Fig 1.
Two different conformational states of DENV2 NS5 and their relationship between the JEV and DENV3 structures.
Superimposed but individually presented structures of JEV (A, PDB entry: 4K6M, chain A), DENV3 (B, PDB entry 4V0Q), and two forms of DENV2 (C and D) NS5 shown in the orientation viewing from the top of the RdRP. Coloring scheme: MTase in cyan, linker in red, RdRP palm in grey, thumb in blue, index in green, middle in orange, pinky in light red, N-terminal extension (NE) in pink, and the signature YGDD sequence in magenta. Zinc ions and SAH molecules are shown as brown spheres and sticks, respectively. Block arrows are used to indicate plausible conformational transitions between structural states together with the straight line and rotation direction and angle associated with each transition. The fully ordered motif G region (residues 404–415) in the pinky finger of the JEV structures is highlighted by thicker ribbon representations.
Fig 2.
A comprehensive comparison of the intra-molecular MTase-RdRP interface shown as stereo-pair images.
A) A comparison between the JEV (top) and the first form of DENV2 (bottom) structures. B) A comparison between the DENV3 (top, two models) and the second form of DENV2 (bottom) structures. C) The binding of the second SAH molecule observed in the form 2 of DENV2 structure. The binding pocket is shown as surface representations with conservation scores projected. Thinner sticks show the moderately different SAH binding mode observed in the other NS5 molecule in the crystallographic asymmetric unit. Composite simulated-annealing (SA) omit electron density maps contoured at 1.2 σ are overlaid with the DENV2 models in panels A and B and the SAH molecule in panel C. For the DENV2 structures in panels A and B, the two NS5 molecules in the crystallographic asymmetric unit were superposed and shown as thick and thin representations with the density maps of the thick model overlaid. All structures in panels A and B were superposed but may be presented separately. The coloring scheme is the same as in Fig 1. For panels A and B, the rotational movements correlate the both structure pairs are indicated.
Table 1.
X-ray diffraction data collection and structure refinement statistics.
Fig 3.
Key elements that may mediate the conformational switches of the flavivirus NS5.
A) Stereo-pair images (wall-eyed) shown in a view looking at the MTase-RdRP linker region. For clarity, the entire RdRP region is shown in grey, linker is in red, and only the C-terminal ten residues including the very C-terminal GTR sequence of the MTase are shown. B) A comparison between the two DENV2 conformational states highlights a likely role of the highly conserved index finger NLS-helix. Left: JEV-mode; right: DENV3-mode. The coloring scheme is the same as in Fig 1. The α-carbon atoms of two important patches of residues are shown as blue spheres. C) The sequence logo plot showing the conservation of the two RdRP interacting regions, the GTR-linker region (top panel), and the NLS-helix that is important for both JEV- and DENV3-mode conformational states, the middle and ring finger regions only critical in JEV-mode states (bottom panel). The triangles indicated key residues involved in the intra-molecular interactions. D) A schematic free energy diagram for all four conformational states of flavivirus NS5. The relative free energy was crudely estimated by the solvent accessible surface area occluded by the NS5 intra-molecular interface interactions. E) A list of DENV2 NS5 mutants that were designed based on both the JEV- and DENV3-mode conformations, with abbreviations and full descriptions, including mutation site and mutation type.
Fig 4.
Virus proliferation analyses of the WT JEV and DENV2 and the viruses bearing the DENV3-mode mutations.
A) Effects of NS5 mutations on JEV replication. IFA of JEV genome-length viral RNA containing E67A, E67D, K68A, K68R, and E67A/K68A mutations in transfected BHK-21 cells at 72 hpt. Monoclonal antibody 4G2 against envelope protein and FITC-conjugated goat anti-mouse IgG were used as primary and secondary antibodies for IFA, respectively. B) Effects of NS5 mutations on DENV2 replication. IFA of DENV2 genome-length viral RNA containing different mutations of NS5 in transfected BHK-21 cells at 96 hpt. Virus production of the supernatants of the transfected cells at each time point post transfection was detected by monolayer plaque assay, and the visible plaques were used to calculate titers.
