Precision engineering of anti-atherosclerotic herbal nanomedicine: from machine learning-aided active components screening to optimized metal-phenolic network codelivery.

Atherosclerosis (AS), a chronic inflammatory disease linked to oxidative stress and lipid imbalance, remains a major cardiovascular threat. Traditional herbs Salvia miltiorrhiza and Carthamus tinctorius exhibit multi-target anti-AS potential, yet their compositional complexity limits clinical translation. This study aimed to systematically identify core anti-AS components from these herbs and enhance their anti-AS efficacy via machine learning-aided screening and nanotechnology-driven codelivery. We initially pioneered a machine learning-aided hybrid strategy integrating network pharmacology and quantitative activity relationship (QSAR) modeling to identify four core anti-AS polyphenols (i.e., salvianic acid A, salvianolic acid B, protocatechuic acid, and hydroxysafflor yellow A). Subsequently, a quaternary metal-phenolic network (SSPH-MPN) was engineered for plaque-targeted codelivery, optimized via the median-effect principle for achieving a synergistic effect based on ROS scavenging efficacy. The optimized SSPH-MPN was characterized by a series of studies, including molecular dynamics simulations, UV, DLS, TEM, FTIR, XPS, and ICP-MS. The anti-AS effect of the optimized SSPH-MPN was evaluated by monitoring oxidative status (ROS levels, antioxidant enzymes SOD, GSH-Px, MDA, T-AOC), inflammatory markers (IL-1β, IL-6, TNF-α), lipid metabolism (DiI-oxLDL uptake, cholesterol efflux, blood lipid levels, lipid accumulation), and plaque areas. The results demonstrated that the optimized SSPH-MPN showed great efficiency in inhibiting lipid uptake and accumulation, and mediating cholesterol efflux in RAW 264.7 cells, and exhibited improved lipid metabolism, attenuated oxidative stress and inflammation, thus acquired diminished plaque area in apoE