Halogenated Chrysins Inhibit Dengue and Zika Virus Infectivity

Dengue virus infection is a global threat for which no specific treatment has not been established. Previous reports suggested chrysin and flavanone derivatives were potential flaviviral inhibitors. Here, we reported two halogenated chrysins, abbreviated FV13 and FV14, were highly potent against DENV1-4 and ZIKV infectivities with the FV13 EC50 values of 2.30 ± 1.04, 1.47 ± 0.86, 2.32 ± 1.46, 1.78 ± 0.72 and 1.65 ± 0.86 µM; and FV14 EC50 values of 2.30 ± 0.92, 2.19 ± 0.31, 1.02 ± 0.31, 1.29 ± 0.60 and 1.39 ± 0.11 µM, respectively. The CC50s to LLC/MK2 of FV13 and FV14 were 44.28 ± 2.90 μM, 42.51 ± 2.53 µM, respectively. Mechanism of drug action studies suggested multiple targets but maximal efficiency was achieved with early post infection treatment. This is the first report showing a high potency of halogenated chrysins for development as a broad-spectrum anti-flaviviral drug.

selected 8 flavonoid derivatives (Fig. 1) and tested their effect on DENV2 NGC in a LLC/MK2 cell-based system. Briefly, the compounds at final concentrations of 10 μM and 25 μM in DMSO were added to DENV2 infected LLC/MK2 cells and the effect on viral particle production was measured by plaque titration of the culture supernatants. Interestingly, two halogenated chrysins, FV13 and FV14, and a chalcone derivative, CH1, strongly inhibited virus production with >99% (Table 1). We also tested cytotoxicity of FV13, FV14, and CH1 to verify the viral inhibition in LLC/MK2 cells ( Table 2). The viabilities of FV13 treated cells were 81.00 ± 2.69% and 59.40 ± 2.42%, at 10, and 25 µM respectively, whereas the values for FV14 treated cells were 60.24 ± 3.31% and 60.86 ± 3.57%. We also examined Vero, THP-1, HepG2, and HEK-293 cell viabilities in the presence of selected compounds ( Table 2). The results suggested that human-derived cell lines, THP-1, HepG2, and HEK-293 were generally tolerant to the compounds with >85% viability, except for 25 µM CH1 to HepG2 and HEK-293. Vero cells, in contrast, were very sensitive to FV13 and FV14, but not CH1. From this data, we decided to further explore the efficacy of FV13 and FV14 as potential candidates of flaviviral inhibitors.
FV13 and FV14 effectively inhibited all dengue serotypes and Zika virus. Next, we examined the efficacy of FV13 and FV14 with DENV1-4 and ZIKV (Table 3). Compound at various concentrations were added to virus infected LLC/MK2 cells and viral titers were accessed by plaque titration of supernatants as previously described. The EC 50 values of both compounds against DENV1-4 and ZIKV infectivity were ranged between 1-3 µM ( Table 3). Note that FV13 and FV14 showed similar efficacy against all tested viruses. Based on these results, we suggested that both compounds could potentially be broad-spectrum flaviviral inhibitors. We further explored cytotoxic concentrations (CC 50 ) of FV13 and FV14 to LLC/MK2 cells and both compounds showed similar cytotoxicity after 48 h incubation period at 44.28 ± 2.90 μM and 42.51 ± 2.53 µM, respectively ( Fig. 2A). Selectivity indices (CC 50 /EC 50 ) ( Table 3) indicated that the compounds would not be categorized as drugs with narrow therapeutic index (NTI-drug) and therefore were suitable for further consideration. We also tested the effect of long term incubation on cytotoxicity (Fig. 2B). Whereas FV13 was similarly non-toxic at both 48 and 120 h, FV14 showed significantly greater cytotoxicity at 120 h (p-value < 0.01) (Fig. 2B). The limited toxicity of FV13 might reflect rapid degradation of the compound. Therefore, we tested FV13 decomposition in DMSO and incubated at room temperature. FV13 NMR signals at 24, 72, and 120 h ( Fig. 3A-C) clearly showed that the compound was stable for at least 120 h at room temperature. Based on the efficacy, toxicity, and stability results, we then chose FV13 to explore its molecular target and mechanism of action.

