Materials

Synthesis

All starting reagents and solvents were used as received (without further purification, distillation, nor dehydration).

Cell line

Vero (African green monkey kidney) CCL-81 cells expressing DENV serotype 4 replicons were used [⁵]. Vero cells were cultured in Dulbecco's Modified Eagle's Medium high glucose (DMEM) (GibcoTM) supplemented with 10 % fetal bovine serum (FBS) (GibcoTM), glutamine 2 mM (Sigma-Aldrich-Merck) and G418 300 μg/mL (GibcoTM) and placed in 5 % CO₂ atmosphere at 37 °C.

Instrumentation

Microwave-assisted reactions were performed in closed-vessel mode on a CEM Discover SP MW-reactor. Reaction progress was monitored by thin-layer chromatography (TLC) and the spots were visualized under ultraviolet (UV) light (254 or 365 nm). Flash columns packed with silica-gel 60, 230-400 mesh particle size and glass preparative plates (20 × 20 cm) coated with silica-gel 60 doped with UV indicator (F₂₅₄) were used to purify the products.

General method for the preparation of the assayed bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f

The corresponding bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f were synthesized using the methodology reported by us [¹⁹]. Thus, in a sealed CEM Discover microwave reaction tube (10 mL) containing a stirring bar, corresponding aldehydes (1.00 equiv.) and amines (0.10 mmol, 1.00 equiv.) were diluted in dry toluene (1.00 mL). These mixtures were stirred and heated using microwave irradiation (65 °C, 100 W) for 5 minutes. Then, ytterbium(III) triflate (0.03 equiv.) was added and heated using microwave irradiation (65 °C, 100 W) for 5 minutes. The corresponding isocyanides (1.20 equiv.) were added and again heated using microwave irradiation (70 °C, 150 W) for 15 minutes. Finally, maleic anhydride (1.40 equiv.) was added, and the new reaction mixture was stirred and heated using microwave irradiation (80 °C, 150 W) for 15 minutes. At the end of the reaction time, the solvent was removed to dryness under vacuum. Then, extractions of the crude were carried out using dichloromethane (3 × 25 mL) to collect the organic phases, which were washed with brine (3 × 25 mL), dried in anhydrous Na₂SO₄, filtered, and concentrated to dryness. The new crude was purified by column chromatography using as stationary phase silica-gel followed by preparative TLC (20 x 20 cm) plates. The solvent mixtures of n-hexane (Hex) and ethyl acetate (EtOAc) in 1:1 or 1:2 (v/v) proportions were employed as mobile phase in both chromatography procedures to isolate the corresponding bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f in 45 to 82 % overall yields (Fig. 2).

Fig. 2 Series of bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f synthesized via MCRs and in vitro assayed against DENV. 

In vitro assays

Vero CCL-81 cell viability

The effect of the six bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f on Vero CCL-81 cell viability was determined as previously described [¹⁹]. Briefly, 24 h before the experiment 25000 cells/per well were placed in a 96-well plate using DMEM supplemented with 8% of FBS and glutamine in an incubator under 5% CO₂ atmosphere at 37 °C. The compounds were added at concentrations of 0.1, 1, 10, 100 μM in a final volume of 100 µL of medium. After 48 h, media was removed and, 100 μL of crystal violet (0.1% crystal violet, 3.7% formaldehyde) were added. After 1 h at room temperature, plates were washed and 100 μL of 10% of acetic acid was added release to the dye. Plates were placed in an absorbance reader at a wavelength of 590 nm (Sunrise, TECAN) and cell viability was determined with respect to the vehicle control (DMSO). The experiments were performed three times in duplicates (n = 6). The vehicle refers to cells that were cultured in DMEM medium plus 0.5 % DMSO, which was the DMSO volume used in the diluent to solubilize the compounds.

Antiviral assay in DENV4 replicon system

The DENV4 replicon reporter assay was performed as previously described [⁵]. Briefly, 24 h before the assay, 25000 cell/per well of Vero cells were seeded in a 96-well plate. 10 µM of the compounds were added and after 48 h, the cells were lysed and Renilla luciferase was detected using Renilla Luciferase Assay System (Promega) and then measured using Luciferase reader (GloMax 20/20 Luminometer, Promega). Percent of expression was calculated normalizing the luciferase signal with the cells treated with vehicle (DMSO) (100 % of expression). Graph and statistical analysis were calculated using in GraphPad Prism version 9.5.1 (San Diego, CA, USA). p-Value was calculated using ordinary one-way-ANOVA test.

