The Limit of Accuracy of Protein Modeling: Influence of Crystal Packing on Protein Structure

https://doi.org/10.1016/j.jmb.2005.05.066Get rights and content

The size of the protein database (PDB) makes it now feasible to arrive at statistical conclusions regarding structural effects of crystal packing. These effects are relevant for setting upper practical limits of accuracy on protein modeling. Proteins whose crystals have more than one molecule in the asymmetric unit or whose structures were determined at least twice by X-ray crystallography were paired and their differences analyzed. We demonstrate a clear influence of crystal environment on protein structure, including backbone conformations, hinge-like motions and side-chain conformations. The positions of surface water molecules tend to be variable in different crystal environments while those of ligands are not. Structures determined by independent groups vary more than structures determined by the same authors. The use of different refinement methods is a major source for this effect. Our pair-wise analysis derives a practical limit to the accuracy of protein modeling. For different crystal forms, the limit of accuracy (Cα, root-mean-square deviation (RMSD)) is ∼0.8 Å for the entire protein, which includes ∼0.3 Å due to crystal packing. For organized secondary elements, the upper limit of Cα RMSD is 0.5–0.6 Å while for loops or protein surface it reaches 1.0 Å. Twenty percent of exposed side- chains exhibit different χ1+2 conformations with approximately half of the effect also resulting from crystal packing. A web based tool for analysis and graphic presentation of surface areas of crystal contacts is available (http://ligin.weizmann.ac.il/cryco).

Introduction

The densely packed environment of globular proteins in crystal structures is likely to affect protein structure. Crystal contacts bury a significant portion of the solvent accessible surface of a protein1, 2, 3, 4 and this might induce structural changes. What are the characteristics and the extent of such changes? A direct way to approach this question would be to compare X-ray structures and those modeled by NMR (see, for example Smith et al.5). Unfortunately, there are relatively few examples of proteins resolved by both methods, nor is it yet clear how best to compare a multitude of NMR models with a single X-ray model. An indirect way is to analyze differences between structures of the same protein in different crystal environments (i.e. when the arrangement of molecules in the crystal lattice is different). This has been done for several small datasets6, 7, 8 or for several specific proteins.4, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 In some of these, different crystal packing resulted in rigid body motion of large structural units9, 17 or loop conformational changes.18 In most others, crystal packing had a role in local structural differences. Sometimes, the structural difference was much larger than that originating from point mutation.11

The few database studies carried out regarding the influence of crystal packing on protein structure deal mainly with side-chain conformation. Bower et al.20 examined the side-chain angle differences between lysozyme crystallized in different space groups in order to derive an upper limit for prediction accuracy of side-chains in modeling programs. They found that about 25% of the lysozyme residues differ in χ1 by >40° in different crystal environments, and about 40% differ in χ1 or χ2. Jacobson et al.21 found that side-chains with close intermolecular contacts tend to have different conformations more often if their crystal environment is different. This might result in an underestimation of predictive accuracy, especially for surface residues. Moreover, it is not clear if regions of protein structures that participate in intermolecular crystal contacts are less or more mobile than other surface regions, since it was variously claimed that the average B-factor is larger in the region of contacts11 as well as the opposite.2

The growth of the Protein Data Bank (PDB) database makes it now feasible to arrive at statistical conclusions regarding structural effects of crystal packing. Here, we analyze proteins whose crystals have more than one molecule in the asymmetric unit or whose structures were determined at least twice by X-ray crystallography. The comparison of different structures of the same protein, in identical or different structural environments, is the main tool available for examining the amount of structural variability associated with crystal packing. However, a major problem is that in almost all cases there is some degree of dependence between structures resolved more than once. The potential limits of accuracy of structure prediction as evaluated by crystal structure comparisons is discussed.

Section snippets

Results

A visual example of the crystal packing phenomenon we are addressing is illustrated in Figure 1 using basic fibroblast growth factor as a case in point. At least two different crystal forms of this protein exist.22, 23 In these two crystal forms, different regions of the protein surface are shown to be involved in crystal contact (Figure 1(a)). In fact, in the case of pancreatic ribonuclease and its six crystal forms, it was found that almost any surface residue can be involved in crystal

Crystal packing effects

The results of our database study show that, on average, crystal environment influences protein structure. A variety of parameters not previously examined at a statistical level were used. These include: local backbone and hinge-like motions, side-chain flexibility, B′ values, positional conservation of associated water and ligands. The variability between proteins resolved in different crystal forms was clearly higher than between those having the same form. However, despite the different

Datasets

Datasets for this study were created by first extracting all pairs of PDB36, 37 chains identical in sequence (February 2004 version) and whose structures were determined by X-ray crystallography to a resolution equal or better than 2.5 Å. Pairs were then separated into those whose partners share, or do not share, the same structural environment (determined from the space group, unit cell dimensions and atomic contacts between the molecules27).

We derived two lists of pairs having different

Acknowledgements

We thank Dr Zippora Shakked for valuable discussions.

References (47)

  • F.K. Bernstein et al.

    The Protein Data Bank: a computer-based archival file for macromolecular structures

    J. Mol. Biol.

    (1977)
  • A.G. Murzin et al.

    SCOP: a structrural classification of proteins database for the investigation of sequences and structures

    J. Mol. Biol.

    (1995)
  • R.A. Sayle et al.

    RASMOL: biomolecular graphics for all

    Trends Biochem. Sci.

    (1995)
  • J. Janin et al.

    Protein–protein interaction at crystal contacts

    Proteins: Struct. Funct. Genet.

    (1995)
  • O. Carugo et al.

    Protein–protein crystal-packing contacts

    Protein Sci.

    (1997)
  • A. Kuglstatter et al.

    Induced structural changes of 7SL RNA during the assembly of human signal recognition particle

    Nature Struct. Biol.

    (2002)
  • L.J. Smith et al.

    Comparison of four independency determined structures of human recombinant interleukin-4

    Nature Struct. Biol.

    (1994)
  • C. Chothia et al.

    The relation between the divergence of sequence and structure in proteins

    EMBO J.

    (1986)
  • T.P. Flores et al.

    Comparison of conformational characteristics in structurally similar protein pairs

    Protein Sci.

    (1993)
  • M.J. Betts et al.

    An analysis of conformational changes on protein-protein association: implications for predictive docking

    Protein Eng.

    (1999)
  • A.A. Kossiakoff et al.

    Variability of conformations at crystal contacts in BPTI represent true low-energy structures: correspondence among lattice packing and molecular dynamics structures

    Proteins: Struct. Funct. Genet.

    (1992)
  • K.V.R. Kishan et al.

    Comparison of the structures and the crystal contacts of trypanosomal triosephosphate isomerase in four different crystal forms

    Protein Sci.

    (1994)
  • V. Cody et al.

    Comparison of two independent crystal structures of human dihydrofolate reductase ternary complexes reduced with nicotinamide adenine dinucleotide phosphate and the very tight-binding inhibitor PT523

    Biochemistry

    (1997)
  • Cited by (0)

    View full text