Sequence-Specific Fidelity Alterations Associated with West Nile Virus Attenuation in Mosquitoes

Abstract

High rates of error-prone replication result in the rapid accumulation of genetic diversity of RNA viruses. Recent studies suggest that mutation rates are selected for optimal viral fitness and that modest variations in replicase fidelity may be associated with viral attenuation. Arthropod-borne viruses (arboviruses) are unique in their requirement for host cycling and may necessitate substantial genetic and phenotypic plasticity. In order to more thoroughly investigate the correlates, mechanisms and consequences of arbovirus fidelity, we selected fidelity variants of West Nile virus (WNV; Flaviviridae, Flavivirus) utilizing selection in the presence of a mutagen. We identified two mutations in the WNV RNA-dependent RNA polymerase associated with increased fidelity, V793I and G806R, and a single mutation in the WNV methyltransferase, T248I, associated with decreased fidelity. Both deep-sequencing and in vitro biochemical assays confirmed strain-specific differences in both fidelity and mutational bias. WNV fidelity variants demonstrated host-specific alterations to replicative fitness in vitro, with modest attenuation in mosquito but not vertebrate cell culture. Experimental infections of colonized and field populations of Cx. quinquefaciatus demonstrated that WNV fidelity alterations are associated with a significantly impaired capacity to establish viable infections in mosquitoes. Taken together, these studies (i) demonstrate the importance of allosteric interactions in regulating mutation rates, (ii) establish that mutational spectra can be both sequence and strain-dependent, and (iii) display the profound phenotypic consequences associated with altered replication complex function of flaviviruses.

Fig 1. Selection for West Nile virus populations with decreased susceptibility to ribavirin.

A. Ribavirin susceptibility, measured as mean log₁₀ pfu/ml reduction in viral titer +/- SD resulting from 50uM ribavirin treatment for 72h on Hela cell culture relative to untreated controls, following passage of parallel series of ribavirin-treated WNV-IC designated as lineage I (black) and lineage II (grey). B. Ribavirin susceptibility in selected clonal populations. Clonal populations were created with lineage I from 20 plaques isolations (pp) and strains with the highest levels of resistance (WNV pp3/9) were chosen for further characterization.

Fig 2. Location of the mutation sites associated with altered fidelity in WNV NS5.

A. Shown are the crystal structures of the N-terminal methyltransferase domain (MTase, colored pink; PDB 2OY0) and the C-terminal polymerase domain (RdRp, colored red; PDB 2HFZ) of the NS5 protein from WNV. The relative positioning of the two domains was based on the crystal structure of the full length NS5 protein of DENV (PDB 4V0R). The protein is represented as ribbon and the identified mutants (T248, V793, and G806) are represented as black spheres. The amino acids V793 and G806 are located in the priming loop that is part of the thumb subdomain. B. The initiation model of WNV RdRp showing the thumb subdomain of the polymerase, rotated 180° compared to the view in A, a 4-mer RNA extracted from the ɸ6-RdRp (PDB 1HI0, green carbon atoms), rNTP modeled at the priming site (P) and the catalytic site (C) based on the complex structure of HCV RdRp (PDB G1X5, yellow carbon atoms), and the active-site aspartates D636 and D669 with the bound catalytic Mg²⁺ ion. The substitutions, V793I and G806R, are expected to impact the interactions of the nearby residues of the priming loop, including W792, T799, W808, and M809. Disruption of these interactions would affect the dynamics of the priming loop, including the conformational change of W800 that is suggested to be required for stabilizing the initiation complex. C. the MTase and polymerase domains shown in (A) are rendered as pink and red surfaces, respectively, to emphasize the expected large interface between the two domains. Residue T248 is colored black; substitution of this residue by an isoleucine could impact the interface between the two domains of NS5 structure. The figure was prepared using program CHIMERA.

Fig 3. Decreased mutagen susceptibility of WNV replication complex mutants.

RdRp and Mtase mutations identified in WNV populations with the highest ribavirin resistance were reverse engineered into the WNV-IC individually (T248I, V793I and G806R) and in combination (V793I/G806R, T248I/V793I/G806R) and assessed for mutagen resistance on Hela cell culture using either 100uM ribavirin or 50uM 5-fluorouracil (5-FU). Bars depict mean reduction in titer +/- standard deviation following mutagen treatment. Significantly higher viral titers (*t-test, df = 5, p<0.05) were measured for WNV-IC relative to all WNV mutants after 72h treatment with both antivirals.

Fig 4. Strain-specific differences in mutational spectra breadth and composition.

WNV mutant swarm characterization by deep-sequencing of nucleotides 1312–3261 (ENV/NS1) following 72h growth on mosquito cell culture. A. Mean distribution of genomic diversity (single nucleotide polymorphisms [SNP]) of 2 replicates of WNV-IC across the sequenced region. The distribution of mutations was similar among strains. B. Mean number of minority SNPs identified for individual WNV strains at frequencies greater than 0.5%. C. Distribution of substitution types among WNV strains.

Fig 5. Evaluation of WNV NS5 variants with an in vitro primer extension assay reveals differences in polymerase incorporation fidelity.

A. Schematic of primer extension assay for evaluating WNV NS5 polymerase activity. B. Evaluation of elongation reactions using either correct or incorrect nucleotide substrates. Shown is the experimental design and denaturing PAGE gel of both correct and incorrect nucleotide addition. Experimental design: WNV NS5 is assembled for 30 min to produce elongation-competent complexes at which point a trap (heparin) for free enzyme is added. The reaction is then rapidly mixed with NTP substrate and quenched at various times. Denaturing PAGE gel: The 15-mer RNA is rapidly extended to 16-mer RNA product in the presence of correct nucleotide substrate, ATP. The presence of both ATP and GTP together promote full extension to 20-mer RNA as the terminal templating bases are competent for correct nucleotide addition. The presence of GTP alone allows for the kinetics of incorrect nucleotide incorporation to be observed (G:U mispair). C. Comparison of the kinetics of misincorporation between WNV-IC, V793I/G806R and T248I NS5. There is no observable difference in the kinetics of GMP misincorporation (G:U mispair). Shown is the percentage of RNA product plotted as a function of time and fit to a single exponential. D. By using an alternative substrate that allows for A:C mispairs to be evaluated, a difference in the efficiency of AMP misincorporation is observed between the three NS5 proteins.

Fig 6. Host-specific effects of WNV replication complex mutations on replicative fitness in vitro.

Growth kinetics of WNV strains in vertebrate (BHK) and mosquito (CxT) cell culture are shown for both 1-step (A; MOI 10 and 8, respectively) and multi-step (B; MOI 0.01) infections. Viral growth kinetics were equivalent in BHK cells and significantly different in CxT cells (repeated measures ANOVA, p<0.001). Specifically, WNV-IC titers were significantly higher in CxT cells at both MOIs relative to all mutant strains (ANOVA, Tukey post-test, p<0.05).

Fig 7. Specific infectivity of WNV replication complex mutants.

WNV RNA (viral particles) and infectious virus (pfu) were quantified following 1-step growth on mosquito (CxT) or vertebrate (BHK) cell culture and specific infectivity (particle/pfu) was compared among WNV-IC, WNV V793I/G806R and WNV T248I. Graphs represent means +/- standard deviation. Significantly decreased infectivity (*) was measured for WNV T248I relative to WNV-IC and WNV V793I/G806R (t-test, df = 4, p = 0.0028).