Antiviral assay Inhibition of DENV4 infection

For the infection inhibition assays of the Dengue virus serotype 2 and 4, 25000 Vero cells were seeded 24 h before infection. Infection was carried out at a multiplicity of infection (MOI) of 0.1 at 37 °C for 1 h. The addition of the compounds is as described above. Carnosine (L-CAR) was used as a positive control for replication inhibition since its direct interaction with the viral protein NS3-NS2B of DENV2 has been reported [²⁰]. 24 h later, RNA was extracted from the collected supernatant using QuickExtract RNA Extraction Kit (Lucigen) following the supplier's recommendations. Briefly, the medium was removed from the cells, washed with 1x PBS, 50 μL cold Quick Extract RNA Extraction Solution was added, and the lysate was transferred to a 1.5 mL, vortexed for 1 min and heated for 2 min at 65 °C. Afterwards, 5.5 μL of DNase Buffer 1, 1.25 μL of RiboGuard RNase Inhibitor, 2.5 μL of RNasefree DNase were added and heated for 15 min at 37 °C. Finally, 2 μL of Stop Solution was added and incubated for 10 min at 65 °C. Subsequently, the RT-qPCR was carried out in CFX96 Touch Real Time PCR Detection System (Bio-Rad) using OneStep RT-PCR Kit (Qiagen) to determine viral load following the supplier's recommendations. Briefly, 10 µL of 5x QIAGEN OneStep RT-PCR Buffer, 2 µL of dNTP Mix, 0.6 µM of forward primer 5´-TTGAGTAAACYRTGCTGCCTGTAGCTC-3´ and 0.6 µM of reversed primer 5´-GAGACAGCAGGATCTCTGGTCTYTC-3´[²¹], 2 µL of OneStep RT-PCR Enzyme Mix, 50 µg of RNA and 1 µL of Eva Green (Biotium). The RT-qPCR conditions were 50 °C for 30 min and 95 °C for 15 min, followed by 40 cycles of 94 °C for 30 sec, 55 °C for 30 sec and 72°C for 30 sec. Fluorescence quantification was performed during the extension step. The viral load of the dengue virus was obtained with the -2^(ΔCT) method [²²] and analyzed by two-way Anova in the Graph Pad V5 software using, p-values ≤0.05 w. The decrement of the viral load in each assay was compared to the vehicle, which was DMSO and was used as the negative control.

In silico studies

The in silico analyzes were carried out specifically on the protein used in the in vitro assays. A homology model of the DENV4 replicon has been computed using the Swiss-model server [²³] and their sequence of amino acids, and so called NS3-t4. Also, it was considered the NS3-NS2B protein with their cofactor (NS2b) for the serotype-4 (PDB code: 2FOM, NS3/NS2B-t4). On the other hand, the bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f, as well as a control molecule (carnosine) [²⁰] were modeled using the Avogadro package [²⁴] and optimized through the generalized gradient approximation functional development by Perdew, Burke, and Erzerhof (PBE) [²⁵] in the Gaussian 09 software [²⁶].

The molecular docking was carried out using the Moldock [²⁷] scoring function implemented in the Molegro Virtual Docker (MVD) package [²⁸]. Finally, with ZincPharmer software [²⁹] a pharmacophoric model was computed.

In vitro assays

The compounds under investigation were the six bis-furyl-pyrrolo[3,4-b]pyridin-5-ones 1a-f that have previously shown moderate activity against the human SARS-CoV-2 [¹⁹]. By evaluating the efficacy of these compounds in the Dengue replicon system and using the viral particle, their potential as antiviral agents against DENV 4 were herein assessed.

