Integrating Computational, Statistical and Biochemical Approaches to Characterize the Molecular-Functional Evolution of Protein

The observable protein functional diversity has been generated through an extensive evolutionary process. Sophisticated phylogenetic analyses are allowing researchers to peer into the distant evolutionary past and reconstruct ancestral molecules. These ancestral molecules can then be synthesized and studied in detail in the laboratory. Detailed laboratory analyses coupled with crystallization of ancestral proteins has allowed us to examine how evolutionary processes can drive changes in protein molecular function by altering protein structure and dynamics [1-5].;However, current approaches rely on costly and time-consuming procedures, making them difficult to extend to large-scale analyses. Without the capacity to examine molecular-functional evolution across large datasets, it is difficult to characterize what aspects of existing detailed studies generalize to larger-scale evolutionary processes, and which are idiosyncratic to the particular systems amenable to detailed laboratory study. As a step towards overcoming this limitation, our main objective is to propose and validate an automated and fast methodology for the characterization of protein molecular-functional evolution by predicting protein-ligand interactions with high accuracy. This methodology was implemented through the following steps: 1) Characterize the structural features explaining different types of protein-ligand interactions; 2) Propose a robust and integrated approach for characterizing the evolution of proteins function and specificity; 3) Apply the computational, statistical, and biochemical approach to characterize the functional evolution of proteins applied to a diverse set of case studies.;Using phylogenetic, statistic, and biochemical information we were able to identify some of the critical changes in structure and function that were responsible for increasing protein-ligand specificity in different protein families. Using the support of experimental binding affinity data, we were able to validate the results of our methodology in this case-study. The present methodology is not limited only to the case studies evaluated in this work, but can be applied in the study of the evolution of protein-DNA/RNA and protein-protein specificity in other protein families. As a high-throughput affinity prediction tool, the proposed approach will accelerate the process of identifying shifts in specificity of protein-ligand interactions by predicting the binding affinity of thousands of current and ancestral protein-ligand complexes, minimizing and guiding the experimental measurements.