Hunting water in the lipocalin drylands

Although interactions with water play a significant role in determining protein structure and function, the precise behavior of water within any protein structural ensemble can be quite difficult to pin down. Despite their importance, these interactions are often transient, making them difficult to pin down. NMR and crystallography have had some success in addressing structural waters, but even these techniques can have substantial difficulty proving that the data genuinely represent proximal interactions with water rather than long-range polarization transfer or some other artifact. An indirect approach is often required in order to resolve these kinds of questions.

Consider the case of β-lactoglobulin (BLG), a member of the lipocalin family. An image of the protein displaying the solvent-accessible surface area, shamelessly stolen from (1), is at right (explore this structure at the PDB). BLG binds nonpolar ligands in that long cavity in the middle (the calyx), which the crystal structure identified as containing several water molecules in a small change (blue spheres). In addition, there were two long-resident structural waters present in other cavities in the protein (red spheres). Qvist et al. wish to establish whether these water molecules are, in fact, present in the calyx in solution.

To do this they use a kind of relaxation-dispersion experiment in which they measure the longitudinal relaxation rate (R₁) of various atoms in water. Dispersion is achieved by using different external fields. Across the frequency range used in this study, the R₁ profile of bulk water is flat due to its very short rotational correlation time. Adding the protein alters the dispersion profile because much of the water becomes part of a "hydration shell" for the protein molecules. In addition, the observed profile will be altered by any waters that are internally bound to the protein. As a result, one would expect that the addition of palmitate, which binds in the calyx (and thus displaces those 5 structural waters) would significantly alter this profile.

In fact this is not what happens. There are slight differences between the profiles when palmitate is added (the authors use NMR to ensure that the palmitate was actually bound), but the results rule out the possibility of water in the pocket with a residence time >10^-8 s. This is inconsistent with the presence of 5 ordered waters in the pocket. In order to support this result, the authors perform molecular dynamics simulations and free energy calculations. From these it is evident that the observed differences in relaxation dispersion could not be due to organized water in the calyx. The authors calculate that burying the waters in the nonpolar calyx would come at a cost of 71 kcal/mol, too much to make such a thing believable even if they were off by an order of magnitude.

So, what was in the calyx in the structure? The authors suggest that some kind of small organic molecule or mixture of such molecules might have been responsible for the observed electron density. This is not an indictment of the structure generally, however: the present study finds solid evidence for the presence of the two structural waters identified elsewhere in the protein. Additionally, Qvist et al. found evidence in the dispersion profiles for additional structural waters, which they thought might be located at the dimer interface. This is consistent with the predictions of a crystal structure of the BLG dimer.

All of this might sound hopelessly arcane, but because it usually must be displaced, the presence of water in a binding pocket has significant consequences for the energetics and kinetics of ligand binding. Understanding how proteins arrange these kinds of binding sites, and what kind of moieties can penetrate them, will be important in designing new proteins, and developing drugs to alter the binding properties of proteins found in nature.

1. Qvist, J., Davidovic, M., Hamelberg, D., Halle, B. (2008). A dry ligand-binding cavity in a solvated protein. Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0709844105