Investigating the functional contributions of phospholipids in selective organ targeting lipid nanoparticles

Abstract

Modular lipid nanoparticles (LNPs) are a promising platform to deliver mRNA to various tissues and cells. Optimization of LNPs for hepatic and extrahepatic tissues often involves substitution of helper lipids or addition of novel lipids not found in conventional four-component LNPs. Among the lipids that comprise LNPs, the functional contributions of phospholipids (PLs) in selective organ targeting (SORT) LNPs remain poorly understood. In this study, we systematically evaluate the roles of PLs within SORT LNPs. Our results demonstrate that PL enrichment enhances cellular transfection efficiency by increasing membrane fusion and endosomal escape. In vivo, we observe that PL-containing SORT LNPs significantly increase protein expression following intramuscular administration in mice, whereas moderate PL inclusion is optimal for intravenous delivery. Cryo-electron microscopy reveals that PL modulation induces distinct morphological rearrangements in LNP structure, which may influence the selective adsorption of plasma proteins, an essential factor in endogenous targeting mechanisms. These findings highlight the fundamental role of PLs in supporting intracellular delivery and guiding organ-specific biodistribution through protein corona formation. A deeper understanding of the structural and functional impact of lipid components, especially PLs, will be crucial for the rational design of next-generation mRNA delivery systems with improved efficacy and precision.

1. Introduction

Lipid nanoparticles (LNPs) have emerged as one of the most clinically advanced platforms for nucleic acid-based therapeutics. Their development has revealed the potential of mRNA as a next-generation drug across a broad range of applications including vaccines, protein replacement therapies, cancer immunotherapies, and gene editing [1-4]. LNPs enable mRNA therapeutics by protecting nucleic acids from enzymatic degradation, facilitating extracellular transport, and promoting cellular uptake [5]. However, many key physical and biological processes involved in effective mRNA delivery, such as endosomal escape and organ-selective distribution, remain poorly understood, limiting broader clinical applications.

Conventional LNPs are typically composed of four lipid components. Ionizable amino lipids are critical for nucleic acid encapsulation [6] and intracellular release via endosomal fusion and escape [7,8]. PEGylated lipids improve colloidal stability, control particle size, and extend circulation time [9,10]. Cholesterol and phospholipids (PLs), often referred to as “helper lipids”, aid in nucleic acid encapsulation, cellular delivery, and structural stability [6,11,12]. PLs, in particular, play an essential role in forming the outer bilayer due to their amphiphilic nature, creating a favorable environment for nucleic acid packaging [13, 14]. These zwitterionic PL molecules are also hypothesized to associate with ionizable lipids, nucleic acids, and water molecules during LNP self-assembly [7,11,15]. Molecular dynamics simulations suggest that PLs localize around internal water pockets containing RNAs and are dispersed throughout the LNP surface [16]. Although originally adopted from small-molecule drug liposome formulations, PLs offer untapped potential for advancing LNP design and functionality [17-19]. Despite their inclusion in approved therapies such as Onpattro and COVID-19 vaccines, PLs have received less attention, as research has largely prioritized the development of novel ionizable lipid and lipid-like materials [20]. This focus has overshadowed the functional contributions of PLs in creating safer, more stable, and more effective delivery vehicles.

Clinically approved LNP formulations include the PL 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) [1]. Due to its saturated alkyl chains and bulky phosphocholine (PC) headgroup, DSPC adopts a cylindrical shape that favors lamellar phase formation and enhances membrane stability [21]. In contrast, phosphoethanolamine (PE)-based PLs such as 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) possess a conical geometry due to their unsaturated chains and smaller headgroup [22]. This molecular geometry promotes the formation of an inverted hexagonal phase, which increases membrane fluidity, facilitates endosomal destabilization, and enhances intracellular mRNA delivery [23,24]. Moreover, recent studies have demonstrated that PL molecular geometry significantly affects LNP transfection efficiency, with PE-based PLs outperforming PC analogs in mRNA delivery both in vitro and in vivo [25]. PLs have also been engineered to enhance the innate fusogenic properties of conical lipids by promoting the formation of a cubic-phase lipid structure, which is known to improve cytosolic nucleic acid delivery [26].

Although clinical applications of I.V.-administered LNPs have been largely restricted to the liver [27], recent advances have been made in delivery to other tissues [28-31]. The development of Selective Organ Targeting (SORT) LNPs expanded mRNA delivery to tissues such as the lungs, spleen, bone marrow, and kidneys by incorporating a fifth lipid component that modulates endogenous targeting [32-35]. This strategy leverages adsorption of different plasma proteins due to the tailoring of physicochemical properties for distinct protein corona formation to achieve organ delivery via an endogenous targeting mechanism of action [32,36,37]. This advancement opened up new avenues of therapeutic opportunities in additional tissues, though it also adds another layer of complexity to an already intricate system.

In this study, we systematically investigate the functional contributions of PLs in SORT LNPs. We show that PL inclusion enhances membrane fusion, endosomal escape, and transfection efficiency. PL-enriched five-component SORT LNPs drive higher production of human erythropoietin (hEPO) protein in mice following delivery of hEPO mRNA, indicating improved efficacy compared to PL-deficient four-component LNPs. Additionally, cryo-EM imaging reveals PL-driven nanostructural rearrangements that influence protein corona formation, which mediates organ targeting. Together, these findings underscore the importance of PLs not only in traditional LNP formulations but also in advancing the capabilities of next-generation organ-selective LNPs.

