Design approaches for new fat-based particles used in mRNA vaccines and treatments: current knowledge and future outlook

Exosomes have a negatively charged surface and are primarily composed of phosphatidylcholine (10–30%), cholesterol (0–42%), phosphatidylserine (PS) (6–15%), and sphingomyelin (8–23%). They also contain smaller proportions of phosphatidylethanolamine, phosphatidylinositol, and ceramides [19, 20]. Derived from biological fluids like cell culture supernatants, blood, urine, and breast milk, the clinical translation of exosome‐based nucleic acid delivery systems faces several challenges, including large‐scale production issues associated with cell culture methods and purification strategies [21]. Moreover, the lack of standardized critical quality attributes (CQAs) during exosome manufacturing further complicates regulatory approval and quality assessment. Insufficient knowledge of exosome formulations and optimal storage conditions has resulted in low encapsulation efficiency of nucleic acids, introducing further uncertainty about their clinical applicability [22, 23]. Meanwhile, LNPs, with established manufacturing protocols, high encapsulation efficiency, and tunable targeting capabilities, have achieved significant clinical success, highlighting their advantages over exosome‐based delivery systems.

PNPs are nanoscale delivery systems constructed from natural or synthetic polymers, including poly (lactic‐co‐glycolic acid) (PLGA), polycaprolactone, polyethylene glycol (PEG), polyethylenimine (PEI), and chitosan [24]. PLGA‐based NPs have been extensively employed to modulate the release kinetics of small‐molecule drugs, prolonging therapeutic duration [25]. Cationic polymers, such as PEI and polyamidoamine, complex with small interfering RNA (siRNA) or mRNA through electrostatic interactions to enhance cellular uptake and endocytosis [26]. Chitosan‐based NPs have been investigated for oral vaccine delivery to enhance mucosal immune responses [27].

Despite these advantages, PNPs face several critical limitations in nucleic acid delivery. High‐charge‐density polymers like PEI cause substantial cytotoxicity and hemolytic risk. Moreover, polymer–nucleic acid complexes often lack stability, particularly for delivering long‐stranded mRNA molecules. PNPs also lack effective endosomal escape mechanisms, such as membrane fusion, crucial for cytosolic release. Moreover, complex polymer synthesis and considerable batch‐to‐batch variability pose significant obstacles to clinical translation and large‐scale manufacturing [28, 29]. Conversely, LNPs, composed of naturally derived or cell membrane–mimicking lipids, offer superior biocompatibility, high encapsulation efficiency, and robust physicochemical stability. LNPs also benefit from well‐established microfluidic manufacturing platforms, enabling scalable and reproducible production suitable for clinical and industrial use.

Hybrid NP systems combine two or more materials, such as lipids with polymers, inorganic NPs, or proteins. For example, lipid–polymer hybrid NPs comprise a polymeric core—such as PLGA—that provides structural stability, encapsulated within a lipid shell to improve biocompatibility and mRNA encapsulation. Lipid‐coated gold or silica NPs enhance imaging, thermal responsiveness, and reactive oxygen species (ROS) modulation [30]. Nevertheless, these hybrid NP systems face challenges like interfacial stability, standardization, scalability, and long‐term biosafety. Similar to exosomes and polymeric NPs, hybrids require complex synthesis protocols and face challenges in achieving batch‐to‐batch consistency, complicating quality control and clinical translation. LNPs have become widely adopted in tumor and gene therapy, and infectious disease prevention due to their biocompatibility, delivery efficiency, and safety profile [31].

Liposomes are spherical vesicles composed of one or more phospholipid bilayers enclosing an aqueous core and have long been utilized as nanocarriers for small molecules, peptides, proteins, and nucleic acids [32]. Their biocompatibility, structural versatility, and ability to encapsulate both hydrophilic and hydrophobic drugs have made liposomes one of the most clinically validated drug delivery systems, exemplified by doxorubicin‐loaded liposomes [33]. For nucleic acid delivery, cationic or ionizable lipid‐containing liposomes can electrostatically bind negatively charged RNA molecules, facilitating cellular uptake and protection against enzymatic degradation.

However, conventional liposomes face several limitations for mRNA delivery, including relatively low encapsulation efficiency, instability during storage, and insufficient endosomal escape capacity compared with advanced LNPs. Additionally, large‐scale reproducibility remains challenging [34]. Despite these drawbacks, liposomes remain an important platform for therapeutic delivery, and continued advances in liposome engineering may bridge the gap toward more efficient nucleic acid formulations.

In summary, while exosomes, PNPs, hybrid NPs, and liposomes each present unique features that may complement or enhance mRNA delivery, their clinical translation remains limited by manufacturing complexity, insufficient endosomal escape efficiency, and concerns regarding safety and reproducibility. Compared with these alternatives, LNPs maintain a clear advantage with their clinically validated performance, efficient intracellular delivery mechanisms, and scalable production platforms. Subsequently, the basic composition and preparation process of LNPs are elaborated in detail.

Buschmann et al. [39] reported that the negative charge of LNPs is not conducive to their accumulation in muscles and lymph nodes, which is necessary to trigger an immune response. The negatively charged LNP tends to be widely distributed in vivo, increasing the risk of adverse effects such as chills and fever in mice. The efficacy of the vaccine is closely related to the charge carried by LNPs. Although cationic lipids are frequently employed to deliver nucleic acid substances as key components of LNPs, their toxicity and short half‐life can adversely affect drug efficacy in vivo [40]. Currently, pH‐sensitive lipids are preferred as they remain neutral under physiological conditions after being assembled into LNPs, but they become positively charged in the acidic endosomes of the nucleus [41]. They offer low toxicity, facilitate endosomal escape, and thus enhance the in vivo efficacy of the loaded drug [42, 43, 44]. Typically, this class of lipids constitutes 30–50% of the total LNP composition, with their structure directly influencing their function [45].

Ionizable lipids generally have a hydrophilic head group, a linkage bond, and a hydrophobic tail group comprising amine groups (quaternary and tertiary amines), heterocyclic rings, and guanidine groups (Figure 1) [46, 47, 48]. The head group, typically positively charged, binds mRNA via electrostatic interactions, protecting it from nucleases. Additionally, the head group fuses with the target cell membrane, aiding cell entry [47]. The linker bonds act as "bridges" between the head and tail of the ionizable lipids and are categorized as nondegradable (ethers and carbamates) or degradable (esters, amides, and thiols). Nondegradable linkages have a high transfection efficiency but increase lipid cytotoxicity due to overstabilization. Meanwhile, degradable linkers are preferred when designing novel lipid structures for their rapid clearance and reduced side effects following multiple administrations [49, 50]. Hydrophobic tails comprise saturated or unsaturated hydrocarbon chains, branched structures, and steroid derivatives, enhancing LNP stability by supporting the phospholipid bilayer structure.

Unsaturated hydrocarbon chains and branched structures enhance membrane fluidity and transfection efficiency. Cholesterol derivatives improve membrane rigidity and structural integrity when used as ionizable lipids tails. Heyes et al. [51] synthesized four ionizable lipids in vitro with trailing alkyl branched chains containing zero (bis (2‐methacryloyl)‐oxyethyl‐disulfide), one (1,2‐dioleyloxy‐3‐dimethylamino‐propane), two (1,2‐diloinoleyloxy‐3‐dimethylaminopropane [DLinDMA]), and three (2‐dilinolenyloxy‐N,N‐dimethyl‐3‐aminopropane) unsaturated double bonds. The findings indicated that the saturation of the ionizable lipid tails influenced the delivery efficiency of LNP, with saturated tails correlating positively with the cellular uptake capacity. Higher saturation of ionizable lipid tails improves cellular uptake but reduces endosomal escape capacity, affecting delivery efficiency.

Hope et al. [52] uniformly adopted a structure analogous to the DLinDMA tail structure within the same alkyl chain class, opting for a ketocyclic structure as the connecting bond. They constructed an ionizable lipid library by designing head group structures composed of varying numbers of methylene groups. An ionizable lipid with two methylene groups in the head (DLin‐KC2‐DMA) exhibited the highest delivery efficiency, and DLin‐KC2‐DMA lipids demonstrated favorable tolerance and delivery performance in rodents and nonhuman primates. The research team further optimized the head group of ionizable lipids based on the structure of DLin‐KC2‐DMA and synthesized 53 novel ionizable lipids. DLin‐MC3‐DMA, with three methylene groups in the head group, a degradable ester linkage, and unsaturated tails, exhibited the highest in vivo delivery efficacy, with an optimal pK_a range of 6.2–6.5 [53].

