Structure-based neural network protein-carbohydrate interaction predictions at the residue level

Carbohydrates dynamically and transiently interact with proteins for cell-cell recognition, cellular differentiation, immune response, and many other cellular processes. Despite the molecular importance of these interactions, there are currently few reliable computational tools to predict potential carbohydrate-binding sites on any given protein. Here, we present two deep learning (DL) models named CArbohydrate-Protein interaction Site IdentiFier (CAPSIF) that predicts non-covalent carbohydrate-binding sites on proteins: (1) a 3D-UNet voxel-based neural network model (CAPSIF:V) and (2) an equivariant graph neural network model (CAPSIF:G). While both models outperform previous surrogate methods used for carbohydrate-binding site prediction, CAPSIF:V performs better than CAPSIF:G, achieving test Dice scores of 0.597 and 0.543 and test set Matthews correlation coefficients (MCCs) of 0.599 and 0.538, respectively. We further tested CAPSIF:V on AlphaFold2-predicted protein structures. CAPSIF:V performed equivalently on both experimentally determined structures and AlphaFold2-predicted structures. Finally, we demonstrate how CAPSIF models can be used in conjunction with local glycan-docking protocols, such as GlycanDock, to predict bound protein-carbohydrate structures.

Effects of active site cleft residues on oligosaccharide binding, hydrolysis, and glycosynthase activities of rice BGlu1 and its mutants

Rice BGlu1 (Os3BGlu7) is a glycoside hydrolase family 1 β-glucosidase that hydrolyzes cellooligosaccharides with increasing efficiency as the degree of polymerization (DP) increases from 2 to 6, indicating six subsites for glucosyl residue binding. Five subsites have been identified in X-ray crystal structures of cellooligosaccharide complexes with its E176Q acid-base and E386G nucleophile mutants. X-ray crystal structures indicate that cellotetraose binds in a similar mode in BGlu1 E176Q and E386G, but in a different mode in the BGlu1 E386G/Y341A variant, in which glucosyl residue 4 (Glc4) interacts with Q187 instead of the eliminated phenolic group of Y341. Here, we found that the Q187A mutation has little effect on BGlu1 cellooligosaccharide hydrolysis activity or oligosaccharide binding in BGlu1 E176Q, and only slight effects on BGlu1 E386G glycosynthase activity. X-ray crystal structures showed that cellotetraose binds in a different position in BGlu1 E176Q/Y341A, in which it interacts directly with R178 and W337, and the Q187A mutation had little effect on cellotetraose binding. Mutations of R178 and W337 to A had significant and nonadditive effects on oligosaccharide hydrolysis by BGlu1, pNPGlc cleavage and cellooligosaccharide inhibition of BGlu1 E176Q and BGlu1 E386G glycosynthase activity. Hydrolysis activity was partially rescued by Y341 for longer substrates, suggesting stacking of Glc4 on Y341 stabilizes binding of cellooligosaccharides in the optimal position for hydrolysis. This analysis indicates that complex interactions between active site cleft residues modulate substrate binding and hydrolysis.

Crystallographic, thermodynamic, and molecular modeling studies of the mode of binding of oligosaccharides to the potent antiviral protein griffithsin

The mode of binding of oligosaccharides to griffithsin, an antiviral lectin from the red alga Griffithsia sp., was investigated by a combination of X-ray crystallography, isothermal titration calorimetry, and molecular modeling. The structures of complexes of griffithsin with 1-->6alpha-mannobiose and with maltose were solved and refined at the resolution of 2.0 and 1.5 A, respectively. The thermodynamic parameters of binding of 1-->6alpha-mannobiose, maltose, and mannose to griffithsin were determined. Binding profiles of 1-->6alpha-mannobiose and mannose were similar with Kd values of 83.3 microM and 102 microM, respectively. The binding of maltose to griffithsin was significantly weaker, with a fourfold lower affinity (Kd = 394 microM). In all cases the binding at 30 degrees C was entropically rather than enthalpically driven. On the basis of the experimental crystal structures, as well as on previously determined structures of complexes with monosaccharides, it was possible to create a model of a tridentate complex of griffithsin with Man9GlcNAc2, a high mannose oligosaccharide commonly found on the surface of viral glycoproteins. All shorter oligomannoses could be modeled only as bidentate or monodentate complexes with griffithsin. The ability to mediate tight multivalent and multisite interactions with high-mannose oligosaccharides helps to explain the potent antiviral activity of griffithsin.

The binding of beta- and gamma-cyclodextrins to glycogen phosphorylase b: kinetic and crystallographic studies

A number of regulatory binding sites of glycogen phosphorylase (GP), such as the catalytic, the inhibitor, and the new allosteric sites are currently under investigation as targets for inhibition of hepatic glycogenolysis under high glucose concentrations; in some cases specific inhibitors are under evaluation in human clinical trials for therapeutic intervention in type 2 diabetes. In an attempt to investigate whether the storage site can be exploited as target for modulating hepatic glucose production, alpha-, beta-, and gamma-cyclodextrins were identified as moderate mixed-type competitive inhibitors of GPb (with respect to glycogen) with K(i) values of 47.1, 14.1, and 7.4 mM, respectively. To elucidate the structural basis of inhibition, we determined the structure of GPb complexed with beta- and gamma-cyclodextrins at 1.94 A and 2.3 A resolution, respectively. The structures of the two complexes reveal that the inhibitors can be accommodated in the glycogen storage site of T-state GPb with very little change of the tertiary structure and provide a basis for understanding their potency and subsite specificity. Structural comparisons of the two complexes with GPb in complex with either maltopentaose (G5) or maltoheptaose (G7) show that beta- and gamma-cyclodextrins bind in a mode analogous to the G5 and G7 binding with only some differences imposed by their cyclic conformations. It appears that the binding energy for stabilization of enzyme complexes derives from hydrogen bonding and van der Waals contacts to protein residues. The binding of alpha-cyclodextrin and octakis (2,3,6-tri-O-methyl)-gamma-cyclodextrin was also investigated, but none of them was bound in the crystal; moreover, the latter did not inhibit the phosphorylase reaction.