I have gone through a simple idea that proteins have different conformations; it is a dynamic system. Moreover, whenever we do ensemble experiments, so we get an average picture. There are some questions-:

Q.1 Often protein structures are reported in PDB (protein data bank), so what do they signify? The most stable conformation or some averaged picture of all conformation or both?

Q.2 In the case of the NMR, proteins have the freedom to move/vibrate through space as they are in the solution phase, but in XRD, they are rigid. So what will be the difference if I get one structure from XRD and one from NMR?

Q.3 Why structure of some proteins has not been resolved yet? I get it that a large protein cannot be solved using NMR; it will be much complex. However, why we cannot use XRD? a) Why is the crystallisation of some protein hard? b) Will there be any other challenges even if we get the crystals of a protein?

Q.4 Investigators often use different buffers to crystalise the given protein, but how they make sure that structure that they are getting in that buffer is the "actual" structure? How they make sure that crystalisation do not affect the "actual" structure of the protein.

From "Actual" Structure, I mean the structure or set of conformations that are retained in the physiological condition (in the working conditions).


1 Answer 1


Dynamics can mean many things, and not all proteins undergo extensive conformational changes during their lifetime. Some are fairly rigid in solution.

XRD or NMR data represents an average over the conformations of a protein construct sampled under experimental conditions. A protein construct is a variant of the protein, selected to allow expression and purification, and possibly also to ensure stability and crystallization (as necessary for XRD).

NMR or XRD data is processed to derive an ensemble of geometries that are consistent with the data and satisfy other criteria, usually imposed by a protein force field or prior statistics about the frequency of geometric features, which allows scoring the quality of a geometry. Before being deposited, the ensemble undergoes quality control by the PDB, amongst other things to verify that there are not too many or extreme violations of expected protein geometric parameters (bond lengths, angles, dihedrals, vdw contacts). Deposited NMR geometries are intended as representative, not exhaustive.

Whether experimental geometries are relevant depends both on the protein and what structural (or dynamical) questions are of concern. Approximating physiological conditions improves the likelihood that the data is relevant.

Proteins under XRD have less conformational freedom, they are more "rigid" (having looked up the definition in Meriam-Webster's dictionary) than those in solution. Different XRD samples (crystals) may still capture views of significantly different conformations, and complementary techniques such as NMR can be used to fill in gaps.

Yes, bigger proteins are more difficult generally to study, but progress has been made. Some of the limits are instrumental, others with sample prep. There is no single answer (too broad). Q4 is a standard problem. You get the structure and then worry whether there are inconsistencies with what is known/expected. You can try different media and see if they generate different structures.

Why is crystallizing proteins hard? Crystallization has to happen spontaneously! The protein must be kept stable (folded) as it crystallizes, which means being restricted to a narrow range of solvent and temperature conditions. There's a lot more that can go wrong versus what needs to go right. Even as you want the protein to associate and form an ordered crystal, you don't want it to form the wrong kind of interactions resulting in a disorderly aggregate.

  • $\begingroup$ Sorry for the mistake I meant why structure of "some" protein has not been resolved yet even though they are not very big. They don't crystallize, why? $\endgroup$ Commented Nov 25, 2019 at 19:58
  • $\begingroup$ @AdityaShrivastav Added an explanation to my answer. $\endgroup$
    – Buck Thorn
    Commented Nov 25, 2019 at 20:31

Not the answer you're looking for? Browse other questions tagged or ask your own question.