Inhibition of dengue virus replication by novel inhibitors of RNA-dependent RNA polymerase and protease activities

Abstract Dengue virus (DENV) is the leading mosquito-transmitted viral infection in the world. With more than 390 million new infections annually, and up to 1 million clinical cases with severe disease manifestations, there continues to be a need to develop new antiviral agents against dengue infection. In addition, there is no approved anti-DENV agents for treating DENV-infected patients. In the present study, we identified new compounds with anti-DENV replication activity by targeting viral replication enzymes – NS5, RNA-dependent RNA polymerase (RdRp) and NS3 protease, using cell-based reporter assay. Subsequently, we performed an enzyme-based assay to clarify the action of these compounds against DENV RdRp or NS3 protease activity. Moreover, these compounds exhibited anti-DENV activity in vivo in the ICR-suckling DENV-infected mouse model. Combination drug treatment exhibited a synergistic inhibition of DENV replication. These results describe novel prototypical small anti-DENV molecules for further development through compound modification and provide potential antivirals for treating DENV infection and DENV-related diseases.


Introduction
Dengue virus (DENV) is responsible of worldwide arthropodborne viral infection, which globally represents a serious human health concern. DENV is the etiological agent of disease in more than hundred countries with up to 3 billion people exposed to the risk of infection in the tropical regions [1][2][3] . DENV has expanded its global range with sustained outbreaks in South America and Asia, with these epidemics accompanied by increased disease severity.
DENV are single-stranded, positive sense RNA viruses belonging to the Flaviviridae family. The DENV family can be viewed as falling in four related, but antigenically distinct, DENV 1-4 serotypes. The carriers of DENV to humans are the mosquitoes Aedes aegypti and Aedes albopictus. The DENV infection causes a variety of illness, including asymptomatic or subclinical disease, dengue fever (DF) symptoms and the most severe dengue haemorrhagic fever (DHF) and dengue shock syndrome (DSS) which cause millions of infections around the world [4][5][6][7][8] . DENV has a 10.7 kb, positive-sense RNA genome with 5 0 -and 3 0 -untranslated regions flanking a polyprotein that encodes three structural (C, prM/M and E) and seven non-structural (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) proteins. prM and E structural proteins are the primary antigenic targets of the humoural immune response in humans [9][10][11][12] . DENV NS3 is a multifunctional protein, which contains protease, helicase and triphosphatase domain. The N-terminal amino acids (residue 1-184) of NS3 is responsible for protease activity. NS2B serves as a cofactor of NS3 protease and forms complex with NS3; the central amino acid hydrophilic domain (residue 49-92) of NS2B is critical for cofactor activity 13 . DENV replication also requires the nonstructural protein 5 (NS5) which is the essential RNA-dependent RNA polymerase (RdRp) activity 14 . DENV NS5 RdRp has proven to be a promising target for direct-acting antiviral (DAA) drug development, because it is structurally conserved among the four DENV serotypes, and NS5 RdRp has no enzymatic counterpart in mammalian cells 15 .
Currently, no licensed antiviral drugs are available to block DENV infection, and while vector control efforts remain the only means to stop the spread of the infection, they have not successfully inhibited annual epidemic outbreaks throughout the tropics 16 . Recently, a live-attenuated DENV vaccine 17 based on the yellow fever virus 17 D backbone was licensed for use in the Philippines, Brazil and Mexico. However, the serotype-specific efficacy of the DENV vaccine is varying, and the long-term protection and safety of this vaccine still need more investigation 18 . In the present study, we have focused on the identification of potential anti-DENV inhibitors by targeting the enzymatic activities of the NS5 RdRp polymerase and NS3 protease in vitro and in vivo [19][20][21] . As part of a continuation of our studies [22][23][24] , we developed new pyrazole derivatives 25 as potential DENV NS5 RdRp inhibitors (Table 1). In addition, we carried out virtual screening (VS) studies on the NS2B/NS3 protease to design and synthesise new DENV NS3 protease inhibitors (Table 2). In sum, we identified five compounds that exhibited anti-DENV replication activity without cytotoxicity; two compounds exhibited anti-DENV activity in DENVinfected ICR-suckling mouse model. Interestingly, combination treatment with compounds, respectively, targeting NS5 RdRp polymerase and NS3 protease, demonstrated a synergistic inhibitory effect on DENV replication.

