Inhibition of Enterovirus 71 by Adenosine Analog NITD008

ABSTRACT

Enterovirus 71 (EV71) is a major viral pathogen in China and Southeast Asia. There is no clinically approved vaccine or antiviral therapy for EV71 infection. NITD008, an adenosine analog, is an inhibitor of flavivirus that blocks viral RNA synthesis. Here we report that NITD008 has potent antiviral activity against EV71. In cell culture, the compound inhibits EV71 at a 50% effective concentration of 0.67 μM and a 50% cytotoxic concentration of 119.97 μM. When administered at 5 mg/kg in an EV71 mouse model, the compound reduced viral loads in various organs and completely prevented clinical symptoms and death. To study the antiviral mechanism and drug resistance, we selected escape mutant viruses by culturing EV71 with increasing concentrations of NITD008. Resistance mutations were reproducibly mapped to the viral 3A and 3D polymerase regions. Resistance analysis with recombinant viruses demonstrated that either a 3A or a 3D mutation alone could lead to resistance to NITD008. A combination of both 3A and 3D mutations conferred higher resistance, suggesting a collaborative interplay between the 3A and 3D proteins during viral replication. The resistance results underline the importance of combination therapy required for EV71 treatment.

IMPORTANCE Human enterovirus 71 (EV71) has emerged as a major cause of viral encephalitis in children worldwide, especially in the Asia-Pacific region. Vaccines and antivirals are urgently needed to prevent and treat EV71 infections. In this study, we report the in vitro and in vivo efficacy of NITD008 (an adenosine analog) as an inhibitor of EV71. The efficacy results validated the potential of nucleoside analogs as antiviral drugs for EV71 infections. Mechanistically, we showed that mutations in the viral 3A and 3D polymerases alone or in combination could confer resistance to NITD008. The resistance results suggest an intrinsic interaction between viral proteins 3A and 3D during replication, as well as the importance of combination therapy for the treatment of EV71 infections.

INTRODUCTION

Enterovirus 71 (EV71), a member of genus Enterovirus in the family Picornaviridae, is one of the major causative agents of diarrhea, rashes, and hand, foot, and mouth disease (HFMD) in children. Besides EV71, HFMD could also be caused by coxsackievirus A16 (CV-A16) infection (1). Infection with EV71 is sometimes associated with severe central nervous system diseases (2). Since its first isolation in 1969 (3), EV71 outbreaks have occurred frequently in western Pacific region countries, including China, Japan, Malaysia, and Singapore. In China alone, EV71 caused over 1 million infections and hundreds of deaths each year from 2009 to 2011 (http://www.chinacdc.cn/tjsj/fdcrbbg/). No effective anti-EV71 drug is available, although several clinical trials of inactivated EV71 vaccine have recently been completed (4, 5). In a phase III trial, an inactivated alum adjuvant EV71 vaccine showed efficacies of 90% against EV71-associated HFMD and 80.4% against EV71-associated diseases (including HFMD, herpangina, meningitis or encephalitis, febrile illness, viral exanthema, and respiratory infection) (4). Despite the promising progress in vaccine development, antiviral therapy should still be pursued as a complementary means of intervention against EV71 infection.

EV71 is a nonenveloped, single-positive-strand RNA virus. The genomic RNA is approximately 7,000 nucleotides in length and contains a single, long open reading frame (ORF) that is flanked by the 5′ untranslated region (UTR) and the 3′ UTR with a poly(A) tail. VPg (3B) is covalently linked to the 5′ end of the viral genome and involved in the initiation of viral RNA replication (6). The 5′ UTR consists of two functional domains, a cloverleaf and an internal ribosome entry site, which contribute to viral replication and translation, respectively. The ORF encodes a single polyprotein that is initially processed by viral proteases into three precursor proteins, P1, P2, and P3 (7). Precursor P1 is further cleaved to yield structural proteins VP0, VP3, and VP1; and VP0 is finally cleaved into VP2 and VP4 during virion formation. P2 and P3 are proteolytically processed into various nonstructural proteins associated with viral replication, such as 2A^pro, 2BC, 2B, 2C, 3AB, 3A, 3B (VPg), 3CD, 3C, and 3D (8, 9). Following viral protein synthesis, the genomic RNA serves as a template for negative-strand RNA synthesis, generating the intermediate double-stranded replicative-form (RF) RNA. The double-stranded RF RNA is then used as a template to synthesize more positive-strand RNA. The nascent positive-strand RNA is encapsidated. Viral particles are released from infected cells (10).