Alnylam conducted a Phase III clinical trial (NCT01960348) in the United States, involving 225 adults aged 18–85 years, which demonstrated that patisiran effectively treats hereditary thyrotropic transprotein amyloidosis [54]. The successful market launch of patisiran was significantly attributed to the ionizable lipid DLin‐MC3‐DMA composition based on LNPs. Pardi et al. [55] reported that SM‐102 (mRNA‐1273) and ALC‐0315 (BNT162b2) share structural similarities, containing saturated alkyl chains in the tails, degradable ester bonds as connecting elements, and hydroxyl groups at the ends of their head groups [56]. These characteristics reduce head group hydration and enhance hydrogen bonding to nucleic acids, thereby improving the transfection capacity [57, 58]. Despite these similarities, they differ in the number of sub‐methyl groups in the tail alkyl branched chain and head group. Consequently, SM‐102 demonstrates superior potency and stability, inducing higher antibody titers in BALB/C mice. The successful conversion of these two ionizable lipids into products further affirmed the feasibility of employing saturated alkyl chains.

Song et al. [59] designed an LNP featuring a novel ionizable lipid (4N4T) with four tertiary amine groups and four hydrophobic chains. This LNP encapsulated the 4N4T‐DS mRNA vaccine encoding the full‐length S protein sequence of SARS‐CoV‐2. Experimental results indicated that the mRNA translation efficiency of 4N4T–LNP was markedly higher than that of SM‐102–LNP, with the 4N4T‐DS mRNA vaccine inducing higher neutralizing antibody titers. The type of neutralizing antibody produced was related to the chemical structure of the ionizable lipid. In conclusion, the design of the head group structure and the modification of the added groups of ionizable lipids can promote the delivery of LNP, the selection of degradable linkages can reduce the toxicity of lipids, and the different types of tail structures can also affect the interaction between LNP and the cell membrane. Consequently, when formulating new ionizable lipids, it is anticipated that integrating components with diverse structural functions will enhance the delivery efficacy of LNP.

Auxiliary lipids constitute 10–20% of LNPs, with 1,2‐dioleoyl‐sn‐glycero‐3‐phosphoethanolamine (DOPE) a prevalent example. Other examples include 1,2‐dioleoyl‐sn‐glycero‐3‐phosphocholine (DOPC) and 1,2‐distearoyl‐sn‐glycero‐3‐phosphocholine (DSPC). DOPE contains ethanolamine phosphate and two unsaturated alkyl chains connected by ester bonds. This structural characteristic stabilizes the hexagonal II (H_II) phase, crucial for membrane fusion, transmembrane transport, and lipid flip‐flop movement [60, 61].

In 2020, Ermilova and Swenson [62] utilized DLin‐MC3‐DMA as an ionizable lipid model in atomic molecular dynamics simulations, revealing that under equivalent DOPE and DOPC concentrations, DLin‐MC3‐DMA predominantly localizes within the DOPC lipid bilayer, enhancing membrane stability. Conversely, Lin‐MC3‐DMA exhibits a propensity to accumulate on the DOPE surface. This distribution disparity can be attributed to proton transfer disruption around the head groups, which impacts the LNPs' transfection efficiency. DSPC differs structurally from DOPE and DOPC, containing saturated alkyl chains and possessing a high melting temperature suitable for constructing highly stable LNPs; however, an overly stable LNP structure is not conducive to endosomal escape [63, 64, 65]. This highlights the tradeoff between cellular‐level stability and transfection potency during transformation studies.

Auxiliary lipids also influence the biodistribution of LNP‐delivered drugs. For instance, LNPs composed of DOPE tend to accumulate in the liver, whereas those composed of DSPC tend to accumulate in the spleen [61]. Importantly, auxiliary lipids are generally nontoxic, nonimmunostimulatory, and nonhemolytic, ensuring minimal interference with LNP delivery. Adjuvant lipids regulate LNP membrane stability and fluidity and do not carry an intrinsic positive charge, thus reducing immune responses to LNP [66]. Notably, Onpattro, Spikevax, and Comirnarty employ DSPC as an adjuvant lipid. Consequently, the rationale design of adjuvant lipids will continue to be a focal point in future studies.

Natural sources of cholesterol are abundant, and cholesterol is a key component of animal cell membranes, typically comprising 20–50% of LNPs [61, 67, 68]. The positioning of cholesterol on the LNP surface influences membrane fluidity, stability, delivery efficacy, and circulating half‐life [69]. In 2019, Paunovska et al. [70] successfully created an LNP database comprising nine cholesterol analogs. They reported that oxidatively modified cholesterol efficiently delivers mRNAs to cells within the liver microenvironment [70]. Furthermore, the hydrocarbon tail of cholesterol exhibits enhanced tolerance to oxidative modification compared with the cholesterol ring.

In 2024, Jung et al. [71] developed an ionizable cholesterol analog, 3β[L‐histidinamide‐carbamoyl] cholesterol (Hchol), as a substitute for cholesterol in LNP assembly. Results from in vivo efficacy experiments revealed that the imidazole moiety of Hchol promotes endosomal escape and increases the delivery potency and stability of LNPs. Xie et al. and Li et al. replaced the cholesterol component with methylprednisolone (MP) to prepare novel MP–LNPs. MP retained its anti‐inflammatory properties while avoiding systemic exposure to the drug to mitigate side effects. Furthermore, delivering MP–LNP‐encapsulated mRNA encoding C3 transferase (MP–LNP–C3) to injury sites effectively repairs damaged neural networks [72]. The precise mechanism by which cholesterol, which is inert to lipid responses, affects LNP recognition by target cells is unclear. A comprehensive analysis of impurities in raw cholesterol materials is crucial for LNP quality control.

The PEG lipids in LNPs typically constitute approximately 1.5%. They are predominantly distributed on the surface of LNPs, impeding LNP aggregation and enhancing their stability. PEGylated lipids also create a spatial barrier on the surface of LNPs, hindering their clearance by macrophages and prolonging their circulation time and half‐life. However, PEGylated lipids may trigger the production of antibodies that can lead to pseudo‐hypersensitivity activated by complement, contributing to adverse reactions caused by mRNA vaccines [73, 74].

In 2010, Akinc et al. [75] demonstrated that apolipoprotein E (ApoE) binds LNPs and mediates the delivery of LNPs to hepatocytes via low‐density lipoprotein receptors (LDLRs). Subsequent studies found that hydrophobic interactions between PEGylated lipids and biofilms correlate with alkyl chain length [75, 76, 77]. Meanwhile, in 2020, Suzuki et al. [77] reported that the shedding rate of PEGylated lipids in LNPs inversely relates to the length of the alkyl chain, with faster‐shedding lipid NPs inducing IgM antibody titers. Hence, rationally designing PEGylated lipids are imperative to mitigate adverse effects.

Furthermore, the PEGylated lipid content influences delivery efficacy, with comparatively low levels of PEGylated lipids exhibiting higher delivery potency [78]. PEG‐sized lipids effectively enhance the delivery potency of LNPs after nebulization [79]. Additionally, they can be biocoupled with ligands or macromolecules to serve specific functions. For example, Parhiz et al. [80] utilized the N‐Succinimidyl‐S‐acetylthioacetate‐maleimide (SATA‐maleimide) coupling technique to couple antibodies to PEGylated lipids, resulting in efficient drug delivery to the lungs. Meanwhile, Singh et al. [81] demonstrated the potential of PEGylated lipids for delivering LNPs via aerosolization by coupling hyaluronic acid to PEGylated lipids, achieving targeted drug delivery to ovarian cancer cells. The functionalization of PEGylated lipids is essential for improving LNP delivery efficacy, targeting precision, and stability. However, further investigation is necessary to understand the allergic reactions associated with PEGylated lipids. Thus, it is essential to explore the mechanisms underlying the action of PEGylated lipids and their role in triggering allergic reactions.

Once the key structural components of LNPs are defined, selecting an appropriate manufacturing process becomes a critical step to ensure delivery efficiency and quality control.

