Arrays of hydrated protein molecules arranged in a regular, repeating manner can form a crystal which acts as a diffraction grating and scatters radiation ending up in a diffraction pattern. The diffraction pattern can be converted into electron density maps and used for three-dimensional structural fits of the protein molecule and the surrounding hydration layer. X-ray and neutron crystallography are primary, indispensable, and direct methods for determination of atomic coordinates of hydrating water molecules (Savage abd Wlodawer 1986). Functionally important water molecules generally reside on the surface or at interface positions. Their determination is possible at resolutions of at least 2 Å (Carugo 1999, Finney 1977). There are more than eighty thousand crystallographic structures deposited in the Protein Data Bank (PDB, Berman et al. 2003) containing atomic level information on water structure of molecular surfaces and interfaces. Despite the large number of PDB entries, there are limitations of crystallographic determination of the hydration structure − few of which are listed below.
1) Whereas the number of crystallographic structure refinement techniques is increasing (e.g. Afonine et al. 2013), assignation of electron density peaks to possible interface water positions is still not a routine job due to inherent mobility of water and high number of degrees of freedom (Badger 1997).
2) Quality of a solved structure depends on molecular size (Finney 1977). For small proteins assignation of electron density peaks to atomic positions is easier than it is for larger macromolecules.
3) Electron density peaks of water are generally smaller than those of the surrounding (protein) interface as measured by X-ray diffraction. Small electron density peaks are consequences of small X-ray scattering of the oxygen atom compared to the surrounding group of atoms in the interface, and very low scattering power of hydrogen atoms. Thus, X-ray crystallography of water structure is focusing on a difficult task of identification of a small effect from water hidden among large effects from the surrounding molecules (Finney 1977).
4) Protein hydration in the crystal is not the same as in solution (Halle 2004a). For small proteins, 30–40% of the solvent-accessible surface is usually buried at crystal contacts (Islam and Weaver 1990), where water molecules often mediate protein–protein interactions.
5) Assignation of electron densities to water molecules is often performed to improve the fit of data during structural refinement. Misleading identification of water sites at this stage was found to be a bad practice (Ladbury 1996).
6) Cryocrystallography used for protection of the protein molecules from damages caused by high energy synchrotron beams suffer from structural cryo-artefacts (Halle 2004b).
While crystallographic methods measure long-time-averaged occupancy of a water positions and provide direct information on hydration structure of protein surfaces, NMR detects only water molecules with residence time of the same magnitude of tumbling time of the molecule in solution (Schoenborn et al. 1995). However, NMR and related techniques can provide useful information on instantaneous time behavior of structural water molecules, such as their residence time on protein surfaces (Halle 2004a).