1′- and 4′-Cyano Modified Adenosine Analogs Against Prototypic Flavivirus RNA-Dependent RNA Polymerases

TL;DR

Researchers tested modified adenosine drugs against flaviviruses, finding they inhibit viral replication through different mechanisms.

Contribution

The study reveals that template-dependent inhibition is an effective alternative to immediate chain termination in antiviral drugs.

Findings

GS-646939 acts as an immediate chain terminator against flavivirus RdRps.

GS-443902 inhibits replication by blocking complementary UTP incorporation.

Remdesivir shows superior antiviral activity compared to GS-7682.

Abstract

Flaviviruses are arthropod-borne RNA viruses associated with significant human diseases globally. There are no effective direct-acting antivirals approved to treat these viral infections. Given its critical role in viral replication, the RNA-dependent RNA polymerase (RdRp) is a logical target for antiviral drug development. Remdesivir (formerly GS-5734), a 1′-cyano modified C-adenosine monophosphate prodrug, was the first US Food and Drug Administration (FDA) approved antiviral for coronavirus disease 2019 (COVID-19) and was also shown to inhibit flavivirus replication. GS-7682, a 4′-cyano modified C-adenosine prodrug, exhibits a broad-spectrum antiviral activity. Here, we determined the anti-flavivirus potency of both remdesivir and GS-7682 and characterized their active triphosphate forms, GS-443902 and GS-646939, respectively, against a panel of purified flavivirus RdRps. These include dengue, Japanese encephalitis, West Nile, yellow fever, and Zika. Enzyme kinetics demonstrate efficient RNA incorporation of GS-443902 and GS-646939. GS-646939 acts as an immediate chain terminator. Conversely, GS-443902 acts through a template-dependent inhibition mechanism by impeding the incorporation of the complementary UTP. Both mechanisms correlate with anti-flavivirus activity, although remdesivir is generally superior. The data demonstrate that immediate chain termination is not necessarily a preferred mechanism of action of nucleotide analogs. Template-dependent inhibition should also be considered, especially for viruses lacking intrinsic proofreading activities.

Figures

Full text

1. Introduction

Flaviviruses are positive-sense, single-stranded RNA viruses associated with human diseases worldwide [1,2]. Important pathogens include dengue virus (DENV), Japanese encephalitis virus (JEV), West Nile virus (WNV), yellow fever virus (YFV), and Zika virus (ZIKV). The transmission of these flaviviruses is sustained in a zoonotic cycle between mosquito vectors and several mammalian hosts, including pigs, birds, and non-human primates. Cumulatively, these viruses infect over 400 million people annually, with more than 95% of these cases attributed to DENV infection [2,3]. These viruses can cause a variety of severe and life-threatening conditions, including meningitis, encephalitis, hepatic failure, febrile syndrome, and congenital abnormalities [2,3]. Vaccines are available against DENV, JEV, and YFV. However, their utility is limited for several reasons, including poor coverage of genotypes, constraints in production and distribution, and antibody-dependent enhancement (ADE) of disease [2,4,5,6]. Efficient direct-acting antivirals for the treatment of flavivirus infections are currently not approved. The viral RNA-dependent RNA polymerase (RdRp) is essential for viral replication, thus presenting a logical target for therapeutic intervention strategies. Nucleotide analogs represent a class of antivirals that target the RdRp catalytic site. The active triphosphate (TP) form of nucleosides can compete with natural nucleoside triphosphate (NTP) pools for incorporation into nascent RNA strands. Sugar moiety modifications can cause a range of inhibitory effects depending on the specific chemical nature and the polymerase target [7,8,9].

Much of the pre-clinical data on nucleoside analogs against flaviviruses is focused on modifications of the 2′ position. Sofosbuvir (SOF), a 2′-C-methyl modified nucleoside prodrug approved for treatment of the related hepatitis C virus (HCV), acts as a non-obligate chain terminator targeting the RdRp. HCV is a hepacivirus and a member of the Flaviviridae family, thus possessing an RdRp that is structurally homologous to that of flaviviruses [10,11]. Recent studies have demonstrated SOF’s antiviral efficacy against several insect-borne flaviviruses both in vitro and in vivo [12,13,14,15]. However, clinical data from human trials are lacking. 2′-acetylene-7-deaza-adenosine (NITD008) was developed to inhibit DENV RdRp [16,17]. Similar to SOF, the incorporation of NITD008 promotes the termination of RNA synthesis, exhibiting potent antiviral effects against flaviviruses in vitro and in vivo. However, NITD008 failed to progress beyond pre-clinical studies due to toxicities [16]. Numerous other 2′ modified nucleoside analogs have also undergone pre-clinical investigation against the flaviviruses [18].