LNPs deliver mRNA therapeutics into host cells where the translated target proteins serve as immunogens to elicit immune responses that protect the host from pathogens. Following the success of COVID‐19 mRNA vaccines, this technology has rapidly expanded to target other viral and bacterial pathogens [98]. mRNA–LNP vaccines that target pathogen‐specific antigens are promising for preventing infectious diseases. Table 2 presents the clinical progress of mRNA–LNP‐based vaccines for preventing infectious diseases.

mRNA vaccines offer advantages such as flexible antigen design, rapid production cycles, and the ability to induce humoral and cellular immune responses. Recent advancements have been observed in the development of mRNA vaccines targeting various viral pathogens. For instance, the quadrivalent seasonal mRNA–LNP vaccine candidate, mRNA‐1010, encodes four full‐length hemagglutinins. Phase II clinical trials have demonstrated that mRNA‐1010 elicits higher hemagglutination inhibition titers against influenza A strains and comparable titers against influenza B strains. However, preliminary Phase III results indicate a higher incidence of adverse events associated with mRNA‐1010 [99, 100]. mRNA‐1345, approved by the US FDA for preventing respiratory syncytial virus (RSV) infection, is Moderna's second approved product and is currently undergoing Phase I trials in children aged 5–24 months (NCT05743881) and RSV‐seropositive children aged 12–59 months (NCT04528719). In another example, mRNA‐1653 encodes the full‐length membrane‐bound fusion proteins of human metapneumovirus and parainfluenza virus. Results from the Phase I clinical trial (NCT03392389) indicate that intramuscular (IM) administration induces robust neutralizing antibody titers, suggesting the potential to prevent infections by both pathogens [101]. In response to the 2022 monkeypox virus (MPXV) outbreak, mRNA‐1769, encoding four conserved MPXV surface proteins (M1, A29, A35, and B6), exhibited protective efficacy in murine models against vaccinia virus (VACV) and is now in Phase I trials (NCT05995275) [102]. BNT166a, a quadrivalent vaccine encoding M1, A35, B6, and H3 antigens, and its trivalent counterpart BNT166c (excluding H3) both confer protection in mice against VACV, MPXV Clade I and Clade IIb infections. BNT166a has been authorized for Phase I clinical evaluation (NCT05988203) [103]. Tian et al. [104] developed a thermostable ionizable LNP (iLNP) platform encapsulating mRNA encoding the MPXV A29L antigen, which elicited potent cross‐neutralizing immune responses against both vaccinia and monkeypox live viruses. The CV7202 rabies vaccine induced high antibody titers in healthy individuals during a Phase I trial (NCT03713086) [105]. The mRNA‐1647 vaccine for human cytomegalovirus (HCMV), encoding glycoprotein B and the pentameric complex, has demonstrated durable and specific antibody responses in seronegative and seropositive individuals [106]. The BNT163 trivalent HSV‐2 vaccine candidate, encoding gD2, gC2, and gE2, is currently in a Phase I clinical trial (NCT05432583) to evaluate dose escalation, immunogenicity, and safety. Last, two clinical vaccine candidates for varicella‐zoster virus—mRNA‐1468 (encoding glycoprotein E, NCT05701800) and BNT167 (encoding glycoprotein E, NCT05703607)—are both in Phase I clinical development [98].

The prevalence of antimicrobial resistance among human bacterial pathogens underscores the need for effective antibacterial vaccines. However, developing vaccines against bacteria faces greater challenges than viruses due to the larger and more complex bacterial genomes, which present a broader range of potential vaccine targets, including nonprotein antigens currently inaccessible to mRNA vaccine platforms. Moreover, while broad‐spectrum protection against viruses is often associated with high levels of neutralizing antibodies, targeting intracellular bacteria typically requires robust CD8⁺ T cell responses. BNT164a1/BNT164b1, developed by BioNTech, is currently the only multiantigen mRNA vaccine candidate for tuberculosis that has entered clinical trials. The BNT164 vaccine comprises antigens expressed during acute, reactivation, and latent stages of infection: Ag85A (Δ1–41), ESAT‐6, M72, VapB47, Hrp1, RpfA, RpfD, and HbhA. Phase I/II trials began in April 2023 in Germany (NCT05537038) and South Africa (NCT05547464) to evaluate dosage and immunogenicity in HIV‐negative and HIV‐positive individuals, with completion anticipated in 2027 [107]. Another promising candidate, mRNA^CV2, developed by the University of Sydney, encodes a CysD/Ag85B fusion protein. Preclinical studies have demonstrated its immunogenicity, inducing a Th1‐biased CD4⁺ T cell response and markedly reducing Mycobacterium tuberculosis burden in the lungs of infected mice. Recent data suggest it may also enhance the immunogenicity and long‐term protective efficacy of Bacillus Calmette–Guérin vaccination [108]. The mRNA–LNP vaccine candidate mRNA‐1982, which target Lyme disease by encoding the outer surface protein A (OspA) of Borrelia spp., is currently undergoing clinical evaluation (NCT05975099). Additionally, researchers at the Israel Institute for Biological Research have developed a bivalent mRNA vaccine that encodes the LcrV and F1 antigens and has demonstrated strong immunogenicity and protective efficacy in preclinical models of pneumonic plague. This vaccine notably induces neutralizing antibodies and robust cellular immunity, thereby providing a dual protection mechanism [109].

Since the COVID‐19 pandemic, mRNA vaccine technology has developed rapidly to combat various viral and bacterial pathogens. With flexible antigen designs, short manufacturing cycles, and the ability to induce potent humoral and cellular immune responses, mRNA vaccines show considerable promise. Several candidates targeting influenza, RSV, CMV, MPXV, HSV, and varicella have advanced to clinical trials with favorable outcomes. However, mRNA vaccines against bacterial pathogens face challenges due to antigenic complexity and the presence of nonprotein targets.

Future advancements in mRNA vaccines are expected to improve mucosal delivery, multivalent design, chronic infectious disease prevention, and the targeting of antimicrobial‐resistant bacteria—ushering in an era of precision, personalization, and broad‐spectrum protection.

Originally developed for the prevention of infectious diseases, mRNA–LNP vaccines have increasingly been applied in immunotherapy due to their protein expression capabilities. At Ruijin Hospital, researchers formulated LNPs using 1,2‐dioleoyl‐3‐trimethylammonium‐propane (DOTAP), phosphatidylcholine, DMG‐PEG 2000, cholesterol, and palmitic acid through self‐assembly, with an optimized formulation ratio of 47.5:10:1.5:36:5 to encapsulate VEGF‐C mRNA. In vitro transfection of H9c2, AC16, and HEK293T cells demonstrated specific expression of the encoded protein. In a C57BL/6 mouse model of myocardial infarction, intramyocardial delivery of VEGF‐C mRNA–LNPs at multiple injection sites promoted lymphangiogenesis, improved cardiac function, and decreased inflammation [110]. These favorable therapeutic outcomes were largely attributed to palmitic acid, which provides carboxyl groups that bind the extracellular matrix and prolong LNP retention within myocardial tissue, enhancing transfection efficiency. Table 2 presents the clinical progress in mRNA–LNP tumor therapeutic vaccines.

