Vero (African green monkey kidney fibroblasts, ATCC-CL 81) and BHK-21 (Baby Hamster Kidney-21, ATCC-CCL 10) cell lines were used for viral infection. They were maintained in Dulbecco's modified minimal essential medium (DMEM - Lonza®, Catalogue No. 12-604Q), supplemented with 10 % fetal bovine serum (FBS) (Biowest®, Catalogue No. S18b-500), and 1 % of penicillin-streptomycin antibiotic solution (Lonza®, Catalog No.17-602F) and incubated at a temperature of 37 °C with a 5 % CO₂ atmosphere. The CHIKV strain was isolated from Colombian clinical samples and characterized by RT-qPCR assays. Viral stocks were prepared by infecting Vero cells to a third passage; briefly, cell supernatants were harvested when 70 % of the cells showed a cytopathic effect, aliquoted, stored at - 80 °C, and used throughout the study. After the virus was propagated, the infectivity titer was determined by inoculation with increasing dilutions of the virus (10⁻¹ up to 10⁻¹⁰) into Vero cells, grown to confluent monolayer in 96 well plates, in eight replicates. After three days of incubation, the highest dilution showing 50 % infection was used to calculate the TCID50/mL by the Reed-Muench formula (Reed and Muench, 1938[49]).

The plant Marcetia taxifolia (A.St.-Hil.) was collected in the Amazonas State of Venezuela and archived (MYF 28418) in the "Herbario Víctor Manuel Ovalles" (Victor Manuel Ovalles Herbarium) of the Pharmacy school of the Universidad Central de Venezuela. The compounds of M. taxifolia were extracted and purified as previously described (Baptista et al., 2016[7]). Briefly, the plant parts were dried, macerated, and separated by organic extraction. The isolated flavonoids were fractionated by chromatographic techniques and characterized by spectroscopic and spectrometric methods. The stock solution of myricetin 3-(6-rhamnosilgalactoside) (MRG), 5,3'-dihydroxy-3,6,7,8,4'-pentamethoxyflavona (PMF), 5-hydroxy-3,6,7,8,3',4'-hexamethoxyflavona (HMF), and myricetin rhamnoside (MR) were prepared by dissolving each compound in DMSO to a final concentration of 10 mg/mL. Subsequently, the stock solutions were stored at -20 °C until use. Further dilutions were made in DMEM at the time of use.

To establish the cell line most susceptible to CHIKV growth, 50.000 BHK-21 cells/well and 35.000 Vero cells/well were plated into 96-well plates 16 to 24 h prior to infection. The culture medium was then removed, and cell monolayers were washed twice with 1X PBS (Lonza®, Catalogue No. 17-516Q). Cells were infected with CHIKV at a multiplicity of infection (MOI) of 0.0001 for one hour for viral adsorption. The cell monolayers were then washed thrice with 1X PBS to remove the non-adsorbed virus. Finally, 200 μL of DMEM 2 % FBS was added, and incubated for up to 106 hours at 37 °C in 5 % CO₂. At 10, 22, 34, 46, 63, 70, and 106 hours, supernatants and infected cells were collected and stored at -70 °C for quantification by plaque assay and RT-qPCR. The plaque assay was performed by seeding 70.000 BHK-21 cells/mL; 16 hours after, each supernatant was serially diluted from 10^-1 to 10^-8 in DMEM supplemented with 2 % SFB (2 % DMEM). Three replicates were performed for each dilution. After one hour of viral adsorption, 1.5 % of carboxymethyl cellulose (CMC) in 2 % DMEM was added to each well. The plates were then incubated in 5 % CO₂ at 37 °C for five days, after which the medium was removed, and the cells were fixed with Crystal Violet staining solution (3.5 % formaldehyde, 0.2 % w/v cristal violet) for one hour. The solution was then removed, the plates were washed with water, dried at room temperature, and finally the lysis plaques formed were counted to calculate the viral titer by adding 200 μL of assay medium. The cultures were incubated for five days and subsequently fixed with formaldehyde, stained with crystal violet and the plaques were counted to determine the viral titer expressed in plaque-forming units (PFU/mL) (Alvarez et al., 2005[5]).

