High-Throughput DNA Barcoding Identifies Lipid Nanoparticles for Lung Delivery of TRAIL-mRNA to Induce Tumor Cell Death in Experimental Metastasis Model

Abstract

Introduction

Lipid nanoparticles (LNPs) have enabled the clinical translation of nucleic acid therapeutics, and their composition plays a critical role in organ targeting and route-specific delivery. Intranasal (IN) administration offers direct access to the lungs, while reducing systemic exposure. However, designing LNPs capable of overcoming respiratory tract barriers remains a challenge.

Methods

We used DNA barcoding (b-DNA) for high-throughput in vivo screening of an LNP library following IN administration to investigate the impact of helper lipid identity and changes in the relative helper lipid/cholesterol composition on lung delivery. Following physicochemical characterization, pooled LNPs were administered intranasally to C57BL/6 mice, and their biodistribution was quantified by next-generation sequencing. Lead formulations were further evaluated for lung transfection in vivo and pulmonary cell transfection in vitro, cell viability, concentration-response relationships, and membrane-disruptive properties were also assessed. To better understand the discrepancies between the in vivo and in vitro performance, LNP stability was investigated in simulated nasal fluid and cell culture media.

Results

DOPE-based LNPs exhibited uniform diameters of approximately 90–110 nm, whereas DSPC- and DOTAP-based formulations showed greater variability in size. While DOPE- and DSPC-LNPs displayed slightly negative zeta potential values, DOTAP-formulations exhibited positive values. DSPC-LNP6 has emerged as the lead formulation for lung delivery and transfection in vivo, despite exhibiting limited transfection efficiency in pulmonary cells in vitro. Notably, the impact of simulated nasal fluid on the physicochemical properties of LNPs was comparable across formulations, suggesting that intrinsic membrane properties contribute to enhanced pulmonary transfection after IN administration. Finally, DSPC-LNP6 successfully mediated TRAIL mRNA delivery and induced tumor cell death in an experimental lung metastasis model.

Conclusion

Collectively, our findings provide critical insights into the development of RNA therapeutics that target the respiratory tract.

Introduction

Intranasal (IN) administration of nanoparticle-based systems is a strategy to target the lungs with reduced systemic exposure.¹ The IN route provides direct access to the lungs, bypassing traditional limitations associated with intravenous (IV) administration of nanomedicines, such as rapid clearance by the reticuloendothelial system.² However, various challenges imposed by this route arise from the biological barriers of the respiratory tract.

Several respiratory diseases can benefit from the advantages of IN administration. Pulmonary fibrosis is a major public health problem with increasing concern due to rising air pollution.³ Lung cancer remains another critical challenge. The lungs also serve as a common metastatic niche for multiple cancers, including breast, prostate, and colorectal cancers, which are among the most prevalent types worldwide. Notably, up to 50% of patients with metastatic cancer develop lung metastasis, a condition associated with poor prognosis.⁴

The successful clinical translation of lipid nanoparticle (LNP) formulations for genetic diseases and vaccines has established these systems as effective nucleic acid delivery platforms for IV and intramuscular administration,^5,6 whereas the design principles governing their performance through IN administration remain to be further explored. This delivery technology enhances RNA stability in the biological environment while protecting it against nuclease degradation and promoting efficient intracellular delivery via endosomal escape after cellular uptake.⁷ However, achieving optimal therapeutic outcomes requires the use of efficient, minimally immunogenic RNA and optimized lipid composition based on its biological application.⁸

The first barrier for LNP-mediated RNA delivery to the lungs through IN administration is mucociliary clearance (MCC). Ciliary movement actively removes particles from the airways, whereas mucus, a hydrogel composed of water, proteins, salts, and lipids, protects the epithelium and maintains hydration. Together, mucus and MCC create a physical and chemical barrier that must be overcome by LNPs to achieve effective pulmonary delivery.⁹

Strategies to enhance nanoparticle diffusion through this barrier include surface modification with poly-ethylene-glycol (PEG), which improves mucus penetration.¹⁰ Another key interaction of LNPs within the respiratory tract is the formation of a lung protein-surfactant corona, analogous to the plasma protein corona observed following IV administration, which can influence LNP fate.^11,12 Recent studies indicate that different helper lipids in LNP formulations can modulate protein corona formation and thereby affect biodistribution after systemic delivery.^13,14 However, the role of helper lipid composition in determining lung targeting and transfection efficiency following IN administration remains poorly understood.