The Vero cell line is considered as the standard model cell in dengue studies due to its infectability in response to dengue infection. This cell line is derived from African green monkey (Chlorocebus sabaeus) renal are classified into four major cell lineages: Vero JCRB0111, Vero CCL-81, Vero 76, and Vero E6. Each one of this cell lineages display specific features making them suitable for the propagation of certain viruses. For example, Vero CCL-81 is capable of propagating the Japanese encephalitis virus under prolonged culture conditions [¹²], West Nile Virus [³⁰], Zika virus and DENV2 [³¹], while Vero E6 is used to propagate SARS-CoV-2 more efficiently [³²]. The cell line expression in DENV replicon and infection used in this study was Vero CCL-81. As each subline has phenotypic and genetic differences [³³]. It was relevant to determine the effect of the compounds on the Vero CCL-81 viability, despite their effect on Vero E6 viability was already investigated [¹⁹].

We observed a viability greater than 80% when Vero cells, expressing the DENV4 replicon, were exposed to the series of six compounds at the concentration of 10 µM (Fig. 3). At 100 µM, cell viability was reduced significantly, with the exception of compound 1f. To note, these results align with the previous findings reported [¹⁹].

Fig. 3 Vero CCL-81 cell viability assays. The experiment was performed two independent times in triplicate and the bars depict the standard deviation of the mean (n = 6). 

Antiviral activity on the Dengue virus replicon system

The antiviral activity on the DENV4 replicon of heterocyclic compounds was studied at 10 µM as this concentration did not affect the viability of Vero CCL-81 cells. Fig. 4 expresses the percentage of luciferase activity in y-axis that is a direct measurement of the viral genome translation and replication. Compounds 1c-f did not reduce the expression of Renilla luciferase. On the other hand, compounds 1a and 1b showed a considerable reduction of the expression of the reporter gene, 44.2 % and 31.6 % respectively.

Fig. 4 Inhibition of DENV serotype 4 replicon replication and translation in Vero CCL-81 cells. The experiment was performed two independent times in triplicate and the bars depict the standard deviation of the mean (n = 6). **** p-Value = < 0.0001 using ordinary one-way-ANOVA test. 

Antiviral activity on Dengue virus serotype 4

In this regard, DENV2 and DENV4 infection inhibition tests were carried out at 10 µM of the compounds (Fig. 5). In the case of DENV2, 1a and 1b did not reduced viral load, nevertheless with L-CAR we observed a reduction as reported (Figure S1 of Supplementary Materials File). The result obtained was that for DENV2 there is no reduction in viral load measured through RT-qPCR. Whereas that, 1a y 1b reduce in the viral load of DENV4 with respect to compounds 1c-1e, L-CAR and negative control, although this trend was observed, it was not statistically significant (Fig 5).

Fig. 5 Inhibition of DENV4 serotype 4 infection in Vero CCL-81 cells. The experiment was performed three independent times in duplicate and the bars depict the standard deviation of the mean (n = 6). 

The results indicate that compounds 1a and 1b have an antiviral effect in the replicon model and DENV4 infection. The specific inhibition observed in 1a and 1b could be due to the genome variability between serotypes and particularly for NS3, the conserved regions of the protein are few [³⁴]. The same behaviorcan be observed for the positive control L-CAR that reduces the viral load specifically for DENV2 in our results but not DENV4.

As 1a is the one with the highest activity, which may be due to interactions directed by the furane heterocycle in compound 1a. Differences between compounds 1a and 1b are found in the furan and tetrahydrofuran heterocycles that decorates pyrrolo[3,4-b]pyridin-5-one core.

This result seems to be in contrast when the series were evaluated against SARS-CoV-2 [¹⁹], in which 1a and 1b did not affect the viral infection. We must take into account that there are structural differences between the two proteases, where the SARS-CoV-2 protease requires its dimeric form to be active, DENV protease requires a protein cofactor (NS2B) to be active. Hence, the active sites differ, and for this reason, we see a difference between the preference of compounds with inhibitory capacity according to the viral protease [³⁵]. Compounds of a similar nature have been tested against other members of the Flaviviridae family e.g. Zika virus and have been shown to interact with and affect the viral protease NS3-NS2B of this flavivirus [³⁶]. Therefore, performed an in silico study to delve into the possible interactions between 1a and 1b and the viral proteases mentioned above.

