Exploring the Mutarotation Mechanism of Glucose in Solution Using Deep Learning Potential

Glucose mutarotation plays a fundamental role in carbohydrate chemistry by governing the interconversion between α- and β-anomers in solution, thereby influencing the physical, chemical, and biological properties of glucose. Two mechanisms have been proposed for the mutarotation of glucose in aqueous solution. However, it remains unclear which pathway predominates under typical conditions, as both have been suggested, and definitive experimental or theoretical evidence distinguishing them is still lacking. To clarify the mutarotation mechanism of glucose, deep learning potential molecular dynamics (DLPMD) simulations were performed to investigate the underlying mechanism and free energy profiles of glucose mutarotation. Compared with previous ab initio molecular dynamics results, the DLPMD simulations provide a more accurate and statistically converged description of the reaction landscape, revealing that mutarotation preferentially proceeds via the ring-opening pathway. This route exhibits a lower activation barrier and avoids the formation of a high-energy C1 carbocation. Within the ring-opening mechanism, the formation of the β-anomer is kinetically favored. These results demonstrate that DLPMD simulations reliably capture both reaction pathways and conformational preferences in aqueous solution, offering a computationally efficient alternative to conventional density functional theory (DFT) methods.