Nucleic acid barcodes represent a powerful strategy to evaluate the biodistribution of multiple LNP formulations simultaneously.^15–18 This approach requires fewer animals and avoids the common limitation of fluorescent labeling, which may not accurately reflect true LNP biodistribution. To date, the use of nucleic acid barcodes to monitor LNP biodistribution following IN administration has not been reported.

Herein, we prepared a library of 18 different LNPs by microfluidic mixing and varying helper lipid identity and molar ratios to assess their impact on lung targeting following IN administration (Figure 1A). Although helper lipids are known to modulate LNP physicochemical properties, protein corona formation, and biodistribution after systemic administration, their role in determining lung delivery and functional transfection following intranasal administration remains poorly understood. To address this knowledge gap, DOPE, DSPC, and DOTAP were selected as helper lipids with distinct physicochemical and membrane properties, including fusogenic and non-bilayer-forming behavior, structural stability, and permanent cationic charge, respectively.^13,14 Next, we validated the top-performing LNP for each helper lipid in vivo to determine whether the barcode-based results translated into enhanced lung transfection. To further support these findings, the in vitro transfection efficiency in pulmonary cells was assessed using nucleic acid and particle concentration-response curves (Figure 1B). Physicochemical characterization of LNPs in complex dispersants, including cell culture medium and simulated nasal fluid (SNF), was performed to understand formulation behavior and performance (Figure 1C). Finally, the top-performing LNP were selected for functional evaluation in a lung metastasis model to assess their therapeutic potential (Figure 1D).

Figure 1.

Schematic workflow for the screening, optimization, and application of TRAIL-LNP in a lung metastasis model. (A) High-throughput screening with b-DNA LNPs administered via the intranasal route. (B) In vitro and in vivo transfection validation of selected LNPs based on helper lipid identity. (C) Physicochemical characterization of selected LNPs in protein-enriched medium. (D) In vivo biological validation of the top-performing LNP encapsulating TRAIL mRNA in a lung metastasis model.

Materials and Methods

LNP Formulation

LNPs were formulated using microfluidic mixing of an ethanolic phase containing lipids with an acidic aqueous phase containing the nucleic acid of interest at a volume ratio of 1:3 and a 10:1 weight ratio of ionizable lipid to nucleic acid, using a custom-made microfluidic device.^18,19 Briefly, stock solutions of C12-200 (100 mg/mL) (synthesized as reported)²⁰ 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE-Lipoid), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC-Lipoid), or 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP-Lipoid), cholesterol (Avanti Polar Lipids) and 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethyleneglycol)-2000] (ammonium salt) (C14-PEG 2000-Avanti Polar Lipids) (10 mg/mL) were prepared in pure ethanol. Lipids were combined at defined molar ratios (Table S1), and the aqueous phase was prepared in acidic citrate buffer (pH 3 – I = 2.65mM). Next, both phases were placed in individual glass syringes and the mixture was prepared using a syringe pump (Harvard apparatus). LNPs were collected in a dialysis bag (20 KDa) and dialyzed against PBS at room temperature for 2 h. Following dialysis, LNPs were sterile-filtered using 0.2 μm filters.

LNP Characterization

LNPs were characterized by dynamic and electrophoretic light scattering (DLS and ELS, respectively) to assess their size, polydispersity index (PDI), and zeta potential (ZP) using a Zetasizer Nano Zs machine (Malvern). Using appropriate cuvettes and 1:100 (v/v) dilution in water, three independent measurements were performed, unless otherwise noticed.²¹ Encapsulation efficiency (EE) of mRNA-LNPs was determined using the Quant-iT RiboGreen RNA assay (Thermo Scientific) following the manufacturer’s instructions. In summary, LNPs were diluted in Tris-EDTA (TE) buffer to measure accessible/unencapsulated mRNA or TE buffer containing 1% Triton X-100 (v/v) to disrupt the LNPs and measure total mRNA. EE was calculated as follows: EE (%): [(total mRNA − unencapsulated mRNA)/total mRNA] × 100.²² LNPs concentration and stability in PBS and Dulbecco’s Modified Eagle’s medium (DMEM) supplemented with 10% of fetal bovine serum (FBS) were performed using the nanoparticle tracker analysis (NTA) using NanoSight Pro (Malvern). Fresh analyses were performed by diluting LNPs in PBS or DMEM-10% FBS at 1:1000 (v/v), and three independent automated captures were made using flow of 3 μL/min at 37 °C.

