November 18, Tue 2014
1:00 pm, MRB 200
Department of Biochemistry and Molecular Biology Associate Director, Interdisciplinary Graduate Program in Biomedical Sciences, The University of Kansas Medical Center
Rheostats and toggle switches for modifying protein function
When genomes are sequenced for personalized medicine, each patient can have up to 10,000 differences in their protein coding regions. To filter for medically-relevant differences, current computer algorithms make predictions that are based, in-part, on how each protein has evolved. Algorithms also include "rules" devised from decades of mutation experiments: Chemically similar amino acids have similar functional outcomes; many amino acid substitutions abolish function or structure; each mutation will have the same outcome in any homolog. However, experiments have been heavily biased to conserved positions. In many proteins, >50% of amino acid positions are not conserved during evolution. If nonconserved positions follow different rules, they may lead to false positive and negative predictions in genome analyses.
We are performing large-scale, quantitative studies of nonconserved positions. In our first study, we recapitulated a multiple sequence alignment (MSA) in experimental space: We mutated 15 LacI/GalR homologs (MSA rows) at 12 nonconserved positions (MSA columns); each position was substituted with 8-12 amino acids. Strikingly, substitutions at nonconserved positions did not follow any of the rules listed above. First, the multiple variants at each position could be rank-ordered to show a "rheostatic" effect on function that spanned orders of magnitude (in contrast to the "toggle" behavior of conserved positions). Further, the rank-order showed few correlations with physico-chemical properties and was not predictable from their occurrence in LacI/GalR evolution. Finally, the outcome of a given substitution differed greatly among homologs. Retrospective inspection of published datasets showed that three unrelated proteins contain positions with rheostatic mutational outcomes. Thus, rheostat positions are likely to occur in many proteins. We are now testing the hypothesis that rheostat positions have a particular pattern of change during evolution. If so, they can be reliably identified in genome analyses, and their substitutions should be classified as having "unknown significance". Long term, we must formulate new rules for correctly predicting functional outcomes that arise from mutating rheostat positions.