Nucleoside analogs with modifications at the 1′ position have been considered as well. Remdesivir (RDV) is a 1′-cyano modified C-adenosine monophosphate prodrug (Figure 1A) [19] derived from the parent nucleoside GS-441524 (Figure 1B). RDV was the first antiviral to receive Food and Drug Administration (FDA) approval for the treatment of COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) [20]. In cell culture, RDV has demonstrated broad-spectrum antiviral activity against several positive-sense RNA viruses and non-segmented negative-sense RNA viruses [19,21,22,23,24,25,26]; however, RDV does not show significant antiviral activity against influenza viruses, Lassa virus, or Crimean–Congo hemorrhagic fever virus [22,26]. Importantly, the half maximal effective concentration (EC_50_) of RDV required to inhibit flavivirus replication in cell culture proved to be in the sub-to-low micromolar range [26]. The incorporation of RDV in its active triphosphate form, herein referred to as GS-443902 (Figure 1C), by the SARS-CoV-2 polymerase outcompetes its natural counterpart adenosine triphosphate (ATP), approximately 2- to 3-fold [27,28]. A steric clash between the 1′-CN of the incorporated RDV-monophosphate (MP) and S861 in non-structural protein (Nsp) 12 results in delayed chain termination [29,30,31]. This inhibitory effect can be overcome at increased NTP concentrations, yielding viral genome copies that contain GS-443902 residues. Subsequently, embedded GS-443902 can inhibit RNA synthesis through a template-dependent mechanism [29]. Previous work has demonstrated that HCV is susceptible to RDV in cell culture [19,26]. Upon investigation, HCV readily incorporated GS-443902; however, no significant inhibition of primer extension reactions was observed [32]. Instead, the termination of RNA synthesis solely relied on a template-dependent mechanism of inhibition. Similarly, the incorporation of GS-443902 by flavivirus RdRps does not result in the termination of RNA synthesis [33]. Despite these observations and the relationship between HCV and flaviviruses, a detailed mechanism of inhibition for GS-443902 against the flaviviruses has yet to be elucidated.

Modifications at the 4′ position have also been associated with antiviral effects against flaviviruses. Balapiravir, a 4′-azidocytidine analog originally developed for HCV, was also explored against DENV [34]. Although treatment demonstrated favorable pharmacokinetics, balapiravir failed to exhibit antiviral potency in a clinical trial. GS-7682 is a 4′-cyano modified C-adenosine analog (Figure 1D) derived from the parent nucleoside GS-646089 (Figure 1E). GS-7682 is likewise associated with a broad spectrum of antiviral activities against multiple viruses, demonstrating its strongest in vitro potency against viruses belonging to the Pneumoviridae and Picornaviridae families [35]. Previously, our group characterized the mechanism of action of the active triphosphate form of GS-7682, herein denoted as GS-646939 (Figure 1F), against several respiratory RNA virus polymerases, providing biochemical explanations for the observed antiviral activity in vitro [36]. However, the antiviral efficacy of GS-7682 and the biochemical mechanism of GS-646939 against flaviviruses have not been explored. Given the prevalence of flavivirus infections and the absence of approved antivirals targeting flaviviruses, a detailed characterization of the 1′-CN and 4′-CN modified nucleotide analogs could contribute to a critical gap in targeted anti-flaviviral drug development. Here, we demonstrate that the 1′-CN and 4′-CN modified nucleotide analogs exhibit different antiviral efficacies against the flaviviruses. This discrepancy may be attributable to their distinct mechanisms of action, providing insights that could support the development of novel antivirals targeting flavivirus RdRps.

2. Materials and Methods

2.1. Viruses

Stable nanoluciferase (Nluc) reporter flaviviruses were generated following a previously described protocol [37]. The study utilized the following viral strains: dengue virus 2 (DENV-2, New Guinea C strain), Japanese encephalitis virus (JEV, 14-14-2 strain 1454), West Nile virus (WNV, NY99 strain), yellow fever virus (YFV, 17D strain YFS11), and Zika virus (ZIKV, Dakar 41525 strain).