For NTA analysis in simulated nasal fluid (SNF) the same acquisition conditions described above were used. The SNF composition was based on previous literature reports.^23,24 SNF is composed of 7.45 mg/mL NaCl; 1.29 mg/mL KCl, and 0.32 mg/mL of CaCl2 adjusted to pH 5.5 with HCl 1M and combined with aqueous mucin in the 5:1 (w/w) ratio. The diffusion coefficient was calculated using Einstein-Stokes equation (D = kB · T / (3π · η · dH)), using NTA mean size of the LNPs, and 0.7 and 0.9 mPa.s of viscosity for the PBS and SNF solutions, respectively. For stability analysis, LNPs were prepared with the same dilution and incubated in a dry bath at 37 °C with shaking at 500 rpm for 24 h, unless otherwise noted. After this period, NTA was performed as described above. Structural and morphological analyses were performed using cryogenic transmission electron microscopy (cryo-TEM). DSPC-LNP6 encapsulating luciferase mRNA was placed in a vitrification system (25 °C, 100% humidity). Vitrified LNP were examined using a Tecnai G2-20 - FEI SuperTwin 200 kV at the Microscopy Center of UFMG.¹⁶

In vitro Experiments

Human lung epithelial cell line A549 (CCL-185) was obtained from ATCC and cultured in DMEM supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin-streptomycin solution (100 U/mL penicillin and 100 μg/mL streptomycin; Gibco).²⁵ For the in vitro transfection efficiency using top-performer LNPs per helper lipid from barcode screening. A549 cells were seeded at a density of 1×10⁴ cells per well in 100 μL using 96-well plates. Two different serial dilutions of the LNPs were prepared in complete cell medium: one for increasing fixed mRNA concentration-response screening (0.0625–1 μg of mRNA/well) and another for increasing fixed nanoparticle concentration-response screening (1 × 10⁶–1×10⁸ of NPs/well). After 6 and 24 h of treatment, the transfection efficiency was evaluated using the ONE-GLO (Promega N1610) luciferase detection kit. The luminescent signal was detected using a Cytation 5 Cell Imaging multimode reader (Biotek). Cell viability was monitored using the Alamar blue (DAL1025 Thermo-fisher) assay following the manufacturer’s instructions, only at 24 h post-LNP treatment. After incubation, fluorescence reads were obtained using a Varioskan.

Hemolysis Assay

The ability of LNPs to disrupt biological membranes was verified using a hemolysis assay.²⁶ Red blood cells (RBCs) were obtained from C57Bl/6 mice and washed twice with PBS by centrifugation at 700 × g for 5 min. The supernatant and clots were removed, and the RBCs were diluted to 4% v/v solution in PBS pH 7.4 or 5.5. The RBCs solution was incubated with the LNPs at a final mRNA concentration of 2 μg/mL for 1h at 37° °C in a final volume of 200 μL. A solution of 1% of Triton X-100 was used for complete RBS lysis and PBS as control. Next, the plate was centrifuged, 100 μL of the supernatant was transferred to another 96-well plate, and the absorbance was read at 540 nm using Varioskan. Hemolysis % was calculated using Triton X-100 lysis as 100%.

DNA Barcoding (b-DNA) Library Preparation

b-DNAs were ordered from Integrated DNA Technologies (IDT) using an architecture based on previous studies.^15,18,27 b-DNAs are single-stranded with 61 nucleotides and a barcode region comprised of 10 nucleotides located at the center of the single strand. The 5′ and 3′ ends contained three phosphorothioate bonds and priming sites for the Illumina adapters. The b-DNA sequences are listed in Table S2.

b-DNA Delivery and LNP Biodistribution

A pooled LNP library containing 18 distinct b-DNA-loaded LNPs was administered to 8-week-old female C57BL/6 mice via the intranasal route at 0.5 μg of b-DNA per LNP.^15,18,27 Mice were sacrificed 4 h post-administration and organs (heart, spleen, lungs, brain, and liver) were collected, snap-frozen in liquid nitrogen, and stored at −80 °C. Prior to DNA extraction, the organs were macerated using autoclaved scissors, and 25 mg of the disrupted tissue was processed for DNA isolation using the PureLink Genomic DNA Mini Kit (Thermo Fisher, Cat. No. K1820-02). Purified DNA was quantified using a NanoDrop OneC UV–Vis spectrophotometer (Thermo Fisher, Cat. No. ND-ONE-W), and its purity was verified using A260/280 ratio and agarose gel electrophoresis.

