A fast method for predicting amino acid mutations that lead to unfolding

Amino acid mutation(s) that cause(s) partial or total unfoldingof a protein can lead to disease states and failure to producemutants. It is therefore very useful to be able to predict whichmutations can retain the conformation of a wild-type proteinand which mutations will lead to local or global unfolding ofthe protein. We have developed a fast and reasonably accuratemethod based on a backbone-dependent side-chain rotamer libraryto predict the (folded or unfolded) conformation of a proteinupon mutation. This method has been tested on proteins whosewild-type 3D structures are known and whose mutant conformationshave been experimentally characterized to be folded or unfolded.Furthermore, for the cases studied here, the predicted partiallyfolded or denatured mutant conformation correlate with a decreasein the stability of the mutant relative to the wild-type protein.The key advantage of our method is that it is very fast andpredicts locally or globally unfolded states fairly accurately.Hence, it may prove to be useful in designing site-directedmutagenesis, X-ray crystallography and drug design experimentsas well as in free energy simulations by helping to ascertainwhether a mutation will alter or retain the wild-type conformation.