Tag Archives: Software

Valence screening of water in protein crystals reveals potential Na+ binding sites

Nayal M, Di Cera E

J. Mol. Biol. 1996 Feb;256(2):228-34

PMID: 8594192

Abstract

Identification of Na+ binding sites in protein crystals is complicated by comparable electron density of this monovalent cation and water. Valence calculations can predict the location of metal ion binding sites in proteins with high precision. These calculations were used to screen 332,242 water molecules in 2742 protein structures reported in the Protein Data Bank (PDB), searching for molecules with Na+/- specific valence values V(Na+) > or = 1.0 v.u., as expected for a bound Na ion. Thirty-three water molecules (<0.01% of the total) were found be have V(Na+) > or = 1.0 v.u. and to be located within 3.5 A from at least two protein oxygen atoms. These water molecules, with a high Na+ -specific valence, do not have valences specific for other cations, like Li+, K+, Mg2+ or Ca2+. They belong to nine different proteins (deoxyribonuclease I, enolase, hen egg-white lysozyme, human lysozyme, phospholipase A2, proteinase A, rubredoxin, thrombin and phage T4 lysozyme) and appear with similar coordination geometry, typically octahedral, in the same place in multiple crystal structure determinations of the same protein. In the case of thrombin, the water molecule singled out by valence calculations is, in fact, a bound Na ion as demonstrated by molecular replacement with Rb+. Valence calculations provide an accurate screening of water in protein crystals and may help identify Na+ binding sites of functional importance.

Predicting Ca(2+)-binding sites in proteins

Nayal M, Di Cera E

Proc. Natl. Acad. Sci. U.S.A. 1994 Jan;91(2):817-21

PMID: 8290605

Abstract

The coordination shell of Ca2+ ions in proteins contains almost exclusively oxygen atoms supported by an outer shell of carbon atoms. The bond-strength contribution of each ligating oxygen in the inner shell can be evaluated by using an empirical expression successfully applied in the analysis of crystals of metal oxides. The sum of such contributions closely approximates the valence of the bound cation. When a protein is embedded in a very fine grid of points and an algorithm is used to calculate the valence of each point representing a potential Ca(2+)-binding site, a typical distribution of valence values peaked around 0.4 is obtained. In 32 documented Ca(2+)-binding proteins, containing a total of 62 Ca(2+)-binding sites, a very small fraction of points in the distribution has a valence close to that of Ca2+. Only 0.06% of the points have a valence > or = 1.4. These points share the remarkable tendency to cluster around documented Ca2+ ions. A high enough value of the valence is both necessary (58 out of 62 Ca(2+)-binding sites have a valence > or = 1.4) and sufficient (87% of the grid points with a valence > or = 1.4 are within 1.0 A from a documented Ca2+ ion) to predict the location of bound Ca2+ ions. The algorithm can also be used for the analysis of other cations and predicts the location of Mg(2+)- and Na(+)-binding sites in a number of proteins. The valence is, therefore, a tool of pinpoint accuracy for locating cation-binding sites, which can also be exploited in engineering high-affinity binding sites and characterizing the linkage between structural components and functional energetics for molecular recognition of metal ions by proteins.