Spatial and kinetic modeling for targeted fusion proteins

Interferon alpha is a cytokine that has shown efficacy in treating cancer, but its use is limited due to severe side effects.  We wish to target the anti-proliferative effects of interferon to cancer cells by following the chimeric activator approach.  This consists of making a fusion protein between interferon alpha and an antibody fragment that binds to EGFRvIII, an epitope specific for glioblastoma cells, and also mutating residues in the receptor binding sites of interferon to weaken its affinity.  The cell specificity of interferon’s activity is then driven primarily by the antibody fragment binding to EGFRvIII.  We have built a model to predict the activity of a given antibody fragment-interferon fusion based on the properties of the peptide linker between the two domains and on the strength of the mutation to interferon’s binding site.  A constrained Brownian dynamics component of the model predicts the effect of the linker properties on cooperative binding kinetics.  Those results are used in an ordinary differential equation model of binding state transitions to make quantitative predictions of experimental results.  Our antibody fragment-interferon fusion construct achieves a 100-fold increase in specificity on target cells, as measured by shift in EC50 in a proliferation assay, and is well-predicted by our model.