Mutagenesis Scanning Uncovers Evolutionary Constraints on Tobacco Etch Potyvirus Membrane-Associated 6K2 Protein

Abstract

RNA virus high mutation rate is a double-edged sword. At the one side, most mutations jeopardize proteins functions; at the other side, mutations are needed to fuel adaptation. The relevant question then is the ratio between beneficial and deleterious mutations. To evaluate this ratio, we created a mutant library of the 6K2 gene of tobacco etch potyvirus that contains every possible single-nucleotide substitution. 6K2 protein anchors the virus replication complex to the network of endoplasmic reticulum membranes. The library was inoculated into the natural host Nicotiana tabacum, allowing competition among all these mutants and selection of those that are potentially viable. We identified 11 nonsynonymous mutations that remain in the viral population at measurable frequencies and evaluated their fitness. Some had fitness values higher than the wild-type and some were deleterious. The effect of these mutations in the structure, transmembrane properties, and function of 6K2 was evaluated in silico. In parallel, the effect of these mutations in infectivity, virus accumulation, symptoms development, and subcellular localization was evaluated in the natural host. The α-helix H1 in the N-terminal part of 6K2 turned out to be under purifying selection, while most observed mutations affect the link between transmembrane α-helices H2 and H3, fusing them into a longer helix and increasing its rigidity. In general, these changes are associated with higher within-host fitness and development of milder or no symptoms. This finding suggests that in nature selection upon 6K2 may result from a tradeoff between within-host accumulation and severity of symptoms.

Keywords: Potyvirus; TEV; bulk selection; mutagenesis; virulence; virus fitness.

Fig. 1.

—Average per site variability, measured as Shannon’s entropy, for the variants library and the WT TEV population.

Fig. 2.

—Comparison of the predicted ternary structures for the WT (in gold) and the mutant (in blue) 6K2 peptides. As measures of structural similarity, the R_dis and dRMS values are reported. The upper most left panel shows the predicted structure of the WT 6K2 sequence; the three α-helices and other relevant motifs are marked with boxes.

Fig. 3.

—In silico evaluation of the potential functional effect of every possible amino acid replacement on each residue of 6K2. Columns represent the 6K2 residues (indicated in the top) and rows the possible changes. Hot colors (red) represent strong functional effects, whereas cold colors (blue) represent neutral changes. Black squares represent no amino acid change. Mutations studies in this work are indicated with asterisks. The table below indicates the scores for each of the 11 6K2 mutants studied. Amino acids involved in α-helices H1, H2, and H3 are indicated with boxes.

Fig. 4.

—Confocal microscopy imaging of Nicotiana benthamiana leaf epidermal cells expressing 6K2::YFP. Each image panel shows the localization of the WT protein and its variants (1–13). The position of nuclei is indicated with an arrow. Red objects are chloroplasts. White scale bar represents 10 μm.

Fig. 5.

—Phenotypic characterization of the different TEV 6K2 mutants. (A) Mean time to symptoms development estimated from the Kaplan–Meier regression. For those genotypes that did not showed symptoms at the end of the experiment, 20 dpi represents the lower bound of the estimated mean time (upper bound being +∞). (B) Mean infectivity 20 dpi (n = 10 plants inoculated). (C) Infectious viral load estimated by means of Chenopodium quinoa local-lesion assay method. In all cases, the dashed vertical line corresponds to the mean phenotypic value of the WT TEV. Mutants are ordered in the ordinate axis to better illustrate the statistically homogeneous groups.

Fig. 6.

—Radar plot summarizing all the structural and phenotypic variables estimated for each 6K2 mutant. Trait values have been normalized to a zero to one scale. In all plots, the WT values are included in brown as reference. The three phenotypic traits (infectivity i, viral load LFU, and mean time to symptoms ST₅₀) are placed at the left side of the heptagon (highlighted in green in the first panel). For those genotypes that did not showed symptoms at the end of the experiment, ST₅₀ → +∞. The four structural traits (SNAP2, transmembrane score TMHMM, dRMS, and the normalized structural similarity index R_dis) are placed at the right side of the heptagon (highlighted in gray in the first panel). Nonviable mutants are indicated in red.