Blocking NS3-NS4B interaction inhibits dengue virus in non-human primates

Abstract

Dengue is a major health threat and the number of symptomatic infections caused by the four dengue serotypes is estimated to be 96 million¹ with annually around 10,000 deaths². However, no antiviral drugs are available for the treatment or prophylaxis of dengue. We recently described the interaction between non-structural proteins NS3 and NS4B as a promising target for the development of pan-serotype dengue virus (DENV) inhibitors³. Here we present JNJ-1802-a highly potent DENV inhibitor that blocks the NS3-NS4B interaction within the viral replication complex. JNJ-1802 exerts picomolar to low nanomolar in vitro antiviral activity, a high barrier to resistance and potent in vivo efficacy in mice against infection with any of the four DENV serotypes. Finally, we demonstrate that the small-molecule inhibitor JNJ-1802 is highly effective against viral infection with DENV-1 or DENV-2 in non-human primates. JNJ-1802 has successfully completed a phase I first-in-human clinical study in healthy volunteers and was found to be safe and well tolerated⁴. These findings support the further clinical development of JNJ-1802, a first-in-class antiviral agent against dengue, which is now progressing in clinical studies for the prevention and treatment of dengue.

Fig. 1. The molecular structure, in vitro pan-genotype and -serotype activity, and mechanism of action of JNJ-1802.

a, The molecular structure of JNJ-1802. b, In vitro antiviral activity in Vero E6 cells against a panel of clinical isolates. Data are mean EC₅₀ values. The asterisk and hash symbols indicate that the DENV strain carries a T108I (*) or T108A (#) mutation in NS4B. Where indicated by a triangle, the mean EC₅₀ in Vero E6 cells was calculated by setting the values below the 0.04 nM threshold at 0.04 nM. A, American; AA, Asian American; AI, Asian I; AII, Asian II; C, cosmopolitan; S, sylvatic. c, Schematic of the DENV NS4B membrane topology^,. JNJ-1802-selected resistance mutations in orange were present in at least 99% of the quasispecies at the end point (passage 42 (sample A); passage 50 (sample B)). Mutations in black were present in less than 50% of the quasispecies at the end point. Mutations in blue appeared transiently and had disappeared at the end of the experiment. The diagram was created in part using the Servier Medical Art library (https://smart.servier.com/). d, JNJ-1802 prevents DENV NS3–NS4B interaction. Three independent co-immunoprecipitation experiments were performed to establish the JNJ-1802 dose–response curve for the NS3–NS4B interaction. Representative western blots are shown in Extended Data Fig. 2a,b. Signal intensity ratios were determined as described in the Methods. Data are mean ± s.e.m. For the comparison of NS3:NS4B–HA ratios between JNJ-1802 treated samples and DMSO control, P values were calculated using repeated-measures one-way analysis of variance (ANOVA) with subsequent Dunnett’s multiple-comparisons test; NS, not significant. EC₅₀, 50% effective concentration; DMSO, dimethyl sulfoxide; IP, immunoprecipitation; TM, transmembrane. Source Data

Fig. 2. In vivo efficacy of JNJ-1802 (b.i.d.) against DENV-1–4 after infection in AG129 mice.

