Protein-Carbohydrate Interactions, and Beyond …

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.

Emergence of Cooperative Glucose-Binding Networks in Adaptive Peptide Systems

Minimalistic peptide-based systems that bind sugars in water are challenging to design due to the weakness of interactions and required cooperative contributions from specific amino-acid side chains. Here, we used a bottom-up approach to create peptide-based adaptive glucose-binding networks by mixing glucose with selected sets of input dipeptides (up to 4) in the presence of an amidase to enable in situ reversible peptide elongation, forming mixtures of up to 16 dynamically interacting tetrapeptides. The choice of input dipeptides was based on amino-acid abundance in glucose-binding sites found in the protein data bank, with side chains that can support hydrogen bonding and CH-π interactions. Tetrapeptide sequence amplification patterns, determined through LC-MS analysis, served as a readout for collective interactions and led to the identification of optimized binding networks. Systematic variation of dipeptide input revealed the emergence of two networks of non-covalent hydrogen bonding and CH-π interactions that can co-exist, are cooperative and context-dependent. A cooperative binding mode was determined by studying the binding of the most amplified tetrapeptide (AWAD) with glucose in isolation. Overall, these results demonstrate that the bottom-up design of complex systems can recreate emergent behaviors driven by covalent and non-covalent self-organization that are not observed in reductionist designs and lead to the identification of system-level cooperative binding motifs.

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 models named CArbohydrate-Protein interaction Site IdentiFier (CAPSIF) that predict 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.

Insights into the effect of protein glycosylation on carbohydrate substrate binding

The role of glycosylation in the binding of glycoproteins to carbohydrate substrates has not been well understood. The present study addresses this knowledge gap by elucidating the links between the glycosylation patterns of a model glycoprotein, a Family 1 carbohydrate-binding module (TrCBM1), and the thermodynamic and structural properties of its binding to different carbohydrate substrates using isothermal titration calorimetry and computational simulation. The variations in glycosylation patterns cause a gradual transition of the binding to soluble cellohexaose from an entropy-driven process to an enthalpy-driven one, a trend closely correlated with the glycan-induced shift of the predominant binding force from hydrophobic interactions to hydrogen bonding. However, when binding to a large surface of solid cellulose, glycans on TrCBM1 have a more dispersed distribution and thus have less adverse impact on the hydrophobic interaction forces, leading to overall improved binding. Unexpectedly, our simulation results also suggest an evolutionary role of O-mannosylation in transforming the substrate binding features of TrCBM1 from those of type A CBMs to those of type B CBMs. Taken together, these findings provide new fundamental insights into the molecular basis of the role of glycosylation in protein-carbohydrate interactions and are expected to better facilitate further studies in this area.

Prediction of protein-carbohydrate complex binding affinity using structural features

Protein-carbohydrate interactions play a major role in several cellular and biological processes. Elucidating the factors influencing the binding affinity of protein-carbohydrate complexes and predicting their free energy of binding provide deep insights for understanding the recognition mechanism. In this work, we have collected the experimental binding affinity data for a set of 389 protein-carbohydrate complexes and derived several structure-based features such as contact potentials, interaction energy, number of binding residues and contacts between different types of atoms. Our analysis on the relationship between binding affinity and structural features revealed that the important factors depend on the type of the complex based on number of carbohydrate and protein chains. Specifically, binding site residues, accessible surface area, interactions between various atoms and energy contributions are important to understand the binding affinity. Further, we have developed multiple regression equations for predicting the binding affinity of protein-carbohydrate complexes belonging to six categories of protein-carbohydrate complexes. Our method showed an average correlation and mean absolute error of 0.731 and 1.149 kcal/mol, respectively, between experimental and predicted binding affinities on a jackknife test. We have developed a web server PCA-Pred, Protein-Carbohydrate Affinity Predictor, for predicting the binding affinity of protein-carbohydrate complexes. The web server is freely accessible at https://web.iitm.ac.in/bioinfo2/pcapred/. The web server is implemented using HTML and Python and supports recent versions of major browsers such as Chrome, Firefox, IE10 and Opera.

Heterologous production of α-chain of Dioclea sclerocarpa lectin: Enhancing the biological effects of a wild-type lectin

Lectins from Diocleinae subtribe species (family Leguminosae) are of special interest since they present a wide spectrum of biological activities, despite their high structural similarity. During their synthesis in plant cells, these proteins undergo post-translational processing resulting in the formation of three chains (α, β, γ), which constitute the lectins' subunits. Furthermore, such wild-type proteins are presented as isolectins or with different combinations of these chains, which undermine their biotechnological potential. Thus, the present study aimed to produce a recombinant form of the lectin from Dioclea sclerocarpa seeds (DSL), exclusively constituted by α-chain. The recombinant DSL (rDSL) was successfully expressed in E. coli BL21 (DE3) and purified by affinity chromatography (Sephadex G-50), showing a final yield of 74 mg of protein per liter of culture medium and specificity for D-mannose, α-methyl-mannoside and melibiose, unlike the wild-type protein. rDSL presented an effective vasorelaxant effect in rat aortas up to 100% and also interacted with glioma cells C6 and U87. Our results demonstrated an efficient recombinant production of rDSL in a bacterial system that retained some biochemical properties of the wild-type protein, showing wider versatility in sugar specificities and better efficacy in its activity in the biological models evaluated in this work.

Synthesis and conformational analysis of a simplified inositol-model of the Streptococcus pneumoniae 19F capsular polysaccharide repeating unit

Carbohydrate mimics have been studied for a long time as useful sugar substitutes, both in the investigation of biological events and in the treatment of sugar-related diseases. Here we report further evaluation of the capabilities of inositols as carbohydrate substitutes. The conformational features of an inositol-model of a simplified repeating unit corresponding to the capsular polysaccharide of Streptococcus pneumoniae 19F has been evaluated by computational analysis, and compared to the native repeating unit. The inositol mimic was synthesized, and its experimental spectroscopic data allowed for verification of the theoretical results.