Supporting data for "Uncovering cellular response to ionizable lipid-induced endolysosomal damage with multimodal imaging"

Abstract

Lipid nanoparticles (LNPs) are the leading non-viral platform for nucleic acid delivery, validated clinically by their use in mRNA vaccines and the siRNA drug patisiran. Nevertheless, their wider use in protein-replacement and gene-editing therapeutics need repeated doses is limited by inflammatory side effects caused by LNPs, which are often associated with ionizable lipid-induced damage to endolysosomal membranes. How cells sense, process, and respond to this damage at the subcellular level remains poorly understood, limiting rational, mechanism-based formulation design and leaving lipid discovery mainly dependent on empirical screening.This thesis addresses this gap using multimodal imaging and cell biology. First, we developed NanoSIMS Stabilizer, an optical-flow-based registration tool built on the RAFT algorithm, to correct rigid and non-rigid distortion in multi-frame NanoSIMS acquisitions and enable high-fidelity subcellular isotope imaging. Applying this tool with deuterium-labeled ionizable lipids and correlative light, electron, and ion microscopy revealed that ionizable lipids localize almost exclusively to endolysosomes at 4 hours post-treatment, with partial redistribution to the endoplasmic reticulum, Golgi apparatus, mitochondria, and particularly lipid droplets by 24 hours, while endolysosomes remained the dominant reservoir at both time points. In tissues, we also validated ionizable lipid endolysosomal retention in the liver and spleen.Mechanistically, LNP treatment triggered endolysosomal membrane damage, exposing glycans recognized by galectins. This damage activated a calcium-dependent, non-canonical autophagy response in which ATG8/LC3B is conjugated directly onto the damaged lysosomal limiting membrane, dependent on ATG7 and ATG16L1 but independent of the canonical ULK1 initiation complex. This response promoted endolysosomal repair and restricted LNP-induced cytotoxicity and inflammation.Building on this mechanism, pharmacological activation of the calcium-driven pathway using the TRPML1 agonist ML-SA5 failed to reduce LNP-induced IL-1β secretion. Instead, LNP damage independently triggered calcium-dependent, cytosolic exposure of sphingomyelin, a second protective repair response. Incorporating sphingomyelin as a fifth lipid component into conventional LNP formulations reduced endolysosomal damage, restored IL-1β secretion to baseline, and simultaneously enhanced mRNA uptake and expression efficiency.Together, these findings establish non-canonical autophagy and sphingomyelin exposure as active, complementary cellular repair responses to LNP-induced endolysosomal damage and demonstrate that engineering LNPs to support these endogenous repair pathways can decouple delivery potency from inflammatory toxicity, overcoming a long-standing trade-off in LNP design. Beyond the finding that incorporating sphingomyelin improves LNP safety and potency, this work also provides mechanistic, cell-biology-driven insights for the rational development of safer, more effective LNP-based therapeutics.