Welcome to the Gray Lab
About Us Research Publications People Facilities Positions Courses Links Wiki
Research: Protein Assembly and Function
Our objective is to understand biomolecular (self-) assembly and function through protein structural calculations and to apply these methods to important problems in biomolecular engineering, nanotechnology, and human health. We will develop comprehensive theory for macromolecular assembly and function including the ability to predict biomolecular behavior and to design tailored biomolecular components or nanodevices. We have three major thrusts in the lab: (1) protein-protein docking, especially focused on applications in therapeutic antibodies, (2) protein allostery, and (3) protein-surface interactions. The projects together can lead to imaginative new devices, such as composite protein sensors which can bind specific proteins or cells and then effect an allosteric conformational change causing an electron transfer into the bound solid surface of a microchip (see figure). The common themes of these projects are the interactions of proteins and the rearrangement of the protein itself in response to interactions. All of these projects help us to understand the physics and thermodynamics of protein structure. Recent advances in protein structure prediction enable these new applications as well as continuing methods and theory development. Ultimately, we strive to be able to predict and design protein assembly and function at the nanometer and atomic scale.

Protein-protein docking and therapeutic antibodies
Protein interactions are crucial for cellular phenomena, and protein-protein docking algorithms can predict the structure of a protein complex from the individual components (Gray 2006). We have developed RosettaDock, one of the leading docking algorithms as noted in the international blind challenge known as CAPRI. Our method simultaneously captures rigid-body and side-chain movements, allowing us to achieve more accurate predictions, especially at the atomic level, compared to other predictors. Several of our models have been the most accurate submitted (Daily et al. 2005).

We are currently tailoring the docking code toward the prediction of therapeutic antibodies. Therapeutic antibodies provide relief from various human diseases including cancer, arthritis, multiple sclerosis, HIV, and infection by anthrax. The structure of a therapeutic antibody in complex with its antigen can yield information about the drug and disease mechanism and allow for improved function through rational design, but structure determination by crystallography or NMR can be difficult and time consuming. Our objective is to develop and test computational protein-protein docking methods for the antibody-antigen class of interactions and to apply these structure prediction techniques to the complexes of novel therapeutic antibodies. We recently determined the structure of monoclonal antibody 806 in complex with the epidermal growth factor receptor (EGFR), and our models were confirmed by experimental measurements in the Wittrup lab. The model identifies the conformation of the EGFR which binds to the drug, and the structure explains the efficacy of the drug, the drug's specificity for cancer cells, and the molecular determinants of the binding affinity (Sivasubramanian et al. 2006). We are currently finishing a second application on an anthrax drug, and a new graduate student project is exploring flexible backbone algorithms to improve our accuracy when using homology antibody models. This work is funded by the NIH.

Protein allostery
Allosteric proteins exist in two structural states which differ in the level of function, and effector molecules modulate the equilibrium between these two states by binding to a site which is distinct from the functional site. Allosteric proteins control fundamental biological processes, and malfunction of allosteric proteins can cause metabolic and genetic diseases. However, the structural basis of protein allostery is not well understood, and it is currently difficult to rationally design therapeutic approaches to control allostery or design novel molecular switches. We have recently completed a systematic study of protein motion by curating a large set of known allosteric protein structures. We statistically characterized structural properties of the motions observed including how much of the protein they comprise, the structural environments which they prefer, and coupling relationships within these flexible regions. The motion calculations generated data which were used toward quantitatively describing allosteric transitions at the mechanistic level and revealing the structural basis of protein allostery. Most importantly, the statistical characterizations of motions suggest key insights about how protein structures organize motions to communicate between sites. For example, most allosteric proteins exhibit significant changes in average structure, approximately 20 percent, in addition to any contribution from dynamics. Furthermore, proteins use constraints like secondary structure and burial to regulate where motions happen in their structures, and proteins can propagate conformational changes through tens of Ångstroms through their structures, enough to bridge allosteric and functional sites in many proteins. This work is currently under review (Daily & Gray 2006). Using our new insights into the nature of allosteric protein conformational change, we are now developing novel protein structure prediction algorithms for allosteric proteins. Eventually we will be able to design novel switches and suggest therapeutic approaches based on allosteric intervention.
Protein-Surface Interactions
In an ambitious, high-risk/high-reward problem funded by the Beckman Foundation, we are attempting to create the first rationally designed, specific protein-surface interaction. We have created methods to capture inorganic surface-protein energetics and conformation within our Rosetta protein structure modeling program, and we have calculated preliminary designs to allow the top7 protein and lysozyme to selectively bind mica and quartz surfaces in specified molecular orientations. In collaboration with the Ostermeier lab, we have determined experimental protocols for purifying the needed proteins without contaminating His tags which would obscure our surface measurements. We are now developing interfacial assays to determine the microstructure of the adsorbed protein, and we will soon be testing our novel designs. If successful, this achievement would be useable as a basis for building supported biomolecular nanostructures.

In a related study, we are attempting to help determine the first known molecular structure of a protein adsorbed to a solid surface. In collaboration with surface scientists at the University of Washington who are measuring distance constraints using solid-state NMR, we will predict the structure of statherin on hydroxyapatite. This is an important interaction for biomaterials such as bone implants. The work is funded by the ACS-Petroleum Research Fund.