(This is an extension of a post on the Native Protein Structures blog dated 12 October 2013)
Let us start our exploration with the recent paper "Crystal contacts as nature's docking solutions" by Evgeny Krissinel (2010), J. Comput. Chem., 31 (1), 133-143 . Because of Dr Krissinel's central role in the development of PISA, the software used in the Protein Data Bank to identify the Biological Unit within a crystal, we might expect this paper to cite examples confirming crystal quaternary structures being the same as solution structures. The first relevant quote in this paper is:-
"PISA (Protein Interfaces, Surfaces and Assemblies) reproduces about 90% of complex structures verified by independent (non-crystallographic) experimental studies (reference given)."
Later, referring to the same work, Krissinel says "...we were able to use only 430 PDB (Protein Databank) entries, reviewed in other studies (two references given), where structure of macromolecular complexes was thoroughly investigated using complementary experimental techniques."
Next, we need to read the cited papers. Unfortunately we find that the papers have not been quoted well. Neither of them deals with structure verification by non-crystallographic methods, apart from confirmation of the number of protein molecules in the complex.
The only other argument advanced in Krissinel's 2010 paper to support the assumption that crystal structures reflect solution structures is:-
"Because the crystalline state represents an energetically optimal arrangement of molecular units, one could expect that favourable protein interactions are preserved by crystal packing. In simple words, this means that crystals are likely to exhibit natural protein contacts, or interfaces, which are formed in protein’s native, “working” environment. This assumption is exploited in most, if not all, studies where structural aspects of PPIs (protein-protein interactions) are inferred from crystals."
As written, this is not a strong argument because it does not consider whether the energetics of protein-protein and protein-solvent interactions are different in the crystal compared to the natural state.
On the positive side, there is much wisdom in the paper. There is a recognition that natural protein-protein interactions differ in their energetic strengths and that the weaker interactions may be disrupted by crystallization. I have dealt with this quantitatively in the following PowerPoint presentation. "The effect of crystallization on protein quaternary structure" (http://f1000.com/posters/browse/summary/1090079).
If we go back to the paper “Inference of Macromolecular Assemblies from Crystalline State” by Evgeny Krissinel and Kim Henrick (2007), J. Mol. Biol. 372, 774–797 , we see a more forthright discussion of the assumption that crystal structures reflect solution structures. In their Introduction there is much discussion of the inherent difficulty of producing experimental evidence that crystal structures reflect solution quaternary structures. No experimental evidence is cited. The only justification offered for the assumption is the statement that “It is reasonable to expect that stable macromolecular complexes do not change during crystallization.” No reason for such an expectation is offered. This should be seen in the context of the lack of any scientifically sound reason, together with the fact that the readership of that journal already believed that crystallization had a negligible effect on protein structure and it was not necessary to give a reason. It was already established practice for protein scientists to consider the various arrangements seen in a crystal and to choose the one that seemed most plausible, declaring it to be the “biological unit”.
To find the source of this belief in crystal quaternary structures we have to go much further back in time.
Don Vanselow 12 October 2013
Note: Free access to preprints of these papers is available at http://www.ebi.ac.uk/msd-srv/prot_int/picite.html . The above quotations are the same in the preprints and in the final published versions. I urge readers of this page to read the the cited references themselves and to follow up on the relevant references cited therein. One of the reasons that the false belief about crystal structures has persisted is that readers of papers have not done so.
Click here to return to the Native Protein Structures blog