Molecular indetermination in the transition to error catastrophe: systematic elimination of lymphocytic choriomeningitis virus through mutagenesis does not correlate linearly with large increases in mutant spectrum complexity

Abstract

Studies with several RNA viruses have shown that enhanced mutagenesis resulted in decreases of infectivity or virus extinction, as predicted from virus entry into error catastrophe. Here we report that lymphocytic choriomeningitis virus, the prototype arenavirus, is extremely susceptible to extinction mutagenesis by the base analog 5-fluorouracil. Virus elimination was preceded by increases in complexity of the mutant spectra of treated populations. However, careful molecular comparison of the mutant spectra of several genomic segments suggests that the largest increases in mutation frequency do not predict virus extinction. Highly mutated viral genomes have escaped detection presumably because lymphocytic choriomeningitis virus replicates at or near the error threshold, and genomes in the transition toward error catastrophe may have an extremely short half-life and escape detection with state-of-the-art cloning and sequencing technologies.

Figure 1

Evolution of LCMV during serial passages in the presence of FU (A–C) and AZC (D). Virus serial passages were obtained by infection of BHK-21 monolayers with 0.01 pfu per cell of LCMV (p0, stock) in the absence (control) or presence of the indicated concentrations of nucleoside analogs (FU and AZC in μg/ml). Each supernatant was assayed for virus infectivity and used as inoculum for the following passage. (A) LCMV dose–response to FU. (B) Virus infectivity titers of the first passage in the presence of increasing concentrations of FU; each value is the mean of three separate experiments. (C) Recovery of LCMV infectivity after mutagenesis with 100 μg/ml FU. Blue lines indicate virus passages in medium containing 100 μg/ml FU; red lines represent incubations without mutagen (control). (D) Evolution of LCMV in the presence of AZC. Multiplicity of infection in series 1/10 of control, 5 μg/ml AZC, and 10 μg/ml AZC was 0.001 pfu/cell per passage.

Figure 2

(A) Schematic diagram showing the positions of mutations found in the genomic regions of control and mutagenized LCMV quasispecies. Each single-stranded genomic segment encodes for two proteins in an ambisense orientation. Positions of start and end codons for each ORF as well as mutations within the L (polymerase), GP, and NP genes are given in the genomic sense. Mutations (base substitutions, deletions, or insertions) found in the analyzed virus populations (see Table 1) are represented as § for untreated virus (control), ↓ for FU-treated virus, and ▾ for AZC-treated virus, irrespective of the passage at which they were found. The hashed areas represent the sequenced genomic regions. (B) Distribution of mutations by genomic regions in LCMV control and nucleoside analog-treated populations. Numbers (n) of mutations and percentages of the total for each treatment are given.

Figure 3

Genetic heterogeneity of LCMV quasispecies replicated in the presence of mutagenic nucleoside analogs. Mutation frequencies (A and B) and Shannon entropy (C) were calculated for the indicated LCMV passages in the absence (control) or presence of FU (A and C) and AZC (B and C). The mutation frequencies (A, tables) were calculated by dividing the number of mutations found in the viral population (and not present in its consensus sequence) by the total number of nucleotides sequenced (7,000–11,000 nt). The normalized Shannon entropy (B) was calculated as −∑_i [(p_i × ln p_i)/ln N] in which p_i is the frequency of each sequence in the quasispecies and N is the total number of sequences compared. Genomic regions analyzed were: L (polymerase), NP, and GP. Passage 9 in the presence of 5 μg/ml AZC is from the 1/10 dilution series.