Predictive Lung- and Spleen-Targeted mRNA Delivery with Biodegradable Ionizable Lipids in Four-Component LNPs

The chemical reagents used in this study were sourced from a variety of suppliers, including Sigma Aldrich (Burlington, MA, USA), Tokyo Chemical Industry (TCI, Tokyo, Japan), Fluorochem Ltd. (Hadfield, UK), and Ambeed (Arlington Heights, IL, USA). Cholesterol and 18:1 PA were both purchased from Sigma Aldrich (Burlington, MA, USA). The 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) was obtained from Avanti Polar Lipids (Alabaster, AL, USA), while the 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG 2000) was acquired from Cayman Chemicals (Ann Arbor, MI, USA). DLin-MC3-DMA (MC3) was purchased from Broadpharm (San Diego, CA, USA). The ionizable lipid DOTAP and 1-Octylnonyl 8-((2-hydroxyethyl) (6-oxo-6-(undecyloxy)hexyl)amino)octanoate (SM-102) were sourced from BOC Sciences (Shirley, NY, USA).

The synthesis of Firefly luciferase mRNA was carried out through in vitro transcription using T7 RNA polymerase and a linearized DNA template. This template comprised the luciferase coding sequence, flanked by 5′- and 3′-UTRs, and included a 100A polyA tail at the 3′-end [25]. The transcription reaction mixture contained ATP, GTP, CTP, N1-methylpseudouridine, and co-transcriptional CAP AG. Following transcription, the mRNA was purified using affinity chromatography with a POROS Oligo(dT)25 column (ThermoFisher Scientific, Waltham, MA, USA). The concentration of the purified mRNA was measured by absorbance at 260 nm, adjusted to 1 mg/mL, aliquoted, and stored at −80 °C. To ensure the quality of the sample, automated electrophoresis was performed using an Agilent 2100 Bioanalyzer G2938B (Santa Clara, CA, USA) to confirm the integrity of the mRNA prior to storage.

Ionizable lipids were prepared through a tandem one-pot reaction. The process began by dissolving thiolactone-sulfur derivatives (0.15 mmol; 1 equiv.) and acrylate (0.15 mmol; 1 equiv.) in 300 µL of tetrahydrofuran (THF) at room temperature, followed by the addition of amine (0.15 mmol; 1 equiv.). After two hours, THF was evaporated under vacuum, and the crude products obtained were directly used for ex vivo screening.

The synthesized lipids were subjected to analysis using high-performance liquid chromatography (HPLC) with a charged aerosol detector (CAD) and further characterized by mass spectrometry. Proton nuclear magnetic resonance (1H NMR) spectra were acquired using a Bruker 400 MHz NMR spectrometer (Billerica, MA, USA). Additionally, HPLC-CAD-MS analysis was conducted with a Vanquish HPLC system equipped with a Vanquish Charged Aerosol Detector and an ISQ EC Single Quadrupole Mass Spectrometer (ThermoFisher Scientific).

CP-LC-1495 was purified using a CombiFlash NextGen 300+ system with gradient elution, transitioning from 100% dichloromethane to 30% of an 80/20/1 DCM/MeOH/NH4OH (aq.) solution. Structural confirmation was performed via mass spectrometry (MS) and nuclear magnetic resonance spectroscopy (NMR). MS-QDa: theoretical [M + H]⁺ = 960.62, experimental [M + H]⁺ = 960.56; ¹H NMR (400 MHz, CDCl3) δ: 6.64 (t, J = 4.9 Hz, 1H), 6.41 (m, 1H), 4.56 (q, J = 7.3 Hz, 1H), 4.08 (m, 4H), 4.00 (m, 2H), 3.30–3.15 (m, 5H), 2.76 (m, 7H), 2.67 (t, J = 6.8 Hz, 2H), 2.57 (m, 10H), 2.23 (t, J = 7.6 Hz, 2H), 2.08 (m, 4H), 1.93 (m, 1H), 1.77 (m, 2H), 1.59 (dt, 8H), 1.44 (m, 2H), 1.30 (m, 28H), 0.88 (m, 12H).

