW (CC 50 = 153.7 ± 1.07 μg/ ml, Supplementary Fig. S2a) and MOME (CC 50 = 21.61 ± 1.08 μg/ml, Supplementary Fig. S2b) showed low cytotoxicity to Vero cells, as determined by standard neutral red assay. These findings suggest that these extracts have some biologically active components in common. The MOMW and MOME were dissolved in DMSO, and subject to liquid chromatography coupled with mass spectrometry (LC/MS) analysis. The base peak chromatogram is shown (Fig. 2c). The compound with mass-to-charge ratio (m/z) of 359.07 was most abundant in the analyte derived from aqueous fraction, and was present in similar quantity in the analyte derived from EtOAc fraction.  and 4643 strains for 1 h, and were overlaid with 0.3% agarose in DMEM/2% FBS containing 0, 39, 78 μg/ ml of MOMW or 0, 10, or 20 μg/ml of MOME. Vero cells were incubated for 5 days. Cells were fixed in 10% formalin and stained with 1% crystal violet solution. Representative plates are shown here. (c) MOMW and MOME, as well as the solvent DMSO were subject to UPLC-MS analysis. The base peak chromatogram is shown. The retention time and m/z for some of base peaks are shown. A representative experiment out of three is shown.  39,78 or 156 μg/ml RA for 16 h. Cellular protein was harvested, and was subject to western blotting with antibodies to 3D and β-actin. The cropped images of the blots are shown. The full-length blots are presented in Supplementary Fig. S9. A representative experiment out of three is shown. (c) RD cells were mock-or infected BrCr at an m. o. i. of 0.05 in the absence or presence of 19, 39, 78 or 156 μg/ml RA for 16 h. Cellular protein was harvested, and was subject to western blotting with antibodies to VP2 and β-actin. The cropped images of the blots are shown. The full-length blots are presented in Supplementary Fig. S10. A representative experiment out of three is shown. (d) RD cells were similarly infected as described in (b), and total RNA was isolated. The level of EV71 genomic copy was determined by quantitative reverse transcription PCR, and normalized to the level of β-actin. Data  Fig. S4c). The levels of VP0 were 42% and 63% lower in similarly treated BrCr-infected RD cells compared to cells infected alone. The discordant decrease in levels of these proteins implies that the processing of VP0 to VP2 may be affected. The ratio of the level of VP2 to that of VP0 decreases in MOM-or RA-treated cells in a dose dependent manner ( Supplementary Fig. S4d) Additionally, RA inhibited synthesis of viral genomic synthesis in EV71-infected RD and Vero cells in a dose-dependent manner ( Fig. 3d and Supplementary Fig. S4e). The level of EV71 RNA in infected RD cells decreased by 71% and 78%, respectively, upon treatment with 78 and 156 μg/ml RA (Fig. 3d). Similarly, its level in infected Vero cells declined by 79% after treatment with 312 μg/ ml RA (Supplementary Fig. S4e). Decreases in viral RNA replication and protein synthesis is accompanied by reduction in progeny virus. Extracellular and intracellular viral particles decreased by 91% and 90% in infected RD cells upon treatment with 78 μg/ml RA (Fig. 3e). The IC 50 of RA for inhibitory effect on EV71 in RD cells was 43.07 ± 1.05 μg/ml ( Supplementary Fig. S4f). These findings validate the anti-EV71 activity of RA.
RA inhibits EV71 infection at attachment and post-attachment stages. As a first step in studying the antiviral mechanism of RA, we conducted a time-of-addition assay to identify the stages at which RA inhibits EV71 infection. RA was administered at different stages of BrCr infection (Fig. 4a), and expression level of viral protein VP1, indicative of viral replication, was studied. When cells were pre-treated with RA, RA failed to suppress EV71 infection, and slightly enhanced it ( RA inhibits viral replication during viral attachment. It is possible that RA may interact with virions and prevent their interaction with cellular receptors. To test this hypothesis, viral particles were incubated with 156 μg/ ml RA on ice for 1 h, and any excess of RA was removed using centrifugal filtration. The remaining viral titer was determined using plaque assay. Viral titer in RA-treated group was 4.28 ± 1.17 × 10 4 PFU/ml, being lower than that of vehicle control group (1.18 ± 0.25 × 10 5 PFU/ml) (Fig. 4c). Such finding suggests that RA may interact with virions, and interfere with their binding to cellular receptors.
