Novel Dengue Virus-Specific NS2B/NS3 Protease Inhibitor, BP2109, Discovered by a High-Throughput Screening Assay

Abstract

Dengue virus (DENV) causes disease globally, with an estimated 25 to 100 million new infections per year. At present, no effective vaccine is available, and treatment is supportive. In this study, we identified BP2109, a potent and selective small-molecule inhibitor of the DENV NS2B/NS3 protease, by a high-throughput screening assay using a recombinant protease complex consisting of the central hydrophilic portion of NS2B and the N terminus of the protease domain. BP2109 inhibited DENV (serotypes 1 to 4), but not Japanese encephalitis virus (JEV), replication and viral RNA synthesis without detectable cytotoxicity. The compound inhibited recombinant DENV-2 NS2B/NS3 protease with a 50% inhibitory concentration (IC₅₀) of 15.43 ± 2.12 μM and reduced the reporter expression of the DENV-2 replicon with a 50% effective concentration (EC₅₀) of 0.17 ± 0.01 μM. Sequencing analyses of several individual clones derived from BP2109-resistant DENV-2 RNAs revealed that two amino acid substitutions (R55K and E80K) are found in the region of NS2B, a cofactor of the NS2B/NS3 protease complex. The introduction of R55K and E80K double mutations into the dengue virus NS2B/NS3 protease and a dengue virus replicon construct conferred 10.3- and 73.8-fold resistance to BP2109, respectively. The E80K mutation was further determined to be the key mutation conferring dengue virus replicon resistance (61.3-fold) to BP2109, whereas the R55K mutation alone did not affect resistance to BP2109. Both the R55K and E80K mutations are located in the central hydrophilic portion of the NS2B cofactor, where extensive interactions with the NS3pro domain exist. Thus, our data provide evidence that BP2109 likely inhibits DENV by a novel mechanism.

Dengue virus serotypes 1 to 4 (DENV-1 to -4) belongs to the family Flaviviridae, a group of enveloped RNA viruses that includes the genera Hepacivirus, Flavivirus, and Pestivirus. The genus Flavivirus consists of arthropod-borne disease agents, such as the yellow fever virus (YFV), Japanese encephalitis virus (JEV), West Nile virus (WNV), and DENV (14). Many members of the genus Flavivirus are important human pathogens and cause significant morbidity and mortality. DENV alone poses a public health threat to an estimated 2.5 billion people living in areas where dengue is epidemic and leads to 50 to 100 million human infections each year (17, 18, 38). DENV infection often leads to dengue fever, life-threatening dengue hemorrhagic fever (DHF), or dengue shock syndrome (DSS) (15, 19, 37). Approximately 500,000 cases of DHF and DSS have been reported in over 100 countries and have caused approximately 25,000 deaths per year (16). Despite the tremendous efforts invested in anti-DENV research, no clinically approved vaccine or antiviral therapy for humans is available for DENV (25, 32, 36). Considering the spread of this epidemic and the severity of DENV, the discovery of an effective anti-DENV drug is now an urgent need.

DENV is an enveloped RNA virus that consists of a 10.7-kb single-stranded, positive-polarity genomic RNA associated with multiple copies of capsid proteins. DENV RNA is translated as a single polyprotein upon entering the host cell and is then cleaved by both the host proteases and the virus-encoded two-component protease (NS2B/NS3pro) into three structural proteins (C, M, and E) and seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) to initiate viral replication (6, 11, 40). The viral protease (NS2B/NS3pro) consists of an N-terminal 180-residue NS3 (NS3pro), which is a trypsin-like protease with a serine protease catalytic triad (His51, Asp75, and Ser135) (2, 13, 44). NS3pro requires the central hydrophilic portion (NS2BH; residues 49 to 95) of NS2B as a cofactor (50) to enhance its proteolytic activity and stabilize its folding (10). NS2BH interacts directly with NS3pro (29) and actively participates in the formation of the protease active site (35). Consequently, the pivotal role of NS2B/NS3pro in viral replication makes it an attractive target for the development of anti-DENV inhibitors.

Viral proteases are excellent antiviral targets, as evidenced by the nine protease inhibitors of human immunodeficiency virus (HIV) currently in clinical use (30) and the numerous protease inhibitors of hepatitis C virus (HCV) in clinical trials (43). By analogy with the successes of HIV and HCV protease inhibitors, efforts have been made to design inhibitors against DENV using dengue virus NS2B/NS3 protease as a molecular target (reviewed in references 23, 36, and 41). Thus far, two main approaches have been taken to develop dengue virus NS2B/NS3 protease inhibitors. The first is the high-throughput screening (HTS) of small-molecule libraries. The other approach is the design of peptidomimetics similar to the protease catalytic substrate. However, there are limitations to these two approaches. For example, no optimal in vitro DENV NS2/NS3 protease assay represents in vivo NS2/NS3 protease activity (36). It is also difficult to design potent inhibitors against dengue virus NS2B/NS3 protease because the active site of dengue virus NS2B/NS3 protease is relatively flat (10) and charged (36). Some progress has been realized in the developing HTS and peptidomimetic approaches (7, 12, 24, 33, 48, 49), although to date there are no dengue virus protease inhibitors in clinical trials.

In the present study, we utilized a previously described recombinant NS2B/NS3 protease (1) for the in vitro HTS assay system (24) to identify a small-molecule protease inhibitor of DENV: BP2109, N,N′-[(9-oxo-9H-fluorene-2,7-diyl)-bis (oxyethane-2,1-diyl)]-bis(N,N-diethyldecan-1-aminium). Enzyme-based protease and dengue virus replicon assays were conducted and suggested that the target of BP2109 in DENV is NS2B/NS3 protease. Sequencing analyses and characterization of BP2109-resistant viruses indicated that BP2109 may have a novel mode of action, interrupting the interaction between the NS2BH and NS3pro of DENV. The data presented here provide an alternative avenue for the development of NS2B-binding inhibitors that target the dengue virus NS2B/NS3 protease.

MATERIALS AND METHODS

Cell lines and virus strains.