2. Results and discussion

2.1. Phospholipid content does not influence the formation of SORT LNPs

Helper lipids are known to support LNP structural integrity, enhance mRNA encapsulation, and promote cellular delivery [15,38]. While most recent studies have focused on conventional four-component LNP formulations composed of ionizable amino lipids, PLs, cholesterol, and DMG-PEG2000 that typically target the liver, the emergence of LNPs containing non-canonical lipids such as five-component SORT LNPs introduces complexity that challenges the direct application of established design principles. More recent efforts have investigated replacing PLs with SORT lipids to enable tissue-specific delivery via the same SORT endogenous targeting mechanisms [39-41]. Although reducing the complexity of five-component SORT LNPs may simplify manufacturing and regulatory tolerability studies, a deeper understanding of lipid-lipid interactions is essential for optimizing organ selectivity and transfection efficiency in these advanced delivery systems. Since chemical and biologically active lipids can induce adverse effects, such as innate immune activation [42,43], mistargeting [44], or reduced transfection efficiency [25,39,40], understanding the fine-tuning required to balance organ specificity and potency is essential for the continued advancement of LNP therapeutics. It is therefore important to understand the potential contributions, or lack thereof, of PLs in the context of organ-selective LNPs.

To this end, we employed a systematic approach by modulating PL content in liver, lung, and spleen SORT LNP formulations and studying functional contributions. The dynamic range of PL content was selected between zero and double the original molar percent inclusion to ensure sufficient ionizable amino lipid content to aid mRNA encapsulation and release (Fig. 1a and b, Supplemental Tables S1-3). Additionally, high PL levels have been reported to prolong circulation time by reducing serum protein binding, which may, in turn, decrease tissue accumulation [45]. To maintain consistent LNP composition and minimize unintended variations, each of the supplemental SORT lipids (DODAP, DOTAP, and 18:1 PA) was adjusted to compensate for changes in DOPE content, as these lipids are primarily believed to influence organ tropism in vivo, resulting in constant amounts of ionizable amino lipid, cholesterol, and PEG lipid (Fig. 1b, Supplementary Tables S1-3).

Fig. 1. Design and characterization of phospholipid (PL) titration series in organ-selective SORT LNPs.

(a) Schematic of the formulation process for liver-, lung-, and spleen-targeting SORT LNPs. A lipid ethanol mixture (including ionizable lipid, cholesterol, phospholipid, PEG lipid, and SORT lipid) was rapidly mixed with mRNA in an acidic aqueous buffer to form LNPs. (b) Visual summary of PL titration strategy for each organ-targeted LNP. Original SORT formulations are highlighted in bold. PL molar content was systematically increased while compensating by reducing SORT lipid content to maintain a consistent 5-component formulation. (c) Hydrodynamic diameter and polydispersity index (PDI) measured by dynamic light scattering (DLS) for all PL titration series (n = 3). (d) mRNA encapsulation efficiency of PL-titrated SORT LNPs, measured via RiboGreen assay (n = 4).

As an initial assessment of LNP formation, we evaluated the size, polydispersity index (PDI), and encapsulation efficiency of all SORT LNP formulations prior to in vitro or in vivo assays. LNPs were prepared via rapid vortex mixing [46] (Fig. 1a), yielding stable, nearly monodisperse nanoparticles ranging from 100 to 150 nm in diameter (Fig. 1c) with favorable encapsulation efficiencies (Fig. 1d). Interestingly, in the liver SORT series, increasing PL content led to a slight decrease in encapsulation efficiency. This suggests that the supplemental ionizable lipid DODAP may enhance nucleic acid complexation and potentially synergize with the primary ionizable amino lipid to improve mRNA encapsulation. For lung-targeting LNPs, high encapsulation efficiencies were observed regardless of PL content, underscoring the beneficial role of positively charged lipids in stabilizing nucleic acid interactions. Both liver and lung SORT LNPs maintained neutral or positive zeta potentials, respectively, with no significant difference observed across PL variants (Fig. S1a and b). In contrast, spleen SORT LNPs displayed reduced encapsulation efficiency at low PL content, likely due to electrostatic repulsion between the highly anionic SORT lipid 18:1 PA and the negatively charged mRNA. This was supported by zeta potential measurements as PL-depleted formulations exhibited increased negative surface charge, while increasing PL content progressively reduced electronegativity (Fig. S1c). Although PLs are traditionally considered nonessential “helper lipids” for LNP assembly, our results indicate they affect physical properties of formulated LNPs. Moreover, the inclusion of supplemental SORT lipids appears to compensate for PL depletion. Given that DODAP, DOTAP, and 18:1 PA share a common lipophilic backbone, they may integrate into and stabilize the outer bilayer structure in the absence of PLs. This structural compensatory effect is evident, as all tested formulations exhibited substantial stability across various conditions. Using dynamic light scattering (DLS), we observed no change in particle size over a 7-day incubation at 25 °C for liver-, lung-, or spleen-targeting LNPs (Fig. S2a). Additionally, serum stability was assessed by incubating liver-, lung-, and spleen-targeting SORT LNPs in low-protein (10 % FBS) and high-protein (50 % FBS) conditions at 37 °C for 120 min. Under both conditions, the particle size distribution of all formulations tested remained unchanged (Fig. S2b-d). These results confirm the exceptional stability of the formulations, with no evidence of aggregation over 7 days or under high serum protein conditions.