RA represses EV71-induced switch between cap-dependent and IRES-dependent translation. The ability of RA to inhibit viral replication during post-absorption phase prompts us to investigate the underlying antiviral mechanism. It is known that EV71 infection induces the switch between cap-dependent and IRES-dependent translation 20,23 . To study whether RA affects this molecular process, RD cells were transfected with a bicistronic plasmid pRHF-EV71-5′UTR (Fig. 5a), infected with BrCr for 1 h, and treated with RA. The ratio of firefly luciferase activity to Renilla luciferase activity (Fluc/Rluc) is indicative of the relative activities of IRES-dependent and cap-dependent translation. It was 27.66% higher in EV71-infected cells than in uninfected cells (p < 0.001) (Fig. 5b). RA mitigated such increase in Fluc/Rluc in a dose-dependent manner (Fig. 5b). It is possible that RA may disturb EV71-induced switch between cap-dependent and IRES-dependent translation. It is known that viral protease-2A pr° hydrolyzes translation initiation factor eIF4G resulting in shutdown of cap-dependent translation 40 . To explore the possibility that RA treatment may inhibit EV71-induced eIF4G cleavage to block the shutdown of cap-dependent translation, we examined the expression level of eIF4G in EV71-infected cells. RD cells were infected with EV71 at a multiplicity of infection (m. o. i.) of 20, and treated with 156 μg/ml RA under conditions depicted in Fig. 4a. EV71 infection led to complete cleavage of eIF4G (Fig. 5c, condition 1). RA pre-treatment did not affect eIF4G cleavage (Fig. 5c, condition 2). When RA was added during either viral adsorption or post-adsorption phases, eIF4G cleavage was partially inhibited (Fig. 5c, condition 3 & 4). When RA was given during both viral adsorption and post-absorption phases, it acted synergistically to inhibit eIF4G cleavage (Fig. 5c, condition 5). These findings suggest that RA inhibits EV71-induced shutdown of cap-dependent translation through preservation of intact eIF4G.
We studied if RA inhibits IRES-dependent translation of enteroviral protein. Initiation of IRES-dependent translation is regulated by ITAFs, such as hnRNP A1. hnRNP A1 re-localizes to cytoplasm during infection, and interacts with IRES within EV71 5′UTR. To study the hypothesis that RA may interfere with this process, we transfected RD cells with an expression plasmid encoding a GFP-tagged hnRNP A1 (pGFP-hnRNP A1); infected the transfected cells with EV71; and analyzed the effect of RA on cytoplasmic translocation of GFP-tagged hnRNP. The GFP-tagged hnRNP A1 was localized to nuclei of mock-infected cells, and translocated from nuclei to cytoplasm in EV71-infected cells (Fig. 5d). Relocation of hnRNP A1 in infected cells was inhibited by treatment with 156 μg/ml RA (Fig. 5d). The percentage of cells showing cytoplasmic accumulation of GFP-tagged hnRNP A1 (i.e. cytoplasmic GFP-positive cells) was quantified using a high throughput imaging technique. The percentage of such cells was 14.39 ± 2.91% in mock-infected group, but it increased to 49.74 ± 4.52% in EV71-infected group (Fig. 5e). RA treatment decreased the percentage of cytoplasmic GFP-positive cells in a dose dependent manner. The percentage of infected cells showing hnRNP A1 relocation declined to 30.92 ± 3.97% upon treatment with 156 μg/ml RA (Fig. 5e). It is possible that RA may downregulate hnRNP A1 translocation and inhibit IRES-dependent translation.
are expressed relative to that of untreated cells. The results are means ± SD of three separate experiments. ***P < 0.001, ****P < 0.0001 vs. infected cells without treatment. (e) RD cells were infected with BrCr at an m. o. i. of 5, and treated with 0, 39 or 78 μg/ml RA for 9 h. Extracellular and intracellular viral particles were harvested for plaque assay for titer determination. The results are means ± SD of three separate experiments. *P < 0.05, **P < 0.01, vs. infected cells without treatment.