Chemistry
Microwave (MW)-assisted reactions were performed on a CEM Discover SP single-mode reactor equipped with Explorer 72 autosampler, controlling the instrument settings by PC-running CEM Synergy 1.60 software (Matthews, NY, USA). Closed vessel experiments were carried out in capped MW-dedicated vials (10 ml) with cylindrical stirring bar (length 8 mm, diameter 3 mm). Stirring, temperature, irradiation power, maximum pressure (P max ), PowerMAX (simultaneous cooling while heating), ActiVent (simultaneous venting while heating), and ramp and hold times were set as indicated. Temperature of the reaction was monitored by an external fibre optic temperature sensor. After completion of the reaction, the mixture was cooled to 25 C via air jet cooling. Organic solutions were dried over anhydrous sodium sulphate. Evaporation of the solvents was carried out on a Bu€chi Rotavapor R-210 equipped with a Bu€chi V-850 vacuum controller and a Bu€chi V-700 vacuum pump. Column chromatography was performed on columns packed with silica gel from Macherey-Nagel (70 À 230 mesh). Silica gel thin-layer chromatography (TLC) cards from Macherey-Nagel (silica gel pre-coated aluminium cards with fluorescent indicator visualisable at 254 nm) were used for TLC. Developed plates were visualised with a Spectroline ENF 260 C/FE UV apparatus. Melting points (mp) were determined on a Stuart Scientific SMP1 apparatus and are uncorrected. Infrared spectra (IR) were run on a PerkinElmer Spectrum 100 FT-IR spectrophotometer equipped with universal attenuated total reflectance (ATR) accessory and IR data acquired and processed by PerkinElmer Spectrum 10.03.00.0069 software. Band position and absorption ranges are given in cm À1 . Proton nuclear magnetic resonance ( 1 H NMR) spectra were recorded a Bruker Avance (400 MHz) spectrometer in the indicated solvent and corresponding fid files were processed by MestreLab Research SL MestreReNova 6.2.1-769 software (Santiago de Compostela, Spain). Chemical shifts are expressed in d units (ppm) from tetramethylsilane. Elemental analyses of biologically evaluated compounds were found to be within ±0.4% of the theoretical values and their purity was found to be >95% by highpressure liquid chromatography (HPLC). The HPLC system used Thermo Fisher Scientific Dionex UltiMate 3000, consisted of a SR-3000 solvent rack, a LPG-3400SD quaternary analytical pump, a TCC-3000SD column compartment, a DAD-3000 diode array  detector and an analytical manual injection valve with a 20 mL loop. Samples were dissolved in acetonitrile (1 mg/mL). HPLC analysis was performed by using a Thermo Fisher Scientific Acclaim 120 C18 column (5 mm, 4.6 mm Â 250 mm) at 25 ± 1 C with an appropriate solvent gradient (acetonitrile/water), flow rate of 1.0 mL/min and signal detector at 206, 230, 254 and 365 nm. Chromatographic data were acquired and processed by Thermo Fisher Scientific Chromeleon 6.80 SR15 Build 4656 software (Waltham, MA, USA).

Molecular modelling studies
All molecular modelling studies were performed on a MacPro dual 2.66 GHz Xeon running Ubuntu 12. The DENV protease structure (PDB ID: 2FOM) 27 was downloaded from the protein data bank. An in-house compound library matching Lipinski's rule of five 28 was used as a training set. No pre-filter was applied. The docking simulations were performed with PLANTS 29 using a 13 Å radius grid sphere. The centre of the binding site was settled according to Tamiri 30 . Molecules were scored by ChemPLP scoring function. 29 Figure 1 was generated by PyMOL 31 .

Ethics statement and experimental animals
Six-day-old ICR strain-suckling mice were obtained from BioLasco Taiwan Co. Ltd (Taipei, Taiwan). All animal studies were conducted in specific pathogen-free conditions and carried out in accordance with the Guide for the Care and Use of Laboratory Animals. The

Evaluation of anti-DENV RNA activity
Huh-7 cells were seeded in 24-well plate and infected DENV at an MOI of 0.2 for 2 h and followed by test compound treatment at concentration of 0, 1, 5, 10, 25, 50 and 100 mM for 3 days. The total cellular RNA was harvested by RNA extraction kit 32 following manufacturer's instrument. DENV RNA and cellular mRNA levels were determined by quantitative real-time reverse-transcription polymerase chain reaction (RT-qPCR) with specific primers 33 . The DENV RNA level was normalised by cellular glyceraldehydes-3-phosphate dehydrogenase (GAPDH) mRNA level. The relative DENV RNA level was calculated by StepOne TM Software v2.2.2 (Applied Biosystems, Foster City, CA, USA) following normalisation of cellular glyceraldehydes-3-phosphate dehydrogenase (GAPDH) mRNA level. The zero dose was defined as 100%. The EC 50 values were calculated from non-linear regression curve fitting using GraphPad Prism7 software (San Diego, CA, USA). Results were obtained from three independent experiments.