2.2. Phospholipids in SORT LNPs can enhance endosomal membrane interactions

Although LNPs overcome extra- and intracellular barriers for nucleic acid delivery, endosomal escape remains a limiting factor for achieving efficient mRNA transfection [47]. Prior studies suggest that the unsaturated alkyl tails of PLs enhance membrane fluidity, thereby facilitating endosomal membrane destabilization. In SORT LNPs, the supplemental lipid component shares similar fatty acid tails with DOPE; however, the absence of a zwitterionic headgroup may impair lipid organization and alter interactions with endosomal membranes. This suggests that PL structure and inclusion in five-component LNPs may significantly influence membrane fusion dynamics and subsequent cell transfection.

To evaluate the role of PLs in endosomal membrane fusion, we assessed the SORT LNP formulation series using a Förster Resonance Energy Transfer (FRET) assay against plasma membrane-mimicking liposomes (PMMLs) and endosomal membrane-mimicking liposomes (EMMLs). In intact liposomes, FRET dye pairs remain closely associated, producing high FRET efficiency. Upon fusion with LNPs, liposomal membranes expand, increasing dye separation and decreasing FRET signal (Fig. 2a) (Fig. S7b). In the context of PMML, membrane fusion efficiency remained unchanged across lung and spleen SORT LNP formulations, regardless of PL content (Fig. 2c and d). The neutral pH and lack of negatively charged lipids in PMMLs likely limit electrostatic interactions, making PL effects negligible in this context. Interestingly, liver SORT LNPs exhibited a slight inverse correlation between PL content and membrane fusion efficiency in PMMLs (Fig. 2b), though the biological significance of this remains unclear. In contrast, EMMLs, which include acidic pH and negatively charged lipids to mimic endosomal conditions, revealed a positive linear relationship between PL content and membrane fusion efficiency for both liver and lung SORT LNPs (Fig. 2b and c). These enhancements are likely driven by PL-facilitated promotion of a hexagonal phase transition, which destabilizes endosomal membranes and aids release of mRNA cargo. Notably, spleen-targeting LNPs showed minimal fusion improvements despite increased PL content (Fig. 2d). This may be attributed to electrostatic repulsion, given that the highly anionic SORT lipid 18:1 PA could disrupt bilayer organization and counteract the potential benefits of PL enrichment. Collectively, these data suggest that PL incorporation enhances endosomal membrane fusion in a formulation-dependent manner. Specifically, PLs may act synergistically with ionizable lipids to promote phase transitions that facilitate endosomal escape, a key step in effective intracellular mRNA delivery.

Fig. 2. PL incorporation enhances endosomal membrane fusion in a formulation-dependent manner.

(a) Schematic representation of the membrane fusion assay using FRET-labeled plasma membrane-mimicking liposomes (PMML) and endosomal membrane-mimicking liposomes (EMML). Membrane fusion efficiency of liver (b), lung (c), and spleen (d) SORT formulation series, measured by FRET, showing differences in fusion efficiency when tested against PMML and EMML. Data were analyzed via simple linear regression at 120 min post-incubation (n = 4). Results are shown as mean ± SD.

2.3. Phospholipid inclusion in SORT LNPs significantly boosts cell transfection efficiency in vitro

Given the improved membrane fusion observed with PL-enriched LNPs related to endosomal escape, we next evaluated their impact on functional cellular transfection efficiency. HEK293 and HeLa cells were transfected with liver, lung, and spleen SORT LNP series formulated with Luciferase mRNA (Fig. 3a). Luciferase activity, measured via mean luminescence intensity, served as a quantitative readout of successful endosomal escape and mRNA translation within cells. Despite all SORT LNP formulations having desirable sizes and significant mRNA encapsulation (Fig. 1c and d), all PL-deficient formulations failed to transfect cells, demonstrating a complete loss of functional mRNA delivery. In contrast, increasing PL content across all SORT LNP series led to a clear, linear enhancement in transfection efficiency, independent of the cell type used (Fig. 3b-d) (Fig. S3d-f). Importantly, none of the formulations, regardless of PL content, had any measurable effect on cell viability compared to untreated controls (Fig. S3a-c), confirming their biocompatibility. Interestingly, the enrichment of PL resulted in a substantial transfection boost. Across all SORT LNP types, doubling the PL molar percentage led to up to a 5-fold increase in transfection efficiency relative to the original formulations (Fig. 3b-d). Moreover, further PL enrichment in 4-component liver-targeting and 5-component lung-targeting LNPs resulted in maximum increases in cell transfection efficiency at 45.5 % and 39.7 % PL (Fig. S5a and b, Table S5), respectively. Spleen-targeting LNPs exhibited an increase in transfection efficiency up to 40 % PL supplementation (Fig. S5c). In addition, to determine whether our findings extend to other phospholipids, we evaluated the effects of supplementing SORT LNPs with the helper lipids DSPC and DOPC. Other than compatibility exceptions for DSPC in Liver SORT LNPs and DOPC in Lung SORT LNPs, all combinations of PLs and SORT LNPs yielded a dose responsive increase in cellular transfection (Fig. S6). Additionally, we further validated our findings by using cholesterol as the compensating lipid component, adjusting its proportion in response to changes in PL molar percent inclusion in our SORT formulations (Table S4). This approach allowed us to maintain constant levels of ionizable lipid, PEG-lipid, and SORT lipid. Consistent with our previous results, PL-depleted liver, lung, and spleen SORT formulations exhibited minimal transfection, whereas PL-containing formulations demonstrated improved potency. Further enrichment of PLs led to the highest transfection efficiency observed across all tested formulations, supporting the conclusion that the enhanced potency is driven by PL enrichment rather than by the reduction of any potential negative effects from the SORT lipid (Fig. S4a-c). Altogether, these findings highlight the essential role of PLs in SORT LNP-mediated mRNA delivery, demonstrating their contribution to both endosomal escape and intracellular mRNA translation.