b-DNA Amplification

For PCR, a stock primer mix was prepared containing 5 µL of MiSeq universal primer (5 µM), 0.5 µL of barcode base primer (0.5 µM), and 94.5 µL of Tris-EDTA (TE) buffer, pH 8.0. PCR reactions were performed using 5 µL of purified DNA (10 ng/µL), 1.0 µL of primer mix, 1.2 µL of sequence primer (5 µM; a unique primer sequence for each organ, as detailed in Table S2), 0.5 µL of 10 mM dNTPs, 1.0 µL of MgCl₂, 0.75 µL of 100% DMSO, and 10.3 µL of nuclease-free water. Thermal cycling conditions were as follows: 98 °C for 48s, followed by 39 cycles at 98 °C for 12s, 67 °C for 22s, and 72 °C for 28s, with a final extension at 72 °C for 5 min. The amplified b-DNA (144 bp) was verified by agarose gel electrophoresis (1.4%) and bands of interest were excised and purified using the PureLink Quick Gel Extraction Kit (Thermo Fisher, Ref. K2100-12).^15,16,18 Purified PCR products were stored at −20 °C until sequencing.

Next Generation Sequencing and Delivery Quantification

Illumina next-generation sequencing (NGS) was used for deep sequencing. A sequencing library was generated by pooling the PCR products using 5 µL of each purified DNA sample at a concentration of 4 nM. The library was then sequenced using the Illumina MiSeq platform at UFMG. FASTQ files were processed using custom Python scripts to quantify barcode abundance in each organ. The b-DNA delivery by LNPs to a specific organ was calculated as the proportion of reads corresponding to each barcode within the total reads obtained for the same organ. This allowed the identification of the top-performing LNP for each organ.

Animal Models and Ethics Statement

All procedures were approved by the Ethics Committee on the Use of Animals of UFMG (CEUA/UFMG – 264/2020, 269 / 2022) and were conducted in accordance with the guidelines from the National Council for the Control of Animal Experimentation (CONCEA) in Brazil. For in vivo imaging procedures, mice were anesthetized with Isoflurane (induction 3% and maintenance 1.5% isoflurane in oxygen). At experimental endpoints, animals anesthetized with isoflurane or with ketamine/xylazine (75 and 8 mg/kg, respectively) administered intraperitoneally were euthanized by cervical dislocation. All animals were housed under the appropriate conditions with ad libitum access to food and water. C57BL/6 mice were provided by Biotério Central at UFMG, while NOD-SCID IL-2Rγnull (NSG) mice were obtained from the Immunopharmacology Laboratory of the Institute of Biological Sciences (ICB)/UFMG.^16,28

In vivo Lung Transfection Efficiency of LNPs After Intranasal Administration

C57BL/6 mice were provided by Biotério Central (UFMG). The animals were housed under a 12 h light/dark cycle with 50–58% humidity and a temperature of 25 °C. For this study, mice received intranasal administration of LNPs containing mRNA at a dose of 0.25 mg/kg. Four hours after administration (the same time point used in the b-DNA screening), animals were injected intraperitoneally with 100 µL of D-luciferin (25 mg/mL). Fifteen minutes later, in vivo bioluminescent images were acquired, followed by organ harvesting and ex vivo imaging using an in vivo imaging system (IVIS, PerkinElmer, USA). Luminescence intensity was quantified using Living Image software (PerkinElmer). The organ contribution to total luminescence was calculated as follows: (organ luminescence)/(total organ luminescence) × 100.^18,29

Liver Function Assay

Blood samples were collected in microtubes to measure serum levels of Alanine Transaminase (ALT), Aspartate Aminotransferase (AST), Gamma-glutamyl Transferase (GGT), and urea. After blood clotting, samples were centrifuged for 20 min at 2000 × g for 15 min at 4 °C. ALT, AST, GGT and Urea levels were determined using Bioclin kits following the manufacturer’s instructions. Absorbance measurements were performed using a Biotek Epoch microplate reader.^18,30