LNPs were prepared via microfluidics using an ethanol-based lipid solution containing ionizable lipid, DOPE (helper lipid), cholesterol, and DMG-PEG2000 (50:10:38.5:1.5 molar ratio) [26]. This was combined with an aqueous phase of mRNA in 10 mM citrate buffer (pH 4) at a molar N/P ratio of 6. Mixing was performed using a NanoAssemblr^® Ignite^TM device (Precision NanoSystems, Vancouver, BC, Canada) with a 3:1 flow-rate ratio (FRR) and 12 mL/min total flow rate (TFR). Post-formulation, LNPs were dialyzed overnight against pH 8 Tris buffer with 15% sucrose. The final product was adjusted to 100 µg/mL mRNA, filtered (0.22 µm), and stored at 4 °C.

Five-component LNPs (cationic DOTAP or anionic 18:1 PA variants) were synthesized via microfluidics. Ethanol-based lipid phases included charged lipid, ionizable lipid, DOPE, cholesterol, and DMG-PEG2000 at molar ratios of 50:11.9:11.9:23.8:2.4 for DOTAP LNPs and 10:21.4:21.4:42.9:4.3 for 18:1 PA LNPs [16]. These were mixed with mRNA in citrate buffer (pH 4) at an N/P ratio of 10 (w/w) using the NanoAssemblr^® Ignite^TM (FRR 3:1, TFR 12 mL/min). Dialysis, filtration, and storage conditions were the same used in the Section 2.5.

Hydrodynamic size, zeta potential, and polydispersity index (PDI) were measured using a Malvern Zetasizer^® Advance Lab Blue Label (Malvern, UK). mRNA encapsulation efficiency was determined via Quant-IT^® Ribogreen assay. Full characterization data are provided in the Supplementary Materials.

All animal experiments were performed in compliance with European and national regulations for the protection of experimental animals (Directive 2010/63/EU and Spanish Royal Decree 53/2013). The experimental protocols were reviewed and approved by the Ethics Committee for Animal Experiments at the University of Zaragoza (PI07/23). The study adhered to the principles of the 3Rs (Replacement, Reduction, and Refinement) to minimize both animal suffering and the number of animals used.

Female BALB/cAnNRj mice, aged 8 to 10 weeks, were purchased from Janvier Labs. Upon arrival at the Centro de Investigaciones Biomédicas de Aragón (Zaragoza, Spain; reference ES 50 297 0012 01), the mice were housed under specific-pathogen-free (SPF) conditions and allowed a one-week acclimation period. The housing environment was carefully controlled, maintaining a temperature of 20–24 °C, humidity levels between 50–70%, light intensity at 60 lux, and a 12-h light–dark cycle. Mice had access to food and water ad libitum.

Sample sizes for each study were determined through statistical power calculations to ensure reliable and reproducible results while minimizing the number of animals used. Before starting the experiments, all animals were monitored for signs of illness, injury, or abnormal behavior that could disqualify them from participation; no animals met these exclusion criteria. To reduce bias, animals were randomly assigned to experimental groups.

Animals were injected via the tail vein with 0.5 mg/kg of luciferase-encoding mRNA-LNPs, which were diluted in Tris buffer containing 15% sucrose to a final volume of 250 µL and administered using a 27G needle.

Four hours after the injection, mice were anesthetized in an inhalation chamber with 4% isoflurane (IsoVet), and anesthesia was maintained at 1.5% isoflurane throughout the procedure. D-luciferin (12507, Quimigen, Alverca do Ribatejo, Portugal), diluted in PBS, was then administered intraperitoneally at a dose of 150 mg/kg. Twenty minutes after the luciferin injection, the mice were euthanized, and their liver, spleen, lungs, heart, kidneys, and intestine were aseptically harvested. Ex vivo luminescence images of the collected organs were acquired using the IVIS Lumina XRMS Imaging System (Revvity Inc., Waltham, MA, USA) following the manufacturer’s instructions.