Fig. 3. Phospholipid inclusion in SORT LNPs boosts in vitro cell transfection efficiency.

(a) Experimental schematic of the in vitro transfection procedure, with measurement of transfection efficiency and cell viability. Quantification of mean luminescence following transfection with liver (b), lung (c), and spleen (d) SORT LNPs in HEK293 (top) and HeLa (bottom) cells (n = 4). Statistical significance was determined using one-way ANOVA with multiple comparisons. The original SORT LNP formulation (bolded) is compared to all other formulations. ****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05.

2.4. PL inclusion in LNPs improves transfection efficiency by increasing endosomal escape

To directly evaluate how PL inclusion mediates endosomal escape in cells, we employed the lipophilic dye DiD as a reporter for LNP-endosome fusion events. At high concentrations, DiD exhibits self-quenching within the LNP membrane. Upon endosomal disruption, DiD diffuses into the endosomal and plasma membranes, leading to fluorescence recovery as the dye becomes dequenched (Fig. 4a) [48,49]. We incorporated DiD (1 mol% of total lipids) into each formulation of the systematic SORT LNP series and tracked fluorescence intensity over 6 h in HEK293 cells as a proxy for endosomal escape. Consistent with prior transfection data, increased PL content was associated with a higher rate and greater total number of escape events across liver, lung, and spleen SORT LNPs (Fig. 4b-d). In liver SORT formulations, PL enrichment led to a rapid rise in fluorescence, suggesting efficient early endosomal disruption and prompt mRNA release. Although the original formulation exhibited some improvement over the PL-depleted variant, both shared similar initial escape kinetics. The advantage of PL inclusion became apparent through an increased cumulative number of escape events (Fig. 4b). PL-depleted liver SORT LNPs, in contrast, demonstrated a gradual, diminished fluorescence recovery, indicative of reduced endosomal disruption. Interestingly, PL-deficient (0 %) lung SORT LNPs exhibited elevated DiD fluorescence despite poor functional delivery of mRNA (Fig. 3c), suggesting premature DiD release unrelated to cellular uptake and endosomal escape. This may result from a higher proportion of the cationic lipid DOTAP in the absence of structural PLs, potentially destabilizing the LNP membrane and promoting non-specific membrane interactions. This hypothesis was supported by minimal cellular uptake of PL-depleted lung SORT LNPs observed in endocytosis assays (Fig. S8b). Nevertheless, PL-enriched lung SORT LNPs achieved a two-fold increase in endosomal escape and a faster initial release rate compared to the original formulation (Fig. 4c). For spleen SORT LNPs, PL inclusion significantly enhanced both transfection efficiency and endosomal escape events (Fig. 4d). Furthermore, rhodamine-labeled LNPs enabled simultaneous tracking of endocytic uptake. Across all tissue-targeting SORT LNPs, PL enrichment correlated with increased rhodamine signal intensity, indicative of enhanced cellular internalization and accumulation (Fig. S8a-c). Together, these findings demonstrate that PL inclusion improves SORT LNP-mediated mRNA delivery by enhancing both cellular uptake and endosomal escape efficiency, which are critical steps for effective intracellular mRNA release and translation.

Fig. 4. Phospholipid enrichment in SORT LNPs improves endosomal escape in cells.

(a) Experimental schematic for in-cell tracking of LNP endosomal escape, measured by the evolution of fluorescence intensity over time using live-cell imaging. Representative data showing endosomal escape for liver (b), lung (c), and spleen (d) SORT LNP series. Left panels show microscopy images (scale bar = 100 μm) of DiD fluorescence intensity at 6 h post-treatment (20 × magnification). Middle panels present the quantified mean fluorescence intensity normalized to cell count per frame (n = 4) as a function of time, with statistical significance indicated at the 6-h timepoint. Right panels show area under the curve analysis of fluorescence intensity over time. Statistical significance was assessed using one-way ANOVA with multiple comparisons, comparing all formulations to the original SORT LNP (bolded). ****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05.

2.5. PL inclusion enhances in vivo mRNA delivery efficacy across multiple tissue-targeting SORT LNPs in vivo

To evaluate the implications of PL inclusion in SORT LNPs in vivo, 6–8-week-old C57BL/6 mice were administered formulations from the PL titration series containing human erythropoietin (hEPO) mRNA to quantify the amount of protein that could be produced and secreted from transfected cells. This approach allowed us to obtain a more clinically relevant comparison across formulations by measuring protein expression over multiple days post-treatment with high accuracy in protein quantification. LNPs were administered via intravenous (IV) and intramuscular (IM) injection at a dose of 0.3 mg/kg. Tissues successfully transfected by LNPs produced and secreted hEPO into the bloodstream. Blood samples were collected at 0, 6, 24, 48, and 72 h post-treatment, and hEPO protein levels were quantified by ELISA (Fig. 5a). In liver SORT LNPs, PL enrichment led to improved transfection efficiency and significantly higher systemic hEPO production following both IV and IM administration (Fig. 5b). These findings support application of PL inclusion to enhance hepatic delivery in mRNA-based therapeutics. Lung-targeting SORT LNPs showed only modest variation in transfection efficiency upon PL modulation following IV administration, with PL enrichment or depletion resulting in a slight reduction in hEPO levels (Fig. 5c). This suggests that moderate PL content is optimal for efficient lung-targeted delivery. We acknowledge that differences between in vitro and in vivo delivery outcomes have been documented, stemming from the additional biological barriers faced by LNPs in vivo [50]. Notably, lung SORT LNPs failed to produce detectable hEPO after IM injection, underscoring the critical role of the protein corona in mediating organ-specific targeting and cellular uptake in vivo. Spleen-targeting SORT LNPs exhibited minimal sensitivity to PL modulation via IV administration. However, IM delivery led to improved hEPO expression in mice receiving PL-enriched formulations, indicating a route-dependent benefit of PL inclusion (Fig. 5d). Overall, these results demonstrate that PL inclusion generally correlates with enhanced in vivo mRNA delivery efficacy across multiple tissue-targeting SORT LNPs. Across all tested formulations in vivo, PL-depleted LNPs (0 % PL inclusion) consistently underperformed relative to their PL-supplemented counterparts, highlighting the value of this lipid component for effective mRNA delivery to the target tissue. Interestingly, the optimal PL inclusion percentage was formulation-dependent. Moreover, the administration route plays a pivotal role in determining transfection efficiency, highlighting the significance of endogenous targeting mechanisms and their correlation with formulation parameters.

