ELEVATE-LNP: Empty lipid nanoparticle enabled validation and AI-guided therapeutic evaluation for efficient formulation optimization for mRNA delivery

Lipid nanoparticles (LNPs) have emerged as a leading platform for mRNA delivery, particularly following the success of SARS-CoV-2 mRNA vaccines. However, identifying effective LNP formulations remains time-consuming because of the large and complex multicomponent formulation space. Here, we present ELEVATE-LNP an Empty-LNP-Enabled Validation and AI-Guided Therapeutic Evaluation platform by integrating a high-throughput parallelized microfluidic preparation with physicochemical characterization and machine learning for mid-stream and downstream LNP formulation optimization when ionizable lipids are shortlisted. A broad formulation space comprising four ionizable lipids, different helper lipids, lipid ratios, and process conditions was first screened using empty LNPs to identify formulations with desirable particle size and polydispersity index (PDI). Machine-learning analysis was then used to determine the formulation and process parameters associated with empty-LNP properties. Selected candidates were subsequently evaluated following mRNA encapsulation by assessing particle size, PDI, encapsulation efficiency, cell viability, and transfection performance. The results revealed lipid-dependent relationships between formulation parameters and the physicochemical properties of empty LNPs. Six features were strongly associated with particle size, while eleven features contributed to variations in PDI. GFP expression in CFBE cells demonstrated that SM-102-based LNPs achieved higher transfection efficiency than CKK-E12-based formulations. In a separate CFTR mRNA evaluation, DOPC-containing LNPs produced higher CFTR protein expression than their DSPC-containing counterparts. By integrating empty-LNP prescreening with subsequent payload-based validation, the ELEVATE-LNP workflow reduces the number of candidates and the amount of nucleic acid required for functional testing. The platform therefore provides a practical approach for narrowing lipid-ratio and formulation-condition spaces within preselected ionizable lipid systems.