Prediction of protein stability, allowing efficient creation of thermostable proteins, is of significant practical interest. Thermostable proteins also provide better starting points for evolution of protein function [Bloom JD et al., (2006) P. Natl. Acad. Sci. USA 103:5869-5874]. We recently used structure-guided recombination of the heme domains of three bacterial cytochromes P450 (CYP102A1, A2 and A3) to create a library of more than 3,000 folded chimeras. The chimeras are composed of eight fragments, each of which was taken from one of the three parents. The chimeric sequences were highly mutated, having an average of more than seventy mutations compared to the closest parent [Otey CR et al., (2006) PLoS Biol., 4:e112]. Here we explore the relationship between sequence and stability by analyzing the thermostabilities of more than 200 chimeric cytochromes P450 from this synthetic protein family. We find that thermostability is accurately predicted by a linear model, which assumes that the individual protein fragments make additive contributions to protein stability. We also analyze the sequences of 620 folded and 335 unfolded chimeras and find that the thermostable chimeras tend to contain consensus sequence elements (both fragments and residues), i.e., those occurring more frequently among folded proteins and less often among unfolded ones. Thus, protein folding and stability can be predicted from the frequencies of sequence elements. Our analysis of this large ensemble of synthetic proteins demonstrates that sequence elements contribute additively to protein stability and provides key support for the idea of consensus stabilization.

Extended Abstract Status: Not Uploaded