Physical forces between molecules link their structure with the function of biologically important complexes. They let us predict the strength and specificity of interactions among proteins, nucleic acids, lipid bilayers, and carbohydrates. We have found that water structuring plays an unexpectedly large role in the close interaction of all biological systems so far investigated. We think that a new and powerful idea has emerged from these studies. For years we have been seeing hydration interactions between such molecules as DNA double helices, polysaccharides, proteins, lipid bilayers, interactions that grow exponentially over the last ten Ångstroms approaching contact. These are usually repulsive forces that reflect the work needed to remove solvating waters from the macromolecular surface. It is now become clear is that small molecules such as salts and neutral molecules are exponentially distributed close to the molecular surface. The water that is difficult to squeeze out between two macromolecules is also the water that is more difficult (but sometimes easier) for a small solute to enter, which has been seen with several salts in the Hofmeister series and several neutral polyhydric or zwitterionic solutes that are known to stabilize native protein structures. The magnitude of the interaction of salts correlates with their known effect on water structure, further showing the importance and ubiquity of water structuring forces on the interaction of molecules in solution. The consequence of this is a realization that small molecules can stabilize or destabilize the macromolecule like the Lilliputians in their great number can stabilize (or de-stabilize) a normal creature. We now see a way to unify the study of macromolecular stability by the many smaller ligands and other species that control them.

To distinguish the contributions of these solution components to specific and nonspecific binding, we have measured the dependencies of both the dissociation rate of specifically bound EcoRI endonuclease and the ratio of nonspecific and specific association constants on water activity, salt concentration, and pH. The specific site dissociation rate can be separated into two steps: equilibrium between nonspecific and specific binding of the enzyme to DNA and dissociation of nonspecifically bound protein. About 90 percent of osmotic dependence of the dissociation rate is due to the specific-nonspecific equilibrium. The residual osmotic sensitivity linked to the dissociation of nonspecifically bound protein depends significantly on the particular osmolyte used, indicating a change in solute accessible surface area. In contrast the dissociation of nonspecifically bound enzyme accounts for almost all the pH and salt dependencies. We observed virtually no pH dependence of the specific-nonspecific binding equilibrium measured by the competition assay. The observed weak salt sensitivity of the nonspecific-specific association constant is consistent with an osmotic, rather than electrostatic, action. The apparent lack of a dependence on viscosity suggests that the rate-limiting step in dissociation of nonspecifically bound protein is a discrete conformational change rather than a general diffusion of the protein away from the DNA. We are now measuring the osmotic dependence of the EcoRI dissociation rate from two noncognate "star" sequences and a nonspecific, "inverted" DNA sequence at very high osmotic stresses. Given that the specific site dissociation rate becomes very slow, equilibrium experiments at these high solute concentrations are difficult. Kinetic experiments take advantage of these slow rates. The number of waters linked to the dissociation rate of EcoRI from nonspecific DNA sites is consistent with the value inferred from specific sequence dissociation. Even more waters are linked to EcoRI dissociation from "star" sites at high osmotic stress. More importantly, the average number of waters sequestered by these noncognate complexes decreases with increasing osmotic stress. The extent of dehydration correlates with binding free energy of the complex. The water sequestered by DNA-protein complexes should not be considered fixed and invariant. It can be removed by applying high enough osmotic stress, but the work necessary to dehydrate complexes will naturally depend on the resulting DNA-protein contacts and structure.