a, Schematic of viraemia studies against DENV-2. b, The effect of JNJ-1802 on viraemia on day 3 after infection in DENV-2-infected mice treated b.i.d. with 60, 20, 6, 2, 0.6, 0.2 mg per kg per day JNJ-1802 (n = 7, 14, 14, 16, 16 and 13, respectively; without anti-flavivirus antibodies), compared with vehicle-treated mice (n = 24). Treatment started 1 h before infection. Undetermined C_t values imputed at a value of 40 (the limit of detection (LOD)) correspond to 2.6 log₁₀ viral RNA copies per ml. Pooled data of three independent studies were analysed using two-way ANOVA with Dunnett’s multiple-comparison test. The LLOQ is 3.7 log₁₀ viral RNA copies per ml. c, Schematic of the survival study. d, The effect of JNJ-1802 on survival in mice treated with anti-flavivirus antibodies (clone 4G2) receiving b.i.d. 20, 6, 2, 0.6 or 0 mg per kg per day JNJ-1802 starting 1 h before infection. Data are from a single study (n = 10 mice per group). Two-sided Fisher’s exact tests were used on day 25 with Bonferroni’s multiple-comparison test. e, Outline of the viraemia/survival studies against DENV-1, DENV-3 and DENV-4. f–k, The effect of JNJ-1802 on survival (n = 10 mice per group, except for DENV-3 vehicle, for which n = 11 mice per group) (f–h) and viraemia (n = 5 mice per group) on day 3 after infection (i–k) in mice challenged with DENV-1/West Pac (f,i), DENV-3/C0360/94 (g,j) or DENV-4/703/4 (h,k), treated b.i.d. with 60, 6 or 0 mg per kg per day JNJ-1802. Treatment started 1 h before infection. The LOD is 1.7 log₁₀ viral RNA copies per ml. Two-sided Fisher’s exact tests were used on day 6 (survival) with Bonferroni’s multiple-comparison test. For viraemia, ordinary one-way ANOVA with Dunnett’s multiple-comparison test (DENV-1) and Kruskal–Wallis tests with Dunn’s multiple-comparison test (DENV-3 and DENV-4) were performed. P values are shown in the figures. HEP, human end point. For a, c and e, the schematics were adapted from ref. . The dagger symbol indicates the day of euthanasia. Source Data

Fig. 3. In vivo efficacy of JNJ-1802 against DENV-2 in NHPs by measuring viral RNA load, NS1 protein, IgM/IgG response and TCID₅₀.

a, Schematic of the viraemia studies using rhesus macaques. b, The effect of JNJ-1802 on viral RNA in plasma of rhesus macaques (R1–R12) treated q.d. with vehicle, or 0.01, 0.18 or 3 mg per kg per day JNJ-1802 (n = 3 per group). Treatment started 1 day before infection. Analyses were performed in triplicate. Data are mean ± s.d. The LLOQ or LOD of 1,286 GCE per ml is indicated by a dotted line in the graphs. Undetectable levels are shown as 10⁰ GCE per ml. c, The levels of NS1 protein in serum samples of rhesus macaques. Analyses were performed in duplicate. Data are mean. An index value of less than 0.9 is considered to be negative, between 0.9 and 1.1 equivocal, and all values greater than 1.1 positive. The area between 0.9 and 1.1 is indicated by a grey horizontal bar. d,e, The levels of IgM (d) or IgG (e) antibodies in the sera of groups of rhesus macaques. IgM/IgG antibody levels are expressed as the mean index value from two independent assays. An index value of greater than 1.0 is presumptive for the presence of IgM/IgG antibodies to DENV. The cut-off value of 1.0 is indicated by a dotted line in the graphs. f, Quantification of infectious virus in plasma samples of rhesus macaques using the 50% tissue culture infective dose (TCID₅₀) assay. Only samples that tested positive by RT–qPCR assay or in the NS1 ELISA, as well as samples with indeterminate outcomes in these assays, were analysed. NA, not applicable. For a, the schematic was adapted from ref. , Springer Nature Limited. The dagger symbol indicates the day of euthanasia. Source Data

Fig. 4. The in vivo efficacy of JNJ-1802 against DENV-1/45AZ5 was examined in NHPs by measuring viral RNA load and IgM/IgG response.

a, Schematic of the DENV-1/45AZ5 study. b, The effect of JNJ-1802 on viral RNA in rhesus macaques (R1–R12) infected with DENV-1/45AZ5 (0.5 ml at 1.2 × 10⁵ PFU per ml) and treated with JNJ-1802 at 6 mg per kg per day once daily (n = 6; right) in comparison to the vehicle-treated group (n = 6; left). Treatment started 3 days before infection. RNA levels are expressed in GCE per ml. Data are mean ± s.d. The LLOQ of 100 GCE per ml is indicated by a dotted line in the graphs. c,d, ELISA data are presented as end-point titres by study day, which is defined as the reciprocal of the highest dilution of the serum that gives a positive signal. Samples were assayed in duplicate. Data are mean. For a, the schematic was adapted from ref. , Springer Nature Limited. The dagger symbol indicates the day of euthanasia. Source Data