mRNA vaccines are being explored as a platform for cancer vaccine development. These vaccines involve the in vitro synthesis of mRNA sequences encoding tumor antigens, which are delivered into host cells to be translated into proteins, thus activating the immune system. Research on mRNA cancer vaccines is primarily focused on two categories of antigens: TAAs and tumor‐specific antigens (TSAs) [111]. Initial efforts concentrated on TAAs, but their broad expression and associated immune tolerance limited therapeutic efficacy. In contrast, personalized vaccines based on patient‐specific neoantigens (TSAs) have demonstrated enhanced specificity and are now a major focus in the field. Once delivered, the mRNA is translated into antigenic proteins in the cytoplasm of antigen‐presenting cells (APCs) and activates Toll‐like receptor (TLR7) and TLR8 signaling pathways. The resulting peptides are presented by MHC class I molecules to activate CD8⁺ T cells, enabling targeted tumor cell killing. Researchers at the University of Pennsylvania found that codelivering IL‐12‐encoding mRNA with antigen‐coding mRNA–LNPs boosts CD8⁺ T cell quantity, function, and efficacy, reducing tumor growth in a melanoma mouse model [112]. Oberli et al. [113] utilized microfluidic technology to formulate LNPs (B‐11; composed of ckk‐E12:DOPE:cholesterol:C14‐PEG2000:sodium lauryl sulfate in a 15:26:40.5:2.5:16 ratio) encapsulating mRNA encoding ovalbumin (OVA), tyrosinase (TYP)‐related protein 2, or the gp100 peptide with a position 27 mutation. In mice bearing B16F10 tumors, these vaccines elicit strong antigen‐specific CD8⁺ T cell responses and reduce tumor growth. Adding lipopolysaccharide to the formulation improved antitumor efficacy and prolonged survival [113]. An LNP‐encapsulated personalized mRNA vaccine, mRNA‐4157, encodes 34 neoantigens (29 MHC I and 5 MHC II epitopes). In the KEYNOTE‐942 (Phase IIb) clinical trial (NCT03897881), mRNA‐4157/V940 combined with pembrolizumab reduced recurrence risk by 44% in melanoma patients [114]. Another candidate, Autogene cevumeran (BNT122), encoding 20 TSAs combined with atezolizumab, showed strong CD8⁺ T cell responses and reduced recurrence risk in patients with pancreatic ductal adenocarcinoma (PDAC). Additionally, BNT116, an RNA–lipid complex encoding six tumor antigens (MAGE‐A3, CLDN6, KK‐LC‐1, PRAME, MAGE‐A4, and MAGEC1), demonstrated favorable tolerability and safety in a Phase I trial (NCT05142189) for non‐small cell lung cancer, either alone or combined with cemiplimab. A Phase II trial (NCT05557591) is ongoing, expected to conclude by 2027 [115]. The BNT111 vaccine encodes four TAAs: NY‐ESO‐1, TYP, melanoma‐associated antigen A3 (MAGE‐A3), and transmembrane phosphatase with tensin homology. A Phase I trial (NCT02410733) demonstrated that BNT111 induced strong CD4⁺ and CD8⁺ T cell responses in melanoma patients. Additionally, the ongoing Phase II trial (NCT04526899) combining BNT111 with cemiplimab has shown encouraging therapeutic outcomes [116].

Chimeric antigen receptor T cell (CAR‐T) immunotherapy also exhibits promise in treating certain hematological malignancies, such as diffuse large B‐cell lymphoma, acute lymphoblastic leukemia, and BCMA‐positive multiple myeloma. The primary CAR gene delivery method is viral transduction, accounting for 94% of assessable CAR‐T products, with lentiviral vectors comprising 53.3%. However, the random integration of lentiviral vectors into the host genome may disrupt essential genes or activate oncogenes, potentially leading to insertional mutagenesis [117, 118]. To address these concerns, Chen et al. [119] employed a microfluidic device to encapsulate human CAR mRNA targeting CD19 into LNPs (DLin‐MC3‐DMA:DSPC:DSPE‐PEG2000:cholesterol formulation at 50:10:1.5:38.5). They optimized transfection protocols to deliver mRNA into T cells, with mRNA entering the cytoplasm within 8 h, resulting in robust CAR expression on the T cell surface. Coculturing with CD19‐positive tumor cell lines (Daudi, Raji, and Nalm‐6), mRNA–LNP‐engineered CAR‐T cells exhibited high surface expression of CD107a—a marker transiently expressed during degranulation of cytotoxic T cells upon target cell engagement. Furthermore, in a 3D tumor spheroid model, the engineered CAR‐T cells demonstrated excellent infiltration and cytotoxicity. Meanwhile, in vivo, these cells home to tumor sites in B cell malignancy mouse models, significantly suppressing tumor growth. In addition to ex vivo modification of T cells, Hunter et al. [120] devised an innovative strategy for in vivo CAR‐T cell generation using targeted LNPs (tLNPs) formulated with an ionizable lipid (L829). By conjugating tLNPs with anti‐CD5 or anti‐CD8 antibodies, the system enabled selective delivery of CAR‐encoding mRNA into specific T‐cell subsets while minimizing off‐target accumulation in the liver. In humanized mouse models, administration of tLNPs rapidly produced functional CAR‐T cells within hours, leading to efficient eradication of B cell malignancies and solid tumors, with nearly complete tumor regression. In cynomolgus monkeys, intravenous (IV) administration resulted in profound but reversible B cell depletion and was well tolerated. This approach eliminates the need for ex vivo cell manufacturing and preconditioning chemotherapy, thereby addressing several key limitations of current CAR‐T therapies and underscoring the transformative potential of mRNA–LNP‐based in vivo CAR‐T engineering [120]. mRNA–LNP technology also holds promise for treating solid tumors. The BNT211 platform combines CAR‐T cells targeting the TAA CLDN6 with an mRNA vaccine encoding CLDN6 (termed CARVac), enhancing efficacy in CLDN6‐positive solid tumors. Preclinical solid tumor models and a Phase I clinical trial (NCT04503278) reported good safety, tolerability, and preliminary antitumor activity [121, 122].

mRNA–LNP technology, originally developed for infectious disease vaccines, has rapidly evolved into a versatile therapeutic platform with broad applications in immunotherapy, particularly in oncology and regenerative medicine. The ability of mRNA to encode specific proteins has enabled its use in targeted therapies, such as delivering VEGF‐C mRNA to promote lymphangiogenesis and cardiac repair following myocardial infarction. In cancer immunotherapy, mRNA–LNP vaccines have progressed from encoding TAAs to highly personalized vaccines based on TSAs, demonstrating robust activation of cytotoxic T cell responses and significant clinical benefits in various cancers, including melanoma, PDAC, and non‐small cell lung cancer. Several candidates, such as mRNA‐4157, BNT122, and BNT111, are undergoing clinical trials with promising outcomes in terms of tumor regression and recurrence reduction. Beyond vaccines, mRNA–LNPs are also being explored for engineering CAR‐T cells, offering a nonviral, transient, and potentially safer alternative to conventional gene delivery methods. LNP‐formulated CAR mRNAs have shown strong expression, tumor infiltration, and cytotoxic activity in both hematological and solid tumor models. These advances highlight the expanding therapeutic potential of mRNA–LNPs, supporting their continued development as a transformative modality in precision medicine.

siRNA molecules can specifically silence the expression of target genes, and LNPs have proven to be effective carriers for siRNA‐based therapeutics. A notable example is the US FDA‐approved product Onpattro, which has demonstrated clinical efficacy and has since been explored for the treatment of various other diseases. Leveraging the unique advantages of LNP technology platforms, gene therapeutics can be selectively delivered to diseased tissues for targeted intervention.

Zhang et al. [123] utilized microfluidic technology to synthesize targeted LNPs composed of SM‐102, DSPC, cholesterol, DMG‐PEG2000, and DMG‐PEG2000‐Maleimide at a molar ratio of 42:13:43:1.5:0.5. These LNPs were subsequently modified with lactoferrin (Lf), enabling selective delivery to respiratory epithelial cells and neurons in Alzheimer's disease (AD) models. The LNPs efficiently encapsulated siRNA targeting β‐site amyloid precursor protein cleaving enzyme 1 (BACE1). In vivo studies revealed that gene therapy significantly reduced escape latency in mice, improved spatial memory, and thereby showed potential to delay the progression of AD [123]. Hofstraat et al. [124] developed a prototype apolipoprotein‐based NP (aNP) that effectively targeted myeloid cells, hematopoietic stem cells, and progenitor cells in the bone marrow. These aNPs were capable of encapsulating siRNA against lysosome‐associated membrane glycoprotein 1 (Lamp1), offering a novel strategy for the treatment of bone marrow malignancies [124]. Cui et al. [125] synthesized LNPs composed of Dlin‐MC3‐DMA, cholesterol, DSPC, and DMG‐PEG, coencapsulating siRNA targeting epidermal growth factor receptor (EGFR) and gentamicin (GM). Results demonstrated that GM enhanced endosomal escape of siRNA, thereby improving gene silencing efficacy. In murine tumor models, the LNP–siEGFR/GM system achieved an 81% knockdown of EGFR mRNA, significantly inhibited tumor growth, and induced tumor cell apoptosis, indicating its promising potential in cancer therapy [125]. Overall, the use of LNPs for siRNA delivery enables efficient gene silencing for therapeutic purposes. Furthermore, the integration of LNP‐based delivery systems with gene‐editing technologies is rapidly advancing, especially in the context of cancer treatment.

