Biochemical Evaluation of the Inhibition Properties of Favipiravir and 2′-C-Methyl-Cytidine Triphosphates against Human and Mouse Norovirus RNA Polymerases

Abstract

Norovirus (NoV) is a positive-sense single-stranded RNA virus that causes acute gastroenteritis and is responsible for 200,000 deaths per year worldwide. No effective vaccine or treatment is available. Recent studies have shown that the nucleoside analogs favipiravir (T-705) and 2′-C-methyl-cytidine (2CM-C) inhibit NoV replication in vitro and in animal models, but their precise mechanism of action is unknown. We evaluated the molecular interactions between nucleoside triphosphates and NoV RNA-dependent RNA polymerase (NoVpol), the enzyme responsible for replication and transcription of NoV genomic RNA. We found that T-705 ribonucleoside triphosphate (RTP) and 2CM-C triphosphate (2CM-CTP) equally inhibited human and mouse NoVpol activities at concentrations resulting in 50% of maximum inhibition (IC₅₀s) in the low micromolar range. 2CM-CTP inhibited the viral polymerases by competing directly with natural CTP during primer elongation, whereas T-705 RTP competed mostly with ATP and GTP at the initiation and elongation steps. Incorporation of 2CM-CTP into viral RNA blocked subsequent RNA synthesis, whereas T-705 RTP did not cause immediate chain termination of NoVpol. 2CM-CTP and T-705 RTP displayed low levels of enzyme selectivity, as they were both recognized as substrates by human mitochondrial RNA polymerase. The level of discrimination by the human enzyme was increased with a novel analog of T-705 RTP containing a 2′-C-methyl substitution. Collectively, our data suggest that 2CM-C inhibits replication of NoV by acting as a classic chain terminator, while T-705 may inhibit the virus by multiple mechanisms of action. Understanding the precise mechanism of action of anti-NoV compounds could provide a rational basis for optimizing their inhibition potencies and selectivities.

INTRODUCTION

Noroviruses (NoV) cause nearly half of all cases of acute gastroenteritis worldwide (1). Following an incubation period of about 1 to 2 days, the clinical symptoms of NoV infection, diarrhea, abdominal cramps, nausea, and vomiting, usually last about 2 to 4 days. These viruses are highly transmissible and cause severe health and economic burdens in developing countries. In the United States alone, NoV infections are responsible for 56,000 to 71,000 hospitalizations annually (2). Nearly two-thirds of all NoV cases occur in long-term care facilities. NoV are a genetically diverse group that belongs to the family Caliciviridae. They are nonenveloped viruses with a positive-sense single-stranded RNA genome of 7,400 to 7,700 nucleotides (nt). NoV typically express three conserved open reading frames (ORFs) named ORF1 to ORF3 (3, 4). ORF1 encodes a polyprotein that is processed by proteolysis to generate six nonstructural (NS) proteins that replicate and transcribe the viral genome. Viruses commonly isolated in cases of acute gastroenteritis belong to two genogroups: genogroups I (GI) and II (GII). Within these two groups, cases of GII genotype 4 (GII.4) infection account for the majority of outbreaks of gastroenteritis. Until recently, medical research on human NoV (HNV) was limited by the lack of infectious tissue culture systems (5–7), so murine norovirus (MNV) was commonly used as a surrogate (8–10). In 2014, a major breakthrough established that human NoV and MNV interact with enteric bacteria to infect B cells, the natural target of infection (11). Despite these advances and the clear unmet medical need imposed by NoV infection, no vaccine, prophylaxis, or therapeutic options are yet available. In Western countries, it is likely that closed communities, such as hospitals, senior care centers, and military facilities, would benefit the most from anti-NoV treatments to prevent virus transmission during epidemic outbreaks.

The RNA-dependent RNA polymerase (RdRp) function of the NoV NS7 protein is an obvious molecular target for small-molecule inhibitors. In recent years, attempts to inhibit the RdRp function of NoV polymerase have been made (12). The nucleoside analog 2′-C-methyl-cytidine (2CM-C) was originally developed against the polymerase of hepatitis C virus (HCV) and advanced as valopicitabine/NM283, an orally bioavailable 3′-valine ester prodrug (13). 2CM-C was later found to inhibit the replication of MNV (14) and human NoV replication through a Norwalk virus replicon (15). Treatment of MNV-infected AG129 mice with 2CM-C prevented virus-induced onset of diarrhea and mortality by reducing the viral load in organs of replication (15). In an NoV transmission model, prophylactic treatment of sentinel mice with 2CM-C protected against transmission of MNV infection (16). The nucleoside precursor 6-fluoro-3-hydroxy-2-pyrazinecarboxamide (T-705; favipiravir) also has activity against MNV (17). Although T-705 was originally developed against influenza virus, the molecule has been shown to inhibit a number of unrelated RNA viruses, including orthomyxoviruses, noroviruses, bunyaviruses, arenaviruses, flaviviruses, and filoviruses (18–32). In vitro, T-705 is efficiently converted to ribofuranosyl 5′-triphosphate (T-705 RTP) by cellular enzymes (33). Much like ribavirin, the precise mechanism of action of T-705 is subject to debate. In influenza virus, it has been proposed that T-705 exerts antiviral activity in its nucleoside triphosphate form (T-705 RTP) by directly interacting with the viral RdRp (34, 35). Treatment of influenza virus-infected cells with T-705 resulted in a significant increase of lethal mutations within the viral genome, a phenomenon also called error catastrophe (36). Recently, similar experiments conducted by Arias et al. in a mouse model of NoV infection showed T-705 also induces viral mutagenesis in vivo and exerts its antiviral effect through a mechanism distinct from that of ribavirin (37). They showed that T-705 is more efficacious than ribavirin against MNV, both in vitro and in vivo. Although these results suggest that T-705 inhibits NoV by interfering with the RdRp function of its polymerase in its triphosphate form, no direct enzyme studies have been reported. Furthermore, there is currently no evidence that T-705 is also active against human NoV.