Fig. 5. Phospholipid enrichment in SORT LNPs enhances protein production in vivo.

(a) Experimental schematic depicting in vivo transfection in C57BL/6 mice. Mice were administered SORT LNPs containing hEPO mRNA (dose of 0.3 mg/kg) via intravenous (I.V.) or intramuscular (I.M.) injection. Blood samples were collected at 0, 6, 24, 48, and 72 h post-injection to quantify hEPO levels using ELISA. (b–d) Protein expression kinetics following I.V. (left) and I.M. (right) administration of liver (b), lung (c), and spleen (d) SORT LNP series. Graphs show hEPO levels over time, with area under the curve (AUC) analysis used to determine total protein expression. Statistical comparisons were performed using one-way ANOVA with multiple comparisons, comparing all formulations to the original SORT LNP (bolded). ****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05.

2.6. Phospholipid inclusion alters LNP morphology and protein corona formation

Until recently, PLs were assumed to function primarily as structural components, responsible for mediating particle self-assembly and providing membrane stability. However, previous studies revealed that PLs can also serve functional roles in intracellular mRNA delivery, indicating that their structural influence may be more intricate than previously understood [51,52]. To investigate the structural consequences of PL variation in SORT LNPs, we obtained Cryo-EM images of liver-, lung-, and spleen-targeting formulations with different PL contents. To better visualize lipid assembly and internal organization, LNPs were prepared without mRNA cargo, as its high electronegativity tends to obscure internal structural features (Fig. 6a). In liver SORT LNPs, PLs influenced the formation of a discrete lipid core structure, presumably associated with the encapsulation domain of the ionizable amino lipid (Fig. 6b). PL-depleted (0 %) liver SORT LNPs exhibited a singular external bilayer lacking an internal lipid core, whereas PL-rich formulations retained a semi-organized lipid core that protruded beyond the outer bilayer. This protruding structure in PL-rich LNPs may represent an inverted hexagonal phase, known to be highly fusogenic, potentially explaining the enhanced endosomal fusion and mRNA transfection observed in functional assays. Lung SORT LNPs displayed a heterogeneous population of multilamellar and amorphous particles, with lamellarity modulated by PL content (Fig. 6c). In PL-depleted conditions, lung SORT LNPs formed unilamellar structures with disorganized internal cores. Conversely, PL enrichment led to an increase in lamellarity, producing multilamellar vesicles extending from the outermost bilayer to the core. Although DOTAP typically favors multilamellar vesicle formation due to its structural and electrostatic properties, the observed increase in lamellarity with PL inclusion, rather than DOTAP, suggests that PLs may displace the SORT lipid within the particle. This lipid rearrangement could contribute to the formation of distinct lamellar domains. Spleen SORT LNPs also exhibited PL-dependent morphological changes. PL-deficient particles displayed polygonal shapes with sharp edges, likely resulting from high membrane fluidity and instabilities in the outer bilayer [53]. As PL content increased, these particles adopted a more spherical morphology, indicating that PL contributes to membrane stability and structural regularity (Fig. 6d).

Fig. 6. Phospholipid content affects SORT LNP morphology and protein corona composition.

Cryo-electron microscopy images of Luc mRNA-loaded liver, lung, and spleen SORT LNPs (a). Cryo-electron microscopy images of empty liver (b), lung (c), and spleen (d) SORT LNPs across varying phospholipid compositions, illustrating internal and external structural changes driven by PL enrichment (scale bar = 50 nm). (e) Schematic of the experimental workflow for protein corona analysis via mass spectrometry following incubation of SORT LNPs in purified mouse plasma. Quantification of key adsorbed plasma proteins (ApoE, Vitronectin, and β-2-glycoprotein 1) on liver (f), lung (g), and spleen (h) SORT LNPs. Statistical analysis was performed using one-way ANOVA with multiple comparisons, comparing each formulation to the original SORT LNP (bolded). ****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05.