Histopathological Analysis

Histopathological evaluation was performed by an independent pathologist blinded to the experimental conditions. Organs were fixed in 4% paraformaldehyde (PFA) in PBS, dehydrated, and embedded in paraffin. The tissues were sectioned into 5 µm-thick slices and stained with hematoxylin and eosin (H&E) for microphotographic analysis. A Histech 3D Slide Scanner was used for the fast scanning of slides. Tissue morphology was assessed using a semi-quantitative scoring system. Inflammatory infiltrates were characterized by the presence of plasma cells, macrophages, and lymphocytes. Additional parameters, including vascular injury, hemorrhage, fibrosis, necrosis, hydropic degeneration, and inflammatory infiltration, were examined on stained slides. Whole-slide examination was performed for all tissue sections, and alterations were graded as follows: 0 = no injury, 1 = discrete (lesion affecting <25% of the tissue), 2 = moderate (25–50%), 3 = severe (50–75%), and 4 = extensive (lesion affecting >75% of the tissue).^16,18

Animal Model of Experimental Lung Metastasis

We used 7-to 8-week-old transgenic NOD-SCID IL-2Rγnull (NSG) mice of both sexes. Animals were bred at the Immunopharmacology Laboratory of the Institute of Biological Sciences (ICB), Federal University of Minas Gerais (UFMG). They were housed in ventilated cages within an animal biosafety level 2 (BSL-2) facility (ICB, UFMG) and maintained under a 12 h light/12 h dark cycle at 24 ± 2 °C. NSG mice had free access to food and water, and we closely monitored them for clinical signs of illness, including ruffled fur, hunched posture, weight loss, and reduced activity.²⁸

Experimental Lung Metastasis in vivo

Colon cancer cells (Colo 205) expressing both luciferase and green fluorescent protein (GFP) were obtained from GeneCopoeia (Rockville, MD, USA) and cultured in RPMI-1640 medium (Gibco) supplemented with 10% (v/v) fetal bovine serum (Gibco) and 1% (v/v) penicillin-streptomycin solution. The pulmonary metastasis protocol was adapted based on previously reported models.²⁸ Colo 205 cells were centrifuged, washed with PBS, and intravenously injected into NSG mice at a concentration of 2×10⁶ cells in 100 µL of PBS. On the day post injection (DPI) 20, mice were monitored for bioluminescent signals in the thorax region using IVIS with image acquisition covering the area from the abdomen downward. On DPI 26, experimental lung metastasis was monitored again using bioluminescence imaging, and animals were randomly assigned to treatment or control groups. Immediately after group allocation, the mice in the treatment group received an intranasal dose of 5 µg of mRNA-TRAIL formulated in DSPC-LNP6 while the control group received PBS. Subsequent treatments were administered at 48 h intervals on days 28, 30 and 32, using the same dose. IVIS imaging was performed on days 29 and 32 to evaluate potential reductions in lung bioluminescence. After in vivo imaging on day 32, ex vivo imaging and flow cytometry of the lungs were performed.

Flow Cytometry

Lungs were collected in 2 mL micro tubes and gently disrupted using autoclaved scissors for digestion with collagenase I at 3mg/mL for 45 min at 37 °C. The reaction was stopped with 500 μL of FBS, and the cell suspension was filtered through cell strainers, followed by erythrocyte lysis. Total lung cells were plated at a density of 1×10⁶ cells in 100 μL of RPMI supplemented with 10% of FBS and 1% of penicillin-streptomycin solution in rounded 96 well plates. Next, cells were treated with Brefeldin A (ThermoFisher) for 4 h. Debris and cell aggregates were excluded based on the forward scatter area (FSC-A) versus forward scatter height (FSC-H) gating parameters. Cell characterization was performed by gating live and dead cells using a live/dead marker (Invitrogen), as well as for cells expressing GFP. After fixation and permeabilization with the FoxP3 staining buffer set (eBioscience) according to the manufacturer’s instructions, intracellular staining was performed: anti-cleaved Caspase-3 (Asp175) (9661, Cell Signaling) and anti-CD253 (TRAIL) (PE, 12–9927-42, Thermo Fischer). Finally, the secondary antibodies were added (Donkey anti-Rabbit IgG, Alexa Fluor 647). Symphony A5 was used for acquisition.^16,28,31 Analysis were done using FlowJo^TM software and TRAIL+ cells were calculated relative to total singlet population. The frequency of GFP+ dead cells was determined post dead cells identification in singlets, and the frequency of caspase-3+ cells was calculated relative to the GFP+ population.

