Conformational Ensembles from Experimental Data and Computer Simulations

Conformational Ensembles from Experimental Data and Computer Simulations

Poster Abstracts

39-POS Board 39 A Simulation-Guided Method to Select Optimal DEER Experiments to Refine Highly Flexible Conformational Ensembles Jennifer M. Hays 1 , Marissa Keiber 2 , Linda Columbus 2 , Peter M. Kasson 3,1 . 1 University of Virginia, Charlottesville, VA, USA, 3 University of Virginia, Charlottesville, VA, USA. 2 University of Virginia, Charlottesville, VA, USA, Determining the structural basis of flexible molecular recognition is experimentally challenging because many techniques that capture multiple conformational populations provide sparse rather than complete data on the conformational ensemble. Selecting a set of optimal experiments to best refine the conformational ensemble therefore remains an important challenge. The binding of Opa Neisserial virulence protein to its human host receptor (CEACAM) exemplifies these flexible recognition processes. Although Opa has long loops that have been shown by NMR to be loosely structured, these same loops still bind CEACAM with high affinity. In order to refine the Opa-CEACAM conformational ensemble, we have developed a model-free, information- theoretic approach for guiding double electron-electron resonance (DEER) experiments that 1) uses a mutual information distance metric to rank pairs of residues based on how well they refine a conformational ensemble and 2) identifies a set of highly informative pairs that perform well under this metric. Specifically, we utilize the data from initial ensemble simulations of Opa 60 to identify a set of maximally-informative and minimally-redundant (mRMR) pairs, measure the distance distributions of those pairs using DEER, incorporate the experimental distributions into restrained-ensemble MD simulations, and demonstrate that the set of high-scoring mRMR pairs better reduces the conformational search space than a set of experimentalist-selected pairs. This systematic approach provides a way to both efficiently refine flexible receptor-ligand complexes and help elucidate fundamental physical principles of receptor-ligand binding.

72