The adsorption of serum proteins to LNPs has been shown to drive organ and cell enriched delivery via an endogenous targeting mechanism of action [28,44,54]. Previous work identified proteins including apolipoprotein E (ApoE), vitronectin (VTN), and β2-glycoprotein 1 (β2GP1) as highly enriched on the surface of organ-targeting LNPs. To assess whether PL-induced structural changes affect protein corona formation, we incubated each formulation with purified mouse plasma, followed by isolation of LNP-bound proteins and mass proteomics to quantify the enrichment of endogenous targeting-relevant proteins (Fig. 6e). In liver-targeting LNPs, an increase in ApoE enrichment was observed between PL-depleted (0 %) and PL-rich (19.1 % and 38.1 %) SORT LNPs, with no statistical difference between the two PL-rich containing LNPs (Fig. 6f). Although structural changes were seen with variations in PL, the overall lipoparticle-like nature of liver SORT LNPs remained unchanged, allowing effective binding to lipoproteins. This observation is further supported by pKa measurements using the TNS assay (Fig. S9) and in vivo biodistribution data (Fig. S10), which show no significant changes as a function of PL content. For lung SORT LNPs, no statistically significant difference in VTN adsorption was observed between PL-depleted (0 %) SORT LNPs and PL-rich benchmark (11.9 %) SORT LNPs (Fig. 6g). However, the further increase of PL from 11.9 % to 23.8 % led to a reduction of VTN adsorption (Fig. 6g), which may be attributed to the internalization or displacement of DOTAP within the multilamellar structures of PL-rich formulations. This structural change observed in the Cryo-EM images (Fig. 6c) may potentially impede the surface exposure of the SORT lipid and limit VTN recruitment, therefore weakening the endogenous targeting mechanism required for lung-specific delivery. Spleen SORT LNPs showed no significant change in β2GPI enrichment across formulations (Fig. 6h), suggesting that structural PL variation does not disrupt the primary targeting mechanism. However, ApoE adsorption increased with PL enrichment from 16.7 % to 33.3 %, which may interfere with splenic targeting by redirecting LNPs toward liver accumulation, likely explaining the diminished IV transfection efficiency observed with PL-rich spleen SORT LNPs (Fig. 5d). Together, these findings highlight the dual structural and functional roles of PLs in dictating LNP morphology, protein corona formation, and potential organ-specific mRNA delivery efficacy. While recent structural studies have shown that the surface composition and internal organization of LNPs play a critical role in intracellular delivery [48,53], it will be important to explore how distinct LNP nanostructures influence extracellular trafficking through specific protein interactions and other potential mechanisms.

3. Conclusions

LNPs have emerged as one of the most advanced and widely adopted non-viral delivery platforms for nucleic acid therapeutics. Despite their clinical success, the mechanistic understanding of functional mRNA delivery and rational LNP design remains incompletely understood. Given the vast and largely untapped chemical space for lipid optimization, a deeper understanding of the functional contributions of individual lipid components is essential for the continued improvement of this technology. In this study, we systematically investigated the role of PLs in organ-specific SORT LNPs. Our findings demonstrate that PLs function beyond their traditional role as structural “helper lipids.” Notably, they are critical to efficient cellular transfection and modulate endogenous targeting in organ-selective LNPs. Across both in vitro and in vivo models, increased PL content correlated with improved membrane fusion, greater endosomal escape efficiency, and enhanced mRNA transfection. Furthermore, modulation of PL levels altered LNP morphology and internal lipid organization, which influenced the surface presentation of SORT molecules and the recruitment of plasma proteins for protein corona formation, further affecting tissue targeting and biodistribution. Together, our results highlight the multifaceted roles of PLs in LNP design and highlight their importance not only for structural integrity but also for driving transfection efficiency and organ specificity. As the field advances, deliberate control over lipid architecture will be crucial for engineering next-generation LNPs capable of delivering mRNA-based therapies with improved precision and efficacy.

4. Materials and methods

4.1. Materials

The ionizable amino lipid 5A2-SC8 was synthesized in our lab following previously reported protocols. The following lipids were purchased from Avanti Polar Lipids: 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1-oleoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (18:1 NBD Lyso PE), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (18:1 Liss Rhod PE), sphingomyelin (Egg SM), bis(monooleoylglycero) phosphate (S,R isomer) (18:1 BMP (S,R)), l-α-phosphatidylinositol (Soy PI), 1,2-dioleoyl-3-dimethylammonium-propane (DODAP), 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP), and 1,2-dioleoyl-sn-glycero-3-phosphate (18:1 PA). Cholesterol and DMG-PEG2000 (Sunbright GM-020) were purchased from Sigma Aldrich and NOF America Corporation, respectively. The ONE-Glo + Tox Luciferase Reporter and Cell Viability Assay Kit was obtained from Promega. The lipophilic dye1,1′-Dioctadecyl-3,3,3′,3′-Tetramethylindodicarbocyanine, 4-Chlorobenzenesulfonate Salt (DiD) (Invitrogen) and D-luciferin, sodium monohydrate salt (GoldBio), were purchased from Fisher Scientific. The hEPO DuoSet ELISA Kit was acquired from R&D Systems.

4.2. LNP preparation

All lipid nanoparticles (LNPs) used for in vitro and in vivo experiments were prepared by rapid vortex mixing according to published protocols [55]. 5A2-SC8, DOPE, cholesterol, DMG-PEG, and the corresponding SORT lipid (DODAP, DOTAP, or 18:1 PA) were suspended in 40 μL of ethanol at a molar ratio of 15:X:30:3:Y, based on the specific SORT formulations. In all proposed SORT LNPs, X = 0, 3, 6, 9, 12, 15, 21, 24, 27, or 30. For liver-targeting LNPs, Y = 30.75, 27.75, 24.75, 21.75, 18.75, 15.75, 12.75, 9.75, 6.75, 3.75, or 0.75. For lung-targeting LNPs, Y = 78, 75, 72, 69, 66, 63, 60, 57, 54, 51, or 48. For spleen-targeting LNPs, Y = 42, 39, 36, 33, 30, 27, 24, 21, 18, 15, or 12.