LNPs were first diluted to a final lipid concentration of 1 g/L in 1× PBS before being mixed with human plasma at a 1:1 volume ratio. The mixture was incubated for 1 h at 37 °C. Plasma samples were provided by the Biobank of Aragon under standard operating procedures approved by the Ethical and Scientific Review Boards within the framework of the Aragon Health Sciences Institute.

After incubation, the LNP/plasma mixture was layered onto a 0.7 M sucrose cushion in equal volume and centrifuged at 15,000× g at 4 °C for 1 h. This process was repeated twice more for a total of three washes. The resulting pellet was resuspended in TRIS-HCl (150 mM pH 8), and excess lipids were removed through TCA/acetone precipitation.

The cleaned pellet was then resuspended in 2× Laemmli buffer and loaded onto a 12% Mini-PROTEAN TGX Precast Protein Gel. The gel was run at 90 V until the proteins migrated approximately 1 cm into the gel. To fix and visualize protein bands, the gel was stained with SimplyBlue Safe Stain for 1 h. Protein bands were excised using a sterile razor blade, placed in ddH2O, and stored at 4 °C until further analysis by mass spectrometry.

The digestion of the polyacrylamide bands was performed in an automatic digester (Intavis, Bioanalytical Instruments, Cologne, Germany). Bands were washed with water, ammonium bicarbonate (100 mM), and acetonitrile (ACN). The samples were reduced by incubation with DTT (10 mM) at 60 °C for 45 min and alkylated by incubation with iodoacetamide (50 mM) at room temperature for 30 min. Digestion was carried out with trypsin overnight at 37 °C (300 ng, Trypsin Gold, Promega, WI, USA). Digestion was stopped by adding 0.5% trifluoroacetic acid and the peptides were extracted sequentially with increasing concentrations of ACN in water.

Proteins were identified in a hybrid trapped ion-mobility quadrupole time-of-flight mass spectrometer (TIMS TOF Flex, Bruker Daltonics, Bremen, Germany) coupled online to an EvoSep ONE liquid chromatograph (EvoSep, Odense C, Denmark). Peptides (200 ng) were loaded onto the EvoSep ONE chromatograph using the 30 SPD (samples per day) chromatographic method.

Peptides were separated on a C18 column (15 cm × 150 μm, 1.5 μm, Evosep) using a linear 44-min gradient and a cycle time of 48 min at a constant flow rate of 0.5 uL/min. Column temperature was set at 40 °C. Data were acquired using data-dependent acquisition mode with PASEF (parallel accumulation serial fragmentation). MS data were collected over an m/z range of 100 to 1700 and a mobility range of 0.60–1.60 V s/cm². During each MS/MS data collection, each cycle was 1.1 s and included 1 MS and 10 PASEF MS/MS scans.

Protein identifications were carried out with Bruker ProteoScape^TM (BPS) software (version 2025b) and were searched against the Uniprot human protein database plus sequences of known contaminants concatenated to a decoy database in which the sequence for each entry in the original database was reversed using ProLuCID algorithm (version 2015). Twenty ppm precursor tolerance and 30 ppm fragment ion tolerance were used. Two missed cleavages were accepted. Carbamidomethylation (+57.02146) of cysteine was considered a static modification. TIMScore was enabled to allow the use of the peptide Collisional Cross Section (CCS) during the scoring process. These search results were validated, assembled and filtered using the DTASelect program (version 1.2.0) with a false discovery rate of 1% at the protein level in all analysis.

Mass-spectrometry-identified proteins were initially filtered to exclude non-human contaminating proteins. The physicochemical properties of the remaining proteins in each sample were computed using the ProtParam module of Biopython [27]. For protein abundance quantification, the data were first filtered to remove proteins with low coverage, and abundance percentages were then calculated using the Normalized Spectral Abundance Factor (NSAF) method.

