Automated and Parallelized Microfluidic Generation of Large and Precisely Defined Lipid Nanoparticle Libraries

Abstract

Lipid nanoparticles (LNPs) are being developed for a broad set of therapeutic applications by changing both the structures of the lipids used to formulate each LNP and their relative proportions. Because lipid synthesis and in vivo screening have been parallelized using combinatorial chemistry and LNP barcoding, respectively, the manual and sequential microfluidic formulation of LNPs remains the primary rate-limiting step during early-stage discovery. In this work, we present a parallelized, automated microfluidic platform capable of generating large, precisely defined LNP libraries in parallel, with throughput on the order of 1000 distinct formulations per hour. Each formulation is defined by varying the reagent flow ratios into one of eight microscale mixers using lithographically encoded fluidic resistors and dynamically controlled external pressure supplies. The microfluidic chip is integrated with custom robotic plate handling for the rapid collection of each distinct formulation. To evaluate this platform, we characterized 96 formulations generated on-chip in terms of both physicochemical properties and transfection efficiency in vitro. We further validated our lead candidate against the state of the art in vivo. We demonstrate the ability to rapidly discover a formulation and scale its production to liters per hour under identical mixing conditions, bridging from early discovery to manufacturing through microfluidic parallelization.