Deep Learning-Driven Saccharide Online Sequencing for Elucidating the Pathological Alterations of Heparan Sulfate in APAP-Induced Acute Liver Injury

Heparan sulfate (HS), one of the mostly negatively charged biomacromolecules anchored on the membrane surface of nearly all mammal cells, plays critical regulatory roles through interacting with a variety of proteins. However, there is still no method capable to directly sequence the domain alterations of HS in pathological states. In the current study, the pathological alterations of HS were elucidated for the first time in APAP-induced acute liver injury by a deep learning-driven chemical derivatization-tandem mass spectrometry strategy, and the sequence changes up to octasaccharides within the bioactive domain “GlcA-GlcNS6S” were successfully decoded. GAG-Explorer, a software incorporated with a comprehensive deep learning model capable of predicting the fragmentation patterns of HS oligomers under actual MS/MS condition was developed to facilitate large-scale sequencing of natural HS structures. The HS alterations in the sequence aspect were elucidated thoroughly in APAP-induced acute liver injury, rather than their compositional changes, which is of great significance for the applications of HS-based therapeutic agents in the biomedical field.