Inhibitors have been reported for distinct targets of EV71 and its closely related rhinovirus, as summarized in a recent review (11). Pleconaril, picodavir, and BPROZ-194 were shown to target viral capsid protein VP1 (12–14). Rupintrivir was found to inhibit the 3C proteases of both human rhinovirus and EV71 (15, 16). Nucleoside analogs (such as 2′-C-methylcystidine, N-6-modified purine, and ribavirin) have been reported to inhibit the viral polymerase of EV71 (17–20). None of these inhibitors has been successfully developed into a clinically approved drug. Therefore, new chemical entities are urgently needed for EV71 antiviral development.

NITD008 is an adenosine nucleoside analog that has been reported to selectively inhibit viruses in the family Flaviviridae (21). Although NITD008 showed efficacy in a dengue virus mouse model, it was not advanced to clinics because of the adverse findings of a preclinical toxicity study (21). Here we report that NITD008 potently inhibits EV71 in cell culture and in a mouse model, demonstrating that this compound could potentially be developed for EV71 therapy. Mechanistically, we characterized the profile of EV71 resistance to the compound. The resistance results showed that mutations in the viral 3A and 3D polymerase regions could confer resistance to NITD008, suggesting intimate cross talk between 3A and 3D during viral replication.

MATERIALS AND METHODS

Viruses, cells, and compound.

Vero (African green monkey kidney) cells were cultured in Dulbecco modified Eagle medium (DMEM; Invitrogen) with 10% fetal bovine serum (FBS), 100 U/ml of penicillin, and 100 μg/ml of streptomycin. The wild-type (WT) recombinant viruses were prepared from the infectious cDNA clone of EV71 (22), stored as aliquots at −80°C, and used for antiviral activity assays and resistant virus selection. EV71 strain 5865/SIN/000009 (S41) (23) was used for an animal study. NITD008 was synthesized as reported previously (21).

Antiviral activity assay.

Vero cells were seeded into 12-well plates (1 × 10⁵/well), incubated for 24 h, and then incubated with viruses at a multiplicity of infection (MOI) of 0.1 PFU/cell for 1 h in DMEM with 10% FBS. After the cells were washed once with culture medium, a different concentration of NITD008 was added to each well. Cell cultures were collected at 48 h postinfection (p.i.), and viral titers in supernatants were quantified by plaque assay. The incubations for each concentration treatment were performed in triplicate.

Cytotoxicity assay.

A 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay kit (a tetrazolium colorimetric assay; American Type Culture Collection) was used to determine the cytotoxicity of NITD008 for Vero cells. We seeded Vero cells at 1 × 10⁴ per well (in 100 μl DMEM plus 10% FBS) of a 96-well plate, incubated them for 6 h, and then treated them with different concentrations of NITD008. After 48 h of incubation, 10 μl of MTT reagent was added and the plate was incubated for 3.5 h. Then 100 μl of detergent reagents was added, and the plate was swirled gently and left in darkness at room temperature for 4 h. The absorbance of each well was measured with a microplate reader (Varioskan Flash; Thermo Fisher) with a 550-nm filter. The cytotoxicity of NITD008 was estimated by comparing the absorbance of the compound-treated cells with that of mock-treated cells.

Generation of NITD008-resistant viruses.

NITD008-resistant viruses were selected by passaging WT EV71 in Vero cells with increasing concentrations of NITD008. Two independent selections were carried out. For the first round of selection, Vero cells were infected with WT EV71 at an MOI of 0.1 PFU/cell for 1 h, and then the infected cells were washed once and incubated with 1 μM NITD008 for another 48 h. The supernatants (designated P1) were collected and stored at −80°C. Viral titers from each selection were determined by plaque assay. The following selections were performed as described above. Different concentrations of NITD008 are indicated for different passages (see Fig. 2A).

FIG 2.

Selection and characterization of NITD008-resistant EV71. (A) Scheme used for selection of NITD008-resistant EV71. The concentrations of NITD008 corresponding to different passages during selection are indicated. (B, C) Characterization of resistance phenotypes of P8 (B) and P18 (C) EV71. The indicated concentrations of NITD008 were added to Vero cells after 1 h of incubation with EV71 at an MOI of 0.1 PFU/cell. The viral titers in culture fluids in the presence or absence of NITD008 were determined at 48 h p.i. by plaque assay. Two independent selections were performed (SelA and SelB). (D) Plaque morphologies of WT and selected resistant viruses.

