Machine Learning Meets Lipid Nanoparticles for Next-Gen mRNA Delivery

Endosomal escape remains a major bottleneck in non-viral mRNA delivery. In this talk, I will present our development of polymer-lipid hybrid nanoparticles (pLNPs) that harness pH-responsive functional groups to trigger dual endosomal escape mechanisms: proton sponge swelling and membrane destabilization. Through systematic design and machine learning-guided optimization, we identified polymer architectures that achieve up to 700-fold enhanced mRNA delivery to the liver compared to standard LNPs. Our optimized pLNPs enabled efficient cytosolic delivery and achieved 8% in vivo base editing at just 0.1 mg/kg. This work establishes a closed-loop framework combining rational design, mechanistic insights, and graph-based ML to accelerate the development of next-generation RNA therapeutics.