Here, we aimed to characterize the molecular interaction between human NoV and MNV polymerases (HNVpol and MNVpol, respectively) and the nucleoside triphosphate forms of 2CM-C and T-705. We show for the first time that T-705 inhibits human NoV replication in the subgenomic replicon assay. In cell-free enzymatic assays, 2CM-C triphosphate (2CM-CTP) and T-705 RTP equally inhibited HNVpol, whereas ribavirin triphosphate was significantly less potent. 2CM-CTP acted as a classic chain terminator, competing with natural CTP and causing immediate chain termination. In contrast, the competitive behavior of T-705 RTP was more pronounced with ATP and GTP. Consistently, T-705 RTP was recognized by MNVpol and weakly incorporated into RNA primarily as an adenosine and guanosine analog. The selectivity of 2CM-CTP and T-705 RTP was assessed against human mitochondrial RNA polymerase (HMRP). Finally, a novel analog of T-705 RTP was synthesized, and its activity was evaluated in parallel against NoV and human RNA polymerases.

MATERIALS AND METHODS

Chemicals.

All ultrapure-grade nucleoside triphosphates were purchased from Affymetrix (Santa Clara, CA). Radiolabeled nucleoside triphosphates were purchased from Perkin-Elmer. Ribavirin triphosphate was purchased from Jena Biosciences (Jena, Germany). Suramin sodium salt was purchased from Sigma-Aldrich. T-705, T-705 RTP, 2CM-C, 2CM-CTP, T-1106 triphosphate (T-1106-TP), and 2′-C-methyl-T-1106-TP (2CM-T-1106-TP) were synthesized at Alios BioPharma, Inc. (South San Francisco, CA).

Human NoV replicon assay.

HG23 cells (human NoV replicon-harboring Huh7 cells) (38) were maintained in Dulbecco's modified Eagle's medium (Sigma-Aldrich, St. Louis, MO) with 5% fetal bovine serum and 0.5 mg/ml of G418, with passaging every 2 or 3 days. Viral RNAs were constantly maintained for up to 50 passages. For antiviral treatment in HG23 cells, 1-day-old HG23 cells in 12- or 24-well plates at ∼25% confluence were treated with various concentrations (mock medium or 0.001 to 10 μM compounds) for 72 h in the presence of 0.5 mg/ml of G418. At the end of compound treatment, the expression of the neomycin resistance gene, the neomycin phosphotransferase (NPT) gene, engineered on the replicon was determined by NPT II enzyme-linked immunosorbent assay (ELISA) (Agdia, Elkhart, IN). Briefly, cells were washed once in phosphate-buffered saline (PBS), solubilized in protein extraction buffer with phenylmethanesulfonyl fluoride, and shaken at room temperature for 30 min. The samples were transferred to antibody-coated 96-well plates and incubated at room temperature for 2 h. The wells were washed three times with PBS with Tween 20. Peroxidase enzyme conjugate was added to the wells and incubated for 2 h at room temperature. The wells were then washed three times with PBS with Tween 20. Tetramethylbenzidine peroxidase substrate solution was added to the wells to visualize the activity. The plate was read at 450 nm on a PerkinElmer (Waltham, MA) Victor3 V multilabel counter. The concentrations resulting in 50% of the maximum effective response (EC₅₀s) were determined using GraphPad Prism 5.

Cell viability assay.

A cell proliferation assay (Promega CellTiter-Glo Luminescent Cell Viability Assay; catalog number G7572) was used to measure cell viability. Assay plates were set up in the same format as the replicon assay. An aliquot of 100 μl of CellTiter-Glo reagent was added to each well and incubated at room temperature for 8 min. Luminescence was recorded using PerkinElmer's multilabel counter Victor3 V. The drug concentration required to reduce viable cells by 50% relative to the untreated cell control value (CC₅₀) was calculated based on percent reductions of luminescence using GraphPad Prism 5.

Expression and purification of recombinant MNV RdRp and HNV RdRp.

The recombinant MNV RdRp and the human GII.4 RdRp were prepared as described previously (39). In summary, Escherichia coli Rosetta cells harboring the pET15bMNVRdRp plasmid were grown at 37°C until the optical density at 600 nm (OD₆₀₀) reached between 0.6 and 0.8. Isopropyl thiogalactoside was added to induce protein expression, and the cells were cultured for 16 to 18 h at 16°C. The cells were harvested by centrifugation; resuspended in lysis buffer containing 100 mM Tris-HCl, pH 7.9, 300 mM NaCl, 10% glycerol, 15 mM imidazole, 5 mM mercaptoethanol, 0.1% Triton X-100, and a protease inhibitor cocktail (Sigma); and then sonicated on ice. The lysate was clarified by centrifugation, and the supernatant was mixed with nitrilotriacetic acid resin. The resin was washed extensively with lysis buffer containing 30 mM imidazole and eluted with lysis buffer containing 200 mM imidazole. The one-step-eluted protein was stored in storage buffer containing 50 mM Tris-HCl, pH 7.9, 300 mM NaCl, 20% glycerol, and 5 mM dithiothreitol (DTT).

For expression and purification of the recombinant HNV RdRp, TOP10 E. coli cells were transformed with pBAD GII.4 RdRp plasmid and then cultured at 37°C until the OD₆₀₀ was ∼0.6. After l-arabinose was added to a final concentration of 0.02% to induce protein expression, the cells grew for 5 h at 37°C. The cells were harvested by centrifugation and resuspended in lysis buffer for cell lysis. Purification was performed using Talon metal affinity resin (Clontech Laboratories). The recombinant protein bound to the resin was washed with lysis buffer containing 20 mM imidazole and eluted with the same buffer containing 100 mM imidazole. The eluted protein was stored in buffer containing 50 mM HEPES, pH 7.9, 150 mM NaCl, 20% glycerol, and 5 mM DTT. The recombinant proteins were stored at −80°C for future use.

Assay development and measurement of NTP Michaelis constants for recombinant HNV RdRp.

