124 / 161
Conformational Ensembles from Experimental Data
and Computer Simulations
Poster Abstracts
120
83-POS
Board 3
Solution X-ray Scattering Data Can Reveal Three-dimensional Atomic-level Structural
Diversity in Ensemble Measurements
Shenglan Qiao
, Gundolf Schenk, Derek Mendez, Sebastian Doniach.
Stanford University, Stanford, CA, USA.
Proteins are flexible molecules whose function has evolved based on their ability to sample a
variety of conformations. Methods for studying their structures and dynamics need to be capable
of probing mixtures of conformations. X-ray scattering by ensembles of particles in solution at
an x-ray free electron laser (xFEL) is well suited for this task, but extracting atomic-resolution
three-dimensional structural information is challenging. Computing angular correlations, a
natural extension of small and wide-angle scattering known as correlated x-ray scattering (CXS),
may be used to infer occupancies in models of 3D structures. We present an approach that uses
modeling to disentangle angular correlations in solution scattering data contributed by distinct
conformations in structurally diverse ensembles; once separated, we can use each angular
correlation for refining the structure of its respective component in the ensemble. To illustrate
our approach, we describe results from both physical and simulation experiments. In a proof-of-
principle experiment, we collected solution scattering data at an xFEL from gold nanoparticles.
Analysis of angular correlations, combined with structural models supported by electron
microscopy data, successfully isolates correlated signals from two structurally distinct
populations; the CXS signals reveal 3D structural differences in the two types of co-existing
nanoparticles. We present simulations showing that, based one of the two atomic models for
beta-2 adrenergic receptor (B2AR) conformations, the molar concentration and angular
correlations of the second conformation of B2AR can be recovered from CXS data simulated
from an ensemble that contains both conformations. By demonstrating its ability to provide
atomic-level structural information of components in a mixture and their respective
concentrations, we establish CXS as a methodology for studying protein structural dynamics in
large ensembles.