Baby hamster kidney (BHK21) cells (ATCC CCL-10) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 4.5 g/liter glucose and 5% fetal bovine serum (FBS). The cells were cultured at 37°C in a 5% CO₂ incubator. Aedes albopictus C6/36 cells (ATCC CRL-1660) were cultured in RPMI 1640 supplemented with 5% FBS. The cells were maintained at 28°C in a 5% CO₂ incubator. Virus-infected cells were grown in medium supplemented with 2% FBS. DENV-2 (Taiwanese strain PL046) (26) and JEV (strain RP-9) (8) were both provided by C. L. Liao (Institute of Microbiology and Immunology, National Defense Medical Center, Taiwan). DENV-2 and JEV stock viruses were prepared in C6/36 cells by infecting them at multiplicities of infection (MOIs) of 0.1 and 0.01 PFU/cell, respectively, with RPMI 1640 medium containing 2% FBS and incubated at 28°C until the beginning of the cytopathic effect. Then, the supernatant was harvested and stored with 20% FBS at −80°C. Virus titers were determined by a plaque-forming assay on BHK21 cells. Normally, the titers of DENV-2 and JEV stock are around 1 × 10⁸ and 1 × 10⁹ PFU/ml, respectively.

E. coli and yeast strains.

Frozen, competent Escherichia coli strain C41, a derivative of BL21(DE3) (31), was purchased from OverExpress, Inc. Standard yeast medium and transformation methods were used (5). Saccharomyces cerevisiae YPH857 was purchased from the ATCC. The genotype of YPH857 is MATα ade2-101 lys2-801 ura3-52 trp1-Δ63 HIS5 CAN1 his3-Δ200 leu2-Δ1 cyh2. Competent yeast cells were prepared using the lithium acetate procedure (5).

Construction of recombinant DENV-2 and JEV NS2B/NS3 protease.

All the primers used for the subsequent PCR synthesis are shown in Table S1 in the supplemental material. Construction of the DENV-2 NS2B/NS3 protease plasmid was performed as described previously (1), with slight modification. It expresses an N-terminal glutathione S-transferase (GST) fusion protein comprising the NS3 protease domain (residues 1 to 184) linked via a flexible glycine linker (Gly₄-Sre-Gly₄) to an NS2B hydrophilic domain (NS2BH) (residues 49 to 92) in E. coli (1, 46). A JEV NS2B/NS3 protease plasmid was also constructed in a similar way. DENV-2 (PL046) and JEV (RP9) RNAs were purified from virus stocks using a QIAamp Viral RNA Kit (Qiagen) and used as templates for synthesizing viral cDNAs. DENV-2 or JEV cDNAs were generated by reverse transcription (RT)-PCR using the reverse primer NS3proXhoI-R or JNS3proNotI-R, respectively, and then used as templates for the subsequent PCRs to synthesize the DENV-2 or JEV DNA fragment. The DENV-2 NS2BH DNA fragment was synthesized by PCR using primers NS2B49EcoRI-F and NS2B92GlyBamHI-R. The DENV-2 Gly-NS3pro DNA fragment was prepared by PCR using primers GlyNS3BamHI-F and NS3proXhoI-R. Two primers, NS2B49EcoRI-F and NS3proXhoI-R, were used to perform overlapping PCR to synthesize a DENV-2 NS2BH-glycine linker-NS3pro DNA fragment using the NS2BH and GlyNS3pro DNA fragments as templates. The DENV-2 NS2BH-glycine linker-NS3pro PCR DNA fragment was digested with EcoRI and XhoI restriction enzymes and cloned into the pGEX4T-1 vector to generate the pGEX4T-NS2B/NS3pro plasmid. To generate the DENV-2 KK protease mutant plasmid, two amino acid substitutions (R55K and E80K) were introduced into the pGEX4T-NS2B/NS3pro plasmid with primers 2B/R55KEcoRI-F and 2B/E80K-R. Similarly, the JEV NS2BH DNA fragment was synthesized by PCR using primers JNS2B54SalI-F and JNS2B94BamHI-R. The JEV Gly-NS3pro DNA fragment was generated with primers, JNS3BamHI-F and JNS3proNotI-R. The JEV NS2BH and GlyNS3pro DNA fragments were used as templates to perform overlapping PCR to synthesize an NS2BH-glycine linker-NS3pro DNA fragment with primers JNS2B54SalI-F and JNS3proNotI-R. The JEV NS2BH-glycine linker-NS3pro DNA fragment was digested by restriction enzymes SalI and NotI and cloned into the pGEX6P-1 vector to generate pGEX6P-JV-NS2B/NS3pro. The sequences of all DENV-2 and JEV protease plasmids were verified by DNA sequencing.

Expression and purification of NS2B/NS3 protease.

The E. coli strain JM109 was independently transformed with the expression vectors pGEX4T-NS2B/NS3pro and pGEX6P-JV-NS2B/NS3pro and grown in a fermentor with 5 liters of 2YT medium (16 g of Bacto tryptone, 10 g of yeast extract, and 5 g of Nacl dissolved in 1 liter distilled water [pH 7.2]) containing 100 μg/ml ampicillin at 37°C until the A₆₀₀ reached 0.7 and then induced by IPTG (isopropyl-β-d-thiogalactopyranoside) with a final concentration of 0.1 mM for an additional 4 h at 25°C. Cells were harvested by centrifugation at 3,000 × g for 15 min and stored at −80°C until they were used. For protein purification, the bacteria were thawed and resuspended in 30 ml of sonication buffer (20 mM Na phosphate, pH 7.0, 50 mM NaCl, 0.1% Triton X-100) per 10 g of bacteria; lysed with 200 μl of 100 mg/ml lysozyme at 4°C for 30 min, 200 μl of 1 M MgCl₂ for 10 min, and 100 μl of 2 mg/ml DNase I for 10 min, followed by probe sonication (10 groups of 5-s pulses) on ice; and then centrifuged at 27,000 × g for 30 min at 4°C. The supernatant was collected, filtered with 0.45-μm filters, and stored at 4°C or purified immediately by passage through a column of Glutathione Sepharose 4 Fast Flow (Amersham) preequilibrated with sonication buffer. The column was extensively washed with sonication buffer and subsequently eluted with elution buffer (50 mM Tris, pH 8.0, 0.5 M NaCl, 0.1% Triton X-100, 40 mM glutathione). The eluted protein was analyzed by 10% SDS-PAGE, dialyzed on a Spectra/Pro 1 Regenerated Cellulose Dialysis Membrane in at least 100× sample volume of dialysis buffer (100 mM Tris, pH 8.0, 20 mM NaCl, 0.2% Triton X-100), subsequently concentrated by employing Amicon Stirred Cells (Millipore) with an Ultracel YM-10 Ultrafiltration Membrane (Millipore), and finally stored at −20°C with 50 mM Tris, pH 8.0, containing 10 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol (DTT), 0.1% Triton, and 50% glycerol.

