From Mechanistic to Statistical Modeling, Then Mechanistic Again: Systems-biology Approaches Uncover Key Host-cell Determinants of Susceptibility to Coxsackievirus B3

Viruses are important agents of disease. Though we know much about the life cycles, genetics, and biochemistry of many viruses, the link between infection and disease remains elusive. There have been studies into the importance of host genetics and immunity in determining outcomes to viral infection, but it has not been considered whether small biochemical changes in host-cell species can alter infection dynamics and resulting disease. These sorts of studies require a broad view of the biological system since viral proteins alter the cell state to favor replication while cell-signaling feedback loops prevent or counteract the virus-induced cell state. My goal is to elucidate how population-level differences in host-cell species alter viral infection and resulting disease.
My work focused on coxsackievirus B3 (CVB3), a causative agent of cardiac inflammation (myocarditis). CVB3 is one of the best studied viruses, with a clearly defined life cycle, extensive information on each of its 11 proteins, and decades of research into its role in myocarditis. Thus, CVB3 is perhaps one the best model systems for taking a systems-biology approach to studying virus–host interactions.
I began by building a computational mechanistic model for the entire CVB3 life cycle with host-cell immune responses overlaid as negative feedbacks. Using the model, we uncovered a sensitivity to the timing of the type I interferon response that is dependent on host-cell resistance to cleavage of the key innate-immune signaling protein MAVS by viral proteinases. Looking further into MAVS, we identified a polymorphism at amino acid 93 that dictates susceptibility to cleavage. We show computationally and experimentally that the polymorphism is able to modulate the severity of CVB3 infection. Thus, MAVS is one host-cell species where a small population-level difference can have a big impact on disease.
I then asked how differences in the abundances of the CVB3 receptors DAF and CAR impact susceptibility to infection. Rather than sample over an artificial range of abundances, I obtained RNA-seq data from 1489 human heart samples. However, RNA-seq is not a good substitute for protein abundances. To obtain protein-level estimates, I developed statistical models able to predict protein abundance from mRNA abundance. I predicted the paired abundances of DAF and CAR. When used in the mechanistic model for CVB3, I identified individuals with varying degrees infection severity and cases where individuals were completely resistant to infection. The results demonstrate how intrinsic differences in protein levels can have large impacts on susceptibility to infection.
This dissertation thus identifies two key host-cell determinants with population-level variability that dictate susceptibility to viral infection. The finding was only possible using a systems-biology approach that blended mechanistic and statistical modeling.