STREAMLINING GLYCOMIMETIC SCREENING THROUGH AN AUTOMATED STRUCTURE-BASED DRUG DESIGN PIPELINE

Glycans form a structurally diverse “glycocalyx” on all cells, encoding vast biological information deciphered by lectins. Interactions between the glycocalyx and such glycan-binding proteins regulate critical immunopathological processes, including bacterial and viral infection, fibrosis, homeostasis and cancer proliferation. Fucosylated glycans play a key role in these processes, as their altered expression is a hallmark of many carcinomas and correlates with patient prognosis. Furthermore, many pathogens exploit these glycoconjugates, using lectins with high affinity for fucose to bind human tissues. Inhibiting these recognition events with glycomimetics represents a promising yet challenging therapeutic strategy, primarily due to the limited druggability of lectins and the typically poor pharmacokinetics of carbohydrate-based ligands. To overcome these limitations, a standard structure-based strategy involves decorating the primary sugar anchor to engage secondary subpockets within the lectin binding site, seeking additional interaction points while balancing their marked hydrophilicity. Following this approach, a novel synthetic methodology was developed to functionalize the C5 position of fucose, aiming to achieve C1 and C5-substituted β-fucosides and exploit suitable subpockets in the N-terminal domain of the bacterial lectin BC2L-C. Given the latent potential of this newly accessible chemotype, the scope was broadened to evaluate its versatility across a wider panel of fucose-binding lectins. To understand the structural requirements for receptors engagement, an automated multi-target docking workflow was implemented via the Schrödinger Suite Python API, ensuring efficiency, customization and reproducibility. This approach was deployed to assess the binding potential of a virtual, combinatorial library of accessible fucosides against multiple selected targets. A thorough analysis of promising docking poses, further refined through molecular dynamics (MD) simulations, enabled the identification of high-potential chemotypes and previously unobserved alternative binding modes for several lectins. Beyond yielding novel design paradigms and prioritizing promising candidates for subsequent experimental validation, this workflow highlights systemic challenges inherent to the computational modeling of glycan-protein interactions. In particular, sugar multimodal binding makes the identification of the biologically relevant pose exceptionally difficult. Crucially, this is directly compounded by the limitations of current scoring functions and the frequent oversimplification of interactions that are natively mediated by complex dynamical water networks. Addressing these complexities is therefore the primary focus of ongoing investigations to further enhance the reliability of rational glycomimetic design.