2.2. Antiviral Assay

Huh7 human hepatocarcinoma cells (1.2 × 10^4^ cells/well, obtained from the lab of Dr. Charles Rice at Rockefeller University, New York, NY, USA) were seeded in white opaque 96-well plates (Corning, Kennebunk, ME, USA). The following day, cells were infected with Nluc reporter flaviviruses at a multiplicity of infection (MOI) of 0.01–0.1 in the presence of 3-fold serially diluted inhibitors. Cells treated with 0.5% DMSO served as controls. At 48 h post-infection, luciferase activity was measured using the NanoGlo substrate (Promega, Madison, WI, USA) according to the manufacturer’s protocol. Luminescence signals were normalized to those of the DMSO-treated wells, and EC_50_ values were determined using a four-parameter nonlinear regression model in GraphPad Prism 10 (GraphPad Software, Inc., Boston, MA, USA).

2.3. In Vitro Metabolism

Huh7 cells were seeded in a 12-well plate at 1.5 × 10^5^ cells per well. After 24 h, cell culture medium was replaced with medium containing 1 µM of compound (RDV or GS-7682) and incubated at 37 °C with 5% CO_2_. After compound addition, cells were then harvested at timepoints of 2, 4, 8, 24, 48, and 72 h, by being washed 3 times with ice-cold tris-buffered saline, scraped into 0.5 mL of ice-cold 70% MeOH, and stored at −80 °C. Extracts were centrifuged at 15,000× g for 15 min, and the supernatants were transferred to clean tubes for evaporation using a miVac Duo concentrator (Genevac, Painter, NY, USA). The dried samples were reconstituted in ammonium dihydrogen phosphate (pH ~7). The LC-MS/MS analysis used a multistage linear gradient from 10% to 50% acetonitrile in mobile phase A, which contained 3 mM ammonium formate (pH 5) and 10 mM dimethylhexylamine in water, at a flow rate of 300 μL/min. Analytes were separated on a 50 × 2 mm, 2.5 μm Luna C18(2) HST column (Phenomenex, Torrance, CA, USA) connected to an LC-40D UHPLC system (Shimadzu, Kyoto, Japan). Detection was performed with a Qtrap 6500+ (AB Sciex, Redwood City, CA, USA) mass spectrometer operating in positive ion mode for GS-443902 and negative mode for GS-646939. Quantification relied on a 6-point standard curve prepared in blank matrices spanning 3 orders of magnitude.

2.4. Nucleic Acids and Chemicals

RNA primers and templates used in this study were 5′-phosphorylated and purchased from Dharmacon (Lafayette, CO, USA). GS-646939 and GS-443902 were provided by Gilead Sciences (Foster City, CA, USA). NTPs were purchased from GE Healthcare (Mississauga, ON, Canada). [α-^32^P] GTP was purchased from PerkinElmer (Revvity, Waltham, MA, USA).

2.5. Expression and Purification of Viral Polymerases

Full-length YFV NS5 was expressed and purified as previously described [38]. DENV-2, JEV, and ZIKV were expressed in insect cells (Sf9, Invitrogen) using the MultiBac (Geneva Biotech, Indianapolis, IN, USA) [39,40], whereby the pFastBac-1 (Invitrogen, Burlington, ON, Canada) plasmid contained the codon-optimized synthetic DNA sequences (GenScript, Piscataway, NJ, USA) coding for full-length NS5 from DENV-2 (AAK67712.1 polyprotein, residues 2492-3391), JEV (NP_775674.1), and ZIKV (ALU33341 polyprotein, residues 2526-3416). The C-terminal residues 3417–3423 for ZIKV were removed. The pET-15b (Novagen) plasmid with codon-optimized synthetic DNA sequences (GenScript, Piscataway, NJ, USA) coding for full-length WNV NS5 (YP_001527887.1) was expressed in Escherichia coli. DENV-2, JEV, WNV, and YFV NS5 were purified using Ni-NTA affinity chromatography based on their respective 8×histidine tag according to the manufacturer’s specifications (Thermo Scientific, Rockford, IL, USA). ZIKV NS5 was purified using strep-tag-affinity chromatography, according to the manufacturer’s specifications (IBA, Goettingen, Germany). The identity of the purified proteins was confirmed by mass spectrometry analysis (Dr. Jack Moore, Alberta Proteomics and Mass Spectrometry, Edmonton, AB, Canada).

