Relationship Between Structure And Function Through Protein Binding Site Alignments

Bioinformatics is an interdisciplinary field that combines mathematical, statistical and computational approaches to study and process biological data. As a branch of bioinformatics, structural bioinformatics focuses on the research of macromolecular structures. We here use means of bioinformatics to investigate the structural features of enzyme active sites seeking to understand the relationship between protein structures and their biological functions. In our research particular enzymes with similar catalytic activities are systematically analyzed; new hypothesis is proposed which later is further validated by quantum chemistry studies. This work provides not only binding specificity between enzymes and inhibitors, but also promising insights in challenging target-specific inhibitor design.Two enzyme categories are thoroughly studied in our research:(β/α)8fold (also known as TIM-barrel fold, named according to triosephosphate isomerases) glycoside hydrolases (GHs) and human protein kinases.TIM-barrel fold is one of the most ancient protein folds. Their structures are conservatively similar among families even though their sequences share low similarities. Many protein families fold into a TIM-barrel fold. Surprisingly they participate in various distinct types of protein interactions. This interesting "single fold-multiple functions" feature leads us to dive deep into the mystery of active site mechanisms. Glycoside hydrolases are widespread in all living organisms participating in various synthesis and degradation reactions. They hydrolyze glycolsidic bonds between two or more carbohydrates. Absence or mutation of the specific glycoside hydrolases can lead to developmental disorders and sometimes even death. Study on mechanism, of glycoside hydrolases has a long history. Interestingly almost one third of the TIM-barrel fold proteins happen to be glycoside hydrolases. After Koshland proposed for the first time the mechanism of glycoside hydrolases in1953, numerous researches including site-directed mutagenesis experiments, enzyme kinetics, computational simulations as well as increasing protein structures from crystallography and NMR make it possible to systematically study on their catalytic mechanisms on higher levels.130glycoside hydrolases with TIM-barrel fold are selected in our research, all of which are with distant evolutionary relationship but catalyze similar biological reactions. Two types of catalytic mechanisms are adopted:the classical retaining mechanism and the substrate-assisted intra-molecular nucleophilic attack mechanism. Both mechanisms, however, involve a certain residue as the general acid/base in catalysis. Based on the multiple active site structural alignments, these enzyme active sites can be clustered into six categories. Detailed investigation then suggests that hydrogen bonds interacting with the general acid/base in these categories play a key role during the reactions. Ab initio quantum mechanics calculations indicate that their presence may reduce the energetic barrier during the reactions by as large as17-20kcal/mol. The large energetic effect is primarily caused by the proton transfer from the general acid to the leaving group before the nucleophilic transition state, which also suggests that these interactions should be considered chemically essential. Furthermore, in the substrate-assisted mechanism the nucleophile group are found to interact with a conserved tyrosine which is convergently evolved between two different enzyme categories. Although this interaction may have no favorable effect on the energetic barrier, it may contribute indispensably to the catalysis by reducing the entropy of the reactants.Researches on human protein kinases are another important complimentary study. By comparing active sites among all available structures of human protein kinases, highly diverged residues locating in the binding pockets are revealed. We further combined binding specificities between kinases and inhibitors to calculate their mutual information, through which we found corresponding residues interacting with inhibitors for different kinases. With the modern machine learning techniques, we also predicted binding activities of specific kinases and inhibitors, which serves as a significant validation of our conclusion.We also applied structural bioinformatics onto the field of drug repositioning. Drug repositioning has been growing in importance in the last few years. However, its rather time-consuming process and huge cost make bringing a novel drug to market extremely difficult. Approved drugs have already undergone a series of in vitro, in vivo experiments and clinical trials. Discovering the new uses of the approved drugs will greatly ease the drug development procedure. To this end we performed a high-throughput systems-level analysis, mapping existing FDA (Food and Drug Administration) approved drugs with the potential for repurposing against targets from the malaria structural proteome. The resulting malaria drugome was used to prioritize potential new anti-malaria candidate targets and highlight some novel FDA approved drugs that have apparent anti-malaria effects for possible use as multi-target therapeutics.