GlycoTorch Vina: Improved Docking of Sulfated Sugars Using QM-derived Scoring Functions

Glycosaminoglycans (GAGs) are a family of anionic carbohydrates that play an essential role in the physiology and pathology of all eukaryotic life. Experimental determination of GAG-protein complexes remains difficult due to the considerable diversity in both carbohydrate linkage, and sulfation patterns. To complement existing methods of structural determination, we present our molecular docking tool, GlycoTorchVina (GTV), which demonstrates a substantial improvement at reproducing low energy conformations of GAGs compared with traditional docking programs. Based on the carbohydrate specific docking program VinaCarb (VC), GTV utilizes rotational energy functions, calculated using density functional theory (DFT), specifically designed for glycosidic linkages found in GAGs. The redocking accuracy of four programs (GTV, VC, AutoDock Vina and Glide) was tested over a set of 10 high-quality crystal structures containing co-crystallized GAGs (tetrasaccharides or longer). GTV outperformed other programs and was able to reproduce the native pose of eight structures and produced top-scoring docked poses that were on average only 1.8 Å RMSD away from the crystal structure. Although imitation of crystal structures is a standard test used for assessing the accuracy of docking programs, we illustrate how the underlying quality of the crystal structure, which is often overlooked during benchmarking, affects conclusions drawn from this approach. Statistical and theoretical investigations into charge-charge (“salt-bridge”) interactions are also presented. Again, DFT calculations were used to derive non-bonded potentials describing salt-bridges, and solvent-mediated charge-charge (“water-bridge”) interactions. These data suggest that water-bridges play an important, yet poorly understood, role in the structures of GAG-protein complexes.