Cell cytotoxicity assay
The Huh-7 cells were seeded in 96-well and treated with test compound at concentration of 0, 10, 25, 50, 100, 150 and 200 mM. After 3 days incubation, the cell viability was determined by CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS assay) as previously reported 34 . Briefly, the MTS assay buffer was added to the cultured plate following removal of the culture medium. After 2 h of incubation at 37 C, the absorbance at 492 nm was measured. The cell viability was calculated from nonlinear regression curve fitting using GraphPad Prism7. The zero dose was defined as 100%. Results were obtained from three independent experiments.

Purification of DENV NS5 protein
The DENV-2 NS5 protein purification was performed as previously described 33 . Briefly, Sf9 cells were infected by the recombinant baculovirus vAc-DENV-NS5 at an MOI of 10 for 3 days. The cells were pelleted by centrifugation at 1000Âg for 5 min at 4 C and washed twice with phosphate-buffered saline (PBS). The cells were suspended in 5 mL of binding buffer (20 mM sodium phosphate, pH 7.5, 300 mM NaCl, 20 mM imidazole) containing protease inhibitor cocktail, and then disrupted by sonication. After centrifugation, the supernatant containing His-tagged DENV-2 NS5 protein was subjected to metal affinity chromatography employing the AKTA prime protein purification system 35 . The eluted DENV-2 NS5 protein was concentrated by an Amicon Ultra-15 30 k centrifugal filter device 36 and then dialysed against PBS overnight at 4 C.

Evaluation of anti-DENV RdRp activity
The cell-based experiments were performed as previously described 33 . Briefly, the Huh-7 cells were transfected with 0.5 mg of p(þ)RLuc-(-)DV-UTRDC-FLuc and DENV NS5 expression vector pcDNA-NS5-Myc followed by compound treatment for 3 days. The RLuc and FLuc activities were analysed by Dual-Glo Luciferase Assay System 37 . The enzyme-based fluorescence-based alkaline phosphatase-coupled polymerase assay (FAPA) was performed as previously described 33 . Briefly, the template was amplified from the cDNA of DENV-2 minus strand 3 0 -UTR and its RNA was synthesised by the T7 Megascript kit 38

Purification of DENV NS2B/NS3pro protein
The DENV NS2B/NS3pro purification was performed as previously described 39 . Briefly, the pET24b-NS2B-NS3pro plasmid was transformed into competent Escherichia coli BL21 (DE3) and grew in kanamycin (50 mg/mL)-containing LB at 26 C until OD600 nm reached 0.6. The isopropyl-b-D-thiogalactopyranose was added into the bacterial culture at 0.5 mM for protein induction. After 12-h incubation, cells were collected and subjected to protein purification following previous description.

Evaluation of anti-DENV protease activity
For cell-based activity assay, the Huh-7 cells were cotransfected with 0.5 mg of pEG(MITA)SEAP reporter vector and DENV protease expression vector pcDNA-NS2B-GSG-NS3-Myc followed by compound treatment for 3 days. The SEAP activity was analysed by Phospha-Light SEAP Reporter Gene Assay System 40 . The enzymebased experiment was performed as previously described 39 . Briefly, the 7-amino-4-methylcoumarin (AMC) fluorophore-linked peptide substrate Boc-GRR-AMC 41 and purified DENV NS2B/ NS3pro protein were used to analyse DENV NS2B/NS3 protease activity. The 100 nM DENV NS2B/NS3pro protein was incubated with 1, 5 or 10 mM tested compound and 10 mM Boc-GRR-AMC in cleavage buffer (200 mM Tris [pH 9.5], 20% glycerol) for 30 min at 25 C. The fluorescence signal was detected at excitation wavelength of 380 nm and emission wavelength of 465 nm, respectively.

Evaluation of anti-DENV activity by ICR-suckling mice
Breeder ICR mice were purchased from BioLasco Taiwan Co. Ltd. The experimental protocol was approved by the Animal Research Committee of Kaohsiung Medical University of Taiwan (IACUC, protocol number 102177). All mice received humane care and fed with standard rodent chew and water ad libitum. Mice were kept under a standard laboratory condition following the Animal Use Protocol of Kaohsiung Medical University. The 6-day-old ICR-suckling mice were intracerebrally injected with DENV-2 followed by intracerebrally injecting saline or test compound at 1, 3, 5 days postinfection (dpi). Mice intracerebrally injected with 65 C heat-inactive DENV-2 followed with saline treatment served as healthy control. The body weight, clinical scores and survival rate were recorded every day. The mice were sacrificed at 6 dpi.

