Extending the scope of coiled-coil crystal structure solution by AMPLE through improved ab initio modelling.

Jens M H Thomas,Owen R Davies,Daniel J Rigden,Ronan M Keegan

doi:10.1107/s2059798320000443

Abstract

The phase problem remains a major barrier to overcome in protein structure solution by X-ray crystallography. In recent years, new molecular-replacement approaches using ab initio models and ideal secondary-structure components have greatly contributed to the solution of novel structures in the absence of clear homologues in the PDB or experimental phasing information. This has been particularly successful for highly α-helical structures, and especially coiled-coils, in which the relatively rigid α-helices provide very useful molecular-replacement fragments. This has been seen within the program AMPLE, which uses clustered and truncated ensembles of numerous ab initio models in structure solution, and is already accomplished for α-helical and coiled-coil structures. Here, an expansion in the scope of coiled-coil structure solution by AMPLE is reported, which has been achieved through general improvements in the pipeline, the removal of tNCS correction in molecular replacement and two improved methods for ab initio modelling. Of the latter improvements, enforcing the modelling of elongated helices overcame the bias towards globular folds and provided a rapid method (equivalent to the time requirements of the existing modelling procedures in AMPLE) for enhanced solution. Further, the modelling of two-, three- and four-helical oligomeric coiled-coils, and the use of full/partial oligomers in molecular replacement, provided additional success in difficult and lower resolution cases. Together, these approaches have enabled the solution of a number of parallel/antiparallel dimeric, trimeric and tetrameric coiled-coils at resolutions as low as 3.3 Å, and have thus overcome previous limitations in AMPLE and provided a new functionality in coiled-coil structure solution at lower resolutions. These new approaches have been incorporated into a new release of AMPLE in which automated elongated monomer and oligomer modelling may be activated by selecting `coiled-coil' mode.

Highlights

The coiled-coil is perhaps the best understood protein fold, and in its ideal form constitutes a highly geometric structure that has been defined computationally (Lupas & Gruber, 2005)
Whilst we found that no single metric is sufficient in isolation, a combination of a high SHELXE CC (>30 and 5–10 above background), a low Rfree (
We find that the unique features and pathologies of each crystal form provide baseline statistics for incorrect solutions that can be markedly different for different crystals even at the same resolution

Summary

Introduction

The coiled-coil is perhaps the best understood protein fold, and in its ideal form constitutes a highly geometric structure that has been defined computationally (Lupas & Gruber, 2005). Within the classic coiled-coil model, the interhelical crossing angle provides a periodicity of seven amino acids across two turns, with a pitch along the coiled-coil axis of 5.1 Aper -helical turn (Lupas & Gruber, 2005; Hartmann, 2017; Lupas et al, 2017) Whilst this classical model proved to be correct, subsequent experimental evidence demonstrated that it represents only part of a large family of highly divergent structures. The modern definition of a coiled-coil encompasses a highly diverse family of elongated -helical structures that exhibit a wide range of geometries and topologies

Methods

Results

Conclusion