Sequencing of drug-resistant viruses.

To identify mutations responsible for drug resistance, RNAs of selected viruses were extracted from virions in culture medium with RNeasy kits (Qiagen) according to the manufacturer's instructions and amplified by reverse transcription (RT)-PCR with the SuperScript III one-step RT-PCR kit (Invitrogen). The complete viral genome was covered by four RT-PCRs with four pairs of primers (22). The RT-PCR products were reclaimed by gel purification and sequenced.

Preparation of recombinant EV71 containing resistance mutations.

EV71 full-length cDNA clones with different mutations were constructed with the pACYC-EV71-FL infectious clone as the template (22). All mutations were introduced into the infectious clone by fusion PCR. Specifically, fusion PCR products containing mutations within the 2B/3A and 3D regions were introduced into pACYC-EV71-FL at the AatII-BsrGI and BsrGI-HindIII sites, respectively. All constructs were confirmed by DNA sequencing. To make recombinant viruses, WT and mutant genomic RNAs were in vitro transcribed from the HindIII-linearized cDNA plasmids with the MEGAscript T7 kit (Ambion) according to the manufacturer's protocols. The RNAs were transfected into Vero cells with DMRIE-C (Invitrogen). After transfection, supernatants were collected at different time points. The culture medium containing viruses was aliquoted and stored at −80°C.

Ethics statement.

The animal experiments described here were approved by the National University of Singapore Institutional Animal Care and Use Committee (protocol 070/10). Nonterminal procedures were performed while animal were under anesthesia, and all efforts were made to minimize suffering.

Mouse infection and drug treatment.

Two-week-old AG129 mice (B&K Universal) were bred and housed under specific-pathogen-free conditions in individual ventilated cages. Mice (n = 8 to 10) were inoculated intraperitoneally (i.p.) with 5865/SIN/000009 (S41) (23) at a concentration of 10⁷ PFU/mouse as reported previously (24). Mice were treated orally with NITD008 (6 N HCl [1.5 eq], 1 N NaOH [pH adjustment to 3.5], and 100 mM citrate buffer [pH 3.5] at a ratio of 0.9:2.5:96.6) at a concentration of 5 mg/kg twice a day (b.i.d.) from the day of infection and for 3 consecutive days. Two control groups were included in the study design; one group was left untreated after infection, while the other was given the vehicle alone. Mice were monitored daily for 20 days for clinical illness and death. Clinical disease was scored as follows: 0, healthy; 1, ruffled hair and hunchbacked appearance; 2, limb weakness; 3, paralysis in one limb; 4, paralysis in two limbs; 5, death. To minimize animal suffering, mice were euthanized if they were paralyzed in two limbs.

Determination of viral titers in organs from infected mice.

Following systemic perfusion with 50 ml of sterile phosphate-buffered saline (PBS), organs and tissues were harvested from untreated infected mice and infected mice treated with NITD008 or the vehicle at three different time points (days 4, 5, and 6) p.i. and weighed. Samples were homogenized in DMEM, and the resulting homogenates were centrifuged at 14,000 rpm for 10 min. Clarified supernatants were filtered through a 0.22-μm syringe filter (Millipore) before determination of the viral titer by plaque assay.

Viral plaque assay.

RD cells (ATCC CCL-136) were seeded into the wells of 24-well plates (Nunc) at a density of 10⁵/well. The growth medium was removed, and the cells were infected with 200-μl volumes of suspensions containing 10-fold serial virus dilutions ranging from 10⁻¹ to 10⁻⁸. Infection was carried out for 2 h at 37°C and 5% CO₂ with rocking at half-hour intervals. One milliliter of 1.2% Avicel in minimum essential medium (FMC BioPolymer) was added, and the cells were further incubated for 3 days. After the overlay medium was decanted, the cells were washed with PBS and fixed with 4% paraformaldehyde. Crystal violet was used to fix and stain the cells. Plaques were then quantified via visual scoring.

Statistical analyses.

Differences between data on antiviral activity in cell culture were statistically evaluated with the open-source R-project software (Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org/). Antiviral activity was assessed with the Kruskal-Wallis rank sum test. Statistical analysis of the animal data was performed with GraphPad Prism, version 5 (GraphPad4 Software). Data are expressed as means ± the standard error of the means (SEM). Kaplan-Meier survival curves and clinical-score curves were analyzed with a log rank test and the Wilcoxon test, respectively. A two-tailed P value of <0.05 was considered statistically significant.