The time dependence of HNVpol RNA synthesis using an HCV 5′ untranslated region (UTR) RNA template was measured by the incorporation of radioactively labeled nucleotides into acid-insoluble RNA products. Each 50-μl reaction mixture contained 40 mM Tris-HCl, pH 7.0, 3 mM DTT, 50 nM 5′ UTR RNA, 0.1 μM [³H]UTP (36.3 Ci/mmol), 2 μM GTP, 2 μM ATP, 2 μM CTP, 0.2 mM MgCl₂, 10% dimethyl sulfoxide (DMSO), 0.5 U/μl RNaseIn (Promega, Madison, WI), and 100 nM HNV RdRp. The reaction mixture was incubated at 30°C, and RdRp measurements were taken every 30 min by quenching the 50-μl reaction mixture with 60 μl of 20% (wt/vol) trichloroacetic acid (TCA) with 0.5 mM ATP. The quenched reaction mixtures were incubated at 4°C for at least 1 h. The reaction mixtures were loaded onto a 96-well filter plate (EMD Millipore, Billerica, MA). The filters on the plate were washed three times with 10% TCA and once with 70% ethanol on the Millipore plate wash station with a vacuum applied. The filters on the plate were air dried, and 40 μl Microscint-20 (PerkinElmer, Waltham, MA) was added to each well. The acid-precipitated RNA products on the plate were detected with MicroBeta Trilux (PerkinElmer). The Michaelis constant (K_m) for each nucleoside triphosphate (NTP) with HNVpol was measured under the same reaction conditions described above, except that other NTP concentrations were held at 2 μM and the concentration of the testing NTP was increased from 0 to 2 μM. [³H]CTP was used in the K_m measurement for UTP. Reactions were quenched after incubation at 30°C for 2.5 h.

Inhibition and competition assays using recombinant HNV RdRp.

Inhibition efficacy was measured by the incorporation of radioactively labeled nucleotides by HNV RdRp into acid-insoluble RNA products. RdRp inhibition assays were performed in 96-well filter plates, and each reaction mixture contained 40 mM Tris-HCl, pH 7, 3 mM DTT, 50 nM 5′ UTR RNA, 0.1 μM [³H]UTP (36.3 Ci/mmol), 0.1 μM GTP, 0.1 μM ATP, 0.1 μM CTP, 0.2 mM MgCl₂, 10% DMSO, 0.5 U/μl RNaseIn (Promega), and 100 nM HNVpol. The reaction mixtures were incubated for 2.5 h at 30°C and quenched with a cold mixture of 20% (wt/vol) TCA and 0.5 mM ATP. The competition assays were performed under similar conditions, except that each competing nucleotide was tested at a low (0.1 μM) or high (100 μM) concentration while other NTPs were kept at 0.1 μM. For UTP competition assays, either [³H]GTP or [³H]CTP was used as the isotope tracer.

Inhibition of de novo initiation and primer extension activities of NoV RdRp by T-705 RTP and suramin.

Two RNA molecules that served as templates for RNA synthesis were designed based on an MNV strain CW1 isolate (GenBank accession number DQ285629), nt 5012 to 5059, and strain GII.4 isolate Hu/GII.4/MD-2004/2004/US (GenBank accession number DQ658413), nt 5044 to 5085 (39). These two RNAs contained viral subgenomic core promoters followed by short template sequences. Both RNAs contained three nonviral nucleotides (GCG) at their 5′ termini to allow the incorporation of radiolabeled [α-³²P]CTP during RNA synthesis in vitro. For brevity, they are referred to as MNV proscript (Mps) and HNV GII.4 proscript (Gps). The two RNA oligonucleotides were synthesized by Trilink (San Diego, CA). RdRp RNA synthesis was carried out in 20-μl reaction mixtures containing 20 mM sodium glutamate (pH 8.2), 12.5 mM DTT, 4 mM MgCl₂, 1 mM MnCl₂, 0.5% (vol/vol) Triton X-100, 0.05 mM GTP, 0.05 mM ATP, 0.01 mM UTP, 33.3 nM [α-³²P]CTP, 50 nM proscript RNA, and 250 nM recombinant RdRp, unless otherwise indicated. The reaction mixtures were incubated at 30°C for 2 h, and then the reactions were stopped by the addition of EDTA (pH 8.0) at a final concentration of 10 mM. Samples were analyzed by electrophoresis with a 24% polyacrylamide urea gel, followed by phosphorimaging (Typhoon 9210; Amersham Biosciences, Piscataway, NJ), and with ImageQuant software (Amersham Biosciences).

Single- or multiple-nucleotide incorporation assay using MNV RdRp.

The RNA primers P5 (5′-GUGAA), P6 (5′-GUGAAU), P7 (5′-GUGAAUG), and P8 (5′-GUGAAUGA) were synthesized by Dharmacon, Inc. (Chicago, IL). The 5′-end ³³P-radiolabeling reaction of the primers was conducted with [γ-³³P]ATP and T4 polynucleotide kinase in forward reaction buffer (Invitrogen, Carlsbad, CA), as suggested in the manufacturer's manual. In a typical single-nucleotide or multiple-nucleotide incorporation assay, the 10-μl reaction mixture contained 40 mM Tris, pH 7.0, 3 mM DTT, 10 mM MgCl₂, 1.33 U/μl RNaseIn (Promega), 0.5 μM primer, 1.8 μM MNV proscript RNA, 2 μM enzyme, and 100 μM each NTP. The reactions were carried out at 30°C for 60 min and were stopped by adding 20 μl of formamide solution containing 50 mM EDTA. Samples were analyzed by electrophoresis with 22.5% polyacrylamide urea gels, followed by phosphorimaging using ImageQuant software.

Nucleotide incorporation by human mitochondrial RNA polymerase.