Statistical Analysis

Statistical evaluation was conducted with GraphPad Prism 10 (GraphPad Software, Boston, MA, USA), and data are reported as the mean ± standard deviation (SD). The normality was evaluated using the Shapiro–Wilk test together with visual assessment of Q–Q plots. For experiments involving one independent factor and three or more groups, differences were analyzed by one-way ANOVA followed by Tukey’s multiple-comparisons test. Experiments involving two independent factors were analyzed by two-way ANOVA, including evaluation of the main effects and their interaction, followed by Tukey’s or Bonferroni-adjusted multiple comparisons, as appropriate. Comparisons between two groups were performed using a two-sided t-test. The number of independent samples and technical replicates in each experimental group (N) is described in the corresponding figure legend.

Results

Characterization and Screening of a Library of DNA-Barcoded-LNPs (b-DNA-LNPs) for IN Administration

To investigate the effects of different helper lipids and their molar ratios within LNPs on IN biodistribution targeting the lungs, we prepared a library of 18 different LNPs, comprising six formulations per helper lipid type, each encapsulating unique barcoded DNA (b-DNA) (Figure 2A). To enable these comparisons, the ionizable lipid and PEG-lipid components were kept constant, whereas the helper lipids (DOPE, DSPC, and DOTAP) and cholesterol content varied across formulations (Figure 2B and Table S1).

Figure 2.

High-throughput screening of LNPs encapsulating b-DNA for identification of lung targeting formulations following intranasal administration. (A) LNPs prepared by microfluidic mixing encapsulating a unique b-DNA were pooled to be administered in C57BL6 mice. At 4 hours post-administration organs were collected and NGS was performed to quantify the delivery of each b-DNA to various organs. (B) LNPs composition exploring variations in helper lipid type and molar ratios within the formulation. (C and D) Physicochemical characterization by DLS and ELS respectively. N=3 independent measurements. (E) Heatmap presenting LNP delivery (bottom) to evaluated organs (left) after LNP pool administration. Within the heatmap comparisons should only be made between the delivery of different LNPs within the same organ. (F) Bar graph representing the lung normalized percentage of b-DNA delivery by LNPs. All Data are presented as mean ± SD (n = 5 animals/group), analyzed by one-way ANOVA followed by Tukey multiple comparison test. ****p< 0.0001.

Physicochemical properties, including hydrodynamic diameter, polydispersity index (PDI), and zeta potential, were analyzed by DLS and ELS. Size measurements revealed diameters ranging from 70 to 175 nm, with PDI values below 0.45 (Figure 2C). While DOPE- and DSPC-based LNPs exhibited negative but near-neutral zeta potential values, DOTAP-LNPs exhibited positive zeta values (Figure 2D). Notably, the LNP pool exhibited a monodisperse size distribution, with a mean size of ~100 nm (Figure S1A). It is important to highlight that the LNP pool was an extemporaneous preparation, in which each LNP formulation was prepared and characterized individually and then combined at the same b-DNA amount for administration into mice.

High-Throughput IN Screening of LNP for Lung Delivery

After characterization of the 18-formulation LNP library, the formulations were pooled and intranasally administered to C57BL/6 mice. A b-DNA dose of 0.5 μg per LNP was selected based on previous literature, which established a concentration-response curve for b-DNA detection in mice using NGS.^15,27 At 4 h post-administration, b-DNA delivery to the lungs and other organs was quantified using NGS (Figure 2E).

Biodistribution analysis showed widespread delivery of DOPE-based LNPs to all evaluated organs except the lungs, in which DSPC-LNP4 and DSPC-LNP 6 exhibited the highest delivery rates, followed by DOTAP-LNP2 (Figure 2F). DSPC-LNP6 exhibited approximately 2.5-fold higher delivery than the top-performing DOPE-based LNP (DOPE-LNP5) and 1.8-fold higher delivery than DOTAP-LNP2 (Figure 2F). DSPC-LNP6 was statistically different from all other LNPs, being the top-performing LNP for lung delivery. A table summarizing the statistical differences in LNP delivery to the lungs is provided in the Supplementary Material (Table S3).

Size and zeta potential are key features of nanoparticle systems and are often correlated with biological performance. Thus, we conducted Spearman correlation analysis to assess whether these characteristics were associated with lung biodistribution (Figure S1B and C). Notably, no strong correlation was found between formulation physicochemical parameters and lung delivery, either when LNPs were clustered by helper lipid type (Figure S1C) or when analyzed without clustering (Figure S1B).