High-throughput screening.

High-throughput screening was done at the Division of Biotechnology and Pharmaceutical Research, National Health Research Institutes, Taiwan. The compound libraries used in this primary screening were purchased from Chemical Diversity Laboratory (San Diego, CA) and had >95% purity. Compounds from a library of compounds with diverse structures were provided as dimethyl sulfoxide (DMSO) stock solutions at 0.1 mM and assayed in 50 mM Tris buffer, pH 9.0, containing 10 mM NaCl and 20% glycerol. GST-CF40-Gly-NS3pro (1 μM) was preincubated with various test compounds (10 μM) at 37°C for 10 min, and the proteolytic reaction was initiated by addition of the synthetic peptide substrate acetyl-TTSTRR-para-nitroaniline (24) to a final concentration of 800 μM for an additional 2 h of incubation time. The reaction was stopped with sodium citrate, pH 4.5 (625 μM), and read at A₄₀₅ on a Wallac Victor2 Multilabel HTS Counter (Perkin Elmer). Aprotinin (3 μM) was used as a positive control. Fifty percent inhibitory concentrations (IC₅₀s) were determined as described previously (24) using Sigma plots and measured at least twice from independent triplicate experiments. The data obtained were reproducible with less than 15% difference in the independent experiments. The standard deviation in each data point was less than 15% of the IC₅₀ presented. K_i values were calculated using the Michaelis-Menten equation for competitive inhibition. The results represent the means ± standard errors of the mean (SEM) from duplicate determinations from three independent experiments.

Cytotoxicity assay.

The sensitivities of cell lines to BP2109 were examined using an MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt]-based tetrazolium reduction assay (CellTiter 96 AQueous Non-Radioactive; Promega G5430). Briefly, BHK21 cells were plated at a density of 1.8 × 10⁴ cells per well in 96-well plates containing 120 μl of culture medium for 6 h. Serially diluted compounds or DMSO (positive control) was added and incubated for an additional 72 h. MTS reagent was then added to each well and incubated for 1 h at 37°C in a humidified 5% CO₂ atmosphere before being read at a wavelength of 490 nm using an enzyme-linked immunosorbent assay (ELISA) plate reader. All data are presented as means ± SEM from three independent experiments.

Plaque-forming assay.

BHK21 cells were plated at a density of 2.25 × 10⁵ cells per well in 6-well plates containing 1 ml of culture medium overnight, and then 0.1 ml of serially diluted virus solution was added to ∼70 to 80% confluent BHK21 cells. After adsorption for 2 h, the virus solutions were replaced with either 0.75% methyl cellulose (Sigma; M-0512) containing DMEM and 2% FBS for DENV-2-infected cells or 1.2% methyl cellulose containing DMEM and 2% FBS for JEV-infected cells. On the sixth day postinfection, the methyl cellulose solution was removed from the wells, and the cells were fixed and stained with crystal violet solution (1% crystal violet, 0.64% NaCl, and 2% formaldehyde) (27).

Viral yield reduction assay in cultured cells.

BHK21 cells were plated at a density of 1.5 × 10⁵ cells per well in 12-well plates containing 0.5 ml of culture medium and incubated for 4 h at 37°C. Compounds (500 μl) were added to the wells for 16 h before the addition of DENV-2 or JEV. The plates were incubated for another 72 h at 37°C in a humidified 5% CO₂ atmosphere. To quantify the viral yield of cells in the presence of BP2109, the supernatants of cells treated with compound were harvested and subjected to viral titer determination by a plaque-forming assay on BHK21 cells. The mean values and SEM were calculated from three independent experiments. The detection limit was set at 10 PFU/ml.

Quantitative RT-PCR.

To quantify viral positive- and negative-strand RNA in infected BHK21 cells, the RNA was isolated from the above-mentioned cell samples using the Qiagen RNeasy Kit as described in the manufacturer's protocol. Viral RNA was reverse transcribed to cDNA using the Invitrogen Thermoscript RT Kit with the specific primers DV2.PS-R and DV2.NS-F for detection of positive- and negative-sense viral RNA based on the manufacturer's protocol and as described previously (47), with slight modification. Real-time quantitative PCR was used to measure the viral RNA. The reaction was conducted in a 10-μl volume comprising 5 μl of cDNA, 1× TaqMan Master Mix (Roche), 200 nM (each) previously described primers DV2.U2-F and DV2.L1-R, and 50 nM hydrolysis probe DV2.P1 (20) (Tib Molbiol). The LightCycler TaqMan Master Kit (Roche Biochemicals) and LightCycler 1.5 instrument (Roche Biochemicals) were utilized in this study under the following conditions: preincubation at 95°C for 10 min, followed by 45 cycles of three-step incubations at 95°C for 15 s (denaturation), 60°C for 30 s (annealing and elongation), and 72°C for 1 s (complete elongation with a single fluorescence measurement). A linear relationship was established between RNA copy numbers per milliliter and the corresponding threshold cycle (C_T) value over 7 log units of RNA concentration (correlation coefficient [r] = 0.99) (20).

Construction of reporter replicon.