Human mitochondrial RNA polymerase was obtained as a recombinant enzyme from Indigo Biosciences (State College, PA; catalog number MV100-10). The oligonucleotides required for the human mitochondrial RNA polymerase assay were custom ordered from Dharmacon, Inc., and contained the following sequences: RNA primer, 5′-UUUUGCCGCGCC-3′; dCMP DNA template, 3′-CGGCGCGGCATGTAAGGG-5′; TMP DNA template, 3′-CGGCGCGGTACGTAAGGG-5′; dGMP DNA template, 3′-CGGCGCGGGTACTAAGGG-5′; dAMP DNA template, 3′-CGGCGCGGATCGTAAGGG-5′. The boldface letters in the sequences indicate the first templating deoxynucleoside monophosphates (dNMPs) for single-nucleotide incorporation. The DNA-dependent RNA polymerase (DdRp) assay with human mitochondrial RNA polymerase was performed under single-turnover conditions, where the enzyme concentration is in excess of the primer/template. Therefore, the ³³P-RNA/DNA primer/template was used at a concentration of 100 nM, together with 320 nM enzyme. The standard 10-μl reactions were carried out at 30°C for 60 s with 100 μM each NTP, 10 mM MgCl₂, 50 mM NaCl, 40 mM Tris, pH 7.5, and 1 mM DTT. The reactions were stopped by adding 20 μl of formamide solution containing 50 mM EDTA. The samples were analyzed by electrophoresis with 22.5% polyacrylamide urea gels, followed by phosphorimaging using ImageQuant software.

RESULTS

Inhibition of human NoV by nucleoside analogs.

The antiviral efficacy of the nucleoside analog 2CM-C was evaluated in cells containing a stable subgenomic replicon of human NoV (38). We measured the expression level of the NPT gene, a selectable marker carried by the NoV replicon, which is dependent on the activity of the NoV polymerase complex. In this ELISA, the presence of 2CM-C reduced the level of NPT expression in a dose-dependent manner, with an EC₅₀ of 8.2 μM (Fig. 1A). No significant changes in cell viability were observed at 2CM-C concentrations up to 100 μM. T-705 also caused a dose-dependent reduction in NPT expression, with an EC₅₀ of 21 μM, which was about 2-fold more potent than ribavirin (Fig. 1B and C). Overall, the rank order of antiviral potency among the three compounds was 2CM-C > T-705 > ribavirin.

FIG 1.

Efficacies of 2′-C-methyl-cytidine, T-705, and ribavirin in the human norovirus replicon assay. (A) 2′-C-Methyl-cytidine inhibits the replication of the human norovirus replicon, with an EC₅₀ of 8.2 ± 0.7 μM and a CC₅₀ of >100 μM. (B) T-705 inhibits the replication of the human norovirus replicon, with an EC₅₀ of 21 ± 2 μM and a CC₅₀ of >100 μM. (C) Ribavirin has an EC₅₀ of 43 ± 4 μM and a CC₅₀ of >100 μM. The error bars indicate the standard deviations of the results from at least two independent experiments.

Inhibition of human NoV polymerase activity by nucleotide analogs.

We developed a cell-free radiometric assay designed to measure the competitive inhibition of HNVpol by nucleotide analogs under steady-state kinetics using a long heterogeneous RNA template derived from the HCV genome (40). In this enzymatic assay, [³H]UTP was used as a tracer to monitor the time-dependent RdRp activity (Fig. 2A). Concentrations of natural nucleotides were adjusted based on the determination of their K_m values, which were 0.02 μM for GTP, 0.2 μM for ATP, 0.004 μM for CTP, and 0.04 μM for UTP (Fig. 2B). The inhibition assay was validated with all four canonical 3′ deoxynucleoside triphosphates (dNTPs) used as chain terminators, which gave concentrations resulting in 50% of maximum inhibition (IC₅₀s) of 8.0 μM for 3′ dCTP, 3.3 μM for 3′ dGTP, 139 μM for 3′ dATP, and 4.3 μM for 3′ dUTP (Table 1). The chain terminator 3′ dATP displayed the highest IC₅₀ value, a common but not well understood phenomenon also observed with other viral RNA polymerases, including those of dengue virus, rhinovirus (unpublished data), and HCV (41–44). Under the same assay conditions, where all four natural NTPs were kept at low concentrations, 2CM-CTP and T-705 RTP inhibited the RdRp activity of HNVpol, with IC₅₀s around 2.5 μM (Fig. 3 and Table 1). In comparison, ribavirin-TP was significantly less potent, with an IC₅₀ of 58 μM.

FIG 2.

Filter-based polymerase assay optimization and K_m measurement. (A) The time dependence of HNVpol RNA synthesis was fitted to a linear equation. Time point measurements were taken every 30 min by quenching the 50-μl reaction mixtures with 60 μl cold 20% TCA containing 0.5 mM ATP. (B) The K_m value for each NTP was measured by incubating the reaction mixtures with all other NTPs at 2 μM and increasing the testing NTP concentration from 0 to 2 μM. Each data set was fitted to the Michaelis-Menten equation. The K_m values for GTP, ATP, CTP, and UTP, were 0.02, 0.2, 0.004, and 0.04 μM, respectively. The error bars indicate the standard deviations of the results from at least two independent experiments.

TABLE 1.

Inhibition efficacies of NTP analogs against HNV RdRp

Analog	IC50 (μM)a
3′ dCTP	8.0 ± 0.6
3′ dGTP	3.3 ± 0.3
3′ dATP	138 ± 31
3′ dUTP	4.3 ± 2.1
2CM-CTP	2.4 ± 0.7
T-705 RTP	2.7 ± 0.9
Ribavirin	58 ± 4.7

^aMean ± standard deviation (SD) of the results of at least two independent experiments.

FIG 3.

Inhibition of HNVpol RNA synthesis by nucleotide analogs. IC₅₀s were measured by adding increasing concentrations of each inhibitor, and quantitative analysis of RNA product inhibition was expressed as percent inhibition (see Materials and Methods). The IC₅₀s for 2CM-CTP and T-705 RTP were 2.4 μM and 2.7 μM, respectively. Ribavirin exhibited weaker inhibition, with an IC₅₀ of 58 μM. The error bars indicate the standard deviations of the results from two independent experiments.

The nucleotide competitive profile of 2CM-CTP was assessed by measuring the effect of varying the concentrations of natural NTPs on enzyme inhibition. At high CTP concentration, the IC₅₀ of 2CM-CTP increased by 14-fold (Fig. 4A). In contrast, high ATP, GTP, or UTP concentrations did not affect the IC₅₀ of 2CM-CTP (Fig. 4B and C and Table 2). When the same experiments were conducted with T-705 RTP, high CTP and UTP concentrations modestly increased the IC₅₀, while ATP and GTP individually caused a more pronounced effect (Fig. 4D and F). The largest loss in potency (21-fold) was observed when both high ATP and GTP concentrations were used in the reaction (Fig. 4E and Table 2).

