Poster W17 - Adipocyte-specific mRNA lipid nanoparticles for cell reprogramming

Introduction: Adipose tissue is a heterogenous, multifunctional organ composed of diverse cell types largely responsible for energy storage and expenditure. With increasingly recognized complexity and endocrine function, adipose tissue represents an underexplored therapeutic target. Non-viral nucleic acid therapeutic approaches offer great potential to reprogram adipocyte cell states, disrupting aberrant adipokine signaling and reducing adipocyte dysfunction associated with obesity and other metabolic diseases. While most lipid nanoparticle (LNP)-based nucleic acid delivery platforms conserve the use of four components: an ionizable lipid, helper lipid, cholesterol, and PEGylated lipid, there lacks systematic understanding of the effects of composition and helper lipid charge type on adipocyte transfection. As adipose tissue is home to abundant immune populations, we also aimed to conduct a comparative analysis against macrophages, which represent the greatest fraction of adipose-tissue immune cells, to identify features of immune vs adipocyte selectivity that can be extrapolated from this screening data. Here, we combine high-throughput experimental screening with machine learning analysis to develop optimized LNP formulations for adipocyte-specific transfection from parallel screening in adipocytes vs. macrophages, unveil the LNP features that drives this selectivity, and improve immune evasion of LNP-mediated therapeutics directed to adipocytes.

Materials and

Methods: To optimize for adipocyte delivery, we screened 649 formulations with SM-102 as the ionizable lipid, helper lipids of cationic, zwitterionic, and anionic charge and varying concentrations of cholesterol and DMG-PEG-2000 (Figure 1a) [1,2]. LNPs were prepared using our previously established method [1,3,4] using a firefly luciferase mRNA (TriLink). An adipose cell line differentiated from 3T3-L1 cells (ATCC) following a 10-day maturation protocol using insulin (INS, 5 μg/ml), dexamethasone (DEX,1 µM), and 3-isobutyl-1-methylxanthine (IBMX, 0.5mM) was used for this study; and adipocyte maturity was validated using Oil Red O Staining and qPCR for pro-adipogenic genes. Murine macrophages RAW 264.7 (ATCC) were used for parallel screening (Figure 1b). Cells were lysed at 24 h after LNP dosing and transfection efficiency was measured by the relative luminescence units (RLU) detected with a microplate reader. The machine-learning pipeline was adopted for cell-preferential LNP discovery to target adipocytes [2] in which we elucidated trends in transfection efficiency using game-theory approaches, namely, comparative Shapley Additive Explanations (SHAP) analysis, which provide unified measures of feature importance.

References:
1. Zhu Y, et al. Nat. Commun. 2022; 13(1): 4282.
2. Cheng L, et al. bioRxiv. [Preprint]. 2023 Dec. 8:2023.12.07.570602.
3. Zhu Y, et al. Nat. Biomed. Eng. 2023 Dec 11; doi: 10.1038/s41551-023-01131-0.
4. Zhu Y, et al. Biomaterials. 2024; 308: 122559.

Results, Conclusions, and Discussions: Of the 649 formulations screened and benchmarked against the Moderna formulation used in the SARS-CoV-2 mRNA LNP vaccine, over 10% of the formulations across all six helper lipids investigated outperformed the Moderna formulation in transfection efficiency alone. The results of the regression analysis direct us to formulations that have higher transfection efficiency in adipocytes when compared to macrophages. SHAP analysis indicated helper lipid content, charged lipid content, and PEGylated lipid content are the top three features of importance (Figure 1c). Understanding such features that drive uptake and selectivity elucidate new parameter space to iterate formulation design as well as direct the synthesis of de novo lipids that can maximize all features deemed most preferential to adipocytes. In addition to guiding iterative design optimization, these results are undergoing in vivo murine validation using a local injection model in the inguinal fat pad. Since adipose tissue is directly accessible through injection, combining local delivery with inherent compositional selectivity may enable the reduction of off-target effects and improve the safety profile relative to systemic delivery approaches. An additional advantage of this platform is the ability for redosing. As we anticipate this platform will guide future development of adipocyte-specific therapeutics, specific LNP formulations developed here have the potential to limit off-target effects through cellular selectivity and reduction of the overall therapeutic dosage needed.

Acknowledgements (Optional):