In vivo Validation of Selected LNPs for Lung Transfection Following Intranasal Administration

Based on b-DNA delivery outcomes, we selected one top-performing LNP per helper lipid (DOPE-LNP5, DSPC-LNP6, and DOTAP-LNP2), which exhibited superior b-DNA delivery to the lungs (Figure 2F), to evaluate mRNA transfection in this organ following IN administration (Figure 3A).

Figure 3.

In vivo and In vitro transfection validation of selected LNPs from b-DNA screening. (A) in vivo experimental design. (B) Ex vivo IVIS images. (C) Quantification of luminescent signal from the organs. (D) Contribution of luminescence signal of the imaged organs. Data are presented as % of total luminescence = (organ luminescence)/(total organ luminescence) × 100. Data are presented as mean± SD analyzed by Two-way ANOVA followed by Tukey multiple comparison test. N=5 animals/group. (E) in vitro experimental design. In vitro transfection screening using increasing fixed mRNA concentration between LNPs at 6 (F) and 24 (G) hour post-LNP treatment. (H) Cell viability at 24 h post-LNP treatment. Data are presented as mean± SD analyzed by Two-way ANOVA followed by Tukey multiple comparison test. N=3 biological replicates. *p<0.05, **p<0.005, ****p<0.0001.

Abbreviation: ns, not significant.

To this end, top-performing LNPs were prepared to encapsulate luciferase-encoding mRNA, and their transfection efficiency was assessed 4 h post-treatment, corresponding to the same time point used in the b-DNA screening. DSPC-LNP6 was the only formulation that induced lung transfection under the tested conditions, while all other organs exhibited luminescent signals comparable to PBS, indicating minimal systemic exposure (Figure 3B and C). In vivo imaging showed that DOPE- and DOTAP-based LNPs induced transfection primarily in the upper respiratory tract, whereas DSPC-LNP6 produced additional signals in the lung region (Figure S2A–C). Next, the contribution of each organ to the total luminescence signal was quantified. For DSPC-LNP6, the lung signal accounted for approximately 70% of the total signal, with only minor contributions from other tissues (Figure 3D).

Finally, we evaluated the treatment-related toxicity of the LNPs following IN administration. Histopathological analysis revealed no major alterations in the pulmonary tissue architecture and no evidence of inflammatory infiltrates after treatment (Figure S3). Despite limited systemic exposure, we also evaluated serum biomarkers, including AST, ALT, GGT, and urea levels (Figure S4A). None of the LNPs altered the hepatic enzyme activity. However, a reduction in the serum urea levels was observed for DOTAP-LNP2 (Figure S4A). Histological examination of the liver revealed no morphological alterations or signs of inflammation with any of the LNP formulations (Figure S4B).

In vitro Transfection Efficiency in Alveolar Type II Pneumocytes

Beyond their respiratory functions, the lungs contain diverse cell populations that influence nanoparticle interactions. In addition to pneumocytes, they constitute an important immunological niche that is enriched with a variety of specialized immune cells. To assess the transfection efficiency of the selected LNPs, we used A549 cells, a human lung adenocarcinoma-derived epithelial cell line commonly used as an alveolar epithelial model because of its type II-like characteristics.^32,33 Two complementary approaches were employed: (1) a concentration–response curve comparing LNP efficacy at increasing fixed concentrations of luciferase-encoding mRNA across formulations and (2) concentration-response curves comparing LNP efficacy at increasing fixed particle concentrations across LNPs (Figure 3E and S5).

Using approach 1, A549 cells were treated with mRNA concentrations ranging from 0.0625 to 1 µg/well, and transfection was assessed 6 and 24 h post-treatment (Figure 3F and G). At 6 h, DOPE-LNP5 induced transfection at all tested concentrations, whereas DOTAP-LNP2 induced transfection starting at 0.125 µg/well. The DSPC-based formulation also produced measurable transfection at the highest concentration tested (Figure 3F). Overall, DOPE-LNP5 demonstrated superior transfection efficiency compared to the other formulations, regardless of concentration or time point.