2.6. Evaluation of GS-443902 and GS-646939 Incorporation and Subsequent Primer- and Template-Strand Inhibition on Viral RNA Synthesis

The following synthetic 5′-monophosphorylated RNA templates were used in this study (the portion of the template that is complementary to the 4-nucleotide (nt) primer is underlined): analog selectivity, pattern of inhibition, and inhibition at position i + 1 experiments for all enzymes were performed using 3′UGCGCUAGAAAAAAp. 3′UGCGCUUUUUAAAAAAAAAAp, 3′UGCGCUUUUUG_443_AAAAAAAAp, and 3′UGCGCUUUUUG_646_AAAAAAAAp were used for the evaluation of nucleotide incorporation opposite an embedded GS-443902 and GS-646939, respectively. RNA synthesis assays of viral RdRps, data acquisition, and quantification were performed as previously reported by us [36,41]. For experiments determining V_max_ and Km, the viral RdRp concentrations (350 nM DENV-2, 380 nM JEV, 390 nM WNV, 1000 nM YFV, and 150 nM ZIKV) and the timepoint for each reaction were optimized to ensure incorporation of [α-^32^P]-GTP and ATP is within the linear range of product formation. For evaluation of single-nucleotide incorporation, the concentration range of ATP, GS-443902, and GS-646939 was optimized to avoid misincorporation at subsequent positions. Reaction mixtures for RNA synthesis assays (final concentrations after mixing in 15 µL reactions) contain the purified viral RdRps, Tris–HCl (pH 8, 25 mM), RNA primer (200 μM), RNA template (2 μM), [α-^32^P]-GTP (0.1 µM), and various nucleotide concentrations were prepared on ice and incubated at 30 °C for 5 to 10 min. RNA synthesis was initiated by the addition of 5 mM MgCl_2_ for reactions containing DENV-2, JEV, WNV, and YFV NS5. RNA synthesis performed by ZIKV NS5 was initiated by the addition of 2.5 mM MnCl_2_. The duration of DENV-2, JEV, WNV, and ZIKV reactions was 10 min, while YFV reactions were 20 min. The reactions were stopped after the given durations by the addition of an equal volume of formamide/EDTA (50 mM) mixture and incubated at 95 °C for 5 min. The 3 µL reaction samples were then resolved by 20% polyacrylamide gel electrophoresis (PAGE) containing 8 M urea, 89 mM Tris base, 89 mM boric acid, and 2 mM EDTA, using 1×TBE running buffer. The [α-^32^P] generated signal was then stored and scanned using phosphorimager screens. To calculate selectivity, the fraction of RNA synthesis products generated at position 6 out of the total signal in the lane (5-nt and 6-nt products) was determined and plotted against the concentration of the nucleotide or nucleotide analog. To calculate incorporation efficiency at position i + 1, the sum of RNA synthesis products generated at position 7 and beyond was divided by the total signal in the lane (all RNA synthesis products generated in a given lane) and plotted against the concentration of UTP. The data were analyzed using GraphPad Prism 10 (GraphPad Software, Inc., Boston, MA, USA). To calculate template inhibition, the sum of RNA synthesis products generated beyond position 10 was divided by the total signal in the lane and normalized as a percentage to the highest product fraction observed for that template. This was then plotted against the concentration of UTP. The ratio of UTP concentrations at which 50% of RNA products are generated is beyond position 10 for template A and R generated fold-inhibition values.

3. Results

3.1. Experimental Strategy

This study aimed to compare the antiviral profiles of RDV and GS-7682 against flaviviruses, assessing their potency in vitro while also differentiating the 1′- and 4′-cyano-driven mechanisms of inhibition for GS-443902 and GS-646939, respectively. The viruses included were DENV-2, JEV, WNV, YFV, and ZIKV. The corresponding NS5 was included to evaluate the incorporation efficiency and mechanism of action of the active triphosphate metabolites.

