Use of random copolymers to determine the helix-coil stability constants of the naturally occurring amino acids

Abstract

Abstract Short-range interactions dominate in determining the conformational preferences of the amino acid residues in proteins, and a variety of procedures (based on this concept) have been developed to predict the conformational states of the residues of a protein molecule. This paper is concerned with two such conformational states, helical and non-helical (or coil) states; the relative preferences of each of the twenty naturally occurring amino acids for these two states can be expressed in terms of the Zimm-Bragg parameters s and σ. In principle, these parameters can be determined from experimental studies of the thermally-induced helix-coil transition in homopolymers of amino acids in water. However, since most homopolyamino acids are insoluble in water, or are not helical, or (if helical) do not melt between 0 and 100°C in water, resort is had to the host-guest technique in which a random copolymer of a water-soluble host residue and a small amount of a guest residue is prepared. Using the helix-coil transition curves of such a copolymer, and the values of s and σ for the host homopolymer, it is possible to compute the values of s and σ over the temperature range of 0 – 70°C for the guest residue which is, in turn, each of the twenty naturally occurring amino acids. In most cases, the random copolymers are prepared from their N-carboxyan-hydrides, using suitable blocking groups to protect otherwise-reactive side-chain functional groups. The water-soluble copolymers are checked for the absence of racemization, α → β shifts, etc., and for the required degree of randomness. For example, for methionine as a guest residue, cyanogen bromide cleavage of the polymer chain yields a series of oligopeptides which indicates that the methionine was incorporated randomly in the chain. When the resulting values of s and σ are compared to the frequencies of occurrence of helical and non-helical conformations in proteins, a good correlation is obtained in most cases. In those cases where the correlation breaks down, the discrepancy provides information about the influence of specific long-range (e.g., electrostatic) interactions on the helical preferences of the given amino acid residues. The statistical weights deduced from the one-dimensional short-range interaction model are then incorporated with medium- and long-range interactions into a model to try to predict the three-dimensional structures of globular proteins.