Search for: All records

The pKa values and associated protonation states of ionizable lipids in lipid nanoparticle (LNP) formulations are strongly dependent on their chemical environment. This phenomenon leads to poorly understood structure-function relationships that influence payload delivery, tissue-selective biodistribution, and manufacturing. For example, the charge- and biodistribution of an mRNA-loaded LNP can vary based on the type of ionizable lipid used, the molar ratio of its components, and its cargo. Yet, the spatial resolution of experimental protonation state measurements is currently limited to the apparent charge of an ionizable lipid averaged over all environments/conformations of an LNP — best represented by its apparent pKa value. Such measurements are too coarse to capture the heterogenous charge distributions of ionizable lipids in LNPs, which influence biocorona formation and interactions with the payload. Similar limitations are inherent to classical fixed protonation-state in silico models that cannot capture the environment-dependent protonation states and pKa values determining local pKa. To address this gap in experimental and computational tools available to accurately determine the local charge distributions in LNPs, this work now incorporates a scalable continuous constant pH molecular dynamics (CpHMD) model to simulate the self-assembly processes of five reported distinct LNP formulations. Parameters for ionizable lipids were generated from fitting fixed lambda-state calculations performed with Hamiltonian replica exchange (HREX) to improve conformational sampling during parameterization. Simulated systems were composed of 100 ionizable lipids (50 mol%), cholesterol (40 mol%), distearoylphosphatidylcholine (10 mol%), and mRNA (20 nucleotides) to model the interior of an LNP. Self-assembly was simulated for 100 ns at different pH values to validate the apparent pKa for each system. The theoretically calculated apparent pKa values matched reasonably well with those measured experimentally (mean absolute error = 0.5 pKa units), and all systems exhibited pH-dependent structures. Overall, this work provides a new computational platform technology to (i) predict the pKa values of ionizable lipids in different chemical environments and (ii) enable a structure-based way to model the heterogeneous, environment-dependent charge distributions of ionizable lipids in LNP systems typically encountered during LNP manufacturing and delivery.