The dengue virus reporter replicon (DV2Rep) was prepared for DENV-2 Taiwanese strain PL046 (26). Cloning of the replicon was based entirely on homologous recombination in yeast cells; recombinant DNA in vitro ligation was not used. The backbone of the reporter replicon is the shuttle vector pRS313 (42). DV2Rep consists of the SP6 promoter, 5′ untranslated region (UTR), core gene (C102), Renilla luciferase (Rluc) gene, FMDV2A cleavage site (FMDV2A), neomycin resistance (Neo) gene, an encephalomyocarditis virus (EMCV) internal ribosome entry site element (IRES), and nucleotides (nt) 2392 to 10724 of the DENV-2 genome (see Fig. 4A) (22, 34). The DNA fragment nt 2392 to 10724 of the DENV-2 genome contains a C-terminal envelope gene (E24), dengue virus nonstructural genes, and the 3′ UTR. Two primers, SacI-Sp6-5UTR-F and C102-R, were used to amplify the DNA fragment SP6-C102, including the SP6 promoter, the 5′ UTR, and the core gene (C102). The Renilla luciferase gene was amplified from pGL4.70-hRluc (Promega Corp.) by using primers Rluc-F and Rluc-2A-R. At the same time, an FMDV2A recognition site was introduced by extended primers. The Neo-IRES fragment was amplified from pHCVrep1b.BB7 (4) by using primer pair 2A-Neo-F and IRES-R. The primers E24-F and D2/3634R were used to amplify a fragment, E24-NS1-NS2A, encompassing the last 24 C-terminal residues of the E protein, NS1, and the first 52 residues of NS2A. The synthesis of the DENV-2 cDNA fragment NS2A-3′UTR, was carried out through RT-PCR using commercial enzymes with the primer pair D2/3530 and 3UTR-XbaI-R, according the manufacturer's instructions. The four amplified DNA fragments SP6-C102, Rluc-2A, Neo-IRES, and E24-NS1-NS2A were used to assemble the SP6-C102-Rluc-2A-Neo-IRES-E24-NS2A DNA fragment by overlapping PCR. The DV2Rep plasmid was constructed through the homologous recombination of three DNA fragments, including SP6-C102-Rluc-2A-Neo-IRES-E24-NS2A, NS2A-3′UTR, and linearized pRS313 vector (digested with SacI and XhoI) in yeast cells. The DV2Rep plasmid was purified from yeast cells and then reamplified in E. coli strain C41. The sequences of wild-type (WT) and mutant DV2Rep constructs were verified by DNA sequencing.

RNA transcription and transfection.

The cDNA plasmid of the replicon was linearized with XbaI. DNA was phenol-chloroform extracted, precipitated, and used as a template for in vitro transcription with an SP6 Message mMachine kit (Ambion). The RNA was quantified with a spectrophotometer and stored at −80°C. RNAs were transfected into BHK21 cells with Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol.

Transient replicon assay.

The transient replicon assay was performed for quantification of compound-mediated inhibition of viral translation and reduction of viral RNA synthesis (9). BHK21 cells were seeded in 24-well plates (2 × 10⁴ cells per well) and incubated overnight. Lipofectamine 2000 (Invitrogen) was mixed with 0.5 μg of wild-type or mutant dengue virus replicon RNA for one-well transfection, and cells were transfected according to the manufacturer's instructions. BP2109 at 8 μM or control medium was added to each well and assayed for luciferase activities at the times indicated. Duplicate wells were lysed for luminometry. To perform the luciferase assays, 10 μl of lysate was mixed with 50 μl of Renilla Luciferase Assay Reagent (Promega). For quantification of compound-mediated inhibition, relative luciferase activity derived from the mock-treated cells was set as 100% (51).

Isolation and characterization of BP2109-resistant virus.

BHK21 cells (6 × 10⁴ cells/well) were seeded in 24-well plates for 4 h and then treated with various concentrations of the tested compound (6 to 10 μM) or 0.1% DMSO (negative control) for 16 h. The cells were infected with parental DENV-2 or viruses from passages 1 to 40 at an MOI of 0.1. At 72 h postinfection, the clear supernatants were harvested from BHK21 cells that were infected with parental DENV-2 in the presence of 6 μM BP2109 and termed passage 1 for titer measurements. The passage 1 virus was used to infect new monolayer cells, which were also incubated in the presence of BP2109. Two independent pools of selection were conducted. The procedure was repeated for 10 passages. Passages 11 to 25 were selected with 8 μM BP2109. The viruses from passages 11 to 25 were harvested from infected cells 96 h postinfection and assayed for the virus titer. A higher dosage of compound (10 μM) was used for an additional 15 passages (passages 25 to 40) of selection. The viruses (passages 25 to 40) were harvested from infected cells 144 h postinfection and assayed for the titer. The susceptibility of P10-, P25-, and P40-resistant variants to BP2109 was confirmed by monitoring the increase in resistance and comparing the viral titer from the resistant variants to the titer of the wild-type virus in the presence of BP2109.

BP2109-resistant DENVs from two independent pools of passage 40 were collected and mixed, and virion RNAs were extracted using QIAamp Viral RNA kits (Qiagen). The viral RNAs were amplified by RT-PCR using SuperScript III one-step RT-PCR kits (Invitrogen). The PCR products of NS2B-NS3 were gel purified and subcloned to the shuttle vector pRS313 (42). Twenty colonies of plasmids were purified from yeast cells and reamplified in E. coli strain C41 for DNA sequencing.

Construction of recombinant NS2B/NS3 KK mutant protease and DENV-2 KK mutant replicon.

To generate the KK protease mutant plasmid, two amino acid substitutions (R55K and E80K) were introduced into the pGEX4T.NS2B/NS3pro plasmid with primers 2B/R55KEcoRI-F and 2B/E80K-R.

To construct a double-mutant reporter replicon (KK replicon), primer pairs D2/3799-F and 2B/R55K-R, 2B/R55K-F and 2B/E80K-R, and 2B/E80K-F and D2/6872-R were used for generating 3799-55K, 55K-80K, and 80K-6872 fragments. The three PCR products were gel purified and joined by overlapping PCR to form a fragment with the double mutant (3799-55K-80K-6872) for homologous recombination with linearized DV2Rep plasmids (digested with MfeI). The KK replicon plasmids purified from yeast cells were then reamplified and maintained in E. coli strain C41. All constructs were sequenced to confirm the presence of the desired mutation and to exclude external changes.

RESULTS

Identification of DENV-2 NS2B/NS3 protease inhibitors by HTS.

The flavivirus protease is essential for the processing of the polyprotein, which is required to generate viral proteins for viral replication and the maturation of infectious dengue virions. Therefore, flavivirus protease is an ideal target for the development of antiviral therapeutic agents. We utilized a high-throughput NS2B/NS3 protease inhibition-based screening assay to identify novel small-molecule protease inhibitors of the DENV protease from a library of compounds with diverse structures (24). In this HTS assay, we started with 41,600 compounds, from which three candidates were discovered to effectively suppress DENV-2 protease activity with calculated IC₅₀s lower than 40 μM. The three candidates, BP2109, BP2444, and BP5700, had IC₅₀s of 15.43 ± 2.12 μM, 20.48 ± 4.12 μM, and 27.00 ± 1.32 μM, respectively (Table 1). Regrettably, both BP2444 and BP5700 were deemed too toxic to BHK21 cells and thus were excluded from further studies. Consequently, BP2109 was the only candidate to demonstrate efficient inhibition of NS2B/NS3 protease-mediated substrate digestion with moderate cytotoxicity to BHK21 cells.

