Glycan array screening is a high throughput method for identifying potential ligands for Glycan binding proteins. Screening data provides specificity information and helps identify the minimal glycan fragment that is responsible for affinity (minimal binding determinant). However, such screening provides no structural insight into the origin of the observed specificity among glycans which contain this minimal binding determinant in their sequence. When applicable, crystallographic and NMR methods can provide 3D structures for these complexes, however these methods are neither rapid nor high throughput. Automated carbohydrate docking offers one approach to generating putative models for these complexes, the accuracy of which may be greatly enhanced by selecting theoretical models that satisfy all of the observed binding and non-binding interactions identified by experimental screening. By considering both binding and non-binding glycans that contain the minimal motif, it is possible to unambiguously identify the optimal orientation of the ligand in the binding site, providing not only a 3D model for the complex, but also a rational explanation for the observed array data.
Given a 3D structure for the minimal glycan determinant aligned in the binding site of the receptor protein, the potential for any glycan containing this motif to bind may be predicted by Computational Carbohydrate Grafting (CCG). In addition, we have generated a library of 3D structures of the known human glycome (Glibrary-3D), and demonstrate that CCG may be used to to identify putative binding glycans within the human glycome.
We illustrate this high-throughput approach for a range of glycan binding proteins, including, antibodies, lectins, and enzymes, and show that the results are consistent with data from crystallography and glycan array screening.