3.2. Anti-Flavivirus Activity and Cytotoxicity of RDV and GS-7682

To investigate the antiviral activity of RDV and GS-7682 against the selected prototypic flaviviruses, human hepatocarcinoma (Huh7) cells were infected with Nluc reporter flaviviruses and treated with the compound. RDV inhibits flavivirus replication in Huh7 cells with EC_50_ values of 0.063 to 0.180 µM (Table 1). The potential cytotoxicity of RDV was also evaluated in Huh7 cells by determining the half-maximal cytotoxic concentration (CC_50_). RDV exhibited CC_50_ values >10 µM. In contrast to RDV, GS-7682 inhibited flavivirus replication with weaker potency, yielding EC_50_ values of 2.15 to 4.57 µM (Table 1). The CC_50_ values for GS-7682 were >10 µM.

3.3. Active Metabolite Formation of GS-443902 and GS-646939 in Huh7 Cells

Given the differences in antiviral potencies that were observed for RDV and GS-7682, we wanted to assess the formation of their intracellular active metabolites, GS-443902 and GS-646939, respectively (Table 2). Huh7 cells were incubated with 1 µM of RDV or GS-7682, and intracellular trisphosphate levels were monitored over 72 h at various timepoints. Treatment of Huh7 cells with either prodrug yielded 27.43 pmol/10^6^ cells and 20.43 pmol/10^6^ cells of the intracellular triphosphates GS-443902 and GS-646939, respectively. Thus, no meaningful differences were found between the trisphosphate levels incubated with the prodrugs RDV and GS-7682, suggesting that the disparity in anti-flavivirus potency is likely not due to triphosphate formation.

3.4. Selective Incorporation of GS-443902 and GS-646939 by Flavivirus RdRp Enzymes

Next, we sought to biochemically investigate GS-443902 and GS-646939 against the full-length NS5 from several insect-borne flaviviruses. Previous studies have shown that the efficient incorporation of a nucleotide analog is an essential parameter for antiviral activity [32,36,42]. Therefore, the difference in antiviral potencies between RDV and GS-7682 could be reflected in their respective incorporation efficiencies. Here, we compared the incorporation of GS-443902 and GS-646939 to ATP (Table 3). Under steady-state conditions, the kinetic parameters V_max_ and Km were used to determine the efficiency of a single incorporation event of ATP, GS-443902, and GS-646939 (see section: Evaluation of GS-443902 and GS-646939 Incorporation and Subsequent Primer- and Template-Strand Inhibition on Viral RNA Synthesis). Nucleotide incorporation was monitored in gel-based assays using short primer/templates (Figure S1) [32,41,43]. [α-^32^P]-GTP incorporation at position 5 enables the visualization of RNA synthesis products at single-nucleotide resolution. ATP, GS-443902, or GS-646939 is added to the reaction at increasing concentrations, facilitating incorporation at position 6, and reactions are stopped after a given amount of time after the addition of the divalent metal ion. Assays with DENV-2, JEV, WNV, and YFV RdRps use Mg^2+^ as a metal co-factor, whereas the ZIKV RdRp requires Mn^2+^ for RNA synthesis activity [44]. To compare nucleotide incorporation rates, selectivity is determined as the ratio of incorporation efficiency of ATP to each analog. A selectivity value greater than one indicates that the analog is incorporated less efficiently than its natural counterpart. Prior work determined that the HCV RdRp incorporated GS-443902 with a selectivity value of 0.93, indicating that GS-443902 and ATP are incorporated with similar efficiencies [32]. Comparatively, DENV-2, JEV, WNV, YFV, and ZIKV RdRps show selectivity values of 11.1, 6.8, 6.8, 10.4, and 6.5, respectively, demonstrating that GS-443902 is incorporated less efficiently than ATP. Similarly, GS-646939 is incorporated 7.0-, 3.4-, 5.1-, 5.8-, and 4.0-fold less efficiently than ATP by DENV-2, JEV, WNV, YFV, and ZIKV RdRps, respectively. In summary, GS-443902 and GS-646939 incorporate with similar efficiencies; however, RDV exhibits stronger antiviral potency against the flaviviruses, suggesting that flavivirus RdRps may have different threshold levels of incorporation for a given analog, which is likely influenced by the analog’s mechanism of action.