TABLE 1.

Chemical structures of BP2109, BP2444, and BP5700 and effects of candidates on DENV-2 enzyme-based NS2B/NS3 protease activity

^aThe IC₅₀s were determined as a 50% drop in the enzyme-based protease activity assay. The values represent the means ± standard deviations obtained from three independent experiments.

Dosage-dependent inhibition of all four DENV serotypes by BP2109.

To eliminate possible BP2109-mediated cytotoxicity, the antiviral activity of BP2109 was tested at concentrations below 15 μM (>98% cell viability). To evaluate the effect of BP2109 on the viral yield, BHK21 cells were pretreated in the presence or absence of BP2109 for 16 h prior to DENV-2 infection at an MOI of 0.1 or 1.0. A plaque formation assay was used to measure the titer of DENV-2 present in the supernatant of infected BHK21 cells at 72 h postinfection in the presence of different concentrations of BP2109. Strong inhibition of DENV-2 amplification by BP2109 was observed in a dose-dependent manner at MOIs of 0.1 and 1.0 (Fig. 1). We observed an enormous (>10,000-fold) reduction in the DENV-2 titer in response to 12 μM BP2109 (Fig. 1), which was not caused by the cytotoxicity effect of BP2109 (100% cell viability). However, BP2109 must be active against all four serotypes of DENV (DENV-1, -2, -3, and -4) for it to be considered a valuable anti-DENV agent, and we found that BP2109 is effective against all four serotypes of DENV (MOI = 0.1) at noncytotoxic concentration of 8 μM (Fig. 2). The viral yields of DENV-1, -2, -3, and -4 were 52-, 475-, 91-, and 585-fold lower, respectively, than those of their nontreated controls. These results indicate that BP2109 is a potent therapeutic compound against DENV.

FIG. 1.

Dose-dependent inhibition of the DENV-2 yield by BP2109. The DENV-2 yield was calculated after treating BHK21 cells with increasing concentrations of BP2109 and then infecting them with DENV-2 at MOIs of 0.1 and 1.0, as described in Materials and Methods. The viral yields in culture medium at 72 h postinfection were determined by plaque formation assays. The mean values and SEM from three independent experiments are plotted.

FIG. 2.

Inhibition of all four serotypes of DENV by BP2109. BHK21 cells were incubated with 8 μM BP2109 and then infected with four serotypes of DENV at an MOI of 0.1. The viral yield in culture medium was determined by plaque formation assays at 72 h postinfection. The mean values and SEM from three independent experiments are plotted. **, P < 0.0001.

BP2109 selectively reduces NS2B/NS3 protease activity and the viral yield of DENV.

To examine the antiviral spectrum of BP2109, the NS2B/NS3 protease from JEV was also constructed and purified to assay for anti-JEV activity by BP2109. In contrast to DENV, even at a high dose of 100 μM, BP2109 showed a minimal inhibitory effect on JEV NS2B/NS3 protease activity (<3%) in enzyme-based protease activity assays (Fig. 3A). Moreover, to evaluate the effect of BP2109 on the JEV yield, BHK21 cells were infected with JEV or DENV (MOI = 0.1) in the presence or absence of 12 μM BP2109. In contrast to DENV, BP2109 did not inhibit JEV production (Fig. 3B). Together, these data indicate that BP2109 specifically inhibits dengue virus replication while having no significant effect on JEV.

FIG. 3.

BP2109 selectively inhibits the NS2B/NS3 protease activity and viral yield of DENV. (A) Selective inhibition of protease activity by BP2109. Purified JEV NS2B/NS3 protease was used for an enzyme-based protease activity assay. The values represent percentages of the control (untreated) value obtained for DENV or JEV. The mean values and SEM from three independent experiments are plotted. (B) Antiviral spectrum of BP2109. BHK21 cells infected with JEV or DENV-2 at an MOI of 0.1 were maintained in the absence or presence of 12 μM BP2109. The viral yields in culture medium at 72 h postinfection were determined by plaque formation assays. The mean values and SEM from three independent experiments are plotted. (**, P < 0.0001).

BP2109 reduces the level of viral RNAs in cells infected with DENV-2.

To evaluate the effect of BP2109 on viral RNA production, TaqMan fluorogenic quantitative RT-PCR was used to measure the amount of viral RNA in the infected cells. After BHK21 cells were treated with 12 μM BP2109 for 16 h, the cells were infected with DENV-2 and harvested for RNA measurements at 72 h postinfection. Our results showed that cells infected at MOIs of 1.0 and 0.1 and treated with BP2109 demonstrated approximately 24- to 35-fold reduction in positive-strand viral RNA levels, respectively (Fig. 4). These results demonstrated that BP2109 reduced the viral RNA replication of DENV-2 by affecting the synthesis of positive-strand DENV-2 RNA.

FIG. 4.

BP2109 reduces the levels of viral RNA in BHK21 cells infected with DENV-2. BHK21 cells infected with DENV-2 at an MOI of 0.1 or 1.0 were maintained in the absence or presence of 12 μM BP2109 for 72 h. Positive-strand DENV-2 RNA accumulations in cells were determined by using TaqMan fluorogenic quantitative RT-PCR. The RNA copy number was determined by using standards that were measured with a spectrophotometer. The mean values and SEM from three independent experiments are plotted.

BP2109 inhibits the replication activity of cells harboring a DENV-2 replicon.

