Genome sequence analysis of in vitro and in vivo phenotypes of Bunyamwera and Ngari virus isolates from northern Kenya

Abstract

Biological phenotypes of tri-segmented arboviruses display characteristics that map to mutation/s in the S, M or L segments of the genome. Plaque variants have been characterized for other viruses displaying varied phenotypes including attenuation in growth and/or pathogenesis. In order to characterize variants of Bunyamwera and Ngari viruses, we isolated individual plaque size variants; small plaque (SP) and large plaque (LP) and determined in vitro growth properties and in vivo pathogenesis in suckling mice. We performed gene sequencing to identify mutations that may be responsible for the observed phenotype. The LP generally replicated faster than the SP and the difference in growth rate was more pronounced in Bunyamwera virus isolates. Ngari virus isolates were more conserved with few point mutations compared to Bunyamwera virus isolates which displayed mutations in all three genome segments but majority were silent mutations. Contrary to expectation, the SP of Bunyamwera virus killed suckling mice significantly earlier than the LP. The LP attenuation may probably be due to a non-synonymous substitution (T858I) that mapped within the active site of the L protein. In this study, we identify natural mutations whose exact role in growth and pathogenesis need to be determined through site directed mutagenesis studies.

Figure 1. Photograph of virus infected Vero cell monolayer showing plaque phenotypes based on plaque size.

Vero cell monolayers were fixed in 0.5% crystal violet solution. Plaque size was measured using a Zeiss microscope.

Figure 2. Growth kinetics of wild type parental and amplified Plaque purified phenotypes of A–B) Bunyamwera and C–E) Ngari virus isolates.

The viral isolates including the parental WT, i.e. mixture of SP and LP, were used to infect 90% confluent monolayers of Vero cells at a multiplicity of infection of 0.01. Aliquots of tissue culture fluid were collected at different timepoints and titers determined by plaque assay. The experiment was replicated thrice. The statistical package R was used for fitting exponential growth data using the Kruskal–Wallis test. The detection of correlated error structure in the growth curve data was carried out as follows; the log-transformed data was fit to linear mixed effects models using R, and an AR1 model was determined to fit the data better than a repeated measures model.

Figure 3. Survival curves of mice following infection with wild type parental and amplified plaque purified phenotypes of A)Bunyamwera (GSA/S4/11232) and B)Ngari (TND/S1/19801) virus isolates.

Groups of mice (n = 12) were inoculated with 10⁹ PFU/ml of virus and observed for signs of clinical illness. The experiment was replicated three times. Survival functions were graphed for the two sets of viruses. Pairwise comparisons of survival curves were made using the Wilcoxon-Breslow test to test for equality of survivor functions.