PROTEIN FOLDING AND CHAIN COLLAPSE

It is currently assumed that the folded structure of a globular protein is controlled in a highly deterministic way by its amino acid sequence. We show here that a very different although not necessarily contrasting viewpoint can be adopted. From statistical treatment of x‐ray results, we suggest that the folding pattern essentially follows the collapse to be expected on statistical‐mechanical grounds for an ideal chain effectively experiencing self‐attraction and comprising identical units, whose conformational properties are obtained as an average over the actual amino acid units. The local details of folding of each protein, obviously dictated byits amino acid sequence, can be regarded as statistical fluctuations. We consider 31 globular fragments belonging to 21 different water‐soluble, nonmembrane proteins. By the theory of chain collapse proposed by two of us [G. Allegra and F. Ganazzoli (1985) J. Chem. Phys. 83, 397], all the average intramolecular distances may be obtained. Accordingly, first an average plot of the mean‐square distances between kth neighboring amino acid units is constructed, starting from the observed crystallographic coordinates. Then the plot is basically reproduced with a wormlike chain model undergoing collapse as a result of intramolecular attractive forces. Agreement is especially good for short amino acid sequences (k ≲ 30), in which case the statistical sampling is more accurate, enabling us to determine the model parameters. The resulting mean‐square radius of gyration is also in good agreement with the experimental average, whereas the unperturbed characteristic ratio is roughly consistent with results from conformational calculations by W. L. Mattice [(1977) Macromolecules 10, 516], based on the rotational isomeric state approach. Copyright © 1990 John Wiley & Sons, Inc.