To identify proteins relevant for lung or spleen targeting, we applied quantitative selection criteria based on their relative abundance across different organ experiments. First, protein concentrations were normalized so that the total abundance across all experiments summed to 100%. Then, selection criteria were established as follows: For lung-enriched proteins: (i) The protein must have a normalized NSAF abundance greater than 0.0015 in at least one lung-targeting LNP sample. (ii) The cumulative normalized abundance across the three lung experiments must exceed 75. (iii) The protein must exhibit a minimum normalized abundance of 20 in each of the three lung-targeting LNPs. (iv) The protein must not exceed a normalized abundance of 20 in any non-lung targeting LNP. For spleen-enriched proteins, the criteria were the following: (i) The protein must have a minimum normalized NSAF value of 0.00075 in at least one spleen-targeting LNP. (ii) The total normalized abundance across the spleen-targeting LNP samples must be at least 50%. (iii) The protein must exhibit a minimum normalized abundance of 20 in each spleen-targeting LNP. (iv) The protein must not exceed a normalized abundance of 30 in any non-spleen targeting LNP.

One of the key challenges in mRNA therapeutics is targeting LNPs beyond the liver, as conventional formulations are prone to being expressed in the liver because of their interaction with the protein apolipoprotein E (ApoE) [28]. In our previous study, we conducted a large-scale screening of 1500 ionizable lipids using our STAAR platform (Figure 1A), where each lipid was formulated into a standard LNP composition (50:38.5:10:1.5 of ionizable lipid:cholesterol:DOPE:DMG-PEG; molar N/P ratio of 6) [17,26] encapsulating luciferase mRNA and administered intramuscularly in mice (0.05 mg/kg of mRNA). This screening provided us with a robust dataset of ionizable lipid performance in the muscle, which we anticipated would likely lead to strong expression in the liver.

However, given the known influence of LNP surface charge on protein corona formation [12,29], we hypothesized that some of our ionizable lipids forming LNPs with zeta potential values outside the typical range for liver targeting might express preferentially in other organs. To test this hypothesis, we retrospectively re-evaluated these lipids, selecting the screened candidates that yielded LNPs with physicochemical properties deviating from the liver-targeting norms. We specifically focused on the zeta potential as this reflects the overall surface charge, which can influence interactions with serum proteins, the biodistribution, and passive organ targeting.

Thus, for this selection, we applied two criteria. First, we focused on LNPs with a zeta potential equal to or greater than −2 mV, inspired by previous findings that suggested that an increased surface charge could enhance lung targeting [10,12]. Second, we included LNPs that demonstrated strong protein expression, specifically those exceeding a total flux of 1 × 10⁷ p/s in the original intramuscular screening, ensuring that only high-performing candidates were considered (chemical structures can be found in Figure S1). Additionally, we incorporated thirteen lipids that exhibited exceptionally high protein expression despite having more negative zeta potentials, allowing us to explore potential outliers with unique targeting properties (Figure 1B).

After intravenous administration of LNPs formulated with the selected ionizable lipids and encapsulating luciferase mRNA, ex vivo analyses revealed that, while most of the selected ionizable lipids still showed liver tropism, eight of them demonstrated lung-selective expression exceeding 10% selectivity (Figure 1C,D). Interestingly, six of these eight lipids shared a distinct structural feature: the presence of three beta-propionate linkers and N,N-dimethylethylenediamine (A1) as the polar head (Figure 1E). Recognizing the potential of this family of lipids as new candidates for pulmonary mRNA therapeutics, we decided to conduct further investigations into their physicochemical and structural properties to better understand their lung-selective behavior.

Having identified a common backbone in the structure of our lipids, we then performed structure–activity studies to explore how chemical modifications in specific moieties of these lipids could affect their extrahepatic delivery performance, particularly in their targeting to the lungs.

Hydrophobic tail regions are known to play a pivotal role in determining the LNP biodistribution, impacting factors such as lipid packing, LNP stability, and their cellular uptake in target tissues [6,30,31]. Thus, in this study, we modified the tail structure of selected lipids by introducing varying lengths, degrees of saturation, and branching patterns, assessing how these changes affected extrahepatic targeting, specifically their lung and spleen selectivity (Figure 2A,B).