3.5. Inhibition of RNA Primer Extension Reactions

Despite the ~1.5- to ~2-fold less efficient incorporation than GS-646939 (Table 3), GS-443902 shows at least 10-fold better antiviral activity (Table 1). This apparent disconnect between the incorporation efficiencies of GS-443902 and GS-646939, and their prodrugs’ inverse patterns of antiviral potency, warrants further investigation into their mechanisms of action. To evaluate the pattern of inhibition, we employed an RNA template that facilitates a single incorporation of ATP, GS-443902, or GS-646939 at position 6 (Figure 2A). The reaction was supplemented with 1 µM of CTP and UTP, thereby providing all components required for RNA synthesis and the generation of a full-template length product. For all five flavivirus RdRps, GS-443902 does not inhibit RNA synthesis when incorporated into the primer-strand as indicated by the formation of a full-length product at position 14 (Figure 2B); however, when compared to the extension of AMP, an increased signal can be seen at the site of GS-443902 incorporation at position 6 (‘i’). These observations align with previous studies, which found that GS-443902-terminated primers can be extended by flavivirus and hepacivirus RdRps [32,33]. Conversely, GS-646939 behaves as an immediate chain terminator, preventing RNA synthesis beyond its incorporation at position “i” (Figure 2B).

Both nucleotide analogs possess a 3′-hydroxyl group, which can facilitate a nucleophilic attack and, therefore, could support subsequent nucleotide incorporation events. Therefore, we assessed the RNA synthesis patterns following GS-443902 and GS-646939 incorporation at increasing concentrations of subsequent NTPs. For GS-443902, the pattern of inhibition was similar among the flavivirus RdRps examined (Figure 3 and Figure S2). The incorporation of ATP at position “i” could be readily extended to generate 14-mer full-template length product at sub-micromolar concentrations of NTPs. Subtle product formation resulting from GS-443902 incorporation at position 6 is observed at low NTP concentrations; however, this product formation disappears as NTP concentrations are increased, and a corresponding increase in full-template length product is formed. This observation holds for DENV-2, JEV, WNV, and YFV RdRps. However, ZIKV RdRp displays a buildup of product formation at position 6, the site of GS-443902 incorporation, which is maintained across all NTP concentrations, indicating that GS-443902 does not appear to be extended as efficiently by this enzyme. We confirmed this observation in an additional experiment, where we monitored product formation at positions 5, 6, and 7 as a function of time with DENV-2 and ZIKV RdRps (Figure S3). It is conceivable that for ZIKV RdRp, GS-443902-terminated primers at position 6 may dissociate faster than subsequent nucleotide incorporation, potentially explaining the incomplete conversion of product 6 to product 7. However, it should be noted that a direct comparison of patterns of RNA synthesis of ZIKV RdRp and the other flavivirus RdRps is not possible, given the difference in the catalytic metal ions. Nonetheless, in all cases, and regardless of the metal ion provided, the incorporated GS-443902 can be extended to generate a full-length RNA product.

GS-646939 demonstrates much more robust inhibition against all five flavivirus RdRps (Figure 4 and Figure S4). Unlike ATP or GS-443902, where 1.4 µM CTP and UTP were sufficient to generate a 14-mer product, GS-646939 inhibited RNA synthesis at the site of incorporation. This effect appears to be overcome for DENV-2 and WNV RdRps when nucleotide substrate concentration was greater than 12–37 µM, as indicated by the appearance of 14-mer products. The disappearance of the signal at position 6 demonstrates that GS-646939 can be extended once nucleotide concentrations reach up to 1000 µM; no additional intermediate RNA products are observed. In contrast, the incorporation of GS-646939 at position “i” by YFV and ZIKV RdRps appears more stringent on subsequent nucleotide incorporation, with a lesser extent of RNA products generated beyond position 6. For JEV RdRp, GS-646939 incorporation results in nearly absolute immediate chain termination and is seemingly not overcome at increased nucleotide substrate concentrations.

To quantify the inhibitory effect on subsequent NTP incorporation, we measured the kinetic parameters for nucleotide incorporation events at position “i + 1” with AMP-, GS-443902-, and GS-646939-terminated primer-strands (Figure 5 and Table S3). Compared to AMP, a 3.7- to 75-fold reduction in subsequent UTP incorporation efficiency was observed following a GS-443902-terminated primer. In contrast, all flavivirus RdRps demonstrate a greater than 1000-fold reduction in UTP incorporation on a GS-646939-terminated primer. These marked differences in patterns of inhibition provide strong evidence that the primary mechanism of inhibition of GS-443902 does not pertain to the effects on the primer-strand, whereas GS-646939 acts as an immediate chain terminator.