FV13 may inhibit multiple targets of the viral life cycle.
To access the stage of the viral life cycle affected by FV13, we performed a time-of-addition assay. Briefly, FV13 (10 μM) was added to DENV2 infected LLC/MK2 cells at various time points as described in Methods section. DMSO was added in parallel as a control treatment. Supernatants were collected for plaque titration and cell lysates were collected for RT-qPCR analysis. The titers of FV13-treated samples showed a 2-log decrease from DMSO-treated baseline when the drug was added between 2-10 hours post-infection (hpi). If the drug was added at 12 hpi or later, only 1-log decrease was observed (Fig. 4A,B). The intracellular viral RNA analysis showed a 1-log decrease in FV13-treated samples with drug addition between 2-8 hpi (Fig. 4C). Inhibition gradually declined at 10 hpi and at 12-24 hpi (Fig. 4D), no titer difference was observed between FV13-and DMSO-treated samples suggesting that FV13 did not directly inhibit viral replication. A 1-log difference between plaque and intracellular RNA titers suggests   Table 3. Selectivity of halogenated compounds. a EC 50 were determined by plaque assay in 96-well plates. b EC 50 was determined by plaque assay in 24-well plates. c Selectivity index (CC 50 /EC 50 ). Means ± standard error of the means (SEM) of two independent experiments, in which each experiment was performed in triplicate, were reported.
that a post-replication mechanism could be the target of this compound. It also suggests that FV13 may inhibit viral infectivity with at least two mechanisms of action. The first molecular target could interact with FV13 earlier than 2 hpi and its effect was still visible until 8-10 hpi. The other target could locate at late post-replication steps such as assembly, maturation, or release because a 1-log discrepancy between plaque and RT-qPCR titers indicated genome replication was unaffected. Further investigations will reveal insights into the actual target of this compound.
Maximal efficacy of FV13 is achieved early post-attachment. We further explored whether the compound would block the viral infectivity by neutralization. We set up the attachment inhibition assay by incubating DENV2 with LLC/MK2 cells at 4 °C for 1 h and FV13 was added before adsorption (pre-attachment), simultaneous with adsorption (co-attachment), and after adsorption (post-attachment) (Fig. 5A). DMSO was used as a mock treatment and added in parallel to FV13 to the experiment. Cells were incubated until cytopathic effects appeared under microscope (Fig. 5B). Supernatants and cells were then analyzed by plaque titration (Fig. 5C) and RT-qPCR ( Fig. 5D), respectively. Results indicated that the compound did not interfere with either pre-or co-attachment steps, but rather at post-attachment with plaque titer reductions of 40.59 ± 2.83% (Fig. 5C) and viral RNA reductions of 64.07 ± 3.37% (Fig. 5D). Therefore, the major targets of FV13 inhibition appear to function after attachment, possibly at the fusion or translation steps. Fusion inhibition [22][23][24]  Halogenated chrysins inhibited replicon replication. Next, we studied FV13 inhibition in a DENV2 replicon replication system to investigate whether translation and replication could be implicated as drug targets. BHK-21/DENV2 replicon cells 28 were treated with 5 and 10 µM FV13 and incubated for 72 h before measuring replication efficiency by quantifying replicon RNA. Ribavirin, a known dengue virus replication inhibitor 29 , was used as a positive control. We found that FV13 efficiently inhibited DENV2 replicon replication by 83.41 ± 2.04%, Values are the means ± standard error of means (SEM) from two independent experiments, in which each experiment was performed in triplicate, were reported. Asterisks indicate statistically significant differences using paired t-test as follows; *p-value < 0.05, **p-value < 0.01, ns = no significance. and 82.71 ± 4.01% when added at 5 µM and 10 µM, respectively (Fig. 6A). The inhibitory effects were similar to those of ribavirin at 5 µM or 10 µM concentrations, 87.37 ± 10.71% and 89.88 ± 4.83%, respectively. Cytotoxicity was also measured and the results showed that most replicon cells (>99%) were viable (Fig. 6B) throughout the experiment. This suggested that FV13 interfered with a sustainable self-replicating viral translation and replication system. Hundreds of viral and cellular factors are involved in viral translation and replication. We primarily explored DENV protease because previous reports indicated as the target of flavonoid compounds 16,30 . However, our in vitro DENV2 NS2B/3 protease assay (Supplementary 2) and molecular docking results (Supplementary 3) indicated that FV13 did not target DENV2 protease. However, several host factors are involved in viral translation step and it is still possible that the compounds target one of these critical factors, subsequently resulting in the inhibition of viral translation.

