A procedure for the prediction of temperature-sensitive mutants of a globular protein based solely on the amino acid sequence.

Temperature-sensitive (Ts) mutants of a protein are an extremely powerful tool for studying protein function in vivo and in cell culture. We have devised a method to predict those residues in a protein sequence that, when appropriately mutated, are most likely to give rise to a Ts phenotype. Since substitutions of buried hydrophobic residues often result in significant destabilization of the protein, our method predicts those residues in the sequence that are likely to be buried in the protein structure. We also indicate a set of amino acid substitutions, which should be made to generate a Ts mutant of the protein. This method requires only the protein sequence. No structural information or homologous sequence information is required. This method was applied to a test data set of 30 nonhomologous protein structures from the Protein Data Bank. All of the residues predicted by the method to be > or = 95% buried were, in fact, buried in the protein crystal structure. In contrast, only 50% of all hydrophobic residues in this data set were > or = 95% buried. This method successfully predicts several known Ts and partially active mutants of T4 lysozyme, lambda repressor, gene V protein, and staphylococcal nuclease. This method also correctly predicts residues that form part of the hydrophobic cores of lambda repressor, myoglobin, and cytochrome b562.