CRISPR/Cas9 in vivo gene‐editing therapy, using mRNA technology and LNPs that coencapsulate Cas9 mRNA and target specific single guide RNAs (sgRNAs), can be applied to various conditions [126]. Intellia Therapeutics' NTLA‐2001 targets transthyretin (TTR) for treating transthyretin amyloidosis (ATTR). The LNP formulation (LP000001:DSPC:cholesterol:DMG‐PEG2k = 23.6:4.0:8.2:4.2) encapsulates Cas9 mRNA and a TTR‐specific sgRNA. Phase I trial data (NCT04601051) showed NTLA‐2001 significantly reduced plasma TTR protein levels in patients with ATTR, encouraging the development of CRISPR/Cas9‐mRNA‐based therapies for cancer [127]. Dan Peer's team [126] employed an LNP formulation (DLin‐MC3‐DMA:DSPC:cholesterol:DMG‐PEG:DSPE‐PEG = 50:10.5:38:1.4:0.1) to codeliver Cas9 mRNA and sgRNA targeting PLK1. A single intracranial injection into an aggressive orthotopic glioblastoma model achieved approximately ∼70% in vivo gene editing, suppressed tumor growth by 50%, and increased survival by 30%. Meanwhile, Mitchell's team [128] at the University of Pennsylvania synthesized 252 distinct siloxane‐based lipids to enhance mRNA endocytosis and endosomal escape. Their Si5‐N14 LNP formulation, codelivering Cas9 mRNA and a GFP sgRNA, enabled efficient pulmonary delivery and potent gene knockout in Lewis lung carcinoma‐bearing mice [128].

LNPs have emerged as a powerful platform for the delivery of siRNA and CRISPR/Cas9 gene‐editing systems, enabling precise gene silencing and editing in vivo. siRNA‐loaded LNPs offer a promising therapeutic avenue for diverse diseases by specifically downregulating the expression of pathogenic genes. A landmark in this field is Onpattro—the first US FDA‐approved LNP–siRNA therapeutic—which has validated the clinical potential of LNPs and spurred further exploration into neurodegenerative disorders, cancer, and hematological diseases. Advanced engineering of LNPs, such as surface modification with ligands like Lf or apolipoprotein, enables targeted delivery to specific cell types including neurons, respiratory epithelial cells, and hematopoietic cells. Codelivery strategies, such as combining siRNA with GM to enhance endosomal escape, have further improved gene silencing efficacy in cancer models. Beyond RNA interference, LNPs have also become a key vehicle for in vivo CRISPR/Cas9 gene editing. Platforms like NTLA‐2001 demonstrated successful gene knockout in clinical trials for transthyretin amyloidosis, while preclinical studies using LNPs to deliver Cas9 mRNA and sgRNA in glioblastoma and lung cancer models achieved high editing efficiency, tumor suppression, and extended survival. These breakthroughs highlight the expanding role of LNPs in next‐generation gene therapies, particularly for diseases where precision and safety are critical.

LNPs were initially developed and optimized for nucleic acid delivery. However, they have recently demonstrated substantial potential for encapsulating and delivering other classes of therapeutics, including proteins, peptides, and small molecules. Their favorable biocompatibility, scalable formulation processes, and ability to enhance drug solubility and stability make LNPs a versatile platform beyond nucleic acid‐based therapies.

The delivery of therapeutic proteins and peptides presents considerable challenges due to their susceptibility to enzymatic degradation, poor membrane permeability, and rapid clearance from the body. LNPs can overcome these limitations by encapsulating proteins and peptides within a protective lipid matrix, thereby extending their circulation time, enhancing cellular uptake, and enabling targeted delivery. Several studies have elucidated the application of LNPs in protein delivery. Proteins are amphoteric macromolecules, and under physiological conditions, most exhibit a net positive charge, behaving as polycations [129]. Meagher et al. [130] replaced the ionizable lipid SM‐102 with 50% of the anionic lipid 1,2‐dimyristoyl‐sn‐glycero‐3‐phospho‐(1′‐rac‐glycerol) sodium salt (DMPG) to formulate LNPs that encapsulated various proteins, including fluorescent proteins (mCherry, EBFP2, and Venus), bovine serum albumin, and monoclonal antibodies (trastuzumab and IgG). The results demonstrated that these LNPs significantly enhanced protein encapsulation efficiency and maintained good stability. Robust intracellular fluorescence signals were detected in A549 and SKBR3 cells, indicating efficient protein delivery. Furthermore, trastuzumab‐loaded LNPs specifically bound to HER2 receptors in HER2‐overexpressing SKBR3 cells, confirming that the antibody retained its functional activity postencapsulation [130]. This study supports the feasibility of intracellular delivery of therapeutic proteins via LNPs and holds promise for expanding protein‐based therapies in cancer, cardiovascular, and metabolic diseases. Further exploration of the influence of various anionic lipids on protein encapsulation efficiency could help optimize LNPs for more effective protein delivery. Mastrobattista et al. [131] designed a five‐component LNP formulation (C12‐200:DOPE:cholesterol:PEG‐DMG:DOTAP = 35:16:46.5:2.5:0.25) to encapsulate ribonucleoprotein complexes consisting of Cas9 protein and sgRNA. In vitro experiments demonstrated gene‐editing activity in HEK293T and HEPA1‐6 cells. Moreover, biodistribution and in vivo gene editing studies in female Ai9 mice following systemic administration revealed that LNPs primarily accumulated in the liver, spleen, and lungs, where they achieved effective gene knockout with minimal toxicity [131].

Drug loading within LNPs typically relies on electrostatic interactions between the therapeutic cargo and lipid components. However, peptides often lack a strong net charge, making it difficult for them to interact effectively with LNPs through conventional electrostatic mechanisms. Instead, peptide backbones are rich in amide bonds, which can act as hydrogen bond donors or acceptors. To address this challenge, the research group led by De Geest [132] developed a novel gallic acid‐derived lipid‐based LNP system. In this design, the ionizable lipid head group was replaced with a 3,4,5‐trihydroxyphenyl moiety to enable hydrogen bonding with peptide molecules. The resulting ionizable lipid, S‐GA‐DOG, was constructed by linking the head group to the hydrophobic alkyl tail via a disulfide bond, allowing for redox‐responsive release. When combined with DSPC, cholesterol, and DMG‐PEG, the resulting LNPs were able to successfully encapsulate therapeutic OVA antigen peptides. These peptide‐loaded LNPs facilitated intracellular delivery and effectively activated CD8⁺ T‐cell immune responses, demonstrating their potential for functional peptide‐based immunotherapies [132]. Gu et al. [133] designed quaternary ammonium‐based ionizable lipids incorporating phenylboronic acid groups to form a glucose‐responsive LNP system. These LNPs were capable of efficiently loading recombinant human insulin along with a glucose‐sensing component. Upon exposure to glucose‐rich environments, such as those mimicking hyperglycemia, the insulin was selectively released. In a type 1 diabetes mouse model, this formulation exhibited prolonged glycemic control. This glucose‐responsive LNP platform holds promise as a novel therapeutic strategy for diabetes treatment [133].

Proteolysis‐targeting chimeras (PROTACs) are capable of selectively degrading target proteins within cells. Chan et al. [134] developed a strategy to deliver PROTAC small molecules by forming electrostatic complexes with LNPs containing a fifth lipid component, DOTAP, thereby achieving efficient cytosolic delivery. Their study demonstrated that the K1–LNP formulation—composed of C12‐200:DOPE:C14PEG‐2000:cholesterol:DOTAP at a molar ratio of 19:19:4:38:20—achieved the highest protein degradation efficiency. This platform enabled rapid and robust intracellular protein degradation, with clearance rates reaching up to 95%. Moreover, it allowed targeting of proteins located in diverse subcellular compartments and offered substrate‐specific, repeatable degradation profiles. The modularity of this approach supports the design of clinically relevant protein degraders for therapeutic purposes [134]. Liang et al. [135] developed a peptide‐modified LNP system functionalized with an Asp–Gly–Arg sequence, enabling the effective codelivery of microRNA‐150 and the antiangiogenic agent quercetin. These LNPs demonstrated selective targeting of endothelial cells during abnormal angiogenesis. In a mouse model, the system showed promising therapeutic potential for age‐related macular degeneration, with no observed retinal toxicity [135]. Van der Meel et al. [136] coencapsulated siRNA targeting the androgen receptor and a lipophilic paclitaxel prodrug derivative within LNPs. Using dual radiolabeling techniques, they quantitatively evaluated the pharmacokinetics and biodistribution of the LNPs following systemic administration in tumor‐bearing mice. This codelivery platform highlights a novel approach for the development of combination therapies, facilitating simultaneous delivery of gene silencers and chemotherapeutic agents [136].