RA suppresses EV71-induced phosphorylation of p38 kinase. Subcellular distribution of hnRNP A1
is regulated by p38 signaling 41 . The ability of RA to suppress EV71-induced hnRNP A1 redistribution raises the possibility that RA may regulate relocation of hnRNP A1 through its effect on p38 pathway. To test this hypothesis, we infected RD cells with BrCr at an m. o. i. of 20; treated the infected cells without or with 156 μg/ml RA; harvested their total protein for immunoblotting analysis of phosphorylated p38 kinase. EV71 infection induced biphasic phosphorylation of p38 kinase. Phosphorylation of p38 kinase slightly increased at around 15 to 30 min post infection. A stronger phosphorylation of p38 kinase in EV71-infected cells was observed during the period from 2 h to 6 h post infection (Fig. 6a). RA treatment repressed EV71-induced p38 phosphorylation (Fig. 6b). To demonstrate the role of p38 kinase in EV71 infection, RD cells were treated with p38 inhibitor SB202190, and infected with EV71. SB202190 inhibited expression of viral capsid protein VP1 in a dose-dependent manner (Fig. 6c). Production of progeny virus decreased by 92% in EV71-infected cells treated with 50 μM SB202190, as compared with that of vehicle control-treated cells (Fig. 6d). These results suggest that EV71-induced p38 phosphorylation can be suppressed by RA. Interestingly, the cleavage of eIF4G in EV71-infected cells was not altered upon SB202190 treatment (Fig. 6c), indicating that EV71-induced eIF4G cleavage is independent of p38 activation.
To study the causal relationship between p38 activity and hnRNP A1 relocation, we studied the effect of SB202190 on hnRNP A1 distribution. RD cells were transfected with pGFP-hnRNP A1, treated with SB202190, and infected with BrCr. SB202190 decreased EV71-induced relocation of hnRNP A1 in a dose-dependent manner (Fig. 6e). Such finding suggests that EV71 may induce cytoplasmic accumulation of hnRNP A1 via p38 signaling.  Supplementary Fig. S11. A representative experiment out of three is shown. (c) Interaction between virions and RA reduces the viral infectivity. The viral particles were incubated with 156 μg/ml RA or 0.15% DMSO at 4 °C for 1 h. The reaction was filtered through filter, and the retentate was analyzed for viral titer. Results are means ± SD of three independent experiments. ****P < 0.001, vs. control treatment group. RA suppresses p38 activation through an antioxidative mechanism. EV71 infection induces oxidative stress in host cells [28][29][30] , which activates phosphorylation of p38 kinase 42,43 . The antioxidant capacity of RA was determined using ferric reducing antioxidant power (FRAP) assay as 2.235 ± 0.035 g Trolox/g RA. It is hypothesized that RA scavenges EV71-induced ROS generation, and suppresses activation of p38. To study this hypothesis, we stained EV71-infected cells with H 2 DCFDA or CellROX Deep Red dye to determine the intracellular ROS generation. ROS levels increased in BrCr-, 1743-, and 4643-infected cells, as compared with that of mock-infected cells. RA treatment reduced ROS generation in EV71-infected cells (Fig. 7a and Supplementary  Fig. S5a,b). To assess whether RA acts to block ROS-induced p38 activation, we treated RD cells with 500 μM H 2 O 2 , and examined the effect of RA on p38 phosphorylation. As expected, hydrogen peroxide induced p38 phosphorylation, but RA treatment mitigated p38 activation. Consistent with this, hydrogen peroxide induced cytoplasmic accumulation of hnRNP A1 in RD cells (Supplementary Fig. S5c). These results suggest that the ROS-induced p38 phosphorylation and hnRNP A1 translocation can be inhibited by RA.
