1774-P: Structural Mechanisms and AI-Driven Prediction of Glucokinase Variants: Advancing Precision Medicine in MODY2

Introduction and Objective: Glucokinase (GK) serves as the key glucose sensor in metabolic homeostasis. Loss-of-function (LOF) mutations in GK cause MODY2 but classifying variants of uncertain significance poses a major clinical challenge. This study combines structural biochemistry and artificial intelligence (AI) to delineate the mechanistic effects of GK variants and forecast their responsiveness to the GK activator (GKA) dorzagliatin. Methods: We compiled a comprehensive database of ~1,000 GK variants from ClinVar, HGMD, and published literature. A two-pronged strategy was adopted: (1) Mechanistic Profiling: Selected mutations in substrate-binding, allosteric, and hinge domains were recombinantly expressed and assessed for enzyme kinetics (S0.5, Vmax, Hill coefficient) and dorzagliatin sensitivity. (2) AI Modeling: A supervised learning model, built on a large language model framework, incorporated structural attributes, physicochemical descriptors, and evolutionary conservation metrics (ESM/SaPro) to forecast enzymatic phenotypes. Results: Structural mapping showed gain-of-function mutations predominantly cluster at the allosteric site. In contrast, LOF variants were spatially diffused, mainly disrupting substrate affinity or thermal stability (e.g., Q26L). Dorzagliatin markedly restored Vmax in certain severe LOF mutants (e.g., N213S) but exhibited reduced efficacy in select allosteric variants (e.g., G448D). The AI model demonstrated 90% accuracy and sensitivity in predicting kinetic profiles and therapeutic responses for novel variants. Conclusion: This work establishes clear structure-function relationships for GK mutations: active-site alterations impair substrates affinity, while allosteric/hinge perturbations disrupt allosteric regulation. Integrating mechanistic insights with AI prediction offers a powerful framework for MODY2 genetic diagnosis and patient stratification for targeted GKA therapy, accelerating precision medicine. Disclosure Y. Yuan: None. Q. Zhu: None. X. Zhou: None. S. Wu: None. F. Zhao: None. L. Chen: None. C. Li: None.