Lipid nanoparticles, or LNPs, have become one of the most important technologies in modern medicine. They carry fragile genetic instructions into cells, allowing researchers to replace missing proteins, silence harmful genes, stimulate immune responses, or edit DNA. The success of mRNA vaccines against COVID-19 demonstrated that these microscopic delivery vehicles can move from laboratory concept to global clinical use. Yet the same technology that efficiently transports RNA also has a powerful biological preference: after entering the bloodstream, conventional LNPs overwhelmingly travel to the liver. A new review of the field describes how scientists are attempting to rewrite that preference, transforming LNPs from liver-seeking carriers into programmable systems capable of reaching the lungs, spleen, brain, kidneys, bones, heart, and pancreas.
The liver bias begins almost immediately after intravenous injection. As LNPs circulate through blood, they encounter thousands of plasma proteins that rapidly attach to their surfaces and create what researchers call a protein corona. Among the most influential proteins is apolipoprotein E, or ApoE. Once ApoE coats an LNP, the particle begins to resemble a natural low-density lipoprotein particle. Hepatocytes, the principal cells of the liver, carry abundant low-density lipoprotein receptors, or LDLRs. These receptors recognize ApoE and draw the coated LNPs into cells through receptor-mediated endocytosis. Experiments in mice lacking functional LDLR have shown that this pathway is essential for the efficient hepatic accumulation of many LNP formulations. The liver also contains highly fenestrated sinusoidal blood vessels and Kupffer cells, resident macrophages that act as biological filters, further increasing the organ’s ability to capture circulating nanoparticles.
This natural pathway has been valuable for treating liver diseases. The first approved RNA interference medicine, Onpattro, uses an LNP formulation to deliver siRNA against transthyretin in hereditary transthyretin amyloidosis. Other clinically advanced therapies use similar principles to deliver mRNA or gene-editing components to hepatic cells. However, liver accumulation becomes a serious obstacle when the intended target lies elsewhere. A particle that disappears into hepatocytes cannot efficiently correct a defective gene in the lung, reprogram immune cells in the spleen, cross the blood–brain barrier, or deliver therapeutic RNA to bone tissue. The review argues that extrahepatic delivery cannot be solved by simply attaching one more targeting ligand. Instead, researchers must coordinate three biological design layers: the chemical composition of the LNP, its measurable physical properties, and the protein corona that forms after administration.
The first layer is molecular engineering. Standard LNPs usually contain an ionizable lipid, a helper phospholipid, cholesterol, and a PEG-linked lipid. Ionizable lipids remain mostly neutral at physiological pH, which helps limit nonspecific interactions and toxicity in the bloodstream, but become positively charged in acidic endosomes. This charge shift allows them to interact with negatively charged endosomal membranes, promote a transition toward inverted hexagonal lipid phases, and disrupt the membrane so that RNA can escape into the cytoplasm. Researchers can alter the ionizable lipid’s head group, hydrophobic tails, branching pattern, or biodegradable linkers to influence pKa, membrane behavior, degradation, and tissue distribution. Small structural changes can have large consequences: piperazine-containing lipids have shown lung enrichment, imidazole-based lipids have favored the spleen, and specialized biodegradable structures have improved delivery to pulmonary tissue.
The supporting lipids are equally important. Helper phospholipids such as DSPC and DOPE influence membrane packing, curvature, stability, and endosomal escape. Cholesterol strengthens the particle structure, but replacing it with sterols derived from bile acids, plants, or endogenous corticosteroids can redirect biodistribution. PEG-lipids form a hydrated surface layer that reduces aggregation and delays uptake by macrophages, but they also create a well-known “PEG dilemma.” Too much PEG can prevent tissue penetration and interfere with protein adsorption, while short PEG-lipid anchors may detach rapidly and expose the surface to ApoE, encouraging liver uptake. Scientists are therefore testing cleavable PEG structures, fluorinated PEG lipids, branched architectures, and PEG-free formulations. Some newer systems simplify the traditional four-component design, while others add a fifth functional lipid to obtain more precise organ selectivity.