RESULTS

Antiviral activity of NITD008 against EV71 replication in cell culture.

To facilitate antiviral activity discovery, we previously developed a stable EV71 reporter virus containing an enhanced green fluorescent protein (eGFP)-encoding gene (eGFP-EV71) that could be used for high-throughput screening of inhibitors of EV71 (22). We tested NITD008 in Vero cells infected with eGFP-EV71. The compound exhibited dose-dependent inhibition of eGFP expression in infected cells (Fig. 1A); under a phase-contrast microscope, equal numbers of cells from wells treated with different concentrations of the compound were observed (data not shown). To confirm the results from the reporter virus, we examined the compound's efficacy with WT EV71 in a viral titer reduction assay. As shown in Fig. 1B, NITD008 inhibited the WT virus at an EC₅₀ (50% effective concentration) of 0.67 μM. At a concentration of 4 μM, NITD008 reduced viral titers by up to 1,000-fold.

FIG 1.

Antiviral and cytotoxic analyses of NITD008. (A) Antiviral activities of NITD008 were analyzed by the eGFP-EV71 infection assay. Different concentrations of NITD008 were added to Vero cells infected with eGFP-EV71 at an MOI of 0.1 PFU/cell. At 48 h p.i., eGFP expression was observed under a fluorescence microscope. Antiviral activity was indicated by the reduction of the number of eGFP-positive cells upon NITD008 treatment. (B) Antiviral activity of NITD008 analyzed by WT EV71 viral titer reduction assay. Different concentrations of NITD008 were added to Vero cells after 1 h of incubation with EV71 at an MOI of 0.1 PFU/cell. At 48 h p.i., culture medium was collected and viral titers were measured with plaque assays. Antiviral activity was indicated by the reduction of viral titers upon NITD008 treatment. (C) Antiviral activities of NITD008 against CV-A16. The antiviral activity assay was performed the same way as that with EV71. (D) Cytotoxicity was evaluated by incubation of Vero cells with the indicated concentration of NITD008. At 48 h posttreatment, the viability of treated cells was measured by an MTT assay and presented as a percentage of that of untreated cells (see details in Materials and Methods). (E) Inhibition of EV71 replication by a combination of rupintrivir and NITD008. The antiviral activities of different concentrations of NITD008, alone or in combination with rupintrivir, against EV71 were tested. Vero cells were infected with Gluc-EV71 at an MOI of 0.1 PFU/cell. The construction of Gluc-EV71 will be reported elsewhere. At 48 h p.i., infected cells were collected for the luciferase assay. Antiviral activity was indicated by a reduction of luciferase activity. Statistical significance (P < 0.01) was detected for the antiviral activity of four treatments.

Since infection with both EV71 and CV-A16 can cause HFMD, we tested NITD008 for activity against CV-A16. The compound showed an EC₅₀ of 0.64 μM (Fig. 1C). To exclude the possibility that the inhibition of viral replication was due to NITD008-mediated cytotoxicity, we performed a cell proliferation-based cytotoxicity assay. As shown in Fig. 1D, the compound showed a CC₅₀ (50% cytotoxic concentration) of 119.97 μM in Vero cells. Even at 16 μM NITD008, the cells retained around 90% viability. The results allowed us to conclude that NITD008 is a potent inhibitor of EV71 in cell culture, with an estimated therapeutic index (TI [CC₅₀/EC₅₀]) of 179.

It has been previously shown that rupintrivir (AG7088) efficiently blocked EV71 replication as a potent 3C proteinase inhibitor (16, 25, 26). Therefore, we evaluated the antiviral effects of different concentrations of NITD008 with or without rupintrivir by the Gluc-EV71 infection assay (Fig. 1E). The Gluc-EV71 infection assay was performed by infecting cells with complete EV71 containing the gene for Gaussia luciferase (Gluc), and Gluc activity was quantified to monitor antiviral activity. Our results indicated that the luciferase activities that resulted from dual treatments (NITD008 plus 0.25 μM rupintrivir and NITD008 plus 0.5 μM rupintrivir) were lower than when NITD008 was used alone (P values were 0.02 and 0.004, respectively), which was also the case for treatment with rupintrivir alone (P values were both 0.03). It is suggested that the combination of rupintrivir and NITD008 had a more inhibitory effect on EV71 replication than either NITD008 or rupintrivir alone.

Selection of resistant viruses.