FIG 4.

IC₅₀ shift of NTP analogs against recombinant HNVpol measured by competition assay. (A) The inhibition percentage was measured in the presence of increasing 2CM-CTP concentrations and either a low (0.1 μM) or a high (100 μM) concentration of CTP. (B) Inhibition by 2CM-CTP using either a low (0.1 μM) or a high (100 μM) concentration of ATP. (C) Quantitative representation of the IC₅₀ fold shift of 2CM-CTP by all four nucleotides. The IC₅₀ fold shift was determined by dividing the IC₅₀ of 2CM-CTP at a high NTP concentration (100 μM) by the IC₅₀ of 2CM-CTP at a low concentration (0.1 μM). (D) The same experiment as in panel A, but with T-705 RTP. (E) Inhibition by T-705 RTP using either a low (0.1 μM) or a high (100 μM) concentration of ATP and GTP. (F) Quantitative representation of the IC₅₀ fold shift of T-705 RTP by all four nucleotides. The error bars indicate the standard deviations of the results from two independent experiments.

TABLE 2.

Inhibition efficacies of 2CM-CTP and T-705 RTP against HNV RdRp measured by competition assays

Assay conditionsa	2CM-CTPb		T-705 RTPb
	IC50 (μM)	Fold shift	IC50 (μM)	Fold shift
0.1 μM GACU	2.4 ± 0.7	1	2.7 ± 0.9	1
0.1 μM GAU + 100 μM C	33.6 ± 5.8	14	9.9 ± 1.1	3.7
0.1 μM GCU + 100 μM A	2.3 ± 0.2	1	24.4 ± 2.9	9
0.1 μM ACU + 100 μM G	2.8 ± 0.05	1.2	24.4 ± 2.9	9
0.1 μM CU + 100 μM GA	NA		55.5 ± 15.9	20.6
0.1 μM GAUC	NA		3.2 ± 0.9	1
0.1 μM GAC + 100 μM U	NA		16.3 ± 1.0	5.1
0.1 μM CAUG	1.8 ± 0.2	1	NA
0.1 μM CAG+ 100 μM U	2.7 ± 0.4	1.5	NA

^aThe [³H[NTP tracers are in boldface. For most assays, [³H]UTP was used as the tracer. [³H]GTP was used to measure the IC₅₀ of 2CM-CTP with a high concentration of UTP, whereas [³H]CTP was used to measure the IC₅₀ of T-705 RTP with a high concentration of UTP. In this way, direct competition between the testing compound and the tracer could be avoided.

^bThe fold shift was determined as the IC₅₀ of the compound tested at a high NTP concentration (100 μM) divided by the IC₅₀ of the compound at a low NTP concentration (0.1 μM). The values are the means ± SD of the results of two to five experiments. NA, not applicable.

Inhibition of norovirus RdRp activity at the initiation step.

RdRp inhibition experiments were conducted using side-by-side HNVpol and MNVpol. The RNA synthesis assays were performed with RNA that contained the specific subgenomic promoter and template that were recognized by each enzyme (Fig. 5A). The products of authentic de novo initiation of RNA synthesis were separated from primer extension RNA species by high-resolution urea-PAGE (39). Under these assay conditions, T-705 RTP significantly inhibited the formation of both the primer extension and de novo RNA products from MNVpol (Fig. 5B). The same effect was observed in the enzymatic reaction performed with HNVpol (Fig. 5C). Suramin has been reported to inhibit NoV polymerase (45), and its antiviral effect was confirmed in this assay (Fig. 5B and C). The same assay format was used to determine the inhibition efficacy of T-705 RTP against HNVpol, and the IC₅₀ was 9.1 ± 1.1 μM (Fig. 5D). In the presence of a dinucleotide GpU primer that mimics the formation of the first phosphodiester bond, the inhibition effect of T-705 RTP was significantly reduced (IC₅₀ = 24.7 ± 0.9 μM) (Fig. 5E), indicating that T-705 RTP binds to the initiation pocket in the RdRp to prevent the formation of the first phosphodiester bond. Together, these results suggest that T-705 RTP inhibits de novo RNA synthesis by the NoV RdRps from their homologous promoter template, likely by affecting both the initiation and elongation steps. Since suramin and its analogs interact with the RNA binding pocket of NoV polymerase, their mechanism of action is likely to be distinct from that of nucleotide analogs (see Fig. S1 in the supplemental material). In order to further support the differences in modes of action, we developed a single-nucleotide incorporation assay to study the interaction between the active site of NoV and the two NTP analogs, 2CM-CTP and T-705 RTP.

FIG 5.

T-705 RTP and suramin inhibited RNA synthesis by recombinant norovirus RdRps from their subgenomic promoters. (A) Schematics of the RNAs, Mps and Gps, used to assess de novo-initiated and primer extension (PE) products by recombinant norovirus RdRps. Mps and Gps contain mouse norovirus and human GII.4 subgenomic promoter and template sequences, respectively. (B) T-705 RTP and suramin inhibited RNA synthesis by the MNV RdRp. The gel image shows the products from de novo initiation and primer extension, using Mps as the RNA template. The 11-nt RNA was the correctly initiated and terminated de novo initiation product generated from Mps, and the PE RNA was the hypothetical product from primer extension of the self-annealed Mps RNA. The RNA synthesis reactions were performed with T-705 RTP dissolved in DMSO at a final concentration of 12 μM and with suramin dissolved in H₂O at a final concentration of 5 μM. The RNA products were quantified and normalized to that of the DMSO or H₂O control, as shown below the gel. PE, primer extension product; DN, de novo-initiated RNA product. (C) T-705 RTP or suramin inhibited RNA synthesis by the HNV RdRp with Gps. The de novo-initiated RNA and correctly terminated RNA generated from Gps is 8 nt. (D) The IC₅₀ of T-705 RTP on HNV RdRp de novo RNA synthesis was 9.1 ± 1.1 μM. (E) When 50 μM GpU dinucleotide was added to the reaction mixture, the IC₅₀ of T-705 RTP on HNV RdRp de novo RNA product synthesis was 24.7 ± 1.0 μM. The error bars indicate the standard deviations of the results from two independent experiments.

