Glycosaminoglycans (GAGs) are a family of highly sulfated amino sugars involved in an expansive range of physiological and pathological processes – including cancer, aging and the immune system. GAGs are found in all eukaryotic cells, covalently bound to proteins in the extracellular matrix. Experimental determination of protein-GAG complexes, by NMR and X-ray crystallography for example, is difficult due to high levels of natural heterogeneity and negative charge making purification from biological sources challenging.1 Therefore, computational approaches, to complement experimental techniques, are highly desirable.
Molecular docking is an attractive computational tool that aims to provide insight into how GAGs and proteins interact at the atomic level. The reliability of docking, when predicting the complexes of carbohydrates, is limited by the parameterisation of these methods, the majority of which were originally intended for small drug-like molecules. Naturally occurring GAGs contain many charged moieties and rotatable bonds – features uncommon to most druglike molecules. To overcome this limitation, novel scoring functions based on molecular mechanics principles were developed to better model these highly charged and flexible molecules, parameterised to Kohn and Sham DFT energy calculations.
Glycotorch, a glycoinformatics and molecular docking tool, aims to improve the modelling and analysis of GAG-protein complexes. Glycotorch Vina, an extension of the popular program, Autodock Vina,2 provides an implementation of these novel scoring functions and boasts a greatly improved ability to reproduce known crystal structures of GAG-protein complexes. A companion website (glycotorch.com) is also in development and provides tools to help users prepare input files and analyse output during docking experiments. Glycotorch Vina also allows users to enforce constraints or additional scoring functions. This feature allows Glycotorch Vina to complement experimental data, such as 1D and 2D NMR observables, to create better models of biological systems and enhance carbohydrate-based drug design.