NITD008-resistant viruses were selected by continuously culturing EV71 for 18 rounds in the presence of gradually increasing concentrations of NITD008 (Fig. 2A), i.e., 1 μM for eight passages, 1.5 μM for one passage, 2 μM for four passages, and 2.5 μM for five passages. Two independent selections (SelA and SelB) were performed to test the reproducibility of the resistance results. For each passage, viral titers in the mock-treated infection and the NITD008-treated infection were determined; the viruses were subjected to resistance analysis with a viral titer reduction assay. Figure 2B and C show two representative data from such resistance assays for passage 8 (P8) and P18, respectively. At P8, the selected viruses exhibited marginal resistance to 1 μM NITD008 compared with the WT virus. In contrast, at P18, both the SelA and SelB viruses exhibited significant resistance to NITD008. Specifically, at a 2.5 μM inhibitor concentration, SelA and SelB viral titers were reduced by approximately 50- and 10-fold, respectively, whereas WT viral titers were suppressed by almost 1,000-fold. These results demonstrate that resistant EV71 virus can be reproducibly obtained in cell culture and that the SelB virus is more resistant than the SelA virus to NITD008 inhibition.

As shown in Fig. 2B and C, in the absence of NITD008, the WT and selected resistant viruses grew to comparable titers. In agreement with this result, similar plaque morphologies of the WT, SelA, and SelB viruses were observed (Fig. 2D). The results indicate that the mutations that occurred in the resistant viruses did not reduce the magnitudes of replication in virus-infected cell cultures.

Identification of resistance mutations.

To identify mutations responsible for drug resistance, we sequenced the complete genomes of P18 SelA and SelB viruses, as well as the P18 WT virus. No mutation was found in the P18 WT virus. In contrast, several sense and nonsense mutations were found in the P18 SelA and SelB viruses (Fig. 3A). The nonsense mutations occurred at VP2, VP3, 3B, and 3C. The sense mutations in the SelA and SelB viruses were located in three regions. (i) An Ala→Val substitution at amino acid position 38 of the 2B protein (A38V_2B) was found in SelB; (ii) two mutations in the 3A protein, M62T_3A and V75A_3A, were recovered from SelA and SelB, respectively; and (iii) a common mutation, V63A_3D, in the 3D polymerase was recovered from both selection viruses, while an additional M393L_3D mutation in the 3D protein was recovered from SelB. A close examination of the sequence chromatograms showed that the two mutations from SelA—M62T_3A and V63A_3D—contained pure adaptive sequences (Fig. 3B), whereas the two mutations from SelB—V63A_3D and M393L_3D—had adaptive sequences in a mixture with the WT sequence (Fig. 3C). In summary, all of the amino acid changes were mapped to the 3A and 3D regions of both resistant viruses, except that an additional A38V_2B mutation was found in SelB.

FIG 3.

Mutations identified in NITD008-resistant EV71. (A) Mutations recovered from resistant viruses were tabulated according to the amino acid positions and the corresponding proteins. Complete-genome sequencing of P18 viruses from SelA and SelB was performed. Sense and nonsense mutations are in red and black, respectively. (B, C) Sequence chromatograms of mutations from resistant viruses. Representative sequencing chromatograms are presented for SelA (B) and SelB (C) viruses for P18 selection. Original mutations are indicated at the top of each panel. Amino acid changes and positions in the corresponding proteins are indicated.

Mutations in both 3A and 3D confer resistance to NITD008.

Using an infectious cDNA clone of EV71, we examined the roles of the mutations recovered from the selection isolates in drug resistance. Individual or combined mutations were engineered into the infectious cDNA clone to generate recombinant viruses. The resistance phenotype of each recombinant virus was analyzed by comparing the viral titers after the virus was cultured in the presence or absence of NITD008. The WT and selected (SelA and SelB) P18 viruses were included as controls. Figure 4A shows the resistance analysis of mutations recovered from a SelA isolate. Mutation M62T_3A or V63A_3D alone was sufficient to confer resistance to a level close to that of the SelA isolate. The M62T_3A V63A_3D double mutation slightly improved the resistance to a level equivalent to that of the original SelA isolate. The results demonstrate that both M62T_3A and V63A_3D play a role in resistance.

FIG 4.