Extended Data Fig. 1. In vitro resistance selection profile of JNJ-1802 against DENV-2.

a, b, The dynamics of emerging mutations were studied using whole virus genome sequencing in two independent experiments (Experiment A (a) and experiment B (b)). Each coloured line shows the kinetics of appearance of a particular mutation during virus passaging in the presence of JNJ-1802; each mutation is shown in the same colour over the different experiments. Emerging mutations were defined as amino acid changes compared to the start virus (DENV-2/RL), which were not present in the viruses that were passaged in parallel without any drug exposure. Whole genome sequencing was performed on DENV variants harvested at every 5^th passage (P) and at the end of the experiments (i.e., P42 for experiment A and P50 for experiment B). One passage represents a one-week time span. The dotted line represents the cut off (15%) for the detection of variants compared with wild-type in the virus population. The increasing EC₅₀ values, as determined by microscopic evaluation of virus-induced CPE, are depicted below the graphs. c, Level of compound resistance conferred by NS4B mutations in DENV-2/16681 induced by JNJ-1802. Fold change (mean ± standard deviation) in compound resistance is calculated as the EC₅₀ of the compound against the mutant sub-genomic DENV-2/16681 reporter replicon divided by the EC₅₀ of the compound against wild-type sub-genomic DENV-2/16681 reporter replicon. Data are from at least three independent experiments. d, Natural occurrence of the NS4B mutations in clinical isolates. *The natural occurrence of the mutations was retrieved from the Virus Pathogen Resource database (www.viprbrc.org; accessed in May 2020). Prevalence values of ≤0.1% are not shown. CPE, cytopathic effect; EC₅₀, 50% effective concentration. Source Data

Extended Data Fig. 2. JNJ-1802 prevents DENV NS3-NS4B interaction.

a, b, Huh-7 cells stably expressing T7 RNA polymerase and DENV-2 NS2B-NS3 were transfected with T7-based expression plasmids encoding NS4A-2K-NS4B(-HA^Ct) (NS4B corresponding to wild-type [WT] or mutant NS4B). Four hours post-transfection, cells were treated with various concentrations of JNJ-1802 or DMSO, collected 14 h later, lysed and used for HA-specific pull-down and Western blot analysis of captured complexes. Representative Western Blot of the dose-response assays. Samples contain either WT NS4B or NS4B T108I (a) and mutant NS4B (V91A or L94F) (b). Please note the higher compound concentrations compared to (a). c–f, Western blot signal intensities of total lysates (input) obtained with an anti-HA-antibody were used to determine the dose-dependent effect of JNJ-1802 on the relative abundance of NS4B-containing species (mean ± standard error of the mean [SEM] of three independent experiments). Repeated measures one-way ANOVA and Dunnett’s multiple comparisons test were used to calculate P values. g, EC₅₀ values (mean ± SEM) for the NS3-NS4B species ratio (a, b) and the change of 2K-NS4B proportion in total lysates (c–f) were obtained by fitting four-parameter dose-response curves to the results from each individual experiment. h, Experimental setup to investigate blockade of NS3-NS4B interaction with regard to the timing of JNJ-1802 addition. As indicated, cells were treated with 10 nM JNJ-1802 or DMSO, collected 8 h later, lysed and used for HA-specific pull-down and Western blot analysis. The diagram was created in part using the Servier Medical Art Library (https://smart.servier.com/). i, Representative Western blot from three independent experiments for the setup in (h). j–k, Effect of early and late treatment on the NS3/NS4B-HA species ratio in samples after HA-pulldown. Results are the mean and SEM from three independent experiments. Paired two-tailed t-test was applied to calculate P values. Uncropped images of a,b and i are presented in Supplementary Fig. 1–3. ns, not significant; Ct, C-terminal. Source Data

Extended Data Fig. 3. In vivo efficacy of JNJ-1802 against DENV-2 infection, dosed once-daily for 3 days.