To further validate the mechanistic function of BP2109, we constructed a subgenomic Rluc-based reporter DENV-2 replicon, DV2Rep (Fig. 5A). Given that a viral protease inhibitor interferes with polypeptide processing and subsequent steps (33), we monitored the reduction rate of reporter gene expression levels as an indicator of the inhibition of BP2109. The DENV-2 reporter replicon was transcribed in vitro and transfected into BHK21 cells to distinguish the inhibition of viral translation from the inhibition of RNA synthesis. Luciferase activity was monitored several times over a period of 48 h posttransfection (Fig. 5B). The luciferase activity reached a peak level at around 48 h posttransfection, and the level of luciferase activity was sustained until 72 h posttransfection in the absence of BP2109 (data not shown). Because the luciferase activity peaked within the first 8 h posttransfection, and also after 24 h posttransfection (representing viral translation and RNA replication, respectively [39]), we measured luciferase activity at 2, 4, 8, 24, 36, and 48 h postinfection. BP2109 had a minimal effect on Rluc signals at 2, 4, and 8 h postinfection, but the signals were significantly reduced, by 48%, 96%, and 99%, at 24, 26, and 48 h postinfection, respectively (Fig. 5B). Thus, our data demonstrated strong suppression of viral RNA synthesis by BP2109. To measure the 50% effective concentration (EC₅₀) of the dengue virus replicon and compare it to the 72-h cytotoxicity data, we transfected DV2Rep RNAs into BHK21 cells in the presence or absence of different concentrations of BP2109 for 72 h and measured the luciferase activities of cell lysates. The EC₅₀ of DV2Rep was measured to be 0.17 ± 0.01 μM (EC₅₀s were determined with the Renilla luciferase assay; standard deviations were calculated from at least 3 independent experiments). The concentration causing 50% cytotoxicity (CC₅₀) of BP2109 was also determined by the MTS method to be 29.28 ± 0.43 μM (CC₅₀ values were determined as a 50% drop in MTS signals; the value is the mean ± standard deviation from triplicate experiments), and the therapeutic index (calculated as the CC₅₀ divided by the EC₅₀) was calculated as 172.

FIG. 5.

Inhibition of DENV-2 RNA replication by BP2109. (A) Schematic diagram of the DENV-2 reporter replicon, DV2Rep. Indicated are the 5′ UTR, the N-terminal 102 amino acids of the C protein (C102), the Renilla luciferase gene (Rluc), the FMDV2A cleavage site, the neomycin resistance gene (Neo), an EMCV IRES element, the C-terminal 24 amino acids of E (E24), the entire NS regions (NS1∼NS5), and the 3′ UTR. The replicon was used to quantify the inhibitory effects of BP2109 on the RNA replication of DENV-2. (B) Analysis of BP2109 using transient replicon assays. BP2109 inhibits the viral replication stages (24 to 48 h) rather than the viral translation stages (2 to 8 h). The replicon was transfected into BHK21 cells, which were then maintained in the absence or presence of 8 μM BP2109. Luciferase activity was monitored at the indicated time points posttransfection. The numbers above the BP2109-treated bars represent the percentages of luciferase signals relative to the mock-treated controls (100%), and the error bars represent the SEM from three independent experiments.

Selection and characterization of BP2109-resistant DENV-2.

In order to identify the antiviral mechanism and the molecular target of BP2109, we obtained BP2109-resistant viruses by serially passaging WT DENV-2 in BHK21 cells at an MOI of 0.1 PFU/cell with increasing concentrations of BP2109 (6 to 10 μM), as shown in Fig. 6. With the intention of deciding whether BP2109-resistant viruses emerged in the presence of BP2109, the virus titer of BHK21 cells infected with parental WT DENV-2 was used for comparison with cells infected with viruses from different passages. The virus titer (approximately 10⁴ PFU/ml) derived from two independent pools of BHK21 cells infected by WT DENV-2 (MOI = 0.1) is similar to that derived from the first 10 or 25 passages 72 h postinfection in the presence of BP2109 (6 or 8 μM). In contrast, in the presence of BP2109 (10 μM), the titer (6 × 10⁴ PFU/ml) derived from BHK21 cells infected with passage 40 virus was found to be higher than the titer (3 × 10³ PFU/ml) derived from BHK21 cells infected with WT DENV-2 144 h postinfection, indicating that the passage 40 viruses had become resistant to BP2109. Moreover, the mixture of viral RNAs from two pools of passage 40 viruses was isolated and amplified by RT-PCR, and the PCR products of the NS2B-NS3 region were cloned into yeast cells. Twenty independent plasmids were purified from the yeast cells, amplified in E. coli, and subjected to DNA-sequencing analyses. The sequencing data revealed that only 10 clones harbored the right inserts (NS2B/NS3pro) without deletion, and half of the 10 clones had accumulated two consensus amino acid substitutions, R55K and E80K, in the region of NS2B, a cofactor of the NS2B/NS3pro complex (Table 2). Collectively, our results indicate that R55K and E80K substitution mutants may confer on DENV resistance to BP2109, and BP2109 is likely to target NS2B and affects protease function, thus reducing viral replication.

FIG. 6.

Schematic representation of the selection process for BP2109-resistant DENV-2. Two independent selections were performed. P1 through P10 were selected with 6 μM BP2109, P11 through P25 were selected with 8 μM BP2109, and P26 through P40 were selected with 10 μM BP2109. Viral RNAs from virus pools in P40 were extracted and amplified by RT-PCR and then cloned into vectors. The NS2B/NS3 region of each clone was sequenced.

TABLE 2.

Sequencing analyses of the dengue NS2B-NS3 genes from clones derived from BP2109-resistant DENV-2

Sequence	Encoded amino acid at position:
	NS2B													NS3
	12	13	24	33	38	52	55	80	84	96	116	125	126	8	17	65	68	69	83	88	95	116	137	145	187	194
Parental	G	M	N	V	L	E	R	E	M	T	I	E	V	P	E	I	S	W	W	K	V	F	S	K	K	H
Clone
1							K	K
2							K	K											C
3							K	K					M				P				D
4							K	K
5							K	K
6								K
7								K													D					R
8							K
9
10	R		D									G				V

Mutations at both R55K and E80K within dengue virus NS2B lead to BP2109 resistance of the dengue virus NS2B/NS3 protease and replicon activity.