Discussion
This is the first report of two halogenated chrysins, FV13 and FV14, showing strong anti-flaviviral efficacies towards the unmodified, naturally derived flavonoids. Active anti-flaviviral (DENV2) compounds accessed by cell-based assay include quercetin (EC 50 of 118.12 µM) 14 13 . Obviously, FV13 and FV14 were 20-100 times more potent than previously reported flavonoids. The compounds also showed broad spectrum activities against all dengue serotypes and a Zika virus (Table 3) making them a strong candidate for further drug development.
Moreover, the compounds showed selectivity indices of 20-40, suggesting an applicable therapeutic safety for animal toxicity study. We also examined the cytotoxicity of FV13, FV14 and CH1 to THP-1, HEK-293, and HepG2 cell lines, which are derived from monocytes, renal cells, and hepatocytes, respectively. Results showed that FV13 and FV14 were non-cytotoxic with >80% cell viability ( Table 2). Both halogenated chrysins were relatively non-toxic similar to other naturally-derived flavonoids. Previous pharmacokinetic studies of chrysin in animals and humans showed poor oral bioavailability [31][32][33] . However, these drugs could be administered by the intravenous route for dengue drug treatment because intravenous fluid replacement is a general practice assigned to patients with impending shock 3 . Moreover, animals can be assigned with broader range of drug since the selectivity indices were 20-40 in cell-based system. Our further investigation will focus on in-vivo toxicity and efficacy studies by examining toxic metabolites, half-life of the drug excretion, and monitoring liver and kidney functions.
Undoubtedly, the halogens at the R3 and R5 positions ( Fig. 1) convey strong biological activities to FV13 and FV14 against flaviviral replication. Moreover, we noticed that FV2, a naturally purified flavone and a precursor of FV13 and FV14, actively inhibited DENV2 at 10 and 25 µM with 43.00% and 92.80%, respectively. Two hydroxyl groups at R2 and R4 positions were noted as common characteristics of the FV2, FV13, FV14, as well as quercetin, fisetin, baicalein, luteolin, apigenin, etc. Therefore, the hydroxyl groups at R2 and R4 and the high EN groups at R3 and R5 positions could be responsible for crucial biological activity towards flaviviral infectivity.
In search of molecular targets and mechanisms of drug action, a series of experiments directed us to at least two targets located at early and late steps of infection. Although the drug targets are still elusive, evidences suggested viral translation or replication [34][35][36] . Possible targets could also locate at post-replication steps given a 1-log difference between plaque titers and intracellular viral copies (Fig. 4A-D). Multiple viral and host proteins involved in translation or replication are possible compound targets. Further investigation (e.g. chemical affinity-tag purification coupled with identification by liquid chromatography and mass spectrophotometry; western blot analysis for viral protein expression; and escape mutant study by whole genome sequencing) are methods of choice to identify the drug target.
In conclusion, two halogenated chrysins were demonstrated for the first time as potential inhibitors of dengue and Zika infectivity with high efficacy and low cytotoxicity.