Fig 5.
Mutations perturbing the JEV-mode interface interactions impaired DENV2 RdRP initiation.
A) Left: A diagram of construct T30/P2 used in all polymerase assays and related NTP-driven reactions to generate products with different lengths. Middle: reaction flow chart of the P2-driven EC formation (IC2 to EC9) and the subsequent single-nucleotide extension (EC9 to EC10). Right: the EC9 was in a form of precipitate and was able to extend to EC10 upon CTP addition under high-salt condition. B-C) The EC9 formation comparison with the WT NS5 for the JEV-mode (B) and DENV3-mode (C) mutants. The relative intensity of the 9-mer was used to estimate the polymerase activities (the WT value for each time point series was set to 100). D) Comparison of the WT and two representative mutants (R3 for the JEV-mode; M_67A/68A for the DENV3-mode) in the multiple-turnover P3 formation and in the single-turnover P9 formation assays. The pppGGA was synthesized when GTP and ATP were provided as the only NTP substrates using the P2-free T30 template and was used as a migration marker. Note that pppGGA migrated at similar position as pGG, but faster than pGGA since it contains two extra phosphate groups at the 5′ end. A chemically synthesized 9-mer loaded with an equal molar amount to T30 was used as a quantitation standard (STD, lanes 62 and 81). The average intensity of the STD bands was set to 1. Based on previously reported evaluation, the intensity-molar amount starts to deviate from a linear relationship when the relative intensity approaches 4–5 under similar experimental settings [12]. Therefore, the intensity reported in lanes 64–65 and 76–77 are underestimated. The P9 product migrated faster than the chemically synthesized STD RNA due to its 5′-phosphate inherited from the pGG dinucleotide.
Fig 6.
The reactivity and stability of DENV2 NS5 EC9 were not apparently affected by both types of mutations.
A) The reaction flow chart for the reactivity test (1) and the stability test (2). (B) A comparison of the EC9 to EC10 conversion for the WT and two representative mutants at 5 or 300 μM CTP concentration. The fraction of the 10-mer intensity was shown in each lane. C-D) A comparison of the EC9 stability upon high-salt challenge for the above three NS5 constructs. The fraction of 10-mer intensity (determined based on gels in panel C) as a function of challenge time was plotted (D) to estimate the apparent EC inactivation rate constant (kina) for each construct.
Fig 7.
Characterization of the initiation kinetics on the WT DENV2 NS5 and representative NS5 mutants.
A-C) The P3 (pGGA) formation under different ATP concentrations by the WT (A), R3 (B), and M_67A/68A (C) constructs. The color-coded icons above the lanes indicate samples from the same reaction mixture that were used to correlate intensity in different gels (see Materials and Methods). The STD samples were used as a reference and NTP concentration range was adjusted for each NS5 construct, both for ensuring that all P3 band intensities were within the linear range of the staining method used. D-F) The relative reaction rates (left) and KM fitting (right) analyses for the WT (D), R3 (E), and M_67A/68A (F) constructs. Left: The adjusted intensity of the 3-mer products as a function of time for WT, R3, and M_67A/68A with five ATP concentrations. (G) An analysis of the relative specificity constants of the R3 and M_67A/68A mutants to the WT. Single-gel based reaction rate correlation analysis (left) for the R3-WT (purple points) and M_67A/68A-WT pairs to correlate three Michaelis-Menten curves for determination of the relative specificity constants (right).
Fig 8.
A comparison of P2 to P9 conversion of the WT DENV2 NS5 constructs confirmed the initiation impairment in the R3 mutant.
A) Reaction flow chart. Two ATP concentrations (300 and 1000 μM) were tested. B) The P9 accumulation was monitored over time for the WT, R3, and M_67A/68A constructs and the STD samples were used for quantitation (set to 1). C) The relative P9 intensity as a function of time was plotted for all three constructs under two ATP concentrations. The overall conversion rate (rateconv) was estimated by fitting each data set to a single exponential rise model.