The versatility of LNPs as a delivery platform has extended far beyond nucleic acid therapeutics, enabling the efficient encapsulation and targeted delivery of diverse therapeutic agents—including proteins, peptides, small molecules, and combination regimens such as PROTACs, siRNA, and chemotherapeutics—thereby offering innovative strategies for the treatment of cancer, metabolic disorders, and other complex diseases.

LNPs play a crucial role in mRNA replacement therapy for inherited metabolic disorders. Khoja et al. [137] successfully encapsulated mRNA encoding arginase within LNPs, and in arginase‐deficient knockout mouse models, the therapy effectively treated arginase deficiency, prevented associated protein malnutrition, and restored normal oligodendrocyte function. Sahay and colleagues [138] developed LNPs using PEG–lipids modified with carboxylic lipids to encapsulate mRNA encoding recombinases, achieving efficient transfection of photoreceptor cells. Moreover, by coencapsulating Cas9 mRNA and sgRNA, gene editing was successfully performed in the retina, correcting blindness‐causing mutations and demonstrating therapeutic potential for inherited retinal diseases [138]. A preclinical study by CureVac demonstrated that LNPs encapsulating mRNA encoding phenylalanine hydroxylase, when delivered intravenously, rescued metabolic defects in phenylalanine hydroxylase‐deficient mouse models [139]. mRNA‐3927, a therapeutic candidate encoding the alpha (PCCA) and beta (PCCB) subunits of propionyl‐CoA carboxylase, delivered via LNPs, restored hepatic metabolic function and showed potential in treating propionic acidemia [140]. In another preclinical study led by Moderna, LNPs encapsulating mRNA encoding argininosuccinate lyase were intravenously administered in a mouse model of argininosuccinic aciduria (ASA). The results showed an effective blockade of glutathione metabolism, confirming the therapeutic potential of mRNA‐based intervention [141]. Translate Bio developed LNP‐formulated mRNA encoding α‐galactosidase A, demonstrating favorable therapeutic outcomes in Fabry disease mouse models [142]. Moderna also investigated LNPs encapsulating mRNA encoding porphobilinogen deaminase (PBGD), and systemic administration in mouse models significantly reduced the levels of δ‐aminolevulinic acid and porphobilinogen, providing the first in vivo evidence that mRNA‐based therapy can effectively address PBGD deficiency [143]. Furthermore, the Moderna team designed LNPs encapsulating mRNAs encoding key functional proteins and enzymes, showing therapeutic potential for treating Crigler–Najjar syndrome and glycogen storage disease [144].

Researchers at the University of Tübingen employed chitosan‐modified PLGA NPs to deliver chemically modified cystic fibrosis transmembrane conductance regulator (CFTR) mRNA. This approach led to abundant expression of CFTR protein in the lungs and significantly improved pulmonary function and pathological features in CFTR‐deficient mice, effectively treating cystic fibrosis [145]. In a collaborative study by Ludwig Maximilian University of Munich and Hannover Medical School, a class of self‐assembled peptide–PEI NPs (T704) were developed to deliver chemically modified CFTR mRNA. The mRNA was efficiently expressed both in vitro and in vivo. Further experiments demonstrated that T704 NPs could also deliver a Sleeping Beauty transposon system to insert the CFTR gene, thereby enabling long‐term correction of CFTR deficiency with excellent safety and minimal toxicity in both cellular and animal models [146]. In another study, IV administration of LNPs encapsulating mRNA encoding SLC25A13 successfully led to hepatic expression of functional aspartate/glutamate carrier protein in mice, correcting metabolic abnormalities. Treated animals also showed improvements in physical development and behavioral deficits, suggesting therapeutic efficacy in citrin deficiency [147]. A study conducted at Ludwig Maximilian University of Munich was the first to demonstrate that repeated administration of chemically modified mRNA could restore the expression of pulmonary surfactant protein B (SP‐B) and prevent respiratory failure in mice. Delivery of SP‐B mRNA achieved over 70% of normal SP‐B protein levels in the lungs, maintaining both pulmonary structure and function [148]. Building on this, researchers at the University of Tübingen developed a lung‐targeted LNP formulation modified with a specific targeting peptide for SP‐B mRNA delivery. Upon intratracheal administration, the system partially restored SP‐B mRNA and protein expression in pulmonary cells and improved survival in SP‐B‐deficient mice [149]. This study highlights the potential of airway‐delivered, lung‐targeted LNP–mRNA systems for treating SP‐B deficiency.

Most inherited metabolic disorders currently lack effective therapeutic options. For certain conditions characterized by high incidence, significant clinical severity, and favorable outcomes when diagnosed and treated early, the combination of mRNA–LNP therapy with existing clinical interventions offers promising potential. This approach may provide a novel therapeutic avenue for overcoming rare genetic diseases. Recent studies have reported encouraging progress in the application of LNPs within these emerging areas.

Opioids are among the most utilized drugs for pain relief, with common examples including morphine, fentanyl, oxycodone, heroin, and methadone. These small‐molecule drugs interact with opioid receptors, reducing neuronal excitability and impairing the release of nociceptive neurotransmitters, thereby alleviating pain. Furthermore, opioids can induce euphoria, contributing to their high potential for abuse and addiction, frequently culminating in respiratory depression and death [150].

Current pharmacological interventions primarily rely on opioid receptor agonists and antagonists. While these therapies offer some efficacy, they are insufficient for individuals facing long‐term opioid dependence. Antiopioid immunotherapy is emerging as a promising alternative, aiming to sequester opioids outside the receptor space to retain analgesic effects while minimizing adverse side effects [151]. De Geest et al. [152] developed a NP vaccine by noncovalently assembling a fentanyl hapten, CD4⁺ T‐helper peptide epitopes, and a TLR7/8 agonist within LNPs. This formulation effectively induces the production of fentanyl‐specific IgG antibodies and subclasses, which selectively bind opioid molecules and restrict their distribution to the bloodstream and peripheral tissues, preventing interaction with central nervous system receptors. In preclinical mouse models, this approach significantly mitigates the effects of opioid overdose [152]. Cao's team [153] developed a morphine‐encapsulating LNP formulation (Tet1–LNP) that utilizes peptide recognition—specifically, Tet1's binding to trisialoganglioside (GT1b) receptors—to selectively activate peripheral opioid receptors in dorsal root ganglia. This strategy markedly reduces side effects associated with central opioid receptor activation and demonstrates potent and sustained analgesic effects with favorable safety in animal models [153].

Recently, LNP‐based immunotherapy against opioids has emerged as a novel approach to treat addiction, offering therapeutic opportunities for chronic pain management and addiction intervention. The application of LNPs across various clinical settings relies heavily on the high‐quality structural design of mRNA. To further enhance vaccine efficacy, strategies are employed to optimize the mRNA molecule, enabling better synergistic interaction with LNPs.

The physical properties of mRNA, particularly its length and secondary structure, directly affect its encapsulation efficiency within LNPs. mRNA molecules of moderate length (typically 1–3 kilobases) are more readily encapsulated due to their favorable flexibility and reduced steric hindrance during the self‐assembly of LNPs [154]. In contrast, longer mRNA transcripts—especially those rich in secondary structures, may interfere with NP formation, leading to heterogeneous particle sizes and decreased encapsulation efficiency. Complex secondary structures such as excessive stem‐loops increase molecular rigidity, occupy more space, and hinder lipid interaction with the phosphate backbone, thereby compromising encapsulation. Codon optimization and RNA folding algorithms (e.g., LinearDesign and mRNAid) have been utilized to reduce secondary structure formation, resulting in improved encapsulation [155]. Selecting synonymous codons that are less prone to forming stable local structures can help minimize folding complexity. High GC content tends to stabilize secondary structures; therefore, appropriate adjustment of GC content is essential to balance RNA stability with structural manageability. These strategies work synergistically with lipid components in LNPs to enhance the efficiency of mRNA encapsulation.

Chemical modification of nucleosides is a foundational approach to improve the pharmacological properties of synthetic mRNA by increasing stability, translation efficiency, and immune tolerance. Unmodified mRNA contains immunogenic motifs that trigger PRRs such as TLR3, TLR7/8, RIG‐I, and MDA5, leading to type I interferon responses and inflammation [156]. Incorporating modified nucleosides reduces this immune activation while maintaining or enhancing protein expression. N1‐methyl‐pseudouridine (m¹Ψ) is the most widely used modification in clinical mRNA vaccines. It suppresses PRR activation, reduces IFN‐α and TNF‐α production, and enhances translational efficiency and molecular stability. Pseudouridine (Ψ) similarly enhances base stacking and dampens innate immune recognition, though with slightly lower translational benefit compared with m¹Ψ [157]. 5‐Methylcytidine, especially when used with m¹Ψ, synergistically promotes translation and reduces immunogenicity by improving resistance to endonuclease degradation [158]. N6‐methyladenosine, a natural epitranscriptomic mark, has been explored for translational regulation and immune evasion [159]. Most current designs favor global substitution of uridine with m¹Ψ or Ψ. Site‐specific incorporation, enabled by enzymatic or solid‐phase synthesis techniques, is being explored for applications requiring precise structural control. However, excessive or mispositioned modifications may impair RNA folding and reduce expression. Thus, modification strategies must balance immune evasion, translation efficiency, and protein integrity.

Incorporation of modified nucleotides, such as m¹Ψ, into mRNA sequences not only reduces activation of innate immune responses but also enhances mRNA stability and improves its physicochemical compatibility with LNPs. These nucleotide modifications alter the charge distribution along the mRNA backbone, which is inherently highly negatively charged and poses challenges for efficient encapsulation and cellular uptake [160]. Ionizable lipids within LNPs carry positive charges at low pH, enabling electrostatic complexation with the mRNA's phosphate backbone. Rational mRNA design should aim to minimize immunostimulatory motifs—such as U‐ or A‐rich regions—and avoid excessive charge clustering, thereby reducing molecular rigidity [161]. Such strategies facilitate more efficient incorporation of mRNA into LNPs, promote uniform NP formation, enhance encapsulation efficiency, and reduce innate immune recognition.

Once the target protein sequence is determined, optimization of the mRNA sequence and structural features can be performed to improve its encapsulation efficiency within LNPs. In addition, rigorous purification methods such as high‐performance liquid chromatography (HPLC) or tangential flow filtration (TFF) are employed to remove impurities generated during in vitro transcription [162]. Among these, double‐stranded RNA (dsRNA) contaminants are particularly problematic due to their higher rigidity, larger size, and distinct charge density compared with single‐stranded mRNA. dsRNA can compete with the target mRNA for lipid‐binding sites, severely disrupting the encapsulation process and reducing the efficiency of mRNA loading into LNPs [163]. Ensuring uniform mRNA transcript length and minimizing molecular heterogeneity also contributes to achieving more consistent and homogeneous LNP formulations. While the design of the mRNA molecule itself is undoubtedly important, the delivery efficiency also depends on the performance of the carrier system. To support the clinical translation of LNPs technology, it is essential to elaborate on the key strategies for optimizing LNPs design, including targeted drug delivery, enhancement of endosomal escape for efficient cargo release, and mitigation of administration‐related adverse effects. These approaches provide a theoretical foundation for expanding the therapeutic applications of LNPs.

LNPs ten to accumulate in the liver when administered intravenously or intramuscularly, which impedes the efficacy of mRNA vaccines [164]. This is due to LNPs adsorbing ApoE in the plasma, binding LDLRs, and entering liver cells [165]. Although this process helps target drugs to the liver, it reduces their systemic efficacy. Consequently, researchers have sought to modify LNPs to enhance organ targeting by altering their surface charge, composition, structure, and size [166, 167]. This field has garnered considerable global interest, advancing our understanding of LNP adaptation to immune organs [168, 169, 170]. To maximize LNP delivery potential, optimizing their design and precision targeting is essential (Figure 3).

LNPs primarily enter cells via endocytosis, requiring mRNA molecules to escape from nuclear endosomes into the cytoplasm to be translated into proteins. However, most mRNAs are transported by endosomes to lysosomes for degradation, with only a small fraction escaping to the cytoplasm. This represents a significant impediment to drug delivery development [218, 219, 220, 221]. Enhancing drug therapeutic efficacy necessitates a profound understanding of LNP interactions with in vivo pathways. This will aid in designing optimal drug therapeutics, allowing more mRNA molecules to escape endosomes. Two theories currently explain the endosomal escape process. As early nuclear endosomes evolve into lysosomes, the pH level changes from 7.4 in the normal physiological environment to 6.5 in early nuclear endosomes, 6 in late nuclear endosomes, and 5 in lysosomes [222]. Endosomal escape by LNPs is primarily associated with the pH‐dependent membrane phase. Ionizable lipids are pivotal in dictating the fundamental properties of LNPs and closely correlate with the mechanism of endosomal escape [223]. When the hydrophobic end of the ionizable lipid is equivalent in size to the head group, LNPs often disperse in saline as a bilayer [112]. Once internalized by the target cells, nuclear endosomes engulf the LNPs. As the nuclear endosome matures and its pH decreases, the pH‐sensitive ionizable lipid head group carries a positive charge, inducing H_II structure formation after interacting with the negatively charged nuclear endosome membranes. This causes membrane damage, releasing the loaded drug into the cytoplasm.Another prevalent mechanism of endosomal escape is the "proton sponge effect" [224]. In the endosomal environment, LNPs carry a positive charge, causing chloride ions to passively diffuse into the nuclear endosomal compartment to achieve membrane equilibrium. This increases osmotic pressure within the endosome, leading to swelling and subsequent rupture, facilitating nucleic acid escape.

Current studies primarily focus on the efficiency of endosomal escape in relation to transfection capacity, while relatively less attention is given to the migration of labeled nucleic acid molecules from endosomes to the cytoplasmic lysate. Although modifying the structure of LNPs can improve endosomal escape, no comprehensive solution has been defined. Thus, exploring viable methods is necessary to achieve optimal LNP endosomal escape (Figure 3).

The size of the hydrophobic group in ionizable lipids influences endosome escape. Unsaturated bonds in the hydrophobic tail create a spatial barrier via the rigid structure of the cis‐double bond, facilitating the formation of the H_II phase and aiding endosome escape [225]. Sato et al. [226] developed a pH‐responsive ionizable lipid library, with CL1H6 demonstrating high endosomal escape efficiency due to its hydrophilic head group with a tertiary amine and two tails with unsaturated bonds. The head group affects the pK_a of LNPs, enhancing endosomal escape [226]. Tanaka et al. [227] developed a self‐degradable lipid molecule that accelerates LNP decomposition, promoting mRNA release into the cytoplasm. Hence, incorporating a hydrophobic tail with unsaturated bonds can improve endosomal escape, potentially increasing with additional unsaturated bond [227]. Based on the four‐component LNP, Syn–LNP was assembled from the ionizable lipid L319 and intramuscularly administered to C57BL/6 mice, significantly improving endosomal escape and mRNA delivery [228]. Moreover, terminally branched ionizable lipids (BEND lipids) were synthesized. Following transfection of HeLa cells, the longest and shortest terminally branched BEND lipids exerted the greatest effects [228]. Furthermore, a mouse model revealed that ionizable lipids E4t‐494 and E6t‐494 exhibit superior delivery performance [229]. Hence, incorporating branched chains at the terminal end of ionizable lipids can effectively enhance LNP's membrane‐disrupting capacity and endosomal escape.