Effects of the incorporation of 2CM-CTP and T-705 RTP on primer elongation.

A single- or multiple-nucleotide incorporation assay was designed to measure the effects of 2CM-CTP and T-705 RTP once they were incorporated into the growing RNA primer strand. Although this assay is commonly used for other RNA polymerases (40, 46–48), its adaptation to NoV polymerase is complicated by the need to use a specific promoter sequence to favor productive enzyme-RNA binding (39). Single-nucleotide extension of a preassembled oligonucleotide primer using HNVpol was not sufficiently efficient for further study (data not shown). However, we found that MNVpol recognized an RNA that contains its subgenomic promoter and template sequences, Mps (Fig. 6A), but not other RNA templates in the presence of short RNA primers (see Fig. S2 in the supplemental material). Since only less than 1% productive enzyme-Mps-primer complex was observed under these assay conditions, pre-steady-state kinetic analysis of single-nucleotide incorporation was not suitable due to a low product-to-substrate ratio (see Fig. S3 in the supplemental material). The progress curves of single-nucleotide incorporation were studied, and the product formation was linear up to 60 min, which was chosen as the end time point for our nucleotide incorporation study (see Fig. S3 in the supplemental material). Using the 8-mer primer as a substrate, the incorporation of natural CTP opposite G resulted in nearly 80% conversion to the specific 9-mer product (Fig. 6B, lane 2, and C), and 20% of the 9-mer was further converted to an 11-mer product when the next correct nucleotide (GTP) was added (Fig. 6B, lane 3). In comparison, the addition of 2CM-CTP opposite G resulted in 26% conversion to the specific 9-mer product (Fig. 6B, lane 4, and C) and less than 1% further conversion to the 11-mer product with the next correct GTP (Fig. 6B, lane 5). This single-nucleotide incorporation assay was base specific, as demonstrated by the lack of recognition (less than 2% extension) of mismatched nucleotides (Fig. 6B, lanes 6 to 8). Similarly, T-705 RTP was not recognized as a substrate at the position opposite G on the template (Fig. 6B, lane 9). Using the 7-mer primer, the addition of natural ATP resulted in about 20% conversion to the specific 8-mer product (Fig. 6D, lane 11, and E), and most of the 8-mer was further converted to a 9-mer product when the next correct nucleotide (CTP) was added (Fig. 6D, lane 12). In comparison, adding T-705 RTP resulted in only 6% primer extension from 7-mer to 8-mer (Fig. 6D, lane 13, and E), and 37% of the 8-mer was further converted to extended products up to 11-mer in the presence of CTP (Fig. 6D, lane 14). Using a different primer, the incorporation of T-705 RTP opposite template C was low but significant, and it did not result in chain termination (see Fig. S4A and B in the supplemental material). Finally, T-705 RTP was not incorporated into the RNA by MNVpol as a uridine analog (see Fig. S4C in the supplemental material). Overall, these results showed that T-705 RTP was recognized by MNVpol as an ambiguous purine analog and that its incorporation into the RNA did not completely prevent primer extension in the presence of the next correct nucleotide. Unlike the cytidine analog 2CM-CTP, T-705 RTP did not cause immediate termination of RNA synthesis by MNVpol.

FIG 6.

Incorporation and chain termination assay using MNVpol. (A) Sequences of the RNA primers (7-mer or 8-mer) and the RNA template (Mps; the full sequence is shown in Fig. 5A) used in the study. (B) Single- or multiple-nucleotide/nucleotide analog incorporation using the 8-mer primer (see Materials and Methods). Lanes 2, 4, and 6 to 9 indicate the one nucleotide/nucleotide analog added in the assay. Lanes 3 and 5 indicate the two nucleotides/nucleotide analogs added in the assay. Water instead of a nucleotide was added in lane 1 as a negative control. (C) The band intensities on the gel in panel B were analyzed, and the percentages of primer extensions, 9-mer versus (8-mer plus 9-mer) for single-nucleotide incorporation and >9-mer versus >8-mer for multiple-nucleotide incorporations, are shown. (D) Single- or multiple-nucleotide/nucleotide analog incorporation using the 7-mer primer. Lanes 11, 13, and 15 to 18 indicate the one nucleotide/nucleotide analog added in the assay. Lanes 12 and 14 indicate the two nucleotides/nucleotide analogs added in the assay. Water instead of a nucleotide was added in lane 10 as a negative control. (E) The band intensities on the gel in panel D were analyzed, and the percentages of the products extended from the 7-mer by incorporation of ATP or T-705 RTP [8-mer versus (7-mer + 8-mer)] and the percentage of the products extended from the incorporated AMP or T-705 RTP (>8-mer versus >7-mer) are shown.

Incorporation of 2CM-CTP and T-705 RTP by human mitochondrial RNA polymerase.

The target selectivity of 2CM-CTP and T-705 RTP was assessed by measuring their levels of incorporation by HMRP, an enzyme known to recognize antiviral ribonucleotide analogs (49). Using a primer-template sequence designed for cytidine analog incorporation, natural CTP was efficiently recognized as a substrate by HMRP, which resulted in 50% conversion from 12- to 13-mer primer (Fig. 7A and D). Adding the next correct nucleotide (ATP) resulted in further primer extension in the case of CTP, but not with 2CM-CTP (see Fig. S5A in the supplemental material). Compared to the baseline of misincorporation opposite dG on the template, T-705 RTP and ribavirin-TP were not significantly recognized as cytidine analogs. However, T-705 RTP was efficiently recognized as a substrate using a primer-template sequence designed for adenosine analog incorporation, with a level of incorporation similar to that of natural ATP (Fig. 7B and D). Adding the next correct nucleotide (UTP) caused almost 100% further primer extension (from 13- to 14-mer) for both ATP and T-705 RTP (see Fig. S5A in the supplemental material). In comparison, the level of incorporation of ribavirin opposite template dT was at or below the level of misincorporation (Fig. 7B). The same pattern was observed when using a primer-template sequence designed for guanosine analogs, with a level of incorporation slightly lower than that of natural GTP (Fig. 7C and D). Here again, the incorporation of T-705 RTP did not cause immediate chain termination of RNA synthesis (see Fig. S5A in the supplemental material). Finally, T-705 RTP was not recognized as a uridine analog by HMRP (see Fig. S5B in the supplemental material). In summary, both 2CM-CTP and T-705 RTP were recognized as substrates for HMRP, but only 2CM-CTP caused immediate chain termination of RNA synthesis. Ribavirin-TP was not an efficient substrate for HMRP, irrespective of the nucleotide base used on the template.