One of the most influential examples of this approach is the Selective Organ Targeting, or SORT, platform. By adding a charged lipid to a conventional LNP, researchers can substantially change its apparent pKa, surface charge, and protein adsorption profile. Cationic or quaternary ammonium lipids have been used to promote lung delivery, while anionic lipids can redirect particles toward the spleen. The mechanism depends partly on the protein corona generated by the altered surface. Lung-targeting formulations often become enriched in vitronectin, a plasma protein that can bind integrin αvβ3 on pulmonary endothelial cells. Spleen-targeting particles may recruit beta-2-glycoprotein I and interact with CD169-positive macrophages. SORT has supported mRNA expression and siRNA silencing in the lungs, spleen, kidneys, and other tissues, although some liver accumulation generally remains and highly charged formulations may raise concerns about toxicity, inflammation, or thrombosis.
A newer strategy called POST attempts to make corona engineering more programmable. Rather than using a peptide as a conventional ligand that directly binds a cell-surface receptor, POST attaches short peptide sequences to the LNP surface to influence which plasma proteins assemble around the particle. In preclinical studies, arginine-rich sequences such as 6R have been associated with lung localization, whereas acidic 6D sequences have promoted spleen accumulation. The peptide acts as a molecular code that changes protein adsorption kinetics and corona composition, allowing endogenous proteins to perform the final tissue-recognition step. Another platform, known as ENDO, uses vitamin D3 as an additional component to recruit vitamin D receptor-associated proteins and produce striking pancreatic selectivity in mice. Reports describe more than 99% selectivity for pancreatic tissue in certain models, including delivery to insulin-producing beta cells. Yet the mechanism requires broader independent validation, and its dependence on vitamin D receptor biology may limit its applicability across organs and species.
The consequences of these engineering strategies are now visible in disease models. In the lung, targeted LNPs have delivered VEGFA mRNA to support vascular repair, BMPR2 mRNA to reverse pulmonary arterial hypertension, CFTR mRNA to restore chloride transport in cystic fibrosis models, and siRNA against RUNX1 to reduce pulmonary fibrosis. In the spleen, optimized particles have delivered cancer-vaccine mRNA to antigen-presenting cells and natural killer cells, generating antitumor immune responses while reducing liver exposure. Brain delivery remains considerably more difficult because of the blood–brain barrier, a tightly regulated network of endothelial cells that excludes most large and hydrophilic molecules. Nevertheless, berberine-inspired lipids, neurotransmitter-derived lipids, furan-based particles, and receptor-targeting peptides have produced measurable brain delivery in animals. Other researchers are bypassing the barrier through lymphatic or receptor-mediated routes, or combining brain penetration with targeting of microglia and tumor cells.
The heart, kidneys, and bones illustrate both the promise and the limitations of the field. In rabbits, intracoronary administration of mRNA-LNPs produced broader cardiac distribution than intravenous or intramyocardial injection and improved heart function after infarction. This is effective, but it is route-dependent enrichment rather than systemic targeting achieved through a programmable biological identity. Systemic cardiac delivery remains inefficient. In the kidneys, LNPs carrying SOD2 mRNA have helped restore redox balance after ischemia–reperfusion injury, while vitronectin-recruiting particles have targeted αvβ3 integrin-positive clear-cell renal tumors. For bone, bisphosphonate-containing ionizable lipids bind hydroxyapatite, the mineral component of bone, and improve delivery to the bone microenvironment. Cartilage-targeting particles have also carried IGF-1 mRNA to damaged joints, promoting chondrocyte survival and tissue repair. These advances suggest that each organ may require a distinct combination of chemistry, size, charge, route, and corona biology.
The central challenge now is translation. Protein coronas are dynamic and highly dependent on the surrounding biological environment. ApoE, vitronectin, beta-2-glycoprotein I, albumin, and other proteins differ in concentration and binding behavior between mice, rats, nonhuman primates, and humans. Even sex, age, strain, disease state, and plasma composition can change the corona and alter organ distribution. Repeated administration introduces another obstacle: anti-PEG antibodies and the accelerated blood-clearance phenomenon can rapidly remove subsequent doses, while immune complexes may trigger complement activation or inflammatory reactions. Manufacturing also matters. Small changes in microfluidic mixing, flow rate, temperature, or component concentration can alter particle size, polydispersity, RNA encapsulation, and biological activity. Freeze-drying could ease cold-chain requirements, but freezing and dehydration may cause aggregation, RNA leakage, and loss of targeting performance. The field is therefore moving toward artificial-intelligence-assisted lipid discovery, proteomic corona mapping, organ-on-chip testing, nonhuman-primate validation, and quality-by-design manufacturing. The long-term goal is no longer merely to make an LNP that carries RNA, but to create a reproducible particle whose chemical structure, physical behavior, protein corona, cellular destination, and therapeutic effect can be predicted before it reaches a patient.
