I am not very familiar with the experimental procedure of x-ray crystallography except that it involves the very delicate matter of producing crystal that contain proteins and then diffracting rays through it to get a pattern that tells us about the shape of the protein.

I am curious though when you crystallize a protein that usually stays in the cellular fluid, does it go through any conformational or size changes. For instance isn't there some pressure applied by the water that would potentially result in protein being smaller then what it might be under less pressure. Or what about hydrophobic\philic effects that play an important role in protein folding. Of course once the protein is folded it is bonded through interactions stronger than hydrophobicity so it is not that delicate. But still a complete change of surrounding environment should count as a big change, should it not? So are there any theoretical or experimental explanations as to whether the protein changes size and\or shape during crystallization? References to both experimental and theoretical work are very welcome. Although I guess experimental evidence would make more sense in this matter since usually potentials are optimized to account for the crystal structure to be the minimum of the energy so it I can't see how theoretical works could potentially help to understand this issue. I guess one way would be to take a native protein structure determined by X-ray and run it through AB initio molecular mechanics where potentials do not depend on parameters obtained from the native states of proteins on PDB database. I don`t know how theoretically sound that would be though.

  • 2
    $\begingroup$ a similar question was asked recently on chemistry.SE: chemistry.stackexchange.com/questions/48788 $\endgroup$
    – marcin
    May 16, 2016 at 12:38
  • $\begingroup$ so its expected that the volume of the protein stays the same, that it does not shrink or expand after crystallization but only side chains might be fixed in one of many confirmation? $\endgroup$
    – Sina
    May 16, 2016 at 23:11
  • 2
    $\begingroup$ The assumption is that it stays (at least somewhat) the same, otherwise the structure would not be biologically relevant. This assumption is not always true, there can be changes as a result of crystal formation/buffer/pH/etc. This is why crystal structures are often verified by targeted mutagenesis or another method like NMR. $\endgroup$
    – Luigi
    May 17, 2016 at 2:29
  • 1
    $\begingroup$ This is a very good question that's not talked about enough. I often find it hard to believe that structures from incredibly extreme solutions could possibly reflect the native protein state. In some cases, at the very least we one could argue that a structure would have very different surface protein dynamics and particularly flexible regions would be forced into non-native conformations. $\endgroup$
    – James
    May 17, 2016 at 7:46
  • $\begingroup$ I would totally agree with this comment. I am also interested in the packing density and whether if crystallization has any effect on packing density. I guess at this point one could do an AB initio molecular mechanics to see if there is any length difference for the bonds. I would assume it must have been done already thousands of times and we would have known if there were significant differences? $\endgroup$
    – Sina
    May 18, 2016 at 11:08

1 Answer 1


Protein crystals are not like crystals of more commonly found substances like salt [NaCl] or diamond [carbon only.] These materials do not include other atoms in their crystal structures. For instance, a crystal of NaCl will contain sodium ions and chloride ions. X-ray crystallography of that material will, after mathematical processing, show electron density peaks of two varieties, each easily distinguished by intensity as either a sodium ion or chloride ion. Any other electron density peaks will be few and far between, as well as clearly identifiable as an interloper.

Because most proteins have hydrophilic regions on the exterior surface of the structure, crystals of proteins actually contain a considerable fraction of water molecules within the crystal itself. These water molecules are part of the crystal because they are interacting with those hydrophilic residues on the tertiary surface of the protein, both by hydrogen bonding, in some cases, and less specific polar interactions in other cases. This is at least one of the reasons why obtaining a crystal of any random protein is not at all a routine endeavor. The large proportion of water in these crystals make them very fragile once they do form, as well as not necessarily likely to form in the first place.

If you go into the technical literature to look at the electron density maps that the more commonly diagrammed structure maps are derived from, you'll be able to actually trace water molecules surrounding the individual protein molecules.

In fact, in the days of development of protein crystallography, assigning correct specific atoms to the various electron density peaks was a decidedly non-trivial task.

Effectively, despite being a crystal, the microenvironment experienced by the protein within the crystal is very much like that in an aqueous environment. The hydrophobic interactions & hydrophilic interactions will not be very different from those in solution.

That is, the crystalline state achieved is not, in fact, "a complete change of surrounding environment" to use the phrase that you've used in your post.

  • $\begingroup$ I think your answer generally explains well why protein crystallisation can be used for structure determination in the first place. However, in order to address the main point of the question it might good to include some caveats of the method (e.g. problems with highly unstructurerd regions & the inability to capture confirmations that are less stable than others or only occur under specific conditions). $\endgroup$
    – Nicolai
    Apr 12, 2018 at 12:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .