AI-driven therapeutic antisense oligonucleotide for processing-deficient progeroid laminopathies.

BACKGROUND: Progeroid laminopathies (PLs), including Hutchinson-Gilford progeria syndrome (HGPS), are rare premature aging disorders in which cardiovascular complications drive early mortality. Antisense oligonucleotides (ASOs) represent a promising therapeutic strategy, yet optimal design principles and their impact on cardiovascular pathology remain insufficiently defined. METHODS: We developed an AI-driven pipeline to design ASOs targeting the 3'UTR of LMNA transcripts to suppress processing-deficient pathogenic lamin A isoforms. Lead candidates were evaluated in patient-derived induced pluripotent stem cell-derived cardiomyocytes (PL-iCMs), 3D cardiac organoids, and LMNA transgenic mouse models. Efficacy, toxicity, and systemic transcriptional responses were assessed using molecular, histological, and serum biochemical analyses. FINDINGS: The optimized ASO, LM2556, selectively reduced progerin and farnesylated pre-lamin A expression while preserving lamin C. LM2556 mitigated cellular senescence and improved structural and functional phenotypes in PL-iCMs and cardiac organoids. In vivo, LM2556 decreased pathogenic lamin A isoforms across multiple tissues without evidence of hepatotoxicity or nephrotoxicity. Long-term administration ameliorated progeroid features, improved cardiovascular pathology, extended median lifespan by 82.86%, and enhanced overall healthspan in LMNA transgenic mice. CONCLUSIONS: These findings establish an AI-based framework for therapeutic ASO design and provide proof-of-concept evidence that targeting the 3'UTR of LMNA transcripts can effectively suppress pathogenic lamin A isoforms and ameliorate systemic progeroid phenotypes. FUNDING: This work was funded by the National Natural Science Foundation of China (82450116 and 82502251).