Materials and Methods
Compound synthesis, purification and stability. Eight flavonoid derivatives consisting of five flavones (FV), one flavanone (FN), and two chalcones (CH) were selected as representatives for primary screening. FV2 (chrysin) was purchased from Sigma Aldrich, St. Louis, USA. FV4 was extracted and purified from Kaempferia parviflora. FV6 was chemically modified from FV2 by methylation. FN2 was extracted and purified from Boesenbergia rotunda. CH1 and CH2 were purchased from Sigma Aldrich, St. Louis, USA. The purity and identity of each compound was verified by 1 H-NMR Tensor 37 infrared spectrometer (Bruker, Massachusetts, USA).
All compounds were stored as solid powder at room temperature. The stock solutions (50 mM) were prepared in dimethyl sulfoxide (DMSO) (Merck, California, USA) and stored as aliquots at −20 °C until use. and incubated with 1 ml of MEM medium supplemented with 1% FBS as previously described. The compounds were prepared in dimethylsulfoxide (DMSO) at the indicated concentrations before addition to viral infected cells. Supernatants were collected after 5 day incubation and analyzed by plaque titration 9 . In primary screening, DMSO-treated cells served as the untreated control. Compounds that inhibited virus production by ≥90% were selected for further characterization.
Compounds were further analyzed for their effective concentration (EC 50 ) against DENV1-4 and ZIKV. Compounds were prepared in DMSO at 8-12 different concentrations and added to virus infected cells as previously described. Supernatants were collected and the EC 50 values were calculated from nonlinear regression analysis. Results were reported as means and standard error of the means (SEM) of EC 50 values from at least two independent experiments in which each drug concentration was tested in triplicate.  Time-of-drug addition study (TOA). LLC/MK2 cells were seeded in 24-well plate and incubated as previously described. Cells were then infected with DENV2 (M.O.I. of 0.1). FV13 (10 µM) was added at various time points, early time points (−1, 0, 2, 4, 6, 8, 10, and 12 hpi) and late time points (12,14,16,20, and 24 hpi) and the plates were incubated for 5 days. Supernatants were collected to determine the viral titer by plaque titration assay and the cells were collected to determine viral RNA content by RT-qPCR. Results were confirmed by three independent experiments in which each time point was tested in duplicate.
Briefly, LLC/MK2 cells were seeded in 24-well plates and incubated as previously described. Cells were then adsorbed by DENV2 (M.O.I. of 1) diluted in maintenance medium at 4 °C for 1 h with continuously gentle rocking. FV13 at 10 µM was added to DENV2 virus preparation for 1 h before adsorption (pre-attachment), during adsorption (co-attachment), and after adsorption plus three washings with cold PBS to remove external viruses (post-attachment). Cells were incubated in maintenance medium at 37 °C, under 5% CO 2 for 2 days before supernatants and pellets were collected. DMSO-treated samples were used as a no-inhibition control. Pictures were taken using an Eclipse TS100 Inverted Routine Microscope (Nikon, New York, USA). Results were confirmed by three independent experiments.
Replicon inhibition assay. BHK-21 cells expressing a DENV2 replicon (BHK-21/DENV2) were maintained in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), and 0.3 mg/ml G418 (Bio Basic Canada, Ontario, Canada). The protocol was adapted from Boonyasuppayakorn et al., 2014 28 . Briefly, cells (5 × 10 4 /well) were seeded into 24-well plates and were incubated for 1 day at 37 °C in a humidified CO 2 chamber followed by addition of the compounds in 1% DMSO at final concentrations of 5 µM and 10 µM. DMSO (1%) alone was used as the no-inhibitor control (0% inhibition) and the reference compound was ribavirin (TargetMol, Massachusetts, USA) at a final concentration of 5 µM or 10 µM. Cells were incubated at 37 °C for 72 h and lysed to quantify DENV2 replicons by RT-qPCR. Data were reported as percent inhibition compared with ribavirin, as a reference compound and DMSO, 0% inhibition. Results were confirmed by three independent experiments. Data Availability. All data generated or analyzed during this study are included in this published article and supplementary file.
The level of bio-containment. We performed all pathogen-related experiments in bio-containment level 2. BSL-2 standard operating procedure was performed to fit the requirement of ISO15189 and ISO15190.