On the conformation of proteins: Towards the prediction of strand arrangements in β-pleated sheets

Abstract

A new method of predicting the arrangements of β-strands within β-pleated sheets in globular proteins is described. From a quantitative analysis of the β-pleated sheets in known structures, folding parameters are derived which represent the probability of finding: 1. (1) the last strand in the sheet at the edge; 2. (2) a sheet which is pure parallel, pure antiparallel or mixed; 3. (3) mixed strands (i.e. parallel to one and antiparallel to its other neighbour); 4. (4) a given number of adjacent strands; 5. (5) the number of residues between one strand and the next strand in the sheet; and 6. (6) the distribution of the type of connections (coil, helical or sheet between parallel or antiparallel strands). The weighted product of these parameters yields an overall parameter S which indicates, without calculating residue-residue interactions, whether a particular arrangement of strands is favourable. To evaluate the approach, a computer program has been used to estimate, for each known structure, the number of alternative arrangements of the strands within a sheet for which S is not less than S for the actual arrangement. For most sheets this number is small showing that S is a good indication of probable strand arrangements. For about half the sheets the observed arrangement is the most probable. We outline an approach to overcome the problem of assigning strands to the different sheets and show that for adenyl kinase, erabutoxin, flavodoxin and Staphylococcus nuclease, the sheetstructure that is actually observed is one of a few probable sheets. Unlike the conventional methods, this approach can predict probable sheet structures of large globular proteins from the results of a successful prediction of the secondary structure alone.