RA suppresses EV71-induced phosphorylation of epidermal growth factor receptor substrate 15 (EPS15). EPS15, a target protein of p38 44 , regulates intracellular trafficking 45 , which is essential to viral  i. for western blotting as described in (a). The cropped images of the blots are shown. The full-length blots are presented in Supplementary Fig. S18. A representative experiment out of three is shown.

RA protects mice from EV71 infection.
To study whether RA protects mice from EV71 infection, seven-day-old specific pathogen-free ICR mice were infected with mouse-adapted virus intraperitoneally, and were subsequently treated with RA (50 mg/kg) or PBS at daily intervals for 14 day. The survival rate of RA-treated mice (66.67%, n = 6) were higher than that of PBS-treated mice (33.33%, n = 6) ( Supplementary  Fig. S6a). Moreover, a higher percentage of RA-treated mice were rehabilitated from illness than PBS-treated mice ( Supplementary Fig. S6b,c). These results suggest RA may offer protection against EV71 infection in vivo.

Discussion
The present study has shown that the MOM exhibits anti-EV71 activity. RA, identified as an antiviral constituent of MOM, inhibits viral replication, and offers protective effect against EV71 infection in vivo. Mechanistically, RA suppresses attachment of virion to host cells, eIF4G cleavage, and cytoplasmic relocation of hnRNP A1. RA acts as ROS scavenger to block ROS signaling and p38 kinase activation, resulting in diminished cytoplasmic hnRNP A1 relocation and EPS15 phosphorylation. Our findings suggest that RA exerts its antiviral effect via multiple mechanisms.
MO originates from Southern Europe, Mediterranean region, Middle East countries and North Africa and is cultivated worldwide now 32 . MO is commonly consumed in food and beneficial drinks. Choi et al. have shown that the MO extract prepared by extraction in water at 40 °C possesses antiviral activities 39 . MO extract prepared by methanolic extraction at 60 °C has shown an anti-EV71 activity in our study. However, methanolic extraction at room temperature yielded an extract with no anti-EV71 activity in Choi's study 47 . A major difference may lie in the temperature at which the extraction was performed. The higher extraction temperature employed in our study may be beneficial to extraction of biologically active antiviral constituents from MO. The study of the antiviral mechanism of MOM necessitates identification of biologically active constituents. The MOM diluted in water was further partitioned with an equal volume of EtOAC. Both layers were further processed, and the resulting fractions retained anti-EV71 activity (Fig. 2). Using ultra performance liquid chromatography coupled with mass spectrometry (UPLC-MS) as well as nuclear magnetic resonance (NMR) spectroscopy, we found that RA was present in both aqueous and organic fractions. Chemically speaking, RA is an ester of caffeic acid and 3,4-dihydroxyphenylacetic acid 48 . It is found in many plants within Labiatae family 49 . Several plants of this family, including Salvia miltiorrhiza (danshen) 50 and Ocimum basilicum 51 , show anti-EV71 activities. Rosmarinic acid has also been identified as an antiviral constituent of Salvia miltiorrhiza 52 . Our finding that RA had antiviral activity against a number of clinical relevant EV71 strains, namely 1743 and 4643 strains, is interesting (Fig. 3a). This raises the hope of developing a therapeutic against EV71, and possibly other neurotropic viruses.
A time-of-addition assay has been used to examine the mechanistic stages at which RA acts to inhibit EV71 infection process. RA acts during viral adsorption and post-adsorption phases (Fig. 4b). Incubation of virus with RA reduced its infectivity (Fig. 4c), implying that RA may bind directly to enteroviral particles and interfere with their attachment to cellular receptor. It is known that MO extract and RA can directly interact with enveloped viral particles. The density of HIV-1 virion increases after treatment with an aqueous MO extract 36 , possibly as a consequence of chemical modification of viral particle by and binding of extract constituents. Additionally, MO extract and RA inhibit attachment of herpes simplex virus type I 35 . It remains elusive how MO constituents and RA act to interfere with viral attachment, and whether a similar mechanism is involved for enveloped viruses and non-enveloped enterovirus.
