Machine learning engineered PoLixNano nanoparticles overcome delivery barriers for nebulized mRNA therapeutics

There continues to be a dearth of competent inhalable mRNA delivery although it holds great potential for addressing a wide variety of refractory diseases. The huge advances seen with parenteral-administered lipid nanoparticle (LNP) have not been translated into nebulized mRNA delivery due to the aggressive nebulization process and insurmountable barriers inherent to respiratory mucosa. Here, we show amphiphilic block copolymers revealed by machine learning (ML) can spontaneously form stabilized nanoparticles (PoLixNano) with the lipids components of LNP and simultaneously impart the PoLixNano with "shield" (shear force-resistant) and "spear" (pulmonary barriers-penetrating abilities) capabilities. We present a ML approach that leverages physicochemical properties and inhaled mRNA transfection profiles of a chemically diverse library of polymeric components to validate the integration of "shield" and "spear" properties as highly predictive indicators of transfection efficiency. This quantitative structure-mRNA transfection prediction (QSMTP) model identifies top-performing amphiphilic-copolymers from more than 10000 candidates and suggests their mucus-penetrating ability outweights the shear force-resistant property in contributing to efficient mRNA transfection. The optimized PoLixNano substantially outperforms the LNP counterpart and mediates up to 1114-times higher levels of mRNA transfection in animal models with negligible toxicities. The PoLixNano promotes overwhelming SARS-CoV-2 antigen-specific sIgA antibody secretion and expansion of TRM cells which collectively confers 100% protection in mice against lethal SARS-CoV-2 challenges and blocks the transmission of Omicron variant between hamsters. PoLixNano also displays versatile therapeutic potential in lung carcinoma and cystic fibrosis models. Our study provides new insights for designing delivery platforms of aerosol-inhaled mRNA therapeutics with clinical translation potential.