AI-driven prediction and experimental validation of nucleolin phase separation during aging
Y Li, Y Xu, Y Sun, B Liu, X Yu, Y SongAbstract
Background
Cardiac aging is a major risk factor for cardiovascular diseases, with cardiomyocyte senescence being a key contributor. Ribosome biogenesis plays a central role in driving cardiomyocyte cell cycle reentry, yet its activators or upstream regulators remain unclear. Nucleolin (NCL), a cardioprotective protein, may exert its function through liquid–liquid phase separation (LLPS).
Aims
To investigate whether activation of NCL LLPS mitigates aging.
Methods
We developed a novel three-stage, iterative artificial intelligence (AI) framework to predict the LLPS potential of NCL. Cardiomyocyte-specific circHIPK3 knockout (CKO) mice were generated, with a lifespan not exceeding 18 months. Using animal models, cellular assays, and molecular techniques—including qPCR, Western blotting, RNA pull-down, fluorescence recovery after photobleaching (FRAP), chromatin isolation by RNA purification (CHIRP), and Ribo-Halo assays—we examined the interaction between NCL and circHIPK3.
Results
AI analysis indicated that NCL possesses LLPS potential, and this capacity declines with aging. In mouse hearts, circHIPK3 expression decreased with age. Its knockout led to cardiac dysfunction, fibrosis, and aging-related phenotypes, including activation of the p21 pathway. Mechanistically, circHIPK3 directly bound to the 5′UTR of NCL mRNA, enhancing its stability and preserving NCL-mediated LLPS, thereby promoting ribosome biogenesis. Overexpression of circHIPK3 restored NCL expression and suppressed the aging-associated p21 pathway in knockout mice.
Conclusion
We identified a novel "circHIPK3–NCL LLPS–ribosome biogenesis" axis that protects against cardiac aging through post-transcriptional regulation. circHIPK3 may serve as a potential biomarker and therapeutic target for age-related cardiac dysfunction.