Following our previously reported screening approach [17], which relied on a one-pot reaction of a thiolactone-sulfur derivative (Ty) with an amine (Ax) and an acrylate (Cz) (Figure 2A), we synthesized a selection of 44 lipids (Table S1) with purities of over 80%. We included a comparison between the crude and purified products in HPLC-CAD (Figure S4) by combining four different thiolactone-sulfur derivatives (T1–T4, bearing linear and branched aliphatic chains) with eleven different acrylates (C1–C11, which contained linear, unsaturated, or branched tails). Along with two biodegradable amines (derived from Ax and Ty), the resulting final lipid structure contained at least two beta-propionate linkages (derived from Ty and Cz) which are known to have superior biodegradability [21] in mildly acidic environments, potentially improving endosomal escape [32]. For this initial assessment, the amine selected was the same as the one identified in the previous section, namely N,N-dimethylethylenediamine (A1, Figure 2B). The crude synthesized lipids were then formulated into LNPs and injected intravenously into mice. The resulting LNPs showed encapsulating efficiencies exceeding 90%, hydrodynamic diameters ranging from 70 to 110 nm, and a narrow size distribution (PDI < 0.3). No clear trends were observed in the LNP encapsulation efficiencies, sizes, or PDIs. However, significant variations in the LNP zeta potential were evident across different ionizable lipid structures, displaying positive, negative, and near-neutral values (Table S2).

To evaluate the organ selectivity ex vivo, we defined organ-specific hit rates based on a 50% protein expression threshold. Consequently, an LNP was deemed organ-selective if at least 50% of its total expression occurred in a single organ (Figure 2D,E). Notably, 36.6% of the lipids demonstrated selectivity for lung tissue, highlighting that a significant proportion of the tested compounds effectively targeted pulmonary cells. Remarkably, our screening revealed that as few as 7.3% of the ionizable lipids tested demonstrated selectivity for hepatic tissue. This key finding establishes a foundation for developing novel strategies for extrahepatic delivery, particularly considering the liver’s predominant role in mRNA-LNP biodistribution (Table S3).

Closer examination of the chemical structures of the tested tails revealed that T1 derivatives barely achieved lung selectivity (only one ionizable lipid in that group; 2.4% hit rate), suggesting that a minimum level of branching is required to enhance targeting to lung tissue (Figure 2E). This lack of selectivity in T1 may also stem from the removal of one of the biodegradable beta-propionate linkages in its final structure, which could impact both its stability and biodistribution. In contrast, the branched thiolactone-sulfur derivatives (T2, T3, T4) were designed to contain eight carbons post-ester linkage, making them structural isomers and allowing us to assess how tail structures influence extrahepatic delivery. Among these, T2 (linear n-octyl chains) exhibited the highest proportion of lung-selective ionizable lipids (14.6% hit rate), followed by T3 (12.2% hit rate), with T4 showing the least lung selectivity of the three groups (7.3% hit rate) (Figure 2E). This suggests a structure–activity relationship in which the linear n-octyl chains of T2 offers an advantageous configuration for lung targeting, while the further branching in T3 and T4 appears less effective for this purpose.

Interestingly, during the ex vivo experiments, in addition to analyzing lung tissue, we also evaluated other organs, including the liver, kidney, intestine, and spleen. Notably, 14.6% of the evaluated lipids demonstrated selectivity for the spleen (Figure 2D). Further analysis of the chemical structures revealed that increased branching within the T3 and T4 regions correlated with a shift in selectivity toward splenic tissue. This data suggests that, while the less-branched chains in the thiolactone-sulfur derivatives favor lung targeting, increased branching shifts the selectivity toward spleen. This shift in organ targeting observed with increasing lipid branching likely stems from changes in the lipid nanoparticle (LNP) physicochemical properties and their subsequent influence on protein corona formation. Less-branched chains, which have a more linear and compact structure, may promote a corona enriched with lung-specific proteins, enhancing pulmonary targeting. In contrast, branching introduces steric bulk and increases the hydrophobicity of the lipid tails, leading to ionizable lipids with a more conical shape and LNPs with reduced surface-charge density and a different packing efficiency [30]. These branched lipids likely recruit a spleen-favoring protein corona, driving uptake by spleen cells, a trend supported by findings on corona-mediated biodistribution [13]. This interplay between branching, LNP structure, and corona composition offers a potential pathway for designing organ-targeted delivery systems.

