Rheostats and Toggle Switches for Modulating Protein Function

Spring 2017
Speaker: 
Liskin Swint-Kruse
Affiliation: 
University of Kansas Medical Center
Wed, 2017-02-01
3:00PM
487 GWC
Host: 
Banu Ozkan

Personalized medicine and protein engineering are hindered by our limited ability to predict the functional outcomes of amino acid changes.  To facilitate this, many computer algorithms have been written.  These often use evolutionary information about amino acid changes that can be found in sequence alignments of protein families.  With these algorithms, catastrophic mutations at conserved positions can be readily identified.  However, predictably identifying substitution outcomes at nonconserved positions remains a challenge. 

We considered three underlying assumptions that are shared by many computational algorithms: (i) Most mutations at “important” amino acid positions will abolish function; (ii) a few substitutions will allow function if they are chemically similar to the wild-type amino acid; (iii) each substitution will have the same outcome in any homolog.  These assumptions were derived from the outcomes of laboratory experiments, which are highly biased to conserved positions.  To assess whether these assumptions apply to nonconserved positions, we perfor¬med a large-scale, quantitative mutagenesis study using homologs in the LacI/GalR family.  We found that the assumptions do not apply for a certain group of nonconserved amino acids.  We named this group “rheostat” positions, based on their most prominent feature: When multiple amino acids are substituted at a rheostat position in the protein sequence, their rank-ordered functions show a progressive effect on the protein’s function. 

Retrospective inspection of published data showed that other, unrelated proteins contain positions with rheostatic mutational outcomes.  We estimate these positions can comprises >25% of some protein sequences, and that their mutation can alter a wide range of functional parameters.  Comparison to algorithm outcomes shows that: (i) Most algorithms that predict outcomes for specific substitutions perform poorly at rheostat positions; and (ii) The locations of rheostat positions might correlate with particular patterns of amino acid change in sequence alignments. To improve predictions about mutational outcomes, we must explicitly consider contributions from rheostat positions and better understand the biophysical phenomena that lead to their unconventional mutational outcomes.
 

    

Arizona State University, P.O. Box 871504, Tempe, AZ 85287-1504

The Center for Biological Physics
Contact Us | Webmaster