Drug resistance analysis of mutations identified in recombinant EV71. The resistance associated with mutations recovered from SelA (A) and SelB (B) isolates was analyzed. Vero cells were infected with different recombinant virus strains at an MOI of 0.1 PFU/cell in the presence or absence of 2.5 μM NITD008. At 48 h p.i., the supernatants were collected and viral titers were measured by plaque assay. The drug resistance assay of each recombinant virus was performed in triplicate, and representative data are shown. Error bars represent standard deviations.

Figure 4B summarizes the resistance analysis of mutations recovered from a SelB isolate. A mutant virus containing A38V_2B alone did not show any resistance, whereas viruses containing the single mutation V75A_3A, V63A_3D, or M393L_3D showed partial resistance. The V75A_3A V63A_3D, V75A_3A M393L_3D, and V63A_3D M393L_3D double mutations slightly increased resistance in comparison with that of the single mutant viruses. The V75A_3A V63A_3D M393L_3D triple mutation further increased resistance to a level similar to that of the original SelB isolate. Addition of A38V_2B to the double mutation V75A_3A V63A_3D or to the triple mutation V75A_3A V63A_3D M393L_3D did not change the level of resistance. These results demonstrate that (i) the mutation in 2B (A38V_2B) does not contribute to resistance, (ii) the mutations in 3A (V75A_3A) and 3D (V63A_3D and M393L_3D) confer resistance, and (iii) combining two and three mutations in 3A and 3D additively increases resistance. Furthermore, Fig. 4A and B clearly indicate that three mutations in SelB (V75A_3A V63A_3D M393L_3D) account for the higher resistance than the two mutations in SelA (M62T_3A V63A_3D).

NITD008 treatment completely protected EV71-infected mice from lethal disease.

The in vivo efficacy of NITD008 was assessed in an AG129 mouse model of EV71 infection (24). Two-week-old AG129 mice were infected i.p. with 10⁷ PFU of EV71 strain 41 (S41). On the day of infection and for 3 consecutive days, the animals were administered NITD008 at 5 mg/kg orally twice daily. Control groups consisted of untreated infected animals and infected animals given the vehicle alone. The animals were observed daily for 20 days p.i., and clinical manifestations were scored. When two-limb paralysis was observed, the animals were euthanized. Untreated infected mice and infected animals treated with the vehicle displayed a 100% mortality rate, with some deaths observed as early as 6 days p.i. (Fig. 5A). These mice displayed a characteristic progression of severity of clinical symptoms, including ruffled hair, a hunched back, limb weakness, and one-limb paralysis, followed by two-limb paralysis, at which point the animals were euthanized (Fig. 5B). Although there was a delay in the manifestation of clinical symptoms in the vehicle-treated infected mice compared to that of the untreated infected group, treatment with the vehicle alone did not result in significant protection, as evidenced by the 100% mortality rate observed by day 11 p.i. (Fig. 5A). In contrast, mice treated with NITD008 remained healthy throughout the course of the experiment and did not manifest any clinical signs of illness associated with severe EV71 infection (Fig. 5A and B). These data thus demonstrate the protective efficacy of NITD008 in a mouse model of severe EV71 infection.

FIG 5.

Survival rates (A) and clinical scores (B) of 2-week-old AG129 mice infected with EV71 and treated with NITD008. Groups of 8 to 10 mice were infected via the i.p. route with 10⁷ PFU of EV71 strain 41. Drug treatment started on the day of infection and consisted of oral administration b.i.d. of NITD008 (5 mg/kg of body weight) or the vehicle alone for 3 consecutive days. The mice were monitored for 20 days. Clinical scores were defined as follows: 0, healthy; 1, ruffled hair and hunchbacked appearance; 2, limb weakness; 3, paralysis in one limb; 4, paralysis in two limbs; 5, death. Data are expressed as means ± SEM. A P value <0.0001 was obtained by the log rank test, and survival curves were significantly different.

Viral titers in organs from drug-treated infected mice were significantly reduced compared to those of the control groups.

To further assess the in vivo protective efficacy of NITD008, at 4, 5, and 6 days p.i., the viral titers in the front limbs, hind limbs, spines, and brains harvested from mice infected with EV71 and treated with NITD008 were quantified. These viral titers were compared to those measured in the control groups (untreated infected mice and infected mice treated with the vehicle). Infectious viruses were detected in the untreated infected mice and infected mice treated with the vehicle in all of the organs assessed (Fig. 6A to D) according to a typical kinetic whereby the viruses accumulated first in the limb muscles before reaching the spine and eventually accumulated in the brain, as reported previously (24). Interestingly, the viral titers measured in the infected mice treated with the vehicle were significantly lower than those measured in the untreated infected mice, suggesting that oral administration of the vehicle alone affects the infectivity of the virus. These lower viral titers correlate with the delayed appearance of the clinical manifestations observed in this mouse group (Fig. 5A and B). However, since eventually all of the mice in the vehicle-treated group displayed the full range of clinical manifestations with a 100% mortality rate, we conclude that the vehicle-mediated reduction in viral titers is not sufficient to protect the animals from EV71 infection effectively.

