UC San Diego

Proteins are linear chains of amino acids that self- assemble, or "fold", into 3-dimensional structures which perform tasks in organisms. Nature has selected for sequences which fold quickly and efficiently into their functional structure. While it is important for organisms to have proteins which fold efficiently, it is essential that proteins also perform their function. These two evolutionary pressures are sometimes in conflict, and perfect optimization of both folding and function may not always be possible. For example, a mutation which slows folding may be selected for if it enhances protein function, therefore increasing the fitness of the organism. As a result, regions of a protein that contribute to slow folding may be critical for function. Identifying regions that contribute to slow folding may be an effective way to predict and identify sites that are critical to a proteins function. This dissertation uses this approach to characterize mitoNEET, a protein which is implicated in diabetes, aging, cancer, and obesity. MitoNEET was discovered because it unexpectedly binds the commonly prescribed diabetes drug Pioglitazone. The protein is a uniquely folded homodimer, and each protomer coordinates two [2Fe-2S] clusters. These iron-sulfur clusters are capable of electron transfer, and the cluster itself may be transferred to acceptor partner proteins. Understanding how this protein regulates its metal centers is critical for better drug design. We used structure- based models to simulate the folding of mitoNEET, and observed that a loop far removed from the metal center creates a constraint which slows the folding of the protein. We predicted that this region was evolutionarily conserved, and that mutations to this site would disrupt the function of the protein. To test this theory, we used mutagenesis to introduce perturbations at this site. We observed that changes to this loop alter the rate of cluster transfer, cluster decay, and the redox potential. This result is striking because properties of metal centers are traditionally thought to be controlled by the small fraction of amino acids which directly surround them. Our work challenges this paradigm, and we feel it opens the door to more intelligent drug design for this class of proteins