Extracting hydrophobic free energies from experimental data: relationship to protein folding and theoretical models.

Solubility and vapor pressure measurements of hydrocarbons in water are generally thought to provide estimates of the strength of the hydrophobic effect in the range 20-30 cal/(mol.A2). Our reassessment of the solubility data on the basis of new developments in solution thermodynamics suggests that the hydrophobic surface free energy for hydrocarbon solutes is 46-47 cal/(mol.A2), although the actual value depends strongly on curvature effects [Nicholls et al. (1991) Proteins (in press); Sharp et al. (1991) Science 252, 106-109]. The arguments to support such a significant increase in the estimate of the hydrophobic effect stem partly from theoretical considerations and partly from the experimental results of De Young and Dill [(1990) J. Phys. Chem. 94, 801-809] on benzene partition between water and alkane solvents. Previous estimates of the hydrophobic effect derive from an analysis of solute partition data, which does not fully account for changes in volume entropy. We show here how the ideal gas equations, combined with experimental molar volumes, can account for such changes. Revised solubility scales for the 20 amino acids, based on cyclohexane to water and octanol to water transfer energies, are derived. The agreement between these scales, particularly the octanol scale, and mutant protein stability measurements from Kellis et al. [(1989) Biochemistry 28, 4914-4922] and Shortle et al. [(1990) Biochemistry 29, 8033-8041] is good. The increased strength of the hydrophobic interaction has implications for the energetics of protein folding, substrate binding, and nucleic acid base stacking and the interpretation of computer simulations.