Modelling influenza virus infection dynamics in cell culture-based vaccine production

State-of-the-art human influenza vaccines are manufactured by using mammalian host cells for virus propagation. During this process, the virus utilises the replication machinery of each individual cell to synthesise viral RNA and proteins. Each virus strain shows specific replication dynamics depending on its success to control the host cell machinery. However, the propagation of infection in the host cell population also determines the overall virus yield. Understanding this multiscale process to predict a strain's production performance and to optimise yields requires a coherent mathematical model emphasising both the individual cell level and the cell population level. Here, we present a quantitative model for the infection of individual Madin Darby canine kidney (MDCK) cells, which is validated against real-time RT-qPCR data. Analysing the time course of viral RNA synthesis reveals strain-specific replication dynamics and allows testing hypotheses concerning the influenza virus reproduction cycle. By extending this single cell approach to a heterogeneous population of cells, the impact of these strain-specific differences in virus replication can be investigated with respect to the process level. With the resulting multiscale model and flow cytometric measurements, infection status and apoptosis induction in a cell population are interpreted in the context of a strain's production performance. This approach elucidates an inherent trade-off between a rapid production of viral RNAs and proteins, and an early induction of apoptosis triggered by the accumulation of these components. Such knowledge can contribute significantly to our understanding of the quantitative aspects of virus-host cell interaction and to the development of new concepts for the optimization of vaccine production