FIG 6.

Virus titers in organs from mice infected and treated with the candidate drug. Mice (four or five per group per time point) were infected i.p. with 10⁷ PFU/mouse and treated b.i.d. with either 5 mg/kg of NITD008 or the vehicle as described in the legend to Fig. 5. The animals were euthanized at 4, 5, or 6 days p.i., and the virus titers in their front limbs (FL) (A), hind limbs (HL) (B), spines (C), and brains (D) were quantified by plaque assay. In the untreated group and the infected group treated with the vehicle, organs could be harvested only at 4 days and at 4 and 5 days p.i., respectively, because of the fast progression of the disease and early euthanasia. Means are shown as a solid lines (drug-treated infected mice), dotted lines (vehicle-treated infected mice), or dashed lines (untreated infected mice).

In contrast, significantly lower virus titers in all of the organs assessed, including the central nervous system, were measured in infected mice treated with NITD008 (Fig. 6A to D). Compared to those of vehicle-treated mice, reductions of average virus titers in the organs of NITD008-treated mice ranged from 2-fold to a complete absence of viral detection (order of magnitude of 5) by plaque assay. The results thus strongly support the in vivo protective efficacy of NITD008 against EV71.

DISCUSSION

The increase in EV71-associated morbidity and mortality in humans has underlined the urgent need to develop an effective therapy for EV71 infection. In this study, we demonstrated the in vitro and in vivo efficacy of NITD008, an adenosine analog, as an inhibitor of EV71. NITD008 showed an EC₅₀ of 0.67 μM, a CC₅₀ of 119.97 μM, and a TI of 179 in cell culture. Besides EV71, NITD008 is equally potent against CV-A16, a closely related enterovirus that also causes HFMD in humans. Importantly, treatment of EV71-infected AG129 mice completely prevented clinical manifestations and death. Using the same mouse model, we recently showed that treatment with ribavirin only delayed the outcome of the disease but did not prevent death (24). Therefore, NITD008 is a more potent inhibitor of EV71 than ribavirin is in vivo. A number of antiviral mechanisms have been reported for ribavirin, including depletion of the intracellular pool of nucleoside triphosphates (27), functioning as a mutagen to increase error catastrophe (28), and potentiating the action of interferon by augmenting the expression of interferon-stimulated genes (29). The exact mechanism of ribavirin inhibition of EV71 remains to be determined.

As an adenosine analog, NITD008 was shown to function as a polymerase active site inhibitor in dengue virus (21). Its triphosphate form competitively inhibits polymerase activity, likely through binding at the nucleoside triphosphate site or incorporation into the RNA product chain. The same mechanism could be inferred to inhibit EV71 RNA synthesis. In line with this inference, resistant mutations were mapped to the 3D polymerase of EV71 in two independently selected resistant isolates. Both isolates contained the same mutation in the 3D polymerase (V63A_3D); in addition, SelB had one extra 3D mutation (M393L_3D). Sequence alignment shows that V63_3D and M393_3D are conserved among various members of the genus Enterovirus (Fig. 7A). Using a reverse genetic approach, we confirmed that these 3D mutations were able to confer resistance to NITD008.

FIG 7.

Conserved analysis of mutations identified within 3A and 3D among members of the genus Enterovirus and structural analysis of mutations within 3D. (A) Representative 3A and 3D sequences from members of the genus Enterovirus were aligned with the ClustalW software (http://embnet.vital-it.ch/software/ClustalW.html). Gray shading indicates the conservative residues of 3A and 3D. (B) Structural positioning and analysis of residues V63 and M393 of the EV71 polymerase. The crystal structure of a postcatalysis and pretranslocation elongation complex of the PV polymerase is used as an illustration (Protein Data Bank entry 3OL9) (34), wherein the 3′-terminal nucleotide (stick representation) of the product strand represents the putative binding/incorporation position of NITD008. Residue V63 (large green spheres) is near the N-terminal hydrogen bonding network that has been suggested to affect polymerase activity through motif A residue D238 (left and top right panels) (30). Residue M393 (large blue spheres) forms hydrophobic interactions with motif E residues K376 and R377, which in turn interact with the −1 to −3 backbone region of the product RNA (left and bottom right panels). The EV71 polymerase equivalent positions are labeled. The polymerase palm domain is gray, the thumb is blue, and the individual fingers are as follows: index, green; middle, orange; ring, yellow; pinky, pink. Wherever visible, the N terminus of the polymerase is shown as a small blue sphere. The picornavirus signature sequence YGDD in motif A is magenta. The RNA template strand is cyan, and the product strand is green. The +1 template nucleotide is orange. Key hydrogen bonds involving the N terminus and D238 are shown as broken green lines.

