| Our objective is to design biosensors based on green fluorescent protein (GFP) to produce a simple, inexpensive platform for disease detection and diagnosis. These biosensors are generated by the removal of one of GFP's eleven beta strands to produce a non-fluorescent truncation mutant, named "leave-one-out" (LOO-) GFP. Reintroduction of the missing strand restores fluorescence; in essence, the truncation produces a binding site which can be computationally redesigned to bind a different sequence. Challenges remain in the development of this platform: computational design has resource constraints, and the LOO-GFP scaffold has stability and solubility issues.;Computational and experimental studies are presented with the goal of improving and simplifying the design process. First, a modification to the energy function of the design algorithm is described which optimizes core packing, suggesting the creation of a tighter binding interface. Second, a new computational algorithm is presented which attempts to quantify changes in void volume morphology with the goal of understanding the relationship between core packing and binding affinity in an immobilized setting. Third, a general computational protein design approach is developed which splits a given design problem into smaller disjoint problems that can be computed independently in isolation. This new design scheme not only computes existing designs more quickly with minimal losses in optimality but also allows for the computation of designs too complex to reasonably run on the same computational resources. Lastly, an experimental study which introduced disulfide bonds into the GFP loop regions is reported. This disulfide engineering experiment was intended to use rational unfolding pathway-based hypotheses to stabilize GFP in the leave-one-out state, but eventually produced new information about its folding pathway. The final chapter of this thesis considers future directions for applications of the leave-one-out design approach and other uses for GFP in biotechnology. |
