AI-Guided Structure-Aware Modeling and Thermal Proteomics Reveal Direct Demethylzeylasteral-ACLY Interaction

Abstract

Target deconvolution of bioactive natural products (NPs) is frequently hampered by the inability of traditional thermal shift assays to distinguish direct ligand binding from indirect proteomic stabilization. Here, we developed an integrated target discovery framework combining MAPS-iTSA thermal proteomics with a structure-aware graph neural network (HoloGNN). This strategy identified ATP-citrate lyase (ACLY) as a high-confidence target of Demethylzeylasteral, a bioactive triterpenoid from Tripterygium wilfordii. Crucially, orthogonal biochemical assays, including surface plasmon resonance and limited proteolysis, validated its direct binding (KD = 9.86 M) and profound enzymatic inhibition (IC50 = 3.84 M). To establish the relevance of this interaction in the context of the source herb, thermal profiling of Tripterygium wilfordii extract together with ACLY-based affinity-ultrafiltration mass spectrometry supported ACLY engagement and identified Demethylzeylasteral as an ACLY-binding constituent. Given the established role of ACLY-mediated lipid metabolic reprogramming in psoriasis, we further evaluated the pharmacological significance of this interaction. Demethylzeylasteral suppressed keratinocyte proliferation, alleviated imiquimod-induced psoriasiform dermatitis, and reduced inflammatory cytokine expression in vivo. Single-cell RNA sequencing further revealed reversal of ACLY-SREBP-associated lipogenic reprogramming in keratinocytes. Collectively, these findings establish ACLY as a functionally relevant target of DEM and provide a robust AI-chemoproteomic paradigm for mechanism-guided NP drug discovery.