Molecular recognition is arguably the single-most-important molecular process. It is the key to the structure-specific association of a macromolecule (protein or nucleic acid) with another molecule and is the basis for a number of subcellular activities. These include protein-ligand binding, catalysis, the action of receptors, the formation and operation of mechanical structures in the cell, the generation of energy and vectorial movement of charge, and sensing. The processes involved in molecular recognition are the same as those in the folding of proteins or nucleic acids and, somewhat more loosely, in the formation of lipid bilayers. Arguably, molecular recognition is more fundamental than any other single process in the cell—that is, more fundamental than replication and translation of DNA, synthesis of proteins, or the operation of signaling networks—since it is the basis of all of them. And, astonishingly, it still is not well understood at the molecular level.
Chemistry and biophysics have led to a picture of molecular recognition, a metaphor for which is a lock and key. In this metaphor, two molecules associate when they have complementary shapes. Complementary shapes maximize van der Waals interactions and make it possible to associate complementary electrostatic charges. Much of molecular recognition is ascribed to the hydrophobic effect, which is the association of nonpolar surfaces in water. The problem with this attractively simple metaphor is that, like many metaphors, it gives a distorted picture. Close complementary fit between associating molecules may improve the enthalpy of interaction, but it is unfavorable entropically. A better metaphor is now believed to be a cow in a tent—that is, a loose fit between molecules that minimizes the Gibbs free energy of association. What is needed for good binding of molecules to one another is the right kind of sloppy fit, but the meaning of “right kind” is not clear.
The problem of molecular association has been clearly posed for 50 years but there is still no resolution. For example, it still is not possible to rationalize quantitatively the Gibbs free energy of the binding of ligand–potein pairs or to predict the structure of new ligands for a protein, even if one has detailed knowledge of the structure of the binding site. One thing that has made the problem so difficult is that while water is clearly a necessary component of molecular recognition, the role of solvent in biology is sufficiently inconvenient that it has been ignored for the most part. For example, it is incorrect to express the molecular recognition problem in terms of protein and ligand. Instead, it must be expressed in terms of protein, ligand, water, and, perhaps, other components of biological media as well.
Understanding the interactions between water and proteins is a problem that will require high-resolution structural methods; new theoretical methods, including new methods in statistical mechanics that can handle the large numbers of particles involved; and thermodynamic analysis. Each type of information will need to be supported by physical tools such as high-resolution X-ray and neutron