fragHAR: towards ab initio quantum-crystallographic X-ray structure refinement for polypeptides and proteins

Abstract

The first ab initio aspherical structure refinement against experimental X-ray structure factors for polypeptides and proteins using a fragmentation approach to break up the protein into residues and solvent, thereby speeding up quantum-crystallographic Hirshfeld atom refinement (HAR) calculations, is described. It it found that the geometric and atomic displacement parameters from the new fragHAR method are essentially unchanged from a HAR on the complete unfragmented system when tested on dipeptides, tripeptides and hexapeptides. The largest changes are for the parameters describing H atoms involved in hydrogen-bond interactions, but it is shown that these discrepancies can be removed by including the interacting fragments as a single larger fragment in the fragmentation scheme. Significant speed-ups are observed for the larger systems. Using this approach, it is possible to perform a highly parallelized HAR in reasonable times for large systems. The method has been implemented in the TONTO software.

Keywords: H atoms; Hirshfeld atom refinement; peptides; quantum crystallography.

Figure 1

The molecular fractionation with conjugate caps (MFCC) procedure for cutting a dipeptide (left) across the peptide bond (shown in black), producing two fragment molecules (right), which are then ‘capped’ with –CH₃C=O (red) and –NHCH₃ (orange) groups, comprised of the neighbour and next-neighbour non-H atoms.

Figure 2

Crystal structures (100 K) of the three peptide model systems with 50% ADP probability ellipsoids. Hydrogen bonds are shown in green. (a) Gly-Ala (GA), (b) Ala-His-Ala (AHA), (c) cyclo-(Ala)₄-(d,l-Pro)₂ (A₄P₂).

Figure 3

Bond lengths between non-H atoms from fragHAR calculations plotted against reference HAR values. Error bars are depicted, but are invisible to the eye on this scale.

Figure 4

X—H bonds (with error bars) in all model compounds for fragHAR calculations versus reference HAR values. Bonds with notable differences are marked with the corresponding H-atom name.

Figure 5

X—H bonds (with error bars) in all model compounds for fragHAR with fragments ‘joined’ across hydrogen bonds versus reference HAR calculations.

Figure 6

Hydrogen ADPs (with error bars) obtained with fragHAR versus those from HAR for the A₄P₂ system. The hydrogen-bonded H41A atom is labelled.

Figure 7

Timing of the fragHAR (green) and HAR calculations (black) for single-processor serial (square) and parallel (circles) calculations for GA (two processors), AHA (four processors) and A₄P₂ (four processors).