To further identify the determinant of BP2109 resistance, we prepared recombinant NS2B/NS3 protease containing the above-mentioned KK double mutations (R55K and E80K). Two amino acid substitutions, R55K and E80K, in the NS2B region reduced the BP2109 susceptibility of the mutant protease. The IC₅₀s for WT and mutant (KK) proteases were calculated as 16.86 ± 0.06 μM and 158.44 ± 1.20 μM, respectively. The double-mutant protease exhibited 9.4-fold-higher resistance than the WT protease (Fig. 7A). The KK double mutations were also introduced into the wild-type DENV-2 reporter replicon, DV2Rep, and similar results were observed in this dengue virus replicon system. The KK mutant replicon showed a resistance level higher than that of the parental replicon. The EC₅₀s for the WT and KK replicons were calculated as 0.17 ± 0.01 μM and 12.55 ± 4.79 μM, respectively (Fig. 7B). The EC₅₀ of the KK mutant replicon was approximately 73.8-fold higher than that of the WT replicon. Interestingly, the EC₅₀s of the single-mutant replicons, R55K and E80K, were 0.15 ± 0.01 μM and 10.43 ± 4.7 μM, respectively. The EC₅₀ of the E80K mutant replicon had an approximately 61.3-fold-higher resistance than the WT replicon. In contrast, the R55K mutation alone did not confer resistance to BP2109 on the dengue virus replicon. To further determine the replication capacities of NS2B mutant replicons, the replication kinetic profiles were obtained by measuring the reporter activities of BHK21 cells transiently transfected with NS2B replicons. The R55K mutant dengue virus replicon displayed a replication kinetic profile similar to that of the wild-type dengue virus replicon (Fig. 7C). However, the E80K or KK mutant dengue virus replicon had lower replication capacity than the WT dengue virus replicon. Compared to the replication kinetic profile of the E80K dengue virus replicon, a slight increase was observed in the replication capacity of the KK dengue virus replicon. In summary, the change at residue 80 to lysine in the NS2B region is crucial for drug resistance. Our results from two different assay systems demonstrate that the residue at position 80 in the NS2B region is the major determinant for BP2109 resistance and indicate that the molecular target of BP2109 is likely NS2B.

FIG. 7.

Double mutations (R55K and E80K) within dengue virus NS2B resulted in BP2109 resistance of the dengue virus NS2B/NS3 protease enzyme and the dengue virus replicon. (A) BP2109 susceptibility of the WT and KK double-mutant NS2B/NS3 proteases. The curves were fitted based on the enzyme activity as percentages of the control value (untreated controls). The IC₅₀s of the WT and KK mutant NS2B/NS3 proteases representing a 50% reduction in protease activity assay were calculated as 15.43 ± 2.21 μM and 158.44 ± 1.20 μM, respectively. (B) Resistance analyses of the WT replicon (DV2Rep), two K single-mutant replicons, and the KK double-mutant replicon. Two amino acid substitutions in the DENV NS2B region (R55K and E80K) were introduced into the replicon construct. The EC₅₀s of the WT and KK mutant replicons were derived from Renilla luciferase activity assays. BHK21 cells were transfected with equal amounts (0.5 μg) of WT and KK mutant replicon RNAs and then treated with BP2109 at the indicated concentrations. The EC₅₀ values of the WT, R55K, E80K, and KK replicons were determined by calculating the luciferase activities at 72 h posttransfection; the EC₅₀s were 0.17 ± 0.01 μM, 0.15 ± 0.01 μM, 10.43 ± 4.7 μM, and 12.55 ± 4.79 μM, respectively. (C) Replication kinetic curves of WT and NS2B mutant replicons. The replicons were transfected into BHK21 cells, and the luciferase activity of each clone was monitored at the indicated time points posttransfection. The mean values and SEM from three independent experiments are plotted.

DISCUSSION

In the present study, we used HTS to identify BP2109 as an NS2B/NS3pro inhibitor with a selective antiviral spectrum and potent inhibitory activity in vitro. We discovered that BP2109 was an effective inhibitor of the viral yields and viral RNA replication of, and cytopathology induced by, four serotypes of DENV (but not JEV) (Fig. 1 to 4). Sequencing analyses of isolated BP2109-resistant viruses revealed that two mutations (R55K and E80K) were frequently found within the cofactor NS2B region (Table 2). Most importantly, the introduction of double mutations (R55K and E80K) into the NS2B/NS3 protease and dengue virus replicon constructs conferred strong resistance to BP2109 (Fig. 7A and B). To the best of our knowledge, this is the first report showing the inhibition of dengue virus replication by a dengue virus inhibitor that is affected by the mutations located in the hydrophilic portion of the NS2B cofactor that extensively interacts with the NS3pro domain, suggesting a novel mode-of-action mechanism.

In the present study, we started the screening process with 41,600 compounds and obtained only one candidate, BP2109, that demonstrated moderate cytotoxicity and efficient inhibition of NS2B/NS3 protease-mediated substrate digestion. The discovery of so few small-molecule inhibitors by using the HTS in our small-molecule libraries is likely due to one of several possible scenarios. First, the design of the NS2B/NS3 protease assay is not optimized for HTS. It has been shown that an in vitro protease activity assay such as ours requires a high pH of 9.0 and 20% glycerol to optimize DENV NS2B/NS3 protease activity (24). Unfortunately, these factors resulted in the protonation of some classes of compounds, which led to false-positive or false-negative hits, as well as higher viscosities that led to frequent pipetting errors in the robotic liquid-handling platforms (36). Second, the truncated NS2B/NS3 protease assay used in the present study may not entirely reflect the physiological conformation of the full-length NS2B/NS3 protease because the designed NS2B/NS3 protease construct lacks the three transmembrane helices of NS2B and most of the NS3 regions, including the helicase, NTPase, and RTPase domains. The three transmembrane helices of NS2B are thought to anchor the NS2B/NS3 protease to the endoplasmic reticulum (ER). Regrettably, there is still no full-length NS2B/NS3 protease assay with an artificial membrane available for HTS. Third, the X-ray crystal structure analysis revealed the fact that the active site of NS2B/NS3 protease is relatively flat and negatively charged (10), which may also play a role in the lower-than-expected HTS hit rate.

