Realization of novel molecular function requires the ability to alter molecular complex formation. Enzymatic function can be altered by changing enzyme-substrate interactions via modification of an enzyme’s active site. A redesigned enzyme may either perform a novel reaction on its native substrates or its native reaction on novel substrates. A number of computational approaches have been developed to address the combinatorial nature of the protein redesign problem. These approaches typically search for the global minimum energy conformation among an exponential number of protein conformations.

We developed a novel algorithm for protein redesign, which combines a statistical mechanics-derived ensemble-based approach to computing the binding constant with the speed and completeness of a branch-and-bound pruning algorithm. In addition, we developed an efficient deterministic approximation algorithm, capable of approximating our scoring function to arbitrary precision. In practice, the approximation algorithm decreases the execution time of the mutation search by a factor of ten.

For the Non-Biologist: A protein is a linear chain of amino-acid building blocks. There are twenty naturally occurring amino-acids and the sequence of amino-acids present in a protein specifies its unique structure. The goal of computational protein redesign is for the computer to suggest small changes to protein structure which are capable of changing a desired structural or functional property of the target protein. During protein redesign, we consider the structural and functional effects of switching a small number of amino-acids from their natural type to a different (mutated) type. Each mutation is scored and the best achievable mutations are enumerated for experimental testing. Our group designs algorithms for protein redesign that are biophysically accurate yet efficient to compute. We are interested in both developing these algorithms and applying them to interesting biological systems such as biosensors.

Students: Maria Mirza
Collaborators: Bruce Donald (Duke U.)

In the next phase of our protein redesign work, we are interested in improving both the accuracy and speed of current methods and in extending the domains into which algorithms for protein redesign might be applied. For example, our collaborator Professor Kevin Truong is an expert in the design of fluorescent biosensors. Fluorescent biosensors measure the energy transfer between a donor and an acceptor fluorophore (figure right). This phenomenon is referred to as FRET (Fluoresence Resonance Energy Transfer). By attaching the donor and acceptor to a third protein (one capable of binding the desired target ligand), it becomes possible to detect changes in the concentration of a target molecule. FRET efficiency is a function of the relative positioning of the donor and acceptor fluorophores. Using protein redesign to optimize the relative positions of the donor and acceptor fluorophores we are exploring the possibility of creating a suite of generic fluorescent biosensors and biosensor templates. Biologists would be able to request on-demand redesigns of specific sensors. These biosensors would be useful both in molecular biology labs where researchers measure changes in concentration of various electrolytes and metabolites as well as in the clinic where a new generation of fluorescent bio-sentinels could monitor for changes in patient status. These projects pose several difficult algorithmic challenges, including scalability, linker design, and scoring of FRET efficiency.

In a more general context, the methods we design for scoring protein-ligand interactions in protein redesign can be applied to modeling other protein-ligand interactions. Our algorithms for efficiently computing ensemble-based measures of protein-ligand affinity will be adopted for use in our Structure-Based Drug Design project.

The following is a bullet-point summary of our previous work on protein redesign. For additional details please see our protein redesign publications.
  • Developed Restricted Dead-End Elimination (rDEE), a method for efficient search given the constraint of a bounded number of residue mutations.
  • Designed and implemented K*, the first ensemble-based algorithm for protein redesign.
  • Development of epsilon-approximation algorithms for evaluating individual partition functions and K* that are capable of pruning the vast majority of conformations from more computationally expensive consideration. These algorithms therefore reduce execution time and make the mutation search computationally feasible.
  • The use of K* to reproduce known adenylation domain binding experiments.
  • The use of K* to predict novel mutation sequences capable of switching substrate specificity of the phenylalanine adenylation domain of the Non-Ribosomal Peptide Synthetase (NRPS) Gramicidin Synthetase A protein.
  • Confirmation of the K* method by the creation of predicted protein mutants in the wetlab and testing of their binding specificity by fluorescence quenching binding assays.
  • Developed and implemented MinDEE an extension of the Dead-End Elimination (DEE) rotamer pruning framework for identifying the global minimum energy conformation (GMEC). MinDEE provides provable guarantees on the correctness of pruned residues when utilized with energy minimization. This extended DEE framework can be used with an A* conformation search algorithm to efficiently compute partition function and K* values.
  • Developed a hybrid MinDEE/K*/A* search for protein design.
  • Protein redesign software is available upon request.

Current Projects

Learn more about each project by clicking through to their project pages.