Normal Mode Flexible Fitting of High-Resolution Structures of Biological Molecules toward SAXS Data

Abstract

We present a method to reconstruct a three-dimensional protein structure from an atomic pair distribution function derived from the scattering intensity profile from SAXS data by flexibly fitting known x-ray structures. This method uses a linear combination of low-frequency normal modes from an elastic network description of the molecule in an iterative manner to deform the structure to conform optimally to the target pair distribution function derived from SAXS data. For computational efficiency, the protein and water molecules included in the protein first hydration shell are coarse-grained. In this paper, we demonstrate the validity of our coarse-graining approach to study SAXS data. Illustrative results of our flexible fitting studies on simulated SAXS data from five different proteins are presented.

Structure modeling from small angle X-ray scattering data with elastic network normal mode analysis

“…Programs have been developed to account for the solvation layer to simulate profiles that agree to the experimental data (Svergun et al, 1995;Yang et al, 2009;Schneidman-Duhovny et al, 2010). In our recent study, we attempted fitting the simulated data from CRYSOL (Svergun et al, 1995) with and without considering the scattering from the solvation layer and found that the inclusion of solvation layer improves fitting quality (Gorba and Tama, 2010). In other studies, it has been shown that the solvation layer does not significantly change the simulated intensity profile, and thus would not affect the fitting .…”

“…Thus a good protocol to terminate the iteration process is a key for successful application. In the recent application to SAXS data, we observed that the rate of change of the scoring function is a good measure to stop the fitting process (Gorba et al, 2008;Gorba and Tama, 2010).…”

“…Connecting springs are also modeled uniformly in its original model. In our approach, thin solvent layer is included as a part of the molecule and those water molecules also create spring network with protein and other water molecules (Gorba and Tama, 2010). For further coarse-graining of large systems, a small component of the molecule (such as residues) can be represented as a pseudoatom.…”

“…Such a coarse-grained model corresponds to about $1/8 of the original number of nonhydrogen atoms, because on average each residue consists of eight non-hydrogen atoms. Consequently, to take into account water molecules, 7/8 of the water molecules that surround the protein are randomly removed so that the density of atoms of the protein and solvent shell are comparable (Gorba and Tama, 2010).…”

Solution structure of RNase P RNA

“…Theoretical attempts have been made to combine elastic NMA with fitting to SAXS data (Gorba and Tama 2010). In that case, subsequent perturbations of a coarsegrain model along normal modes are evaluated against the P(r) function to guide the overall fitting procedure.…”