The required mRNA (Luc or hEPO) was diluted in 120 μL of 10 mM citrate buffer at pH 4.5 for liver- and lung-targeting SORT LNPs, while a pH of 3.0 was used for spleen-targeting LNPs. For all reported LNPs, the lipid-to-mRNA ratio (wt/wt) was maintained at 40:1. The mRNA-containing aqueous phase was rapidly added to the lipid mixture under constant vortexing and mixed for 30 s. The formulated LNPs were then incubated at room temperature for 15 min to allow stable self-assembly and subsequently diluted in 1X PBS or dialyzed as needed.

For in vitro experiments, the formulated LNPs were diluted to the required concentration using sterile 1X PBS. Whereas for in vivo experiments, LNPs were dialyzed against 1X PBS at room temperature for at least 1 h, after which the volume was adjusted to the desired concentration.

4.3. mRNA synthesis

Luciferase (Luc) mRNA for this study was provided by ReCode Therapeutics. hEPO mRNA was generated via in vitro transcription (IVT) using the MEGAscript SP6 Transcription Kit, starting from a hEPO plasmid DNA (pDNA) with a pCS2 backbone. Briefly, linear pDNA containing optimized 5′ and 3′ untranslated regions (UTRs) and poly(A) sequences was first obtained through enzymatic digestion. IVT reactions were then performed following standard protocols, incorporating N1-methylpseudouridine-5′-triphosphate modification. Finally, the mRNA was capped with a Cap-1 structure using Vaccinia Capping Enzyme and 2′-O-methyltransferase.

4.4. LNP characterization

Hydrodynamic radius, polydispersity index, and zeta potential were measured using a DynaPro ZetaStar (λ = 785 nm; detection angles of 90° and 163.5°), coupled to an Arc HPLC system (Wyatt/Waters) for automated sample processing. mRNA encapsulation efficiency was quantified using the QUANT-iT RiboGreen assay (Thermo Fisher Scientific) with slight protocol modifications. A standard curve was prepared, ranging from 10 to 0 ng/μL using the corresponding mRNA, and LNP samples were diluted to a theoretical concentration of 2.5 ng/μL in a black 96-well plate (5 μL per well; n = 4). To quantify free/unencapsulated mRNA, 50 μL of Ribogreen reagent (diluted 1000X) was added to each well, followed by a 5-min incubation under constant shaking. Fluorescence intensity was then measured using a Tecan EVO plate reader (λex = 480; λem = 520). Subsequently, 50 μL of 2 % Triton-X (wt/wt) was added to each well and fluorescence intensity was measured again to determine the total mRNA content after dissolution of the LNPs. The percentage of encapsulated mRNA was calculated as follows.

$$EncapsulationEfficiency=\frac{totalmRNA-freemRNA}{totalmRNA}\times 100$$

The colloidal stability of all proposed formulations was assessed using dynamic light scattering (DLS). Formulated LNPs at an mRNA concentration of 2.5 ng/μL were incubated at a constant temperature of 25 °C for 7 days. The LNP hydrodynamic diameter was measured at 0, 1, 3, 5, and 7 days post-formulation using Malvern Zetasizer Nano ZS. Similarly, LNP serum stability was evaluated by incubating the particles in low (10 % FBS) and high (50 % FBS) protein conditions at 37 °C for 120 min. DLS measurements (n = 3) were taken at 0, 15, and 120 min after FBS addition.

Apparent (global) pKa was determined using the well-established TNS assay, following previously described protocols [36]. Briefly, a series of buffer solutions containing 10 mM HEPES, 10 mM MES (4-morpholineethanesulfonic acid), 10 mM ammonium acetate, and 130 mM NaCl were prepared with pH values ranging from 2.0 to 11.0. LNPs were prepared at a total lipid concentration of 60 μM and incubated with TNS reagent (2 μM) in each buffer condition for 5 min at room temperature under agitation (250 rpm) in a 96-well black-bottom plate. Fluorescence intensity was measured using a Tecan plate reader (λEx = 321 nm; λEm = 445 nm) and normalized to the signal at pH 2.0. The inflection point (IC50) of the fluorescence response curve was calculated using GraphPad Prism to determine the apparent pKa of each LNP formulation.

4.5. In vitro membrane fusion FRET assay

LNP membrane fusion efficiency was assessed by tracking fusion events in vitro using endosomal-mimicking liposomes. Plasma membrane-mimicking liposomes (PMMLs) and endosomal-mimicking liposomes (EMMLs) were prepared using the thin-layer hydration method followed by liposome extrusion (Nucleopore membrane size 0.1 μm).

To prepare PMMLs, lipids DOPC, DOPE, NBD Lyso PE, Liss Rhod PE, SM, and cholesterol were mixed in chloroform at a molar ratio of 20:18:1:1:20:30. EMMLs were prepared by mixing DOPC, DOPE, NBD Lyso PE, Liss Rhod PE, BMP, and Soy PI in chloroform at a molar ratio of 50:18:1:1:10:10. Both lipid mixtures were dried in a vacuum oven to form a thin lipid film and subsequently rehydrated either PBS at pH 7.5 or 20 mM citrate buffer at pH 6.0 to form PMMLs and EMMLs, respectively. The suspensions were sonicated using a bath sonicator for 15 min to ensure homogeneity, followed by extrusion through Avanti’s mini extruder. This method produced consistent and nearly monodisperse liposomes, ranging in diameter from 100 to 120 nm (Fig. S7a).