DENV-2 luciferase reporter replicon-based kinetics assays showed that viral RNA replication was completely blocked by BP2109 (>90%) without the suppression of viral translation (Fig. 5). It is interesting that BP2109 appeared to be much more potent in the dengue virus replicon-based assays (EC₅₀ = 0.17 ± 0.01 μM) than in the enzyme-based protease assay (IC₅₀ = 15.43 ± 2.12 μM). This inconsistency is likely due to our artificial-enzyme-based protease assay, which is performed using the soluble recombinant NS2BH-NS3pro protease complex consisting of only the NS2BH cofactor (residues 49 to 92) and the NS3 protease domain (residues 1 to 184). The NS2BH-NS3pro protease complex activity may not reflect the physiological function of the full-length NS2B/NS3 in vivo. Because the three transmembrane helices of NS2B that anchor the active NS2B/NS3 protease to the ER are absent (36), the in vivo environment of the protease is completely different from the in vitro artificial assay system. The finding that glycerol or gelatin enhances NS2B/NS3pro protease complex activities also supports this hypothesis (24). Recent studies on HCV serine protease (3), in which the helicase domain of NS3 enhances enzyme-based NS3-NS4A protease activity, also support our hypothesis. In addition, the recently reported crystal structure of a complete NS3 molecule of DENV (2VBC) (28) suggests that the protease domain of NS3 increases its affinity for nucleotides, participates in RNA binding, and enhances helicase activity. Therefore, it is likely that the two enzymatic domains of DENV NS2B/NS3 are likely to be highly interdependent in the viral life cycle. Collectively, these results may explain why BP2109 appeared to be more potent in the replicon-based assays than in the enzyme-based protease assay.

Because our screening assay was a DENV-2 protease-based HTS, we expected BP2109 to be more selective for DENV-2 than JEV in both the protease activity assay and the viral yield reduction assay. This was indeed the case: BP2109 had an inhibitory effect on the viral yield of DENV-1 to -4 (Fig. 2) but virtually no effect on the cytopathology and proliferation of JEV (Fig. 3). Despite this result, we further aligned and investigated the region of the NS2B/NS3pro protease that may be responsible for the different susceptibilities to BP2109 of DENV and JEV (see Fig, S1 in the supplemental material). In general, the amino acid sequences of NS2B/NS3pro full-length NS2B (residues 1 to 130) and the N terminus of NS3 (residues 1 to 184) are more conserved among DENV than those of NS2B/NS3pro between DENV and JEV. This may explain the different susceptibilities of DENV and JEV to BP2109. The degree of conservation in the amino acid sequences of NS2B/NS3pro between DENV-1 and DENV-2, DENV-3, or DENV-4 was shown to be 65% identity (82% similarity), 71% identity (86% similarity), or 62% identity (81% similarity), respectively. Furthermore, phylogenetic tree analyses of the amino acid sequences of DENV NS2B/NS3pro revealed that two clusters exist, with DENV-2 and DENV-4 in one cluster and DENV-1 and DENV-3 in another (data not shown). Therefore, the better inhibitory effect of BP2109 on the virus yield of DENV-2 or -4 than of DENV-1 or -3 (Fig. 2) may result from the degree of conservation of amino acid sequences of NS2B/NS3pro among DENVs. Further studies are required to identify the binding region of BP2109 to DENV-2 NS2B/NS3pro and to explain the different susceptibilities of DENV-2 and JEV to BP2109.

Although the R55K or E80K mutation was not found to be dominant in a population of isolated BP2109 drug-resistant DENV-2 derived from either passage 10 or 25 viruses in the presence of 6 or 8 μM BP2109, respectively (data not shown), the R55K (6 out of 10 clones) or E80K (7 out of 10 clones) mutation was shown to be a dominant mutation in the passage 40 viruses after 15 more passages of selection in the presence of 10 μM BP2109 (Table 2). Two accumulated K substitution mutations (R55K and E80K) in the region of NS2BH were identified from 5 out of 10 clones derived from BP2109-resistant DENV-2 (Table 2). The KK mutant was approximately 10-fold less susceptible to BP2109 in the enzyme-based protease assay (Fig. 7A) and at least 73-fold less sensitive in the transient dengue virus replicon assay (Fig. 7B). The transient dengue virus replicon assay further showed that the E80K single mutation contributed 61.3-fold-higher resistance to BP2109, whereas the other R55K single mutation had no effect on resistance to BP2109 (Fig. 7B). This result indicates that the R55K mutation enhanced the resistance capacity of the E80K dengue virus replicon to BP2109, although the R55K dengue virus replicon alone was not resistant to BP2109. In Fig. 7C, the replication curves of various dengue virus replicon mutants further demonstrate that the KK dengue virus replicon has better replication capacity than the E80K replicon, which explains our observation that KK mutations are frequently detected in the isolated BP2109-resistant viruses (Table 2). Taken together, the E80 residue located at the C terminus of DENV-2 NS2BH determines resistance to BP2109, while the R55K mutation plays a role in enhancing the replication capacity of the E80K mutation.

Our finding that BP2109 blocks dengue virus RNA synthesis without suppression of viral translation (Fig. 5) led us to believe that BP2109 inhibits DENV by interfering with the function of the NS2B cofactor and inhibiting NS2/NS3 protease activity, followed by the suppression of polypeptide processing and blockage of viral RNA replication. A similar inhibitory mechanism has been suggested in an interesting study of a novel noncompetitive inhibitor of WNV NS2B/NS3pro that was discovered by HTS of a 65,000-compound library (21). The noncompetitive inhibitor may inhibit NS2B/NS3pro by a mode of action similar to that of BP2109, although no detailed mechanistic study was performed to determine its mode of action. Supporting evidence from previous studies demonstrated that NS2BH (amino acid residues 50 to 80 of NS2B) affects the formation of the active site (35), interacts with NS3pro of DENV strongly and directly (29), and actively participates in the formation of the S2 and S3 subpockets in the protease active site (10). Furthermore, recent DENV homology-modeling work suggests that residue E80 of NS2B may be located in the S2 pocket (45), implying that BP2109 may influence the unique mode of interaction between the NS3 protease and its cofactor by binding to the C terminus of the NS2BH domain. Thus, we proposed that the mode of action by which BP2109 inhibits DENV is by interrupting the interactions between NS2B and the NS3 protease rather than by acting as an enzyme inhibitor that abolishes proteolytic activities by directly binding to the active site of an enzyme.

In summary, the identification and characterization of BP2109 represents the first step in the development of this compound for potential anti-dengue virus therapy. In addition, based on recent experimental, crystallographic, homology-modeling, and mutagenesis studies of NS2B/NS3, we believe that BP2109 has a novel mode of action involving interfering with the interaction between a protease enzyme and its cofactor. Further work may be required to optimize BP2109 and to synthesize more active derivatives of BP2109, which will be useful for quantitative structure-activity relationship studies.