A general method to predict the effect of single amino acid substitutions on enzyme catalytic activity

Over the past thirty years, site-directed mutagenesis has become established as one of the most powerful techniques to probe enzyme reaction mechanisms. Substitutions of active site residues are most likely to yield significant perturbations in kinetic parameters, but there are many examples of profound changes in these values elicited by remote mutations. Ortholog comparisons of extant sequences show that many mutations do not have profound influence on enzyme function. As the number of potential single natural amino acid substitutions that can be introduced in a protein of N amino acids in length by directed mutation is very large (19 * N), it would be useful to have a method to predict which amino acid substitutions are more likely to introduce significant changes in kinetic parameters in order to design meaningful probes into enzyme function. What is especially desirable is the identification of critical residues that do not contact the substrate directly, and may be remote from the active site. We collected literature data reflecting the effects of 2,804 mutations on kinetic properties for 12 enzymes. These data along with characteristic predictors were used in a machine-learning scheme to train a classifier to predict the effect of mutation. Use of this algorithm allows one to predict with a 2.5-fold increase in precision, if a given mutation, made anywhere in the enzyme, will cause a decrease in kcat/Km value of ≥ 95%. The improved precision allows the experimentalist to reduce the number of mutations necessary to probe the enzyme reaction mechanism.