FIG 7.

Incorporation of T-705 RTP as an ATP and a GTP analog by human mitochondrial RNA polymerase. (A) NTP analog incorporation opposite dGMP on the template. (B) NTP analog incorporation opposite TMP on the template. (C) NTP analog incorporation opposite dCMP on the template. (D) The intensities of the bands on the gels in panels A, B, and C and Fig. S3A in the supplemental material were analyzed, and the percentages of the products extended from the 12-mer by incorporation of NTP analogs [13-mer versus (12-mer + 13-mer)] and the products extended from the incorporated NMP analogs (>13-mer versus >12-mer) are shown.

Toward new nucleotide analogs with improved target selectivity.

Since 2CM-CTP and T-705 RTP displayed low levels of discrimination by HMRP, we hypothesized that a molecule combining the ribose modification of 2CM-C with the base modification of T-705 might be less efficiently recognized by the human enzyme. Instead of modifying T-705 RTP, we used T-1106-TP, a des-fluoro version that is easier to synthesize (Fig. 8). T-1106-TP potently inhibited HNVpol RdRp activity, with an IC₅₀ of 0.27 μM (Fig. 8A). T-1106-TP was also efficiently recognized by HMRP, both as an adenosine, with 62% incorporation opposite dT (see Fig. S6A in the supplemental material), and as a guanosine analog, with 49% incorporation opposite dC (Fig. 8B; see Fig. S6B in the supplemental material). In comparison, 2CM-T-1106-TP inhibited HNVpol RdRp with an increased IC₅₀ of 21.8 μM. However, 2CM-T-1106-TP was also less efficiently recognized by HMRP, with only 3.9% incorporation opposite dT and 1.4% incorporation opposite dC, a level close to that of background noise (see Fig. S6A and B in the supplemental material).

FIG 8.

Novel compounds with improved selectivity. (A) Structures of compounds and anti-HNVpol activities of T-1106-TP and 2CM-T-1106-TP. The triphosphate form of 2CM-T-1106, which has a 2′-methyl modification on the ribose with a modified base from T-705, inhibited HNVpol, with an IC₅₀ of 21.8 μM, while T-1106-TP inhibited HNVpol, with an IC₅₀ of 0.27 μM. (B) Incorporation of T-1106-TP and 2CM-T-1106-TP by human mitochondrial RNA polymerase. The intensities of the bands on the gels in Fig. S4 in the supplemental material were analyzed, and the quantitation of the percentages of the products extended from the 12-mer by incorporation of NTP analogs is shown. 2CM-T-1106-TP was incorporated by HMRP at background level.

DISCUSSION

Currently, no vaccine, prophylaxis, or treatment options are available to address the medical and economic burden of NoV infection (2). So far, only one small molecule, nitazoxanide, has been tested in the clinic as a potential anti-NoV drug. Nitazoxanide, which was originally developed and commercialized as an antiprotozoal agent, has been used successfully to treat subjects suffering from NoV gastroenteritis by reducing the duration of symptoms compared to a placebo control (50, 51). In these two phase 2 studies, adults in the treated cohorts received one 500-mg nitazoxanide tablet twice daily for three consecutive days. While its mechanism of action is not well understood and probably involves multiple host factors, nitazoxanide also inhibits a broad range of other RNA and DNA viruses, including respiratory syncytial virus, parainfluenza virus, coronavirus, rotavirus, hepatitis B virus, hepatitis C virus, dengue virus, yellow fever virus, Japanese encephalitis virus, and human immunodeficiency virus (for a review, see reference 52). Nitazoxanide is currently in clinical trials to evaluate its use in the treatment of HCV and influenza virus infections.

The only two nucleoside analogs known to inhibit NoV replication are 2CM-C and T-705 (favipiravir) (14, 15, 17). Based on research conducted on other RNA virus polymerases (35, 40), we and others have hypothesized that 2CM-C and T-705 might interfere with NoV replication by inhibiting the viral RNA polymerase. However, there has not been any detailed biochemical study on NoV polymerase to support the mechanism of action of the two molecules. Furthermore, the antiviral effect of T-705 against human norovirus has never been tested. Therefore, we expressed and purified recombinant HNV and MNV polymerases and developed biochemical methods to study the interaction between NoV polymerase and nucleotide analogs. We showed that 2CM-CTP and T-705 RTP inhibit HNV RdRp activity, with IC₅₀s in the low micromolar range, whereas ribarivin-TP was significantly less potent (Fig. 3 and Table 1). Similar IC₅₀s were obtained for 2CM-CTP and T-705 RTP when tested against MNVpol (see Table S1 in the supplemental material), despite the fact that there is only 57% sequence identity between the human and mouse viral enzymes (see Fig. S7 in the supplemental material). The broad antiviral spectra of many nucleotide analogs can be explained by the fact that, although the overall polymerase sequences may vary significantly, the residues involved in nucleotide binding are highly conserved. In a cell-based assay using the subgenomic replicon of HNV, ribavirin was also less potent than 2CM-C and T-705 (Fig. 1). This result is consistent with a prior report of weak inhibition of ribavirin in the HNV replicon assay (53). More recently, Arias et al. also observed that T-705 inhibits MNV replication more efficiently than ribavirin, both in vitro and in vivo (37). The fact that 2CM-CTP and T-705 RTP potently inhibited HNVpol led us to study their potential competition with natural NTPs. As expected for an unmodified nucleobase, 2CM-CTP displayed a clear CTP competition profile (Fig. 4A to C and Table 2). In contrast, T-705 RTP competed mostly with ATP and GTP, with the greatest effect observed when high GTP and ATP concentrations were combined (Fig. 4D to F and Table 2). The complex competitive behavior of T-705 RTP is reminiscent of its inhibition properties against influenza A virus RNA polymerase (35) and suggests that T-705 RTP may be ambiguously recognized as a guanosine and adenosine analog leading to the inhibition of RNA synthesis during primer elongation. In addition, we found that T-705 RTP also interfered with de novo initiation of RNA synthesis by HNVpol, most likely by competing with GTP at the initiation site (Fig. 5). The 5′ end of the NoV genome is covalently linked to VPg, a viral protein that is used as a primer for the polymerase during minus-strand RNA synthesis. We have not attempted to measure the effect of T-705 RTP on VPg-primed RNA synthesis, and therefore, it would be interesting to perform such studies in combination with cell-based measurements of VPg-primed RNA synthesis (54).

