Model-selection-based approach for calculating cellular multiplicity of infection during virus colonization of multi-cellular hosts.

Abstract

The cellular multiplicity of infection (MOI) is a key parameter for describing the interactions between virions and cells, predicting the dynamics of mixed-genotype infections, and understanding virus evolution. Two recent studies have reported in vivo MOI estimates for Tobacco mosaic virus (TMV) and Cauliflower mosaic virus (CaMV), using sophisticated approaches to measure the distribution of two virus variants over host cells. Although the experimental approaches were similar, the studies employed different definitions of MOI and estimation methods. Here, new model-selection-based methods for calculating MOI were developed. Seven alternative models for predicting MOI were formulated that incorporate an increasing number of parameters. For both datasets the best-supported model included spatial segregation of virus variants over time, and to a lesser extent aggregation of virus-infected cells was also implicated. Three methods for MOI estimation were then compared: the two previously reported methods and the best-supported model. For CaMV data, all three methods gave comparable results. For TMV data, the previously reported methods both predicted low MOI values (range: 1.04-1.23) over time, whereas the best-supported model predicted a wider range of MOI values (range: 1.01-2.10) and an increase in MOI over time. Model selection can therefore identify suitable alternative MOI models and suggest key mechanisms affecting the frequency of coinfected cells. For the TMV data, this leads to appreciable differences in estimated MOI values.