Machine Learning-Informed Nano Co-Assembly Inhibits Fibroblast Activation Protein and Improves Drug Delivery in Fibrotic Tissue.

Abstract

Nanoparticle-based drug delivery faces persistent challenges, including complex fabrication processes and limited lesional accumulation. Here we introduce SP-13786 (SP), a precise small-molecule inhibitor of fibroblast activation protein (FAP), as a universal and effective excipient enabling facile co-precipitation into stable nanoparticles (SCAN) with diverse hydrophobic drugs. Screening of 861 compounds revealed a broadly enhanced colloidal stability and drug loading by SP. Corresponding simulations and explainable machine learning (XML) showed SCAN assembly hinges on balanced aromaticity, rigidity, and nitrogen-mediated interaction, offering interpretable framework for co-assembly nanomedicine. Biological assessment demonstrate that SCAN enhances drug delivery and therapeutic efficacy in FAP-positive cells, therefore attentuate the fibrosis-induced drug penetration barriers, increasing drug accumulation within the fibrotic tissue. The improved bioavailability correlate with superior therapeutic outcomes in multiple disease models with progressive fibrosis. Overall, we establish SP as a versatile nanotherapeutic platform combining simplicity in preparation, mechanistic insights provided by XML, and broad applicability for diseases characterized by pathological fibrosis and impaired drug delivery.