Library approaches to biophysical problems

The sequence of a protein specifies its three-dimensional structure, but different sequences can specify the same fold, and the stabilities of these variants can differ vastly. What is the basis of protein stability? What are the sequence determinants of native-like thermodynamic properties? Nearly every biological process, from metabolism to signal transduction, to the structural integrity of the cell, involves proteins. Yet our understanding of the fundamental forces that stabilize structurally diverse proteins still amounts mainly to rules-of-thumb, derived from a relatively small number of mutants of a relatively small number of well-studied proteins. The answers to these questions are critical for a panoply of reasons: many common diseases are the result of protein mutations that lead to instability; many therapeutically or industrially interesting proteins are insufficiently stable for administration or process conditions; the identification of ever more sequences from genomics efforts makes the prediction of structure from sequence increasingly important; and, despite a few recent successes, protein design is still in its infancy, and even fairly successful computational approaches generally neglect protein backbone motion and employ simple potential functions of which the validity is difficult to assess. In fact, there is no generally valid way to calculate the stability of a protein, or even to quantitatively predict the effects of point mutation. The problem of protein stability has been of interest at least since the first three-dimensional structures of proteins were determined. Why, then, after all this time, do we not have a rigorous understanding of how sequence leads to structure and stability? The answer is, at root, that protein sequence space is unimaginably vast, and that our exploration of that space is vanishingly small. Even the smallest model proteins, such as ubiquitin or rop, with about 50 residues, are single points in a 20 50-point sequence space matrix – a collection that would have the mass of a trillion suns if each sequence were represented only once. Even if we restrict the problem considerably (for example, to all the hydrophobic core variants of a given small model protein) we are still faced with the biophysical analysis of billions of variants, and methods such as CD, NMR, calorimetry and X-ray crystallography are not suited to handle these kinds of numbers. Of course, protein biophysicists are not the only scientists that face nature's bewildering diversity. Screens and selections applied to large collections (ÔlibrariesÕ) of variants of organisms, proteins, nucleic acids, small …