LNPs and liposomes were added to a black 96-well plate at a 10:1 total lipid molar ratio (LNPs at 100 μM; liposomes at 10 μM). Fusion events were monitored by measuring changes in NBD emission intensity (535 nm) due to decreased FRET efficiency every 20 min for 4 h at 37 °C. Minimum and maximum NBD intensities were determined by using untreated liposomes and 2 % Triton-X, respectively. Lipid fusion efficiency was calculated as:

$$FusionEfficiency=\frac{F{I}_{LNP}-F{I}_{\text{min}}}{F{I}_{max}-F{I}_{\text{min}}}\times 100$$

Fusion efficiency was reported at 120 min as a representative time point for all experimental conditions.

4.6. In vitro luciferase transfection efficiency

HEK293 and HeLa cells were seeded in an opaque white 96-well plate at densities of 20,000 and 15,000 cells per well, respectively, in 100 μL of supplemented DMEM medium (10 % FBS). Twenty-four hours after seeding, cells were transfected with the proposed LNP formulations, delivering 15 ng of Luc mRNA per experimental well (n = 4). Transfection efficiency and cell viability were assessed 24 h post-treatment using Promega’s ONE-Glo + Tox kit. An untreated control was used as a reference to quantify the viability of cells after LNP treatment. Luminescence intensity was normalized to the relative viability, and the mean luminescence intensity was reported for each of the treatments.

4.7. Live-cell tracking of LNP endosomal escape

Clear-bottom black 96-well assay plates were precoated with poly-d-lysine (Gibco) for 2 h at 37 °C, then washed three times with UltraPure distilled water. Once dried, HEK293 cells were seeded at 30,000 cells per well in 100 μL of supplemented DMEM medium and incubated overnight. LNPs were prepared following the previously described protocol, with the addition of the lipophilic dye DiD at 1 % of total lipid moles. Live-cell imaging was performed using a Nikon SoRa spinning disk microscope equipped with a culture chamber maintained at 37 °C and 5 % CO₂. Imaging began at the time of cell treatment with 100 ng of mRNA (n = 4), and images were captured every 30 min for 6 h. The relative change in DiD fluorescence intensity was quantified using ImageJ software with mean fluorescence intensity normalized to cell count (Hoechst-stained nuclei) per image. To track LNP endocytosis, test formulations included 0.5 % 18:1 Liss-Rhod-PE and were imaged using the same procedure. Rhodamine signal accumulation was quantified using the same ImageJ macro.

4.8. In vivo transfection efficiency

All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Texas Southwestern Medical Center (UTSW) and complied with all applicable local, state, and federal regulations. Mice were maintained in a specific pathogen-free barrier facility under standardized environmental conditions (22 ± 2 °C, 30–50 % relative humidity, 12-h light/dark cycle) and received standard chow (Teklad Global 2916). Female mice aged 6–8 weeks, bred in-house, were used in all experimental procedures.

LNPs were prepared as previously described, using hEPO mRNA as the reporter gene to transfect mice. Eight-week-old mice were treated with the proposed SORT LNP formulation at a dose of 0.3 mg/kg of hEPO mRNA, either IV or IM. Retroorbital blood collection was performed at 0, 6, 24, 48, and 72 h post-LNP treatment. After each collection, blood samples were incubated at room temperature for 30 min in BD Microtainer serum separator tubes, followed by centrifugation at 2000×g for 15 min. The serum fraction was collected and stored at −80 °C until further analysis. Secreted human erythropoietin (hEPO) protein was quantified using the hEPO DueSet ELISA kit (R&D Systems).

4.9. LNP biodistribution

LNP biodistribution, based on mRNA accumulation, was quantified using LNP-encapsulated Cy5-labeled mRNA (Cellerna). LNPs were formulated using the vortex mixing method as previously described. Female mice (Envigo), 6–8 weeks old, were administered 0.2 mg/kg of LNPs via IV injection. One hour post-injection, major organs were harvested for ex vivo imaging. Fluorescence intensity (λEx = 640 nm; λEm = 690 nm; exposure time = 90 s) was recorded using an AMI-HTX imaging system (Spectral Instruments Imaging).

4.10. Cryo-EM microscopy

Selected Liver, Lung, and Spleen SORT LNP formulations were prepared through microfluidics mixing using a t-mixer (0.25 mm Bore size). Using the same formulation parameters mentioned before, 3.2 mg total lipids in 450 μL of Ethanol were mixed with 1350 μL of citrate buffer to produce empty SORT LNPs. Alternatively, 80 μg of Luc2 mRNa was used to formulate mRNA loaded liver, lung, and spleen SORT LNPs. The resulting LNP batch was concentrated using Amicon Ultra – 2 mL (50 kDa) spin filters down to 60 μL. 6 μL of LNP sample was cast to glow discharged lacey carbon grids (300 mesh) in 3 iterations, then blotted and plunged froze in liquid ethane using a Vitrobot Mark IV at 10 C, 90 % humidity. Sample grids were imaged using Talos Arctica 200 kV (Bioquantum Energy Filter with k3 Detector) at 100,000× magnification. 30 frames per stack image were acquired using dose fractionation mode, and the built-in frame alignment tool was used to generate a final image.

4.11. Protein corona assay

In vitro protein corona formation and isolation were performed following the previously reported protocol [36]. Briefly, formulated LNPs (5 μg total mRNA in 200 μL) were mixed in equal volumes of purified mouse plasma (C57BL/6) and incubated at 37 °C for 15 min. After incubation, protein-bound LNPs were separated from unbound proteins using differential centrifugation in a 0.7 M sucrose cushion at 17,000×g for 60 min. Pelleted LNPs were washed 3 times with PBS, and proteins were purified using a ReadyPrep 2-D cleanup kit. Purified proteins were submitted to the UTSW Proteomics core for mass proteomics analysis.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.biomaterials.2025.123671.

Data availability

Data will be made available on request.