Virtual screening studies
In search for allosteric inhibitors of the NS2B/NS3 protease, we carried out virtual screening (VS) and docking studies. We focused our VS studies against the allosteric site , of the enzyme 42,43 . To our knowledge, only few protease allosteric inhibitors have been reported 30 , and many of them are natural compounds that do not possess drug-like properties 30 . We performed structure-based screening studies on an open conformation of the NS2B/NS3 protease structure 44 (pdb code: 2FOM) 27 available at the protein data bank webserver 45 . This structure of the NS2B/NS3 protease was previously used in VS campaign in search for new allosteric inhibitors 43 .  (Table 1 and Figure 2, panel A).

DENV NS3 protease inhibitors
Huh-7 cells were infected with DENV-2 and then treated with protease inhibitors for 3 days and DENV RNA synthesis was evaluated with specific qRT-PCR primers targeting viral NS5 gene. Compounds 4 and 5 strongly inhibited the DENV-2 DNA replication with EC 50 s of 4.60 ± 0.03 mM and 7.28 ± 0.13 mM, respectively (Table 2 and Figure 3, panel A). In addition, a cell-based DENV protease reporter assay was used to characterise the protease specificity of compounds 4 and 5. Huh-7 cells were transfected with pEG(MITA)SEAP and DENV protease expression vector pcDNA-NS2B-GSG-NS3-Myc and then treated with 4 or 5. Both compounds showed strong inhibitory activity against the DENV protease with EC 50 of 6.71 ± 0.20 mM and 7.92 ± 0.62 mM, respectively (Table 2 and Figure 3, panel B). We further performed an enzyme-based assay to evaluate the inhibitory specificity of compounds 4-5. The data showed that compounds 4-5 specifically suppressed DENV protease activity with EC 50 of 4.72 ± 0.3 mM and 6.90 ± 0.11 mM, respectively ( Table 2). In conclusion, the results indicated that both compound 4 and 5 suppress DENV replication by directly inhibiting DENV protease activity.    Synergistic effect of NS5 RdRp inhibitor 3 and NS3 protease inhibitor 4 against DENV replication To examine whether combinational treatment of NS5 RdRp inhibitor 3 and NS3 protease inhibitor 4 could synergistically inhibit DENV replication, DENV-2-infected Huh-7 cells were cotreated with 3 and 4 at the indicated concentration for 3 days. As shown in Figure 6, a synergistic inhibition of DENV replication was observed with the 3 and 4 combination (lines [6][7][8][9], as compared to singlecompound treatment (lines 2-5).

Conclusions
We sought to design and characterise potential anti-DENV inhibitors by targeting the viral enzymatic, NS5 RNA-dependent RNA polymerase and the NS3 protease activities. We identified five potential compounds demonstrating anti-DENV replication activity without cytotoxicity. Two compounds targeting DENV NS3 protease and NS5 RdRp also exhibited anti-DENV activity in ICR-suckling mouse model of DENV infection. Combination treatment exhibited enhanced inhibition of DENV replication. Taken together, these results demonstrate that both classes of small molecules inhibited DENV replication and represent prototypical DAA molecules for further development through compound modification. Figure 5. DENV protease inhibitor 4 protected ICR-suckling mice from DENV infection. Six-day-old ICR-suckling mice were intracerebrally injected with heat-inactive DENV (iDENV, n ¼ 6) or active DENV (DENV, n ¼ 6). Mice-receiving DENV were treated with 1 mg/kg of compound 4 (n ¼ 6) at 1, 3, 5 dpi. Panel A, clinical scores, Panel B, body weight, and Panel C, survival rate were recorded every day. Disease severity was scored as follow: 0: healthy, 1: slightly sick (reduced mobility), 2: inappetance, 3: weight loss and difficult to move, 4: paralysis, 5: death. Each group included six mice. Error bars denote the means ± SD.  (4) inhibitors. Huh-7 cells were infected with DENV-2 at a multiplicity of infection (M.O.I) of 0.2 and followed by treatment of 3 and 4 with indicated concentration for 3 days. The DENV RNA level was analysed by RT-qPCR with specific primer targeting viral NS5 gene, and relative viral RNA levels were normalised against cellular GADPH mRNA levels. Error bars denote the means ± SD of three independent experiments. Ã p < .05; ÃÃ p < .01.
University and Apath, LCC, USA) for kindly providing human hepatoma Huh-7 cells.