Quantitative Prediction of Therapeutic Fusion Protein Dynamics

Design tools for synthetic biological devices are not yet to the point where they can provide reliable predictions even for simple systems.  We address this problem for the case of fusion proteins engineered for therapeutic targeting.  In particular, we address fusions consisting of two well-characterized domains separated by a linker, where one domain binds to a target epitope and the second exhibits desired biological activity.  A two-component modeling approach lets us make quantitative predictions that can be checked against our experimental results to validate the success of the model.  The model is composed of a constrained Brownian dynamics (CBD) simulation component that predicts the effect of linker configuration on fusion protein activity and an ordinary differential equation component that uses details of the experimental setup, kinetic information about the protein domains and results from the CBD component to generate predicted dose-response curves.  We demonstrate application of the combined model to our therapeutic constructs via in vitro cell-based assays. We observed a linker length-dependent effect on targeting efficiency, as well as a moderate effect due to modulating the binding strength of the biologically active domain.  We also generated a two-dimensional landscape mapping binding strength to predicted targeting efficiency and observed the existence of a plateau of maximal targeting which falls off steeply at high binding strengths.  The ability to predict optimal values for targeting of both linker configuration and binding strength will allow us to rationally generate candidate therapeutic designs.