AI-driven discovery of dual antiaging and anti-AD therapeutics via PROTAC target deconvolution of a super-enhancer–regulated axis
Yuan Sun, Sai Liu, Long Chen, Zheng Zhou, Xin Yin, Yiting Shi, Haotian Li, Jinran Li, Yi Lü, Wei Jiang, Yongjun Zhao, Tucheng Dai, Tingting Yao, A. Li, Xinyu Bi, Beiyu Zhang, Xiaoxia Shen, Zheying Zhu, Guangji Wang, Xinuo Li
Abstract
The lack of safe, durable therapeutics that act against both biological aging and Alzheimer's disease is an unmet clinical need. To bridge this gap, we devised an artificial intelligence (AI)-enabled approach that pairs rapid compound triage with mechanistic target deconvolution. Our AI-driven screening highlighted melatonin (MLT) as a promising candidate. Serum profiling of 161 human individuals confirmed an age-related fall in circulating MLT level, while subsequent in vivo and in vitro experiments showed that MLT rescues cognition, suppresses neuroinflammation, and alleviates senescence phenotypes. Proteolysis targeting chimera (PROTAC)-guided chemoproteomic deconvolution next pinpointed the histone acetyltransferase p300 as MLT's target. Integrated Cleavage Under Targets and Tagmentation, single-cell RNA sequencing, and spatial transcriptomics revealed that MLT-bound p300 cooperates with specificity protein 1 (SP1) at a brain and muscle ARNT-like protein 1 super-enhancer, elevating histone H3 lysine-27 acetylation and reengaging a circadian-epigenetic program that links redox resilience to neuroprotection. By combining AI-driven discovery with PROTAC-based target mapping and super-enhancer-centric mechanistic resolution, our study identifies MLT as a dual-action candidate and sets out a reproducible "AI-to-clinic" paradigm for multitarget drug innovation in aging-related neurodegeneration.