Several mechanisms may be held accountable for the anti-EV71 activities of RA at post-adsorption phase. RA suppresses the EV71-induced subversion of host cell translation and initiation of viral translation. Cleavage of eIF4G and PABP by 2A pr° and that of PABP by 3C pr° are implicated in this process 24 . It is probable that RA inhibits the viral proteases. For instance, 2A pr° is a cysteine protease homologous to the trypsin-like family of serine protease 53 . RA can react with the active site of another cysteine protease caspase 3 54 . It is possible that RA may inhibit viral proteases in a similar way. Additionally, RA may suppress viral translation and reduce the expression of viral proteases. Initiation of viral translation involves association of IRES-specific trans-acting factors (ITAFs) and ribosomal subunits with type I IRES 19,20 . An ITAF, hnRNP A1, relocates from nucleus to cytoplasm during infection, and binds to stem loops II and VI of EV71 IRES to enhance translational initiation 21,22 . The inhibitory effect of RA on hnRNP A1 translocation may lead to reduction in viral translation and replication. Interestingly, the discordance in the expression levels of VP0 and VP2 implies that the processing of VP0 is adversely affected by MOM and RA. It is probable that RA and biologically active constituents of MOM may inhibit viral protease and VP0 cleavage.
Subcellular distribution of hnRNP A1 can be regulated by ROS. Arsenite that induces oxidative stress causes hnRNP A1 relocation to cytoplasm [55][56][57] . We have also found that treatment of RD cells with 500 μM hydrogen peroxide induces cytoplasmic relocation of hnRNP A1 (Supplementary Fig. S5c). EV71 infection is known to instigate mitochondrial ROS generation and/or NADPH oxidase activation 28,29 . The increase in ROS may activate downstream signaling and hnRNP A1 translocation. Given the potent antioxidative capacity of RA (Fig. 7a), it may scavenge ROS to block such molecular events.
Activation of MAPK-related pathways has been implicated in pathogenesis of EV71 58,59 . Increased phosphorylation of p38 kinase was observed during the early (15 min to 30 min) and late (2 to 6 h p. i.) stages of EV71 life Scientific RepoRts | 7: 12264 | DOI:10.1038/s41598-017-12388-2 cycle in RD cells (Fig. 6). It is likely that p38 kinase activation is essential to EV71 replication. Pharmacological inhibition of p38 kinase significantly reduced viral translation (Fig. 6e). Consistent with this, p38 kinase signaling cascade plays a regulatory role in subcellular distribution of hnRNP A1 41,58 . Activation of this pathway enhances hnRNP A1 relocation, while its inhibition thwarts the process 41 . In agreement with this, we have shown that treatment with p38 kinase inhibitor blocked EV71-induced cytoplasmic relocation of hnRNP A1 (Fig. 6e). Our finding that RA treatment suppressed EV71-induced activation of p38 kinase is intriguing (Fig. 6). It implies that RA acts on p38 kinase or upstream of it. It has been shown that ROS can activate p38 kinase cascade 60,61 . Hydrogen peroxide treatment induced phosphorylation of p38 kinase, which was inhibited in the presence of RA (Fig. 7b). As aforementioned, RA acts as ROS scavenger, and in this manner, may prevent p38 kinase activation. It is not unprecedented. It has been shown that antioxidant treatment suppresses ROS-induced p38 kinase activation 61 . Additionally, RA is known to regulate the expression of antioxidative enzymes. For example, administration of RA to aging mice induces expression of superoxide dismutase, catalase and glutathione peroxidase in their liver and kidney 62 . These findings suggest that RA may repress activation of p38 kinase and translocation of hnRNP A1 through an antioxidative mechanism.