The cytotoxicity of M. taxifolia flavonoids was assessed on BHK-21 cells by performing an MTT assay. Briefly, cells were seeded in 96-well plates at a cell density of 30.000 cells/well and incubated for 24 h at 37 ° C. Cells were incubated with serial dilutions ranging from 250 to 0,24 μg/mL of each of the four compounds for 24, 48, and 72 h at 37 °C with three replicates set up in each group. Cells were then incubated with 10 μL of MTT reagent (Roche®, Catalogue No. 11465007001) for an additional 4 h at 37 °C. The formazan crystals were solubilized in DMSO and measured using the Multiskan^TM FC microplate reader (Thermo Scientific) at 570 nm. Percentage cytotoxicity or viability was calculated compared to untreated cells. In addition, the concentration that inhibited 50 % of cell growth (CC50), which reduces the metabolic activity of cells by 50 %, was calculated by logarithmic interpolation, following a previously described procedure (Żesławska et al., 2018[59]).

To determine the antiviral activity of MR and MRG on CHIKV, BHK-21 cells were used. For this purpose, 30,000 cells were seeded in each well of a 96-well plate and incubated for at least 16 hours at 37 °C 5 % CO₂. Subsequently, two-fold serial dilutions obtained from the CC50 of MR and MRG (150 to 1.17 μg/mL for MR and 75 to 0.58 μg/mL for MRG) were added in triplicate. Then, 100 μl of a CHIKV suspension was added as described above and incubated for 72 h. The cell control was untreated and non-infected cells, and the virus control wells were untreated but infected cells (with the same virus concentration). No pharmacological control was used as there are no specific antiviral drugs approved to treat this infection. After treatment, the infection was allowed to proceed for 72 hours and cells were observed for the presence of the cytopathic effect (CPE). At the end of incubation, culture supernatants were collected, divided, and analyzed by the Reed and Muench method (Reed and Muench, 1938[49]) and RT-qPCR, as previously described.

CHIKV RNA copy number was determined using a quantitative RT-qPCR approach. Viral RNA was isolated from previously infected Vero or BHK-21 cells using a commercial viral RNA extraction kit according to the manufacturer's instructions (BMG-VR6568-02 Biomiga Inc). For RT-qPCR, a one-step PCR was performed at a final volume of 5 μL containing: 1 ng/μL of RNA per reaction, 2.5 μL of 2X qMax Green One-Step Mix, 0.25 µL of 20X RTase Blend, and 0.4 μM of each primer. Primers were specific to the CHIKV unique domain of nsP1 (CHIKV-F 5'-ACGTGGATATAGACGCTGACAG-3' and CHIKV-R 5'-GCATGGTCATTTGATGTGACC-3'). Amplification was performed using the QuantStudio 3 thermal cycler (Applied Biosystems) under the following thermal cycling conditions: 1 cycle at 50 °C for 10 min for reverse transcription, one cycle at 95 ° C for 2 min for initial denaturation and polymerase activation, and 40 cycles of 95 °C for 5 sec, annealing temperature at 62 °C for 30 sec, and elongation at 72 °C for 30 sec. A melting curve analysis was then performed at 65 °C to 95 °C to verify the assay's specificity. During the runs, two negative controls were included. In the first, the RNA sample was replaced with RNase-free water, and in the second, RNA from uninfected cells was used. The absolute amounts of viral RNA were calculated with a standard curve generated with a 10-fold serially diluted viral RNA extracted from CHIKV inoculum of known titer at 3.16 x10⁸ TCDI50/mL. All samples were analyzed in duplicate.

Molecular docking studies were carried out with the "X" domain, known as the macrodomain, present in viruses of the Coronavidiridae and Togaviridae families, which is activated as adenosine diphosphoribose phosphatase in Chikungunya viruses (Malet et al., 2009[37]). The nsP3 macrodomain (PDB 3GPG) was selected for molecular docking, and MR and MRG compounds were isolated from M. taxifolia.

For the nsP3 macrodomain, a grid box of dimensions 20x, 22y and 20z (Å) with a spacing of 1 Å was defined. The 3GPG crystallographic structure had a resolution of 1.8 Å and adequate metric validation. Calculations with the MR and MRG ligand structures defined these lattice box dimensions as the most accurate docking parameters. The 3D structures of MR and MRG were obtained using the PubChem database (Kim et al., 2016[30]), and their charges were added using the Biovia Discovery Studio (3DS, 2020[1]). Molecular docking was assessed using AutoDock Vina (ADV) software (Morris et al., 2009[40]). A maximum energy difference of 5 kcal/mol between the best and worst poses and a completeness of 8, which is the number of evaluations that take place in the local optimization of a conformer, were used.

