Protein-protein interactions are essential to almost all biological processes. Engineered proteins with novel binding properties are important tools for cellular and molecular research and can be used as therapeutic agents in favorable cases. The objective of this research is to develop and test computational methods for designing new protein-protein interactions. This is a challenging problem for many reasons: given two proteins it may not be clear how they should be docked to promote binding, many proteins undergo side chain and backbone rearrangement upon binding, and large free energies of desolvation must be overcome by designing favorable interactions across the interface. To address these issues, we will test three design strategies that make use of the Rosetta molecular modeling program and are based on structural features observed in naturally occurring protein interactions.
In Aim 1, we will design complexes that are mediated by interacting -strands. We will choose scaffold proteins with solvent-exposed -strands and use them for either homodimer or heterodimer design. Hydrogen bonding between the edge strands of the two proteins will establish the relative positioning of the proteins and compensate for desolvation energies. Sequence optimization and backbone refinement of residues surrounding the interacting strands will be used to further stabilize the interaction.
In Aim 2, we will use metal binding to template protein-protein interactions. Pairs of amino acids that form one-half of a zinc-binding site will be built onto the surface of one (to design homodimers) or two proteins (to design heterodimers). The proteins will then be docked against each other to form the metal binding site and the surrounding residues will be redesigned to form favorable interactions across the interface. Metal binding will provide both affinity and specificity to the target interaction.
In Aim 3, we will examine loop- mediated interactions. Surface loops on the fibronectin domain will be redesigned to bind target proteins. Iterative optimization of loop conformation and sequence will be used to search for low energy sequence/structure pairs that form favorable interactions with the partner. To lower the complexity of this problem, we will first consider cases where one of the fibronectin loops is not designed from scratch, but rather is based on sequences that are already known to bind the target protein. For all three aims we will use biophysical binding measurements, site-directed mutagenesis and high-resolution structure determination (NMR or X-ray) to evaluate the computational predictions. By pursuing this project we will extend the capabilities of computational protein design and test our understanding of the primary determinants of affinity and specificity at protein-protein interfaces.
We are developing computer-based methods for designing protein-protein interactions. Protein-protein interactions are central to many biological processes, and the ability to create interactions from scratch would allow one to develop therapeutic agents as well as novel and critical tools for basic cell and molecular research.
Showing the most recent 10 out of 68 publications