One of the main goals of our study was to develop a single-nucleotide incorporation assay to measure the incorporation of T-705 RTP into RNA by NoV polymerases. Prior attempts to develop such an assay were unsuccessful (55), most likely due to the specific template sequence needed for productive enzyme-RNA interactions (39). Because of the stringent promoter sequence requirement of NoV polymerase, we designed a primer-template substrate that mimics the promoter region of the MNV subgenome and allows the enzyme to incorporate NTPs one base at a time (see Fig. S2 in the supplemental material). Our nucleotide incorporation reaction involved a complex kinetic pathway, including enzyme-RNA binding to form a productive complex, nucleotide binding and incorporation, enzyme dissociation from the product, and rebinding to another RNA molecule. The rate-limiting step in this nucleotide incorporation assay was not clear; therefore, the incorporation percentages measured at 60 min in the assay indicated only if a nucleotide was incorporated and if it could be further extended after its incorporation. In this assay, 2CM-CTP was incorporated opposite template G and prevented the incorporation of the next correct nucleotide (Fig. 6A to C), a property shared with several other 2′-modified chain terminators of RNA synthesis (56). For example, 2CM-CTP has been previously shown to be an immediate chain terminator of HCV polymerase (57). In contrast, T-705 RTP was incorporated by MNVpol opposite template U, resulting in only partial inhibition of further primer extension (Fig. 6D and E). The incorporation of T-705 RTP opposite template C was weak in comparison to that of natural GTP (see Fig. S4 in the supplemental material). The lack of immediate and complete chain termination observed with T-705 RTP is also consistent with previous reports for influenza A virus polymerase when tested under comparable assay conditions (35). One hypothesis for the mechanism of action of T-705 is that, once incorporated into the growing viral RNA, T-705 ribonucleoside monophosphate (RMP) induces lethal mutagenesis in NoV (37) and in influenza virus (36). This antiviral effect is likely different from that of ribavirin. Unlike T-705, ribavirin inhibits inosine-5′-monophosphate dehydrogenase and causes a significant decrease in the intracellular GTP pool (34, 58, 59), which could explain its mutagenic effect without involving any interaction with the polymerase. Until now, no resistance mutants in influenza virus and NoV induced by T-705 selection pressure have been identified. Further evidence for T-705 targeting the RNA polymerases of other RNA viruses was provided by the recent identification of a specific resistance mutant selected in the RdRp of chikungunya virus, although the effect of the mutant on virus sensitivity to T-705 was relatively modest (60). Abdelnabi et al. also reported low-level resistance to T-705 conferred by mutating a conserved lysine in the F1 motif of coxsakievirus RNA polymerase (61). To fully demonstrate the involvement of viral RNA polymerases in T-705 inhibition, further work is needed to characterize the effects of the resistance-associated variants of the corresponding viral polymerases.

The second part of this study consisted of measuring the selectivity of 2CM-CTP and T-705 RTP against human RNA polymerases. As previously reported (49), we confirmed that 2CM-CTP was not selective of viral RNA polymerases because it also interacted with HMRP (Fig. 7A). As with viral RNA polymerases, incorporation of 2CM-CTP by HMRP resulted in immediate chain termination (Fig. 7D; see Fig. S5A in the supplemental material). T-705 RTP was also an efficient substrate for HMRP (Fig. 7B and C), but its ambiguous base pairing and incorporation did not result in immediate chain termination of human mitochondrial RNA (Fig. 7D; see Fig. S5A in the supplemental material). The effect of one or multiple T-705 RMP incorporation events on mitochondrial functions is unknown, and the relevance of these findings to potential clinical safety concerns requires further study. Nevertheless, we and others have established a correlation between the selectivity of other ribonucleotide analogs against HMRP and their toxicity in cells (49). For this reason, we attempted to design an analog of T-705 that would not interact with HMRP. The molecule 2CM-T-1106-TP combines the modified ribose of 2CM-CTP with the modified nucleobase of T-1106, the des-fluoro version of T-705 (Fig. 8A). Unlike T-1106-TP, 2CM-T-1106-TP was highly discriminated by HMRP, with undetectable incorporation levels (Fig. 8B; see Fig. S6 in the supplemental material). At the same time, the inhibition potency of 2CM-T-1106-TP toward HNVpol was considerably reduced compared to that of T-1106-TP (Fig. 8A), and therefore, the parent nucleoside was not tested in cell-based assays. This result shows that, although it is possible to design analogs of T-705 containing small structural changes to alleviate interaction with human polymerases, further work will be needed to discover other analogs of T-705 that also retain inhibition efficacy against viral RNA polymerase targets. Toward this goal, molecular modeling of the interactions between nucleotide analogs and the active site of norovirus polymerase may provide some guidance for the optimization of drug activity (62, 63). In general, understanding the precise mechanisms of action of anti-NoV compounds gives a rational basis for optimizing their inhibition potencies and selectivities. As reported here, the combined evaluation of on- and off-target effects through detailed biochemical analysis could also be used in standard screening assays to evaluate new analogs of T-705 and 2CM-C for NoV and other antiviral applications.