Molecular dynamics assays were performed to assess the stability of the ligand-receptor complexes resulting from the molecular docking assays. The CHARMM-GUI platform was used to prepare the different inputs, and the Gromacs 2021.1 software was used for molecular dynamics (Abraham et al., 2015[3]; Jo et al., 2008[25]; Lindahl et al., 2021[35]). Each protein was preprocessed using the PDB reader tool (Jo et al., 2014[23]). On the other hand, the resulting docking ligands with the highest affinity energy were selected and shifted to mol2 format using OpenBabel (O'Boyle et al., 2011[43]). Next, the ligand .mol2 files were loaded into the Ligand Reader & Modeler tool to generate the parameters and topology files (Kim et al., 2017[29]). Finally, the ligand-receptor complexes were integrated into a single .pdb file that was used in the “Solution Builder” tool to create the system used as input for Gromacs (Lee et al., 2016[32]). The water box was cubic, fitted to the protein size and with 10 Å edge spacing. Each system was neutralized using KCl ions placed by the Monte-Carlo method at a concentration of 0.15 M. Each system was subjected to 5000 steps of energy minimization of steeper decays to eliminate the steric overlap. Then, all systems were subjected to an NVT (constant number of particles, volume and temperature) equilibrium phase for 125000 steps, using the V-rescale temperature-coupling method, with a constant coupling of 1 ps at 303.15 K (Bussi et al., 2007[13]). Subsequently, molecular dynamics was carried out for 100 ns using the CHARMM36m force field (Vanommeslaeghe et al., 2010[55]). Gromacs utilities were used to evaluate the root-mean-square deviation (RMSD) of complexes, and that of each protein and ligand, the root-mean-square fluctuation (RMSF) and hydrogen bonds. The data were plotted using the GRACE program.

Data were analyzed with Student´s t-test and One-way ANOVA test using GraphPad Prism Software version 8.4.0. A p-value < 0.05 was considered statistically significant. All experiments were performed at least three times with three replicates. For each experiment, the mean and standard deviation of the mean are shown. The RT-qPCR data from antiviral assays were transformed using a base 10 logarithm. Then, using the dose-response algorithm, the Ct obtained was normalized to determine the effective concentration 50 (EC50).

To evaluate the viral susceptibility of BHK-21 and Vero cells to the Colombian CHIKV strain, plaque and RT-qPCR assays were performed. CHIKV replicated in the two cell lines with similar log growth rates, indicating that the viral yield increased during the incubation (Figures 1A and B(Fig. 1)). Growth kinetics analyzed by plaque-forming unit assay showed no statistical difference between viral yields in both cell lines, maintaining ∼6 log viral titer up to 106 hours post-infection (Figure 1A(Fig. 1)). Likewise, replication kinetics assessed by RT-qPCR showed an increase in viral yield in both cell lines over time (Figure 1B(Fig. 1)). Notably, viral RNA copies detected in BHK-21 were significantly higher than those in Vero cells at all time points analyzed (Figure 1B(Fig. 1)), indicating that the BHK-21 cell line is more sensitive to CHIKV infection than Vero cells and support a higher level of CHIKV RNA replication.

Before determining the antiviral activity of M. taxifolia flavonoids, BHK-21 cells were used to evaluate their cytotoxic effects, taking into account the higher susceptibility of BHK-21 cells compared to the Vero cell line. HMF and PMF exhibited cytotoxic activity toward BHK-21 cells even with exposure to low concentrations of each compound (Table 1(Tab. 1)). HMF displayed the highest cytotoxicity with the lowest CC50 value at 48 (4.98 μg/mL) and 72 hours (5.4 μg/mL) among all flavonoids tested. PMF also showed high cytotoxic action for BHK-21 cells with a CC50 value of 34.94 and 43.61 μg/mL at 48 and 72 hours, respectively. In contrast, no significant cytotoxic effect was observed after exposure of BHK-21 cells to MR and MRG concentrations up to 250 μg/mL. Therefore, MR and MRG were selected to study their antiviral activity against CHIKV.