3.6. Template-Embedded GS-443902 Demonstrates an Inhibitory Effect

The formation of a full-length template product when GS-443902 is incorporated into the growing primer-strand implies that newly synthesized flavivirus genome copies may contain analog residues embedded throughout. To assess this scenario, RNA templates were designed to monitor UTP incorporation opposite a template-embedded GS-443902-MP at position 11 (Figure 6A). For DENV-2 RdRp, on template “A” containing the natural nucleotide, full-length RNA synthesis was generated at UTP concentrations as low as 0.45 μM (Figure 6B). Under the same conditions, an embedded GS-443902 reduced full-length product formation. On template “GS-443902”, an intermediate product was observed at position 10; however, this point of inhibition can eventually be overcome with increased UTP concentration. Inhibition was quantified by comparing the signal corresponding to RNA products greater than 10 nt to the total signal in that lane to determine the product fraction. This was then normalized to the highest product fraction observed for each respective template. The normalized percentages could then be plotted against the concentration of UTP to determine the concentration required to meet the 50% inhibitory threshold. For DENV-2, relative to template “A”, template “GS-443902” required an increased UTP concentration of ~148-fold to meet the 50% threshold (Figure 6C). The investigation of JEV, WNV, YFV, and ZIKV RdRps against an embedded GS-443902 similarly revealed an inhibitory effect, with elevated UTP concentrations of ~599, ~317, ~370, and ~73-fold, respectively, required to produce 50% of RNA products made past position 10 (Table S4).

Although the full-length 14-mer product is generated following an incorporated GS-646939 at position “i” with elevated NTP concentrations, quantification of nucleotide incorporation at position “i + 1” indicates that the dominant mechanism of action is via immediate-chain termination. Due to this, it is anticipated that a template-embedded GS-646939 is an unlikely scenario for all flaviviruses; however, a previous study found that the 4′-cyano modification elicited a different pattern of inhibition compared to its 1′ counterpart against other viral RdRps [36]. Therefore, we evaluated the effects of an embedded GS-646939 residue. For DENV-2 and WNV RdRps, RNA synthesis opposite GS-646939 generates an intermediate product at position 10, indicative of an inhibitory effect (Figure S5).

To conclude, GS-443902 and GS-646939 display distinct mechanisms of action against flavivirus RdRps. GS-443902 does not display significant inhibitory effects when incorporated into the primer-strand; however, its antiviral effect is evident when acting as a templating base. In contrast, GS-646939 acts via immediate chain termination. Together, these distinct mechanisms may provide a reason for their difference in anti-flaviviral potency and be exploited for future drug development efforts.

4. Discussion

Nucleoside analogs have long shown promise as an effective antiviral drug strategy, particularly against significant viruses like HIV, HCV, and SARS-CoV-2. Currently, infections with HCV are successfully treated by multidrug therapies including sofosbuvir [45,46]. Despite their significant impact on global human health and the extensive research into analogs targeting the phylogenetically related HCV, there are currently still no approved antivirals for flavivirus infection. Here, we studied the antiviral potency of remdesivir (RDV), a 1ʹ-cyano modified C-adenosine nucleotide monophosphate prodrug, and GS-7682, a 4ʹ-cyano modified C-adenosine nucleotide phosphoramidate analog, and the mechanisms of action of their active nucleotide triphosphate metabolites, GS-443902 and GS-646939, respectively.

For each flavivirus polymerase investigated in this study, the selectivity data demonstrate that GS-646939 is more efficiently incorporated than GS-443902 by roughly 1.5- to 2-fold, albeit both analogs efficiently compete with ATP. Both GS-443902 and GS-646939 act as poor substrates for the h-mtRNAP, with selectivity values of 508 and >1000, respectively [36,41]. Poor incorporation by h-mtRNAP is indicative of low cytotoxicity, which aligns with our observed CC_50_ values.