In silico studies

The first results regarding the docking studies are presented in Fig. 6, which demonstrates that the synthesized molecules are docked into the NS3 and NS3-NS2B DENV4 serotype.

Fig. 6 Studied molecules docked into the (A) modeled NS3 DENV4 serotype-4 surface, (B) the NS3-NS2B DENV4 serotype-4. 

Analyzing the results of Table 1, the interactions between the selected ligands and the NS3-t4 and the NS3/NS2B-t4 correlated with the experimental results, considering that the best ligand is 1a. In this order, the best interactions were between 1a and NS3-t4, which present -4.95 kcal/mol, compared with the value of NS3/NS2B-t4, -4.57 kcal/mol. In this order, it was decided to analyze the interactions with NS3-t4 with the studied molecules.

The results agree with the experimental data, see Table 2, mainly considering the better interacting molecule 1a. Highlight that these results were computed using the NS3 free protein, without the possible co-factor, proving that this is a possibility of interaction between 1a-f compounds with the NS3, which interacts before coupling NS3 with the co-factor.

Regarding the kind of interactions between 1a and NS3-t4, Fig. 7 shows the key interactions. As well as Fig. 8 shows the interactions between NS3 protein and carnosine. Carnosine presents better interactions than 1a because the last presents few steric interactions, see Fig. 7 and 8. At the same time, Carnosine presents more hydrogen bond interactions than 1a.

Fig. 7 (A) Hydrogen bond, (B) Electrostatic, and (C) Steric interactions between NS3 and 1a. Blue lines depict the Hydrogen bond interactions, and Red and green dotted lines show the repulsive and attractive interaction, respectively. 

Fig. 8 (A) Hydrogen bond, (B) Electrostatic, and (C) Steric interactions between NS3 and carnosine. Blue lines depict the Hydrogen bond interactions, and Red and green dotted lines show the repulsive and attractive interaction, respectively. 

Fig. 8 shows that carnosine presents electrostatic interactions only with Lys 388, in contrast to 1a, which presents electrostatic interactions with Asp 609, Arg599, and Glu490; see Fig. 6.

In the case of hydrophobic interactions, the better-interacting cavity presents two hydrophobic surfaces surrounding a hydrophilic surface, see Fig. 91a presents interactions with both surfaces (hydrophobic) and, in the center, interacts with the hydrophilic surface (Fig. 9(A)). Carnosine only interacts with one hydrophobic surface and the hydrophilic surface in the center of the cavity, see Fig. 9(B). The last can be the cause of finding a higher inhibitory effect with 1a against serotype DENV4, which is more clearly shown in Fig. 9(C), founding that carnosine, despite presenting a higher Ligand efficiency in the case of the hydrophobic interactions, needs some hydrophilic fragment to promote a better target-ligand interaction.

Fig. 9 Hydrophobic interactions between NS3 and A) 1a, B) carnosine, and C) both molecules. Red surfaces show hydrophilic interactions and blue surfaces show hydrophobic interactions. 

Finally, a pharmacophoric model based on 1a interactions was performed considering the whole ligand-receptor interactions. Fig. 10 shows that one molecule needs mainly six zones to better interact with NS3, these two hydrophobic-aromatic sites, other hydrophilic, and three hydrogen acceptor fragments. In agree with Fig. 9, one ligand proposed to link with NS3 needs only two hydrophobic fragments and one hydrophilic. Besides, Fig. 10 depicts carnosine and 1a (Fig. 9(B), and 10(D), respectively) over pose into the pharmacophoric model, highlighting the need for two hydrophobic sites in the molecule to better interact with the NS3 target. Note that compounds that present higher LE and are not active can be due to the hydrophilic interactions and the need to modify their bioactive structure to interact with the pharmacophoric sites, such as the aromatic-hydrophobic site. Also, the bioactive pose is proposed around some factors, such as the whole functional groups of the molecule and the interactions of this with the target.

Fig. 10 (A) Pharmacophoric model for NS3 receptor (B) Pharmacophoric model with 1a, and carnosine overlapping (C) 1a, and (D) Carnosine into the pharmacophoric model. The purple sphere depicts the hydrophobic-aromatic fragments, the orange sphere is the hydrogen acceptor fragments, and the green sphere is the hydrophilic fragments.