Macromolecular Modeling Blog ™

PLOS Computational Biology. 2008 Sep 26;4(9)

Close atomic packing is an important metric for characterizing protein structures. While the average value of packing density in the protein interior is close to 0.75, it may not be uniform over the whole structure, the density varying in the range 0.66 to 0.84 (Richards et al.,Gerstein et al.,Fleming et al.). The localized defects in packing show up as cavities (Connolly), and when present they can reduce the stability of the structure (Lee et al.). Although the surfaces that form the interface in protein-protein interaction have complementary shape, an issue that has not been addressed is whether the interface can harbor cavities, and their features relative to those present in the protein interior.

The analysis was conducted on a data set of 97 monomeric proteins, 122 homo-dimers (obligate interfaces) and 183 hetero-dimers (non-obligate interfaces), mostly determined to a resolution of 2.5 Å or better. CASTp was used to provide a full description of the protein pockets and cavities, including volume, surface area, protein atoms that line the concavity, etc. 

These are the main findings of the analysis:

* The total volume of cavities increases with the size of the protein (or the interface), though the exact relationship may vary in different cases. The authors could derive a linear relationship between the total volume of the cavities present and the total protein volume. The observed minimum and maximum values were 0.06 and 2.26%, respectively. Cavities are usually located close to the protein surface—considering the cavity lining atoms, most of them belong to the surface of the molecule.

* The larger cavities tend to be less spherical, solvated, and the interfaces are enriched in these. For a comparable number of atoms the interface has about twice the number of cavities relative to the tertiary structure. Cavities were found to cover 10% of a typical interface (Hubbard et al.). Comparing the number of cavity lining atoms to the total this analysis found 5.5% atoms of the tertiary structure and 13.8 and 10.5% of homodimeric and hetercomplex interfaces form cavities. 

* On average 15Å cubed of cavity volume is found to accommodate single water, however, each additional water requires an extra volume of ~40–45Å. The average number of hydrogen bonds involving water in the solvated cavities is 2.6 with protein atoms, and 3.4 if hydrogen bonding with other water molecules is also included—matches with the typical value of 3 hydrogen bonds made by a buried water molecule reported in literature. Polar atoms/residues have a higher propensity to line solvated cavities. 

* Relative to the frequency of occurrence in the whole structure (or interface), residues in beta-strands are found more often lining the cavities, and those in turn and loop the least. The higher involvement of beta-sheet residues in lining the cavities may have implications for the energetics of interaction. It has been observed that for protein-protein interactions, those having interfaces mostly made up of beta-sheet have, on average lower free energy of binding compared to those having beta- or alpha-beta(mixed) classes of interfaces (Guharoy and Chakrabarti, unpublished). This observation may be understood in terms of the lowest packing efficiency of interfacial beta structures, leading to lower van der Waals contacts and therefore lower binding free energies as well.

* Interface Cavities and Water: Just 37% and 51% of the water molecules in interfacial cavities, in homo-dimers and hetero-dimers, respectively, have direct hydrogen bond contact with both the subunits. A smaller number (10% and 5%) of water molecules do not form any bond with either subunit. However, if contacts are considered (instead of hydrogen bonds) made with both the sides, a greater number (72% and 84%) of water molecules bridge the two subunits.

* Cavities Containing Ligand or Cofactor Molecules: In about 30% cases only 1–3 atoms of a much larger ligand are found to be inside the cavity, which are usually < 20Å cubed in volume. These cavities cannot be considered as having a small molecule entrapped. Heterocomplex interfaces have just two cases where molecules used in the crystallization procedure found their way into the cavity. In general, biologically relevant molecules are not found in interface cavities. In the tertiary structure, there is an example of Mg ion being located in a volume of 16Å cubed; cavities having Ca ion usually have a volume in the range of 17–18Å cubed; a K ion is observed in 20Å cubed. Water molecules usually accompany the ligand in the cavity.

The authors conclude that a comprehensive understanding of the features of cavities in protein interiors and interfaces, as presented here, would facilitate protein design experiments.

By Nir London.