Structurally, both V63_3D and M393_3D are situated near sites known to regulate RNA synthesis. Immediately next to the C terminus of residue V63_3D, G64_3D is known to play a critical role in the hydrogen bonding network involving the N terminus and polymerase motif A (Fig. 7B) (30), and mutations of this residue have been documented to alter the polymerase fidelity of picornaviruses and increase resistance to ribavirin (31, 32). Similarly, the side chain of residue M393_3D is spatially adjacent to and has hydrophobic interactions with conserved motif E residues K376_3D and R377_3D, which interact with the −1 to −3 regions of the product RNA strand (Fig. 7B) (33, 34). Therefore, residues V63_3D and M393_3D, both of which have the potential to regulate RNA synthesis by the polymerase, could be used by the virus to overcome the selective pressure of NITD008. Interestingly, the previously identified poliovirus (PV) 3D temperature-sensitive mutation (M394T) is two residues away from the EV71 3D M393 site (35). Structurally, M395 (PV 3D M394 equivalent) participated in a different set of hydrophobic interactions within the thumb domain. Although M393L (EV71) and M394T (PV) likely regulated RNA synthesis through different mechanisms, the identification of these two mutations at the thumb-palm junction indicated the sensitivity of this region in the regulation of polymerase activities.

Besides the 3D polymerase mutations, distinct single mutations in the 3A protein were identified, M62T_3A from a SelA isolate and V75A_3A from a SelB isolate. Sequence alignment shows that V75_3A is conserved among different members of the genus Enterovirus; whereas M62_3A is conserved between EV7 and CV-A16 but different in PV (Fig. 7A). Functional analysis showed that such a single mutation alone (M62T_3A or V75A_3A) was able to confer partial resistance (Fig. 4). Our findings agree with a previous report that resistance mutations against amiloride (an inhibitor of the 3D polymerase of coxsackievirus B3) were mapped to both the 3A and 3D proteins, although the contribution of the 3A mutation (I54L) to the resistance phenotype was not experimentally determined in that study (36). Amiloride was shown to directly inhibit the polymerase activity of coxsackievirus B3 by suppressing VPg uridylylation and RNA elongation (37).

How could a 3A mutation(s) contribute to resistance to an inhibitor of 3D polymerase? 3A is a membrane protein that induces endoplasmic reticulum membrane alterations to form a viral replication complex. Both 3A and its precursor 3AB are indispensable for viral replication (38). Several lines of evidence indicate that 3AB modulates 3D polymerase activity through direct 3AB-3D interaction. For example, 3AB is required to deliver VPg (3B) to the 5′ end of the viral RNA during the initiation of viral replication (39). 3AB stimulates 3D polymerase activity, possibly by increasing the utilization of RNA primer 3′-hydroxyl termini for chain elongation (40–42). 3AB may also recruit 3D to the replication complex (43). In the case of EV71 resistance to NITD008, it is possible that a mutation(s) in 3A could compensate for the compound-mediated loss of 3D polymerase activity through direct 3AB-3D interaction, resulting in resistance. However, the precise mechanism by which the 3A mutations confer such resistance remains to be elucidated.

Our results suggest that NITD008 could be further characterized for preclinical development. The compound showed full protection of EV71-infected mice when administered at 5 mg/kg (Fig. 5). In the dengue virus mouse model, 25 to 50 mg/kg of NITD008 was required to reduce viremia by 10-fold (21). Reducing viremia by 10-fold has been considered the benchmark of antiviral efficacy for dengue virus (44). If so, the potential safety window of NITD008 for the treatment of EV71 could be 5- to 10-fold greater than that for dengue virus. Such a rationale argues that careful preclinical profiling should be performed to define the feasibility of NITD008 for clinical development for EV71 therapy.