Deep-learning-predicted sinomenine ameliorates metabolic dysfunction-associated steatotic liver disease by delaying cellular senescence via NES

Metabolic dysfunction-associated steatotic liver disease (MASLD) is the most common metabolic liver disorder worldwide, and its burden is growing as populations age and obesity rates rise. Current therapies only partially correct metabolic disturbances and fail to reverse hepatocyte senescence or restore functional capacity. Here, we built a deep-learning efficacy prediction system (DLEPS) powered by L1000 transcriptomic data to screen bioactive compounds and identified the natural alkaloid sinomenine (Sino) as a potent MASLD modulator. Mechanistically, Sino suppresses the transcriptional activity of NF-κB and thereby downregulates NES expression. NES emerged as an important intermediate mediator linking NF-κB signaling to hepatocyte senescence. Reduced NES expression was associated with lower Smad3 phosphorylation and reduced P53/P21 pathway activation, delays cellular senescence, inhibits the senescence-associated secretory phenotype (SASP), and restores normal mitochondrial structure and metabolic function. To enhance hepatic delivery, we further engineered galactose-lipid-membrane-coated Sino nanoparticles (Gal-SiNPs), which enhanced hepatic accumulation and retained therapeutic activity at a relatively lower administered dose. Collectively, deep-learning-guided discovery of Sino reshapes hepatocyte function via the NES -senescence pathway, and its nano-targeted formulation shows promising therapeutic potential for MASLD therapy and merits further translational exploration.