Viral titer and nsP1 RNA copy numbers were determined to evaluate the antiviral activity of MR and MRG against CHIKV after 72 hours of treatment. BHK-21 cells infected with the Colombian strain of CHIKV were exposed to MR or MRG in a concentration range of 150 to 1.17 μg/mL for MR and 75 to 0.58 μg/mL for MRG, obtained from the CC50. Remarkably, both flavonoids exerted an important antiviral activity toward CHIKV. MR exhibited a dose-dependent antiviral activity, significantly reducing viral titers with doses from 150 to 18.8 μg/mL (Figure 2A(Fig. 2)). Similarly, MRG showed a significant decrease in the viral titer at a concentration of 37.5, 9.4 and 2.3 μg/mL (Figure 2B(Fig. 2)). Exposure to each compound for 72 h post-infection resulted in a reduction of viral titer from 8.00 to 5.42 log10 TCID50/mL mean at 150 μg/mL for MR and a reduction from 8.00 to 6.50 log10 TCID50/mL mean at 37.5 μg/mL.

Consistent with these findings, RT-qPCR analysis also showed a significant decrease in CHIKV RNA log copy number for MR and MRG treatments at all concentrations tested compared to untreated cells in a dose-dependent manner (Figure 2C and D(Fig. 2)). MR dramatically reduced viral load (80.78 ± 0.46 %) at 150 μg/mL, whereas MRG decreased by only 52.46 ± 0.12 % at the highest concentration tested. However, when we determined the effective concentration (EC50) of each flavonoid against CHIKV by RT-qPCR (Figure 2E and F(Fig. 2)), MR showed a 50 % higher effective concentration (EC50) (49.11 ± 0.12 µg/mL) than MRG (22.43 ± 0.69 µg/mL), suggesting that MRG is a most effective antiviral compound.

Docking assays showed high affinity of MR and MRG with the nsP3 macrodomain, with an affinity energy of -9.1 and -8.5 kcal/mol, respectively. The MR ligand formed 6 hydrogen bonds with the residues: Ser110, Asp31, Val33, Arg144, and Tyr142, one Carbon-Hydrogen bond with Gly42, additionally 4 hydrophobic interactions with the Ala22 and Val33 residues (Figure 3(Fig. 3) and Table 2(Tab. 2)). Furthermore, the MRG presents 5 hydrogen bonds with residues Val33, Arg144, Thr111, Ser110, and Leu108, 6 Carbon-hydrogen bonds with Gly32, Leu109, Gly112, and Thr111, one Pi-donor-hydrogen bond with Val33 and 6 hydrophobic interactions with Val33, Ala22, Arg144 and Trp148 (Figure 4(Fig. 4) and Table 2(Tab. 2)). The stability of the complexes was confirmed by means of 100 ns of molecular dynamics, observing that the position of the ligand in the predicted binding site is maintained (Figure 4A(Fig. 4)). A twist in the MR ligand is observed at 80 ns, but it is a rearrangement and remains in the binding site; the MRG-nsP3 complex remains more stable throughout the simulation.

The stability of the MR and MRG is directly related to the ability to create hydrogen bonds, as can be seen in Figure 4B(Fig. 4), while the MR presents a lower amount of H-bond interactions. Figure 3(Fig. 3) shows the H-bonds formed by the ligand-receptor complex, and it is observed that the availability of -OH groups act both as donors and acceptors of H-bonds, while the protein presents interaction zones marked with red and blue colors for negative and positive polar interaction surface respectively. Finally, it is observed that the protein does not present relevant structural modifications as a result of molecular dynamics. The RMSF shows us the fluctuation of the position of the amino acids, and the whole protein shows a movement of less than 0.5 nm, except for the terminal groups that usually present a wide movement as they are disordered regions. This is also corroborated by the radius of gyration which has a fluctuation of less than 0.2 nm (1.5 to 1.7 nm), showing that the protein has not presented changes in size and conformation.

See also the Supplementary data and the Supplementary information.

This work was supported by Minciencias under research grant No. 124380864546 - contract CT. FP 80740- 152-2019; Universidad Antonio Nariño (UAN) under grant No. 2022206 and its Young Researcher Program. All authors acknowledge to the editorial fund and the Vicerrectoría de investigación (VCTI) from the UAN. We are grateful to Professor Jaime Castellanos from Universidad El Bosque for providing the virus strain.

The authors declare no conflict of interest.