Solvents, interfaces and protein structure.

Abstract

Mean packing densities in protein interiors are comparable to those of most organic solids but the variations between small regions may be substantial. Packing defects may be related to allowed structural fluctuations. Molecular surface areas can be correlated with free energies of transfer between different solvents. The proportionality factor will depend, in general, on the nature of the solute and both the solvents. The changes in solvent-protein interfacial area on chain folding are large and the implied changes in free energy from this solvent-squeezing effect are correspondingly large. The strong tendency to minimize surface area is reflected in the globular shape of most protein molecules or domains in larger structures. The formation of isolated units of secondary structure from an extended chain represents about one half of the eventual total area change. The tendencies of amino acids to form beta-sheets correlate well with the rank-ordered list based on non-polar area change for each residue type. The calculated area changes for helix and sheet formation are not identical in rank order. The rank-ordered list for alpha-helix formation correlates satisfactorily with the probability list prepared from actual structures if glutamic acid and tyrosine are removed. What special characteristics unrelated to surface area these two amino acids might have is not clear. Tertiary structure formation from preformed secondary structural units can be rank ordered on area change and possible nucleation sites can be identified. A prediction scheme for helix-helix interactions is proposed. The hydrophobic force begins to be felt when two helices are about 0.6 nm (6 A) from their final contact positions. Interfacial surface tension is a logical parameter to relate free energy and solvent contact area, but this macroscopic parameter must be used with great caution. It is suggested that water in the deep grooves, characteristic of the active sites of many enzymes, may have a substantially higher fugacity than bulk water as indicated, at least qualitatively, by the Kelvin equation based on surface curvature. Such water would be more easily displaced than its plane surface counterpart and could contribute significantly to ligand-binding energy. This factor would be in addition to the usual solvent entropic effects associated with surface area reduction on association.