Local dynamics of proteins and DNA evaluated from crystallographic B factors.

Bohdan Schneider,Alexandre G de Brevern,Jean Christophe Gelly,Jiří Černý

doi:10.1107/s1399004714014631

Abstract

The dynamics of protein and nucleic acid structures is as important as their average static picture. The local molecular dynamics concealed in diffraction images is expressed as so-called B factors. To find out how the crystal-derived B factors represent the dynamic behaviour of atoms and residues of proteins and DNA in their complexes, the distributions of scaled B factors from a carefully curated data set of over 700 protein-DNA crystal structures were analyzed [Schneider et al. (2014), Nucleic Acids Res. 42, 3381-3394]. Amino acids and nucleotides were categorized based on their molecular neighbourhood as solvent-accessible, solvent-inaccessible (i.e. forming the protein core) or lying at protein-protein or protein-DNA interfaces; the backbone and side-chain atoms were analyzed separately. The B factors of two types of crystal-ordered water molecules were also analyzed. The analysis confirmed several expected features of protein and DNA dynamics, but also revealed surprising facts. Solvent-accessible amino acids have B factors that are larger than those of residues at the biomolecular interfaces, and core-forming amino acids are the most restricted in their movement. A unique feature of the latter group is that their side-chain and backbone atoms are restricted in their movement to the same extent; in all other amino-acid groups the side chains are more floppy than the backbone. The low values of the B factors of water molecules bridging proteins with DNA and the very large fluctuations of DNA phosphates are surprising. The features discriminating different types of residues are less pronounced in structures with lower crystallographic resolution. Some of the observed trends are likely to be the consequence of improper refinement protocols that may need to be rectified.

Highlights

Crystallographic B factors (Trueblood et al, 1996; Rupp, 2009) represent the uncertainty in atom positions in the refined model that results from the superposition of atomic vibrations and crystallographic disorder
An analysis of scaled B-factor distributions in over 700 crystal structures of protein–DNA complexes showed that the dynamics of biopolymer residues, amino acids and nucleotides, as well as ordered water molecules, is first of all a function of their neighbourhood: amino acids in the interior of proteins have the tightest distribution of their displacements, residues forming the biopolymer interfaces have an intermediate distribution and residues exposed to the solvent have the widest distribution (Fig. 1)
The lowest B factors and a relatively low variance of their distributions was observed for buried amino acids, and these residues have another property that distinguishes them from all others: their backbone and side chains show virtually identical distributions

Summary

Introduction

Crystallographic B factors (Trueblood et al, 1996; Rupp, 2009) represent the uncertainty in atom positions in the refined model that results from the superposition of atomic vibrations and crystallographic disorder. The definition of B factors implies that they are a measure of local atomic movements. To determine whether and to what extent they can be used as a realistic gauge of the local molecular dynamics at the subnanometre scale, we decided to analyze their distributions in over 700 crystal structures of protein–DNA complexes selected from a larger well curated. Received 11 March 2014 Accepted 20 June 2014 doi:10.1107/S1399004714014631 2413 research papers data set previously used for the analysis of the structural properties of protein–DNA complexes (Schneider et al, 2014). A simple statistical analysis of B-factor distributions in these structures confirmed several expected features of the local dynamics of proteins and DNA, and revealed some surprising observations. To the best of our knowledge, a comprehensive bioinformatics analysis of the behaviour of B factors based on a large and structurally diverse ensemble of hundreds of structures at a wide range of crystallographic resolutions has not been performed as yet, and we believe that the present bioinformatics study offers some generally valid conclusions about the behaviour of B factors and its relationship to biomolecular dynamics

Methods

Results

Conclusion