a, Schematic outline of a viraemia study using AG129 mice adapted from Kaptein et al. b, Effect of JNJ-1802 on viraemia on Day 3 post-infection (p.i.) in mice orally treated once daily (q.d.) with a dose of 30, 3 or 0.3 (n = 8 for all groups) mg/kg/day JNJ-1802, compared to vehicle-treated mice (n = 8), with the first dose administered 1 h before infection. Data are from a single study. Undetermined cycle threshold values were imputed at a value of 40 (=limit of detection [LOD]), corresponding to 2.6 log₁₀ viral RNA copies/mL. For statistical analysis, ordinary one-way ANOVA was used and Dunnett test to correct for multiple comparisons. p <0.0001 for 30 mg/kg/day versus vehicle; p <0.0001 for 3 mg/kg/day versus vehicle and p = 0.0983 not-significant (ns) for 0.3 mg/kg/day versus vehicle. LLOQ, lower limit of quantification, is 3.7 log₁₀ viral RNA copies/mL. 4G2, anti-flavivirus antibody, clone 4G2. Source Data

Extended Data Fig. 4. In vivo efficacy of JNJ-1802 against DENV-2 infection, dosed twice daily for 6 days.

a, Schematic of the in vivo kinetics study. Compound administration started 1 h before infection and continued twice daily (b.i.d.) for 6 consecutive days. Each treatment group (n = 8) was equally divided into sub-groups A and B for blood collection on alternating days. b, Weight curves (mean values ± s.d.) of AG129 mice in the different treatment groups. c–e, Inhibitory effect of JNJ-1802 on viraemia in mice treated b.i.d. with 20, 2 or 0.2 mg/kg/day JNJ-1802 as compared to vehicle-treated mice. Data (mean ± s.d.) are from a single experiment. f, Schematic of the in vivo kinetics study in a therapeutic setting with treatment starting on Day 4 or 5 after DENV-2 challenge. In the control groups (vehicle and JNJ-1802) treatment started on the day of infection. Each treatment group (n = 10) was equally divided into sub-groups A and B for blood collection on alternating days. g, h, Inhibitory effect of JNJ-1802 on viraemia in AG129 mice with treatment (60 mg/kg/day, b.i.d. for 6 consecutive days) starting on Day 4 (g) and Day 5 (h) p.i. compared with the two control groups, vehicle and JNJ-1802. Data (mean ± s.d.) are from a single experiment. For both experiments (c–e; g, h), undetermined Ct values were imputed at a value of 40 (=limit of detection [LOD]), corresponding to 2.6 log₁₀ viral RNA copies/mL (c-e) or 3.3 log₁₀ viral RNA copies/mL (g-h). i, Mean AUC and 95% CIs were determined for each group depicted in (c–e; g, h). If CIs did not overlap, groups were considered to differ significantly. LLOQ, lower limit of quantification, is calculated to be 3.7 log₁₀ viral RNA copies/mL (c-e) or 4.4 log₁₀ viral RNA copies/mL (g-h). Schemes are adapted from Kaptein et al.. Source Data

Extended Data Fig. 5. Individual and mean JNJ-1802 plasma concentration-time profiles in non-human primates.

a, Weight curves (mean values + standard deviation) of rhesus macaques in the different treatment groups (n = 3 per group, except for Day 3, n = 1 for 0.01 mg/kg/day and 0.18 mg/kg/day groups and n = 2 for 3 mg/kg/day group). b, Mean JNJ-1802 plasma concentration-time profile for the DENV-2 study in NHP. Rhesus macaques (n = 3 per group) were treated prophylactically with different doses (3, 0.18 or 0.01 mg/kg/day, once daily) of JNJ-1802, and subsequently experimentally infected with 10² TCID₅₀ of DENV-2/16681. Dashed line indicates 3x protein-binding adjusted 90% effective concentration (pbaEC₉₀) against DENV-2/16681 (8.2 ng/mL). c, Weight curves (mean values + standard deviation) of rhesus macaques in the different treatment groups (n = 6 per group). d, Mean JNJ-1802 plasma concentration-time profile for the DENV-1 study in NHP. Rhesus macaques were prophylactically treated with JNJ-1802 at 6 mg/kg/day once daily (n = 6, except for Day 1 n = 2) and subsequently infected with DENV 1/45AZ5 (0.5 mL at 1.2x10⁵ plaque forming units [PFU]/mL). NHP, non-human primates. Source Data