GS-443902 has a heterogeneous effect on primer-strand RNA synthesis that is largely dependent on the nature of the viral polymerase [32,36]. In the case of picorna-, corona-, filo-, pneumo-, and paramyxovirus RdRps, GS-443902 displays a delayed chain termination effect; however, the extent and position of this inhibition vary. In the case of HCV RdRp, GS-443902 causes slight inhibition at the site of incorporation, but this effect is easily overcome at the low concentrations of NTPs required to generate a full-template length product [32]. Similarly, the flavivirus RdRps all display subtle inhibition immediately upon incorporation (i) when GS-443902 is incorporated, which is overcome at low NTP concentrations to generate full-length template products [33]. To this end, GS-443902 demonstrates a uniform mechanism of inhibition against the flavivirus RdRps by impairing UTP incorporation when embedded in the template-strand. This template-dependent effect is likely due to the conserved RdRp motif F, rather than a virus-specific residue, as previously described [32]. While this effect can be overcome at elevated NTP concentrations, due to sufficient incorporation rates, this mechanism could describe the sub-micromolar EC_50_ values observed for RDV against flaviviruses.

GS-646939 acts as an immediate chain terminator. Chain termination with the JEV RdRp appears nearly absolute, whereas increasing NTP concentrations allow the continuation of RNA synthesis against the other flavivirus RdRps, as indicated by the generation of full-template length products. A template-embedded GS-646939 is not anticipated, given the poor efficiency of nucleotide incorporation following GS-646939. Interestingly, the processivity of JEV RdRp appears diminished relative to the flaviviruses, as indicated by a much lower V_max_ and Km for nucleotide incorporation. Perhaps the JEV RdRp is less efficient overall, owing to the pronounced termination of GS-646939.

Although GS-443902 and GS-646939 demonstrate similar efficiencies of incorporation by the flavivirus RdRp, the antiviral potency of their prodrugs (RDV and GS-7682, respectively) differ significantly. RDV exhibits EC_50_ values in the sub-micromolar range (0.063–0.18 µM), whereas GS-7682 shows EC_50_ values ranging from 2.15 to 4.57 µM.

While analog incorporation efficiency often correlates with antiviral potency, this relationship is multifaceted and therefore influenced by many factors. To determine if this effect was mediated by a disparity in analog bioavailability, the intracellular concentrations of both GS-443902 and GS-646939 were measured over 72 h in Huh7 cells treated with the prodrugs RDV and GS-7682, respectively. We observed similar triphosphate levels of both analogs, indicating that triphosphate formation was likely not the reason for the differences in antiviral potency. This discrepancy between the efficiency of single nucleotide incorporation and antiviral potency suggests that flavivirus RdRps have different threshold levels of incorporation for a given analog, which is influenced by the analog’s mechanism of action and potentially other external factors.

Previously, GS-7682 activity against SARS-CoV-2 presented a similar example of when incorporation efficiency alone did not dictate antiviral efficacy [36]. Modeling and biochemical analysis showed that GS-646939 may elicit its immediate chain termination by hindering translocation once incorporated. Therefore, despite efficient incorporation by the SARS-CoV-2 RdRp, preventing translocation would provide ample opportunity for the exonuclease to access the incorporated GS-646939, resulting in its weaker antiviral activity. While flavivirus RdRps do not possess intrinsic nuclease properties, factors like pyrophosphorolysis and nucleotide-mediated excision of chain-terminating analogs could reduce an analog’s efficacy, as was observed with the closely related HCV [47,48]. Therefore, inhibition caused by chain-terminating nucleotide analogs is vulnerable to excision, thus potentially limiting their antiviral capacity. By contrast, GS-443902 elicits its inhibitory effect through a template-dependent mechanism, which is not susceptible to excision. Despite the similar incorporation efficiencies of the two analogs, the mechanism of action plays a prominent role in eliciting an antiviral effect. These results highlight the intricate relationship between biochemical and cell-based assays, emphasizing the need to consider both approaches in the evaluation of novel nucleotide analogs and their mechanisms of action. While most previous studies in the field focused on the inhibition of primer extension reactions, this study highlights the importance of template-dependent inhibition and its consideration in subsequent antiviral therapeutic development efforts.

5. Limitations and Bias

Experimentally, this study relies on the use of Huh7 cells for the evaluation of the metabolism and antiviral potency of RDV and GS-7682, which may not completely and adequately reflect the nature of these compounds in an in vivo setting. Also, benchmarks such as sofosbuvir were not included in this study, given the differences in base moieties. Moreover, direct comparison of patterns of inhibition between ZIKV and the other flavivirus RdRps is limited due to the different requirements for metal cofactors.