Using crystallographic water properties for the analysis and prediction of lectin-carbohydrate complex structures

Understanding protein-ligand interactions is a fundamental question in basic biochemistry, and the role played by the solvent along this process is not yet fully understood. This fact is particularly relevant in lectins, proteins that mediate a large variety of biological processes through the recognition of specific carbohydrates. In the present work, we have thoroughly analyzed a nonredundant and well-curated set of lectin structures looking for a potential relationship between the structural water properties in the apo-structures and the corresponding protein-ligand complex structures. Our results show that solvent structure adjacent to the binding sites mimics the ligand oxygen structural framework in the resulting protein-ligand complex, allowing us to develop a predictive method using a Naive Bayes classifier. We also show how these properties can be used to improve docking predictions of lectin-carbohydrate complex structures in terms of both accuracy and precision, thus developing a solid strategy for the rational design of glycomimetic drugs. Overall our results not only contribute to the understanding of protein-ligand complexes, but also underscore the role of the water solvent in the ligand recognition process. Finally, we discuss our findings in the context of lectin specificity and ligand recognition properties.