The beta-beta-alpha fold: explorations in sequence space.

The computational redesign of the second zinc finger of Zif268 to produce a 28 residue peptide (FSD-1) that assumes a betabetaalpha fold without metal binding was recently reported. In order to explore the tolerance of this metal-free fold towards sequence variability, six additional peptides resulting from the ORBIT computational protein design process were synthesized and characterized. The experimental stabilities of five of these peptides are strongly correlated with the energies calculated by ORBIT. However, when a peptide with a mutation in the beta-turn is examined, the calculated stability does not accurately predict the experimentally determined stability. The NMR solution structure of a peptide incorporating this mutation (FSD-EY) reveals that the register between the beta-strands is different from the model structure used to select and score the sequences. FSD-EY has a type I' turn instead of the target EbaaagbE turn (rubredoxin knuckle). Two additional peptides that have improved side-chain to backbone hydrogen bonding and turn propensity for the target turn were characterized. Both are of stability comparable to that of FSD-1. These results demonstrate the robustness of the ORBIT protein design methods and underscore the need for continued improvements in negative design. Copyright 2001 Academic Press.