A similar pattern of branching impacting organ selectivity was observed in the aliphatic chains of the acrylate component (Cz). Although this moiety appears to have a smaller influence on selectivity compared with the thiolactone-sulfur derivatives, five out of six lipids exhibiting high spleen selectivity (>50% total protein expression) contained branched acrylates (Figure 2D). This finding reinforces the notion that increased branching within lipid structures tends to enhance spleen targeting, highlighting the cumulative effect of branched components in directing organ-specific delivery.

Apart from high percentages for organ selectivity, most of the tested lipids achieved protein expression levels exceeding 1 × 10⁶ p/s in the lungs, underscoring their potential for targeted pulmonary delivery (Figure 2F). Notably, lipid A1T2C3 (also known as CP-LC-1067) stood out with a high protein expression level, reaching 5.32 × 10⁷ p/s. This dual success in selectivity and expression positions CP-LC-1067 as a promising candidate for lung mRNA delivery.

Recognizing the potential of this family of lipids, we sought to further explore structural variations within the ionizable lipid family by modifying its polar-head group while keeping the T2 component constant, aiming for a more refined lung- or spleen-targeting efficiency. For this study, we selected two polar heads. The first one incorporated an imidazole substituent as the ionizable moiety, which was chosen based on previous reports indicating its potential for spleen targeting [33,34]. The second polar head included an azetidine residue, which features an ionizable amine within a cyclic structure. This latter polar head was selected as it represents the closest cyclic analog to amine A1 (Figure 3A).

Applying the same previously established one-pot synthetic method (Figure 2A), 22 new ionizable lipids were synthesized, formulated into LNPs, and tested intravenously in mice by ex vivo luminescence analysis (Figure 3A). Employing the same hit-ratio rule as before (over 50% selective expression in a single organ), a remarkable 95% of the LNPs were found to have extrahepatic selectivity, with only one (A3T2C10) out of the 22 LNPs predominantly expressed in the liver (Table S4; Figure 3B,C). Strikingly, amongst the 95% of extrahepatic-targeted LNPs, there was a strong distinction between both polar heads, as most of the imidazole-based lipids (A2 polar head) showed targeting to the spleen (Figure 3B), a result that aligns with previously reported findings [33], whereas the azetidine lipids (A3 polar head) showed a clear affinity for the lungs (Figure 3C). Zeta potential analysis of the LNPs revealed a distinct pattern that correlated with each polar-head group (Figure 3D). Notably, a clear trend was observed based on the zeta potential, with LNPs exhibiting selectivity for splenic tissue demonstrating a negative zeta potential (range −19 to −9.7 mV; Figure 3D and Table S2), whereas those preferentially targeting pulmonary tissue displayed a slightly positive surface charge (range 0.95 to 3.5 mV; Figure 3D and Table S2). This observed correlation between the zeta potential and organ selectivity further supports the role of surface charge in directing LNP biodistribution, reinforcing the potential of rational lipid design for extrahepatic delivery. This distinction in surface-charge characteristics likely influences the interactions between LNPs and biological components, ultimately affecting their biodistribution and organ targeting.

Beyond selectivity, expression levels are equally critical, as low expression may compromise the therapeutic efficacy of LNPs, making high selectivity alone insufficient for effective organ-targeted delivery. Notably, ionizable lipids with A2 and A3 not only exhibited strong organ specificity but also achieved high luciferase expression levels, with the selective organ total flux surpassing 1 × 10⁷ p/s for nine of the ionizable lipids (Figure 3D). Particularly, lipid A3T2C7 (also known as CP-LC-1495) not only achieved a high level of lung selectivity with a remarkable 97.1% but also showed a strikingly high luminescence signal of 1.21 × 10⁸ p/s, highlighting its potential for targeted lung delivery therapeutics. Moreover, several ionizable lipids exhibited spleen-tissue targeting with high selectivity (>80%) and high efficiency (approximately 1 × 10⁷ p/s; Figure 3D). Additionally, cytotoxicity experiments in relevant lung (A549), spleen (Jurkat), and liver (HepG2) cell lines resulted in cell viabilities above 90% for A549 along with HepG2 and above 75% for Jurkat cells (Figure S3), even when using LNP-mRNA concentrations four times higher than those reported in previous studies [23].