EPS15, another target of p38 kinase, plays an important role in EV71 infection 63 . EPS15 co-localizes with α-adaptin and clathrin 64 , and is believed to modulate clathrin-mediated endocytosis (CME) and membrane trafficking 45 . Recent findings have suggested that phosphorylation regulates the biochemical activity of EPS15 45 . EPS15 can be phosphorylated by p38 at Ser796 44 , and this phosphorylation site is implicated in regulation of endocytic process. Silencing of EPS15 has a suppressive effect on enteroviral infection 63 . EPS15 may be involved in CME of virus, and in membrane trafficking essential to EV71 replication. It has been recently shown that enterovirus utilizes the endocytic machinery to distribute cholesterol to replication organelles, where cholesterol facilitates the replication and 3CD pr° processing 46 . The antiviral activity of RA may be in part accounted for by its ability to inhibit p38 kinase and EPS15 phosphorylation.
Based on our findings, we propose a model for antiviral activity of RA ( Supplementary Fig. S7). RA acts in multiple ways to inhibit viral replication. RA binds to viral particles to interfere with their attachment to receptors and internalization; suppresses eIF4G cleavage; removes ROS and inhibits activation of p38 kinase, blocking hnRNPA1 translocation and EPS15 phosphorylation. were extracted with MeOH (5 L) at 60 °C for 4 hours twice. Crude extract of MO was filtered through Waterman No. 1 filter paper (GE Healthcare Life Sciences; Chicago, IL, USA). The liquid fraction was concentrated with a rotary evaporator to give dark green syrup named MOM (with a weight of 94.33 g). MOM was diluted in water and further partitioned with equal volume of EtOAc. The water and EtOAc layers were separated, and concentrated with a rotary evaporator to give dark brown concentrate (MOMW) and dark green powder (MOME), respectively. MOMW was diluted in water for cell studies and in DMSO for chemical analysis. MOME was dissolved in DMSO. The MOMW, MOME and solvent DMSO were chromatographed on an ACQUITY UPLC BEH C18 column (1.7 μm, 2.1 × 100 mm, Waters, USA) using a gradient of 5% ACN to 50% ACN. Each fraction was subject to electrospray ionization tandem mass spectrometry (ESI-MS/MS; HDMS-G1, Waters, USA) in negative ion mode. For purification of biologically active compounds of MOMW, the syrup was diluted with water. It was subject to chromatography on Diaion HP-20 column, followed by chromatography on Sephadex LH-20 column. The eluents were water, water/methanol (1:1, V/V) and methanol. The eluate MOMW-1-1 was further evaporated to obtain a light-brown syrup. The sample was analyzed by proton nuclear magnetic resonance spectroscopy (Bruker Avance III-400 NMR spectrometer; Bruker Daltonik GmbH, Bremen, Germany).  66 . Virus stocks were propagated in Vero or RD cells as previously described 30 . Mouse adapted EV71 strain, MP4, was prepared from an infectious clone, MP4/y5 67 . Infectious clone was digested by MluI, and viral RNA was prepared using the MEGAscript T7 Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA). RD cells were set in 6-well plates at 4 × 10 5 per well and incubated overnight. Three microgram of viral RNA was transfected into RD cell using lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Virus particles were harvested at 24 h p. i. in three freeze-thaw cycles. The MP4 virus was further propagated in RD cells once before animal study. The titer of resulting viral stock was determined by plaque assay. Viral plaque assay and modified plaque assay screening for antiviral activity. The titer of virus was determined by plaque assays with Vero or RD cells. The viral supernatant was serially diluted with serum free medium in a ten-fold manner. The monolayer cells at a confluence of 80% were infected with diluted viral supernatants, and the viral titer was quantified in the form of plaque forming unit (PFU) per ml as previously described 30 .

Cells and viruses. Human rhabdomyosarcoma cells (RD
For plaque reduction assay, RD and Vero cells in six-well plates were respectively infected with 100 and 200 PFU of virus for 1 h at 37 °C. After removal of unabsorbed virus, cells were overlaid with 0.3% agarose in medium supplemented with 2% serum and an indicated concentration of extract or RA. In the control group, mock infected cells were treated with the same concentrations of test samples. The plaque size and number were determined for assessment of the antiviral activity of extract or RA. Determination of the genomic copy number of EV71. For quantification of intracellular viral RNA, total cellular RNA was extracted from EV71-infected cells using TRIzol reagent (Thermo Fisher Scientific, USA) according to the manufacturer's instructions. Concentration of total cellular RNA was determined by spectrophotometer (Nanophotometer, Implen, Germany). The relative copy number of EV71 was quantified by quantitative reverse transcription polymerase chain reaction (qRT-PCR) as previously described 30 .