Although PEGylated materials prolong the blood circulation time of LNPs, they also create a protective barrier due to the shielding effect of PEG, hindering LNP binding to target cells and reducing their uptake. Specifically, PEG compromises the ability of LNPs to fuse with cellular and nuclear endosomal membranes, impeding the release of loaded drugs into the cytoplasm of target cells. To address these limitations, we developed pH‐sensitive PEGylated lipids that undergo cleavage under reducing conditions, improving LNP endosomal escape and enhancing therapeutic efficacy and safety [230, 231]. The LNP surface was modified with anti‐HER2 F(ab′) antibody fragments or PEI polymers, along with lipids, peptides, and other endosome‐interacting molecules. The optimized LNPs exhibited significantly improved endosome escape [232, 233, 234].

Cholesterol's structural properties also impact the endosomal escape ability of LNPs. Introducing eLNP composed of C‐24 alkyl phytosterols into LNPs significantly improves their gene transfection ability. Unlike conventional spherical LNPs, the polyhedral structure of eLNPs enhances cellular uptake and endosomal escape [235]. Substituting cholesterol with naturally occurring phytol enhances LNPs delivery. Similarly, substituting with hydroxycholesterol influences cholesterol recognition by NPC1, improving delivery efficiency [236, 237]. Furthermore, small molecule manipulation of the endosomal recycling pathway can enhance transfection efficiency. For example, annexin A6 (ANXA6) regulates the endocytic pathway, membrane reorganization, and vesicle transport through its lipid‐binding activity on Ca²⁺. However, inhibiting ANXA6–lipid interactions may promote mRNA release from endosomes. Indeed, NAV and ES5—small molecules that specifically inhibit ANXA6–lipid interactions—significantly enhance LNP the delivery efficiency by 1.5‐2‐fold [238]. Furthermore, NPC1 is pivotal in regulating the primary pathway of siRNA delivery. In NPC1‐deficient cells, target genes are silenced, facilitating LNP endosomal escape [239]. Additionally, NPC1 binds to cholesterol in the endocytic pathway, impeding endosomal escape [240].

The structural characteristics of LNP are pivotal in determining its functional capabilities. The incorporation of unsaturated tail groups into ionizable lipids, the introduction of environmentally responsive PEGylation, the addition of polymers and antibodies to enhance membrane rupture, and the improvement of LNP's ability to escape endosomes through active modulation of endosomal recycling pathways provide a theoretical foundation for subsequent studies.

The success of LNP‐based mRNA vaccines against COVID‐19 is a major milestone in infectious disease prevention. As clinical applications expand to cancer, rare genetic disorders, autoimmune conditions, and cardiovascular diseases, safety concerns have become increasingly prominent. Key safety issues include the intrinsic immunogenicity of LNPs, hepatotoxicity, and potential inflammatory responses.

Safety remains one of the most pressing challenges for the successful clinical translation of LNP‐based therapeutics. While the global rollout of mRNA–LNP vaccines has demonstrated that these systems can be safe and well tolerated in the short term, their broader use in chronic diseases and repeated dosing scenarios exposes a range of potential toxicological concerns [296]. One of the most widely discussed issues is the acute activation of the innate immune system, which may occur through pattern recognition receptors such as TLRs or the complement system. This can manifest as systemic inflammatory responses or, in severe cases, complement activation‐related pseudoallergy [204]. In addition, hepatotoxicity has been observed in some preclinical studies, raising concerns about the accumulation and metabolism of ionizable lipids in the liver [297]. Nonbiodegradable lipid components pose further risks, as repeated administration may lead to their persistence in tissues, resulting in long‐term off‐target effects [298]. Another barrier is the immunogenicity associated with PEGylated lipids. Although PEG is often employed to enhance circulation time, repeated dosing can elicit anti‐PEG antibodies, which accelerate blood clearance and may induce hypersensitivity reactions, thereby compromising both safety and efficacy [299].

These challenges are particularly relevant for indications requiring chronic treatment, such as genetic diseases or cancer immunotherapy, where frequent administration is unavoidable [300]. Moreover, data on long‐term safety remain scarce, especially in vulnerable populations such as children, the elderly, and immunocompromised patients [301]. Addressing these issues requires the rational design of biodegradable lipids that can be safely metabolized, coupled with systematic toxicological assessments in relevant animal models. Development of predictive biomarkers for adverse events and improvements in preclinical models—such as organ‐on‐chip platforms or humanized models—may provide deeper insights into the mechanisms of toxicity and immunogenicity. Overall, the establishment of a robust safety profile will be essential to broaden the therapeutic potential of LNP‐based medicines beyond vaccines.

Efficient, scalable, and reproducible manufacturing represents another critical barrier for the translation of LNP‐based therapeutics. The formulation of LNPs typically relies on rapid mixing methods, such as microfluidics or ethanol injection, which enable precise control of NP size and encapsulation efficiency at laboratory scale [302]. However, scaling up these processes to industrial production remains a significant challenge. Consistency in particle size distribution, encapsulation efficiency, and surface properties must be maintained across large batches to meet regulatory standards, but variability in raw materials—particularly in ionizable lipids, and nucleic acid payloads—can compromise reproducibility [303]. Moreover, manufacturing of GMP‐grade ionizable lipids is costly and technically demanding, with complex purification steps that increase production costs and limit global accessibility. Another practical obstacle lies in the supply chain and logistics. Most currently approved mRNA–LNP vaccines require stringent cold‐chain storage and transport conditions, with temperatures as low as −70°C, posing serious challenges for distribution, especially in low‐ and middle‐income countries [304]. Although advances in lyophilization, thermostable LNP formulations, and alternative storage buffers are emerging, their application in large‐scale commercial products remains limited [305].

To overcome these hurdles, a shift toward robust quality‐by‐design strategies is needed, integrating advanced process analytical technologies (PAT) to ensure real‐time monitoring and control of CQAs. Additionally, automation, high‐throughput formulation platforms, and closed‐system bioprocessing are being explored to reduce variability and improve scalability. Collaborative efforts between academia, industry, and regulatory bodies are also necessary to establish standardized manufacturing protocols and benchmarks that can accelerate technology transfer from the laboratory to GMP facilities. Ultimately, addressing these manufacturing barriers is essential not only for cost‐effectiveness and commercial viability but also for ensuring equitable global access to LNP‐based therapeutics.

The regulatory pathway for LNP‐based therapeutics remains complex and relatively underdeveloped compared with traditional drug modalities, posing a significant barrier to clinical translation. Unlike small molecules or conventional biologics, LNPs are hybrid systems that combine a therapeutic payload (such as mRNA, siRNA, or protein) with a nanostructured delivery vehicle, meaning that both the cargo and the carrier must be evaluated simultaneously. Regulatory agencies such as the US FDA and the European Medicines Agency (EMA) currently require detailed physicochemical characterization of LNP formulations, including size, polydispersity, charge, encapsulation efficiency, and stability. In addition, comprehensive evaluation of biodistribution, clearance pathways, and potential immunogenicity are mandated. However, standardized assays and validated analytical methods for NP evaluation remain limited, creating variability between studies and complicating regulatory submissions [306]. Another challenge lies in the regulatory evaluation of novel lipid components. Many ionizable lipids and helper lipids designed for next‐generation LNPs are chemically unique and lack historical safety data, making it difficult to classify them as conventional excipients. This uncertainty can lead to delays in approval, as additional toxicological and pharmacokinetic studies are often required. Furthermore, guidance for combination products—where the therapeutic activity results from the interplay between the drug payload and its carriers still evolving, leaving sponsors with limited clarity on submission strategies.

Beyond technical hurdles, global harmonization of regulatory frameworks is lacking. Differences in requirements between the US FDA, EMA, and regulatory bodies in Asia create additional barriers for multinational clinical development. To address these issues, international collaboration will be crucial in establishing harmonized guidelines and reference standards. Regulatory innovation, such as adaptive approval pathways and rolling review processes, may also accelerate clinical translation while ensuring patient safety. In the long term, fostering stronger dialogue between regulators, researchers, and industry stakeholders will be essential for balancing innovation with rigorous safety and efficacy evaluation.

In summary, despite remarkable progress in advancing LNP‐based therapeutics, unresolved challenges in safety, manufacturing, and regulatory frameworks continue to impede their broad clinical application. Addressing these translational barriers through innovative design, standardized production, and harmonized oversight will be pivotal for unlocking the full potential of LNPs. Looking ahead, future research directions offer promising strategies to overcome these limitations.