Optimized microfluidic formulation and organic excipients for improved lipid nanoparticle mediated genome editing

Abstract

mRNA-based gene editing platforms have tremendous promise in the treatment of genetic diseases. However, for this potential to be realized in vivo, these nucleic acid cargos must be delivered safely and effectively to cells of interest. Ionizable lipid nanoparticles (LNPs), the most clinically advanced non-viral RNA delivery system, have been well-studied for the delivery of mRNA but have not been systematically optimized for the delivery of mRNA-based CRISPR-Cas9 platforms. In this study, we investigated the effect of microfluidic and lipid excipient parameters on LNP gene editing efficacy. Through in vitro screening in liver cells, we discovered distinct trends in delivery based on phospholipid, cholesterol, and lipid-PEG structure in LNP formulations. Combination of top-performing lipid excipients produced an LNP formulation that resulted in 3-fold greater gene editing in vitro and facilitated 3-fold greater reduction of a therapeutically-relevant protein in vivo relative to the unoptimized LNP formulation. Thus, systematic optimization of LNP formulation parameters revealed a novel LNP formulation that has strong potential for delivery of gene editors to the liver to treat metabolic disease.

Fig. 1. Optimization of LNP conditions for co-delivery of gene editing cargo. (A) Schematic describing the overall study design, where microfluidic and lipid excipient parameters were optimized to enhance LNP gene editing efficacy. (B) Depiction of the two nucleic cargos trialed and compared across microfluidic flow rates. (C) and (D) Size (left axis, bars) and PDI (right axis. pink dots) of LNPs produced at different total flow rates (0.3–3.6 mL min⁻¹) encapsulating either GFP mRNA or SpCas9 mRNA and GFP sgRNA. (E) and (F) mRNA encapsulation efficiency of LNPs produced at different flow rates. (G) GFP transfection resulting from delivery of GFP mRNA to HepG2 cells via LNPs produced at different flow rates. (H) Gene editing resulting from delivery of Cas9 mRNA and GFP sgRNA to HepG2-GFP cells via LNPs produced at different flow rates. One-way ANOVA with post hoc Dunnett's test was used for statistical comparison (ns = non-significant, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001); all data reported as mean ± SEM (minimum n = 3).

Fig. 2. Optimization of phospholipid structure for co-delivery of gene editing cargo. (A) Structures of phospholipids incorporated into LNPs. (B) Size (left axis, bars) and PDI (right axis, pink dots), (C) RNA encapsulation efficiency (EE%), and (D) zeta potential of LNPs produced with different phospholipids to encapsulate SpCas9 mRNA and GFP sgRNA. (E) Gene editing resulting from delivery of Cas9 mRNA and GFP sgRNA to HepG2-GFP cells via LNPs produced with different phospholipids. (F) Viability of cells treated with LNPs produced with different phospholipids. One-way ANOVA with post hoc Dunnett's test was used for statistical comparison relative to DOPE LNPs (ns = non-significant, * = p < 0.05, ** = p < 0.01, **** = p < 0.0001); all data reported as mean ± SEM (minimum n = 5).

Fig. 3. Optimization of cholesterol structure for co-delivery of gene editing cargo. (A) Structures of cholesterols incorporated into LNPs. (B) Size (left axis, bars) and PDI (right axis, pink dots), (C) RNA encapsulation efficiency (EE%), and (D) zeta potential of LNPs produced with different cholesterols to encapsulate SpCas9 mRNA and GFP sgRNA. (E) Delivery of Cas9 mRNA and GFP sgRNA to HepG2-GFP cells via LNPs produced with different cholesterols. (F) Viability of cells treated with LNPs produced with different cholesterols. One-way ANOVA with post hoc Dunnett's test was used for statistical comparison relative to cholesterol LNPs (ns = non-significant, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001); all data reported as mean ± SEM (minimum n = 5).

Fig. 4. Optimization of lipid-PEG structure for co-delivery of gene editing cargo. (A) Structures of lipid-PEGs incorporated into LNPs. (B) Size (left axis, bars) and PDI (right axis, pink dots), (C) RNA encapsulation efficiency (EE%), and (D) zeta potential of LNPs produced with different lipid-PEGs to encapsulate SpCas9 mRNA and GFP sgRNA. (E) Delivery of Cas9 mRNA and GFP sgRNA to HepG2-GFP cells via LNPs produced with different lipid-PEGs. (F) Viability of cells treated with LNPs produced with different lipid-PEGs. One-way ANOVA with post hoc Dunnett's test was used for statistical comparison relative to DMPE-PEG LNPs (ns = non-significant, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001); all data reported as mean ± SEM (minimum n = 5).

Fig. 5. Delivery of gene editing cargo in vitro and in vivo with excipient-optimized LNPs. (A) Table depicting formulations for excipient-optimized LNPs using the top-performing phospholipids (DOPE, DSPC), cholesterols (stigmastanol, β-sitosterol), and lipid-PEGs (DMG-PEG, DTA-PEG). Formulations are labeled R1–R8. Base formulation is represented as R0. (B) Delivery of Cas9 mRNA and GFP sgRNA to HepG2-GFP cells via LNPs produced with different combinations of excipients. (C) Viability of cells treated with LNPs produced with different combinations of excipients. (D) Experimental groups for in vivo assessment of LNP gene editing efficacy. PBS was used as a negative control. MC3 LNPs are an FDA-approved LNP formulation. R1, R3, R4, and R8 LNPs were the top-performers from the in vitro screen and were compared against R0 LNPs. All LNPs were formulated to encapsulate SpCas9 mRNA and TTR sgRNA. (E) C57BL/6 mice were treated with LNPs at a dose of 1 mg kg⁻¹ and sacrificed after 5 days. Serum was collected before and after LNP treatment and analyzed for TTR protein via ELISA. (F) AST and (G) ALT levels in the serum of mice treated with LNPs relative to PBS-injected controls. (H) Next-generation sequencing analysis for insertions and deletions (indels) at the expected locus for TTR gene editing in liver genomic DNA of PBS-, R0 LNP-, or R4 LNP-treated animals. One-way ANOVA with post hoc Dunnett's test was used for statistical comparison relative to R0 LNPs unless otherwise denoted (ns = non-significant, * = p < 0.05, *** = p < 0.001, **** = p < 0.0001); all data reported as mean ± SEM (minimum n = 5).