Protein folding: evaluation of some simple rules for the assembly of helices into tertiary structures with myoglobin as an example

A computer program designed to fold a peptide chain consisting solely of helical segments and connecting links of known length is described and evaluated. This study is a detailed extension of certain aspects of the earlier work of Ptitsyn &; Rashin (1975). Possible interaction sites on the helices are sequence dependent and are calculated as described by Richmond &; Richards (1978) using probable changes in solvent contact area as a guide. The helices are then paired according to the list of potential sites, with each helix being paired at least once. The lists of pairings are then examined geometrically, each site having a defined dihedral helix axis angle, a specified inter-helix axis distance, and defined rotations, when required, about each helix axis. Two simplified filters are used: (1) lengths of connecting links must be equal to or greater than the end-to-end distances of the helices; and (2) non-paired helices must not collide. With myoglobin as a test example and only six of the eight helices being considered, a conformation space consisting of more than 3 × 10⁸ structures was surveyed. The two filters reduced the acceptable structure list to 121. Slight readjustment of the parameters in the filters would have reduced this to 20 structures. Of these 20, one closely resembles the actual distribution of helices in myoglobin. The possible utility and pitfalls of this approach as part of an overall protein folding program are discussed.