An award is made to Johns Hopkins University to develop a new method to model higher order mutation interactions, by combining evolutionary genetics, network models, and protein sequence analysis. The study of how the effects of individual mutations combine to impact the overall function of a protein is a key question in evolutionary biology. It is also a very difficult question, as the effects multiple protein mutations can combine non-additively. In this case, higher order interactions (beyond pairwise) between mutations need to be considered, and the number of interactions increases exponentially as a function of the total number of protein mutations involved. Therefore, if a model?s parameters are mutation interactions, these parameters will far exceed the number of possible experimental observations needed to determine them. As a result, existing bioinformatics methods are limited in their exploration of higher-order evolutionary interactions. Specifically, detailed modeling of how protein mutations combine as steps in the evolutionary path of a protein toward new function is currently lacking. In this project, systematic evolutionary analysis of protein sequence will be conducted to ensure that only residue positions most relevant for evolving new functions are considered. Next, mutated positions will be linked together in a network model. This modeling framework dramatically reduces the number of modeling parameters and incorporates higher order interactions in a tractable fashion. Because nodes in the network correspond to mutated protein positions, any complex combination of mutations can be represented as a path through the network. Using graph theory, metrics will be developed to assess the relationship between path importance within the network and the successful adaptation of a protein to its environment. This work focuses on a simple biological model system -- the evolution antibiotic resistance by the enzyme TEM beta-lactamase in Escherichia coli bacteria. In this system, there is a direct correlation between the evolution of a new function (antibiotic resistance) and survival at the level of the whole organism (bacterial growth). Computational predictions will be systematically tested by introducing mutations of interest into bacterial cultures and measuring survival under exposure to a given antibiotic.
This work promotes close interaction between the computational sciences and biology communities: It combines expertise in computational/statistical modeling of mutations in proteins and applied evolutionary genetics in microbial systems. The broader use of this work will be to anticipate the emergence of drug resistance in clinically relevant proteins. It will also have great utility for protein engineers who seek to design proteins with new or improved functions. Furthermore, it will contribute to the design of therapeutic regimens for diseases driven by bacteria or viruses, in which the evolution of drug resistance is commonplace. The educational goals of the project include new course components for undergraduates and graduates at the universities where the project investigators teach and outreach to underrepresented minority students in science and engineering. More information about the project can be found at: http://karchinlab.org/.
