A Boltzmann model predicts glycan structures from lectin binding

Abstract

Glycans are complex polysaccharides involved in many diseases and biological processes. Unfortunately, current methods for determining glycan composition and structure (glycan sequencing) are laborious and require a high level of expertise. Here, we assess the feasibility of sequencing glycans based on their lectin binding fingerprints. By training a Boltzmann model on lectin binding data, we are able to predict the approximate structures of 90 +/- 5 % of N-glycans in our test set. We further show that our model generalizes well to the pharmaceutically relevant case of Chinese Hamster Ovary (CHO) cell glycans. We also analyze the motif specificity of a wide array of lectins and identify the most and least predictive lectins and glycan features. These results could help streamline glycoprotein research and be of use to anyone using lectins for glycobiology.

LeGenD: determining N-glycoprofiles using an explainable AI-leveraged model with lectin profiling

“…Lectins are particularly useful for identifying glycan epitopes and for linkage stereochemistry. Although several studies have attempted to profile glycans holistically 75 or to predict single glycans from lectin binding patterns 33 , mapping these patterns to an actual glycoprofile is a highly under-constrained problem, wherein nigh infinite glycoprofiles could have the same lectin profile (Fig. 1).…”

“…We hypothesized that, while these lectins can recognize various structures, such as internal complex N -glycan structures, there is a need for more specific carbohydrate-binding proteins to enhance the model’s overall performance. In relation to this, another possible explanation for the limited impact of PHA-E and PHA-L on LeGenD’s predictive performance is linked to the findings of Yom et al 33 , who analyzed lectin-glycan binding patterns. They determined that lectins exhibit a stronger affinity towards terminal glycan motifs because they are least susceptible to steric interference.…”

“…Lastly, although successful work has been done in elucidating lectin binding motifs, many are not inferred while accounting for all possible glycan structures. In extreme cases, they may be restricted to single glycan structures or assays to test binding, including limited sampling of all possible glycans, or from a single type of array, and do not account for external factors [33][34][35] . Therefore, it is challenging to leverage lectin-based approaches for the comprehensive elucidation of intact glycoform structures in a given glycoprotein, since they do not account for many intracellular constraints that help shape the orientation of the sugars in complex glycan structures.…”