The striking selectivity and efficiency of spleen- and lung-targeting ionizable lipids, coupled with their excellent cytotoxicity profiles, prompted us to further investigate the protein corona’s role in driving organ-specific delivery to the lungs and spleen of the top-performing ionizable lipids.

The composition of the protein corona that forms around LNPs upon exposure to biological fluids is a key factor in their organ selectivity and biodistribution [28,35]. Given this well-established relationship, we sought to investigate whether the differences in our extrahepatic lipid library were reflected in distinct protein corona compositions, which could potentially explain their selective accumulation in non-hepatic tissues. With that purpose, we selected and purified the two top-performing ionizable lipids for lung targeting (A1T2C3, also named CP-LC-1067, and A3T2C7, also named CP-LC-1495) and the highest-performing lipid for spleen targeting (A2T2C9, also named CP-LC-1465). As controls, we included MC3-DOTAP and MC3-18:1 PA, well-established SORT formulations known for targeting the lung and spleen, respectively, with previously characterized protein coronas [12]. Additionally, we incorporated liver-targeting LNPs using SM-102 as the ionizable lipid and a control without LNPs. Subsequently, to mimic the in vivo environment, we incubated LNPs with human plasma, allowing the natural formation of a protein corona similar to that occurring in the bloodstream (Figure 4A). After purification and LNP protein-corona isolation, we performed mass-spectrometry-based proteomic analysis to identify and quantify the adsorbed proteins.

Examination of key physicochemical properties of the enriched proteins in different selected LNPs samples, including of the isoelectric point, hydropathy (GRAVY index), secondary structure, charge, and stability, showed that lung-targeting LNPs exhibited a notable increase in charge and hydropathy, while spleen- and liver-targeting LNPs displayed distinct trends in protein stability and secondary structure (Figure 4B). Since zeta potential was identified as a key factor in organ targeting by our ionizable lipid library, we further analyzed its influence on related protein properties such as isoelectric point (pI) and overall charge. We found that in lung-targeting LNPs, the isoelectric point distribution of adsorbed proteins was slightly shifted toward pI ~5.5 compared with liver- and spleen-targeting LNPs (Figure 4C). This may be attributed to their higher zeta potential, which likely influences protein binding. Notably, vitronectin [12], a key mediator of lung uptake, has a pI of ~5.5, suggesting a potential explanation for this enrichment. In contrast, the protein-charge distribution was less affected by lipid composition (Figure 4D). This may be due to the LNP charge being better described as charge density rather than absolute charge. While surface charge impacts protein adsorption, it operates within a complex electrostatic environment, which could explain the subtler differences in protein-charge profiles compared with pI trends. These findings suggest that lipid composition influences not only protein adsorption but also the physicochemical characteristics of the corona, impacting organ-specific delivery.

To further investigate protein-corona-driven targeting, we analyzed the top 30 most abundant proteins in lung- and spleen-targeting LNP coronas (Tables S5–S10). Additionally, to provide a clearer interpretation of the results, we focused on proteins that were highly enriched in one organ-specific LNP corona while being absent or barely detected in the other (Figure 4E,F). Results showed that the predominant proteins absorbed in spleen-targeting LNPs are actin, myosin, and the immunoglobulin lambda-like protein. These proteins exhibit a low correlation and are typically nonspecific and are often found in negative mass spectrometry controls. Therefore, the mechanism underlying spleen targeting remains unclear. However, the protein-corona composition of lung-targeting LNPs was clearly distinct from that of liver- and spleen-targeting LNPs. Notably, vitronectin and prothrombin, both extensively reported as key protein-corona components influencing lung targeting, were among the most enriched proteins [12,13,27]. These findings suggest that lung-targeting LNPs in our ionizable lipid library follow previously described mechanisms of proteincorona-mediated lung selectivity.

Data is contained within the article or. Supplementary Materials