Infectivity inhibition assay. EV71 (BrCr) was diluted to 1 × 10 5 PFU/ml in 10 ml, and incubated with or without 156.25 μg/ml RA for 1 h on ice. The medium containing unadsorbed virus was transferred to Amicon Ultra centrifugal filter unit with Ultracel-100 membrane (#UFC910024, Merck Millipore Darmstadt, Germany), and RA was removed by centrifugation at 5000 × g at 4 °C for 30 min. Viral particles in the retentate were resuspended in 1 ml serum-free DMEM. The titer of virus was quantified by plaque assay.
Bicistronic reporter assays for detection of IRES activity. The biscistronic construct pRHF-EV71-5′UTR contains 5′UTR region of EV71 between Renilla luciferase (Rluc) and firefly luciferase (Fluc) genes, and has been used for assay of IRES activity 68 . The plasmid DNA (0.5 µg) was transfected into RD cells with Lipofectamine 2000 according to manufacturer's protocols. Twelve hours after transfection, cells were un-or infected with 6 × 10 6 PFU of virus for an hour. One hour later, DMEM/2% FBS supplemented with indicated final concentrations of RA was added to each well. After 6 h, cell lysates were harvested and activities of Rluc and Fluc were measured with dual-luciferase reporter assay system (Promega, Madison, USA) and GloMax 20/20 single tube luminometer (Promega, Madison, USA) according to the manufacturer's instructions. hnRNP A1 translocation assay. To assess translocation of nuclear protein hnRNP A1, we employed the construct pGFP-hnRNP A1, which encodes a green fluorescent protein (GFP) tagged hnRNP A1 protein. RD cells were transfected with 0.5 µg of pGFP-hnRNP A1 plasmid using Lipofectamine 2000. Two days later, the cells were infected with or without 6 × 10 6 PFU BrCr in serum-free medium at 37 °C for 1 h. After virus adsorption, DMEM containing 2% FBS and 78 or 156 μg/ml of RA was added, and the infected cells were incubated for another 6 h at 37 °C. Infected and mock treated cells were fixed with 10% formalin for 30 min, and the cell nuclei were stained with Hoechst 33342 (Thermo Fisher Scientific Inc., Waltham, MA, USA) in PBS at room temperature. The fluorescent images were obtained using IN Cell Analyzer 1000 (GE Healthcare Life Sciences, USA). The region of nucleus was defined as a blue fluorescent region. The translocation of hnRNP A1 from nucleus to cytoplasm is visualized as an increase in the intensity of green fluorescence in cytoplasm over that in nucleus. The percentage of cells showing hnRNP A1 relocation was quantified from 30 random fields per well.
For confocal microscopic study, RD cells were cultured in poly-D-lysine-coated 35 mm glass bottom culture dish (MatTek Corporation, MA, USA), and treated as described above. The cells were analyzed with Zeiss LSM 780 system (Carl Zeiss Microimaging GmbH, Heidelberg, Germany) as previously described 30 . Western blotting. Cellular lysate was separated by SDS-PAGE and analyzed by western blot as previously described 28 . Antibodies used in this study were listed in Supplementary Table S1.
Ferric reducing/antioxidant power (FRAP) assay. The antioxidant power of RA was measured as previously described 69 . This method is based on the ability of antioxidant to reduce ferric-tripyridyltriazine complex to the blue-colored ferrous form, which has an absorption maximum at 593 nm. The absorbance was analyzed by ELISA reader (VersaMax, Molecular Devices LLC, CA, USA).

Detection of Cellular Reactive Oxygen Species (ROS). For determination of the intracellular ROS
formation, cells were stained with cell-permeable fluorogenic dyes, CellROX Deep Red reagent or H 2 DCFDA as previously described 28,30 .