At 24 h post-treatment, DSPC-LNP6 induced transfection from 0.25 µg/well of mRNA, reaching maximal expression at 1 µg/well (Figure 3G). Temporal transfection profiles differed among formulations. For DOPE-LNP5, the lowest mRNA concentration resulted in relatively low luciferase expression at 6 h; however, it resulted in the highest expression at 24 h among all the concentrations tested. DOTAP-based LNPs exhibited temporal behavior similar to that of the DOPE formulation, while DSPC-LNP6 exhibited a distinct profile. Luciferase expression increased proportionally with both mRNA concentration and time (Figure 3F and G). Cell viability was also monitored at 24 h post-treatment, revealing that only DOTAP-LNP2 exhibited toxicity, which became evident at mRNA concentrations of 0.5 µg/well and above (Figure 3H).

Using approach 2, we performed a similar assay, in which the concentration-response curve ranged from 1×10⁶ to 1×10⁸ NPs/well, corresponding to a range similar to that used in the fixed mRNA screening. Notably, at a fixed particle concentration, the amount of mRNA delivered varies between LNP formulations due to multiple factors, such as final LNP volume, mRNA recovery after mixing, and EE. No significant differences in particle concentration were found between DOPE- and DOTAP-based LNPs (Figure S5A and B), indicating that the corresponding mRNA doses were comparable across these formulations.

Consistent with approach 1, DOPE-LNP5 outperformed all other formulations at both concentrations and time points (Figure S5C and D). At 24 h post-treatment and at a particle concentration of 1×10⁸ NPs/well, DSPC-LNP6 showed higher transfection efficiency than DOTAP-LNP2. This likely reflects the higher mRNA dose delivered by DSPC-LNP6 (0.5 µg/well) compared with DOTAP-LNP2 (0.3 µg/well), corresponding to an ~1.64-fold increase in administered mRNA. Curiously, at 24 h and at a lower particle concentration (0.5x10⁸ NPs/well), where the difference in mRNA concentration between DSPC-LNP6 and DOTAP-LNP2 was similar, the DOTAP-based formulation outperformed DSPC-LNP6 (Figure S5D). The cell viability measured using approach 2 was unaffected by any of the formulations (Figure S5E). This is likely because, for DOTAP-LNP2, the maximal particle concentration corresponded to only 0.3 µg/well of mRNA, below the toxicity threshold observed in approach 1 (Figure S5E). These results indicate that both particle and mRNA concentrations, as well as the intrinsic ability of the LNP membrane to destabilize endosomes, influence LNP performance and the temporal profile of protein expression.

Physicochemical Properties of LNPs in SNF and Supplemented Cell Culture Medium and Their Cell Membrane-Disruption Ability

Because of the discrepancy between the in vivo and in vitro transfection outcomes of the selected LNPs, we conducted a physicochemical analysis to better understand this difference. DOPE-LNP5 exhibited a slight increase in size when the nucleic acid cargo was changed to mRNA (Figure S6A), whereas the zeta potential of DOTAP-LNP2 shifted from a positive value of ~ 23 mV to a less positive value of ~10 mV (Figure S6B). All LNPs showed high mRNA EE exceeding 70% (Figure S6C). Cryo-TEM analysis revealed the presence of both traditional and bleb LNP structures in DSPC-LNP6 cells (Figure 4A).

Figure 4.

Characterization of LNPs in complex dispersants and their cell membrane disruption ability. (A) Cryo-TEM of DSPC-LNP6 highlights canonical and bleb LNP structures. (B) Diffusion coefficient obtained and calculated based on NTA analysis. (C) Size, (D) Particle concentration assessed by NTA and (E) Zeta potential by ELS of LNPs in PBS or SNF at 37 C. (F) Size, (G) Particle concentration obtained by NTA and (H) Zeta potential by ELS of LNPs in PBS or DMEM supplemented with 10% of FBS and 1% penicillin/streptomycin at 37 C. (I) Hemolysis capacity of LNPs at different pH values. Data are presented as mean ± SD analyzed by Two-way ANOVA followed by Bonferroni post-test. (A–G) n=3 independent measurements. (H) n=3 biological replicates. *p<0.05, ***p<0.0005 and ****p<0.0001.

Next, we verified the physicochemical properties of LNPs in SNF-containing mucin, the most abundant protein in the mucus. The addition of mucin to acidic SNF resulted in high polydispersity index, with values of around 0.5 (Figure S6D); therefore, characterization was performed using the nanoparticle tracking analysis (NTA). This analysis revealed an increase in particle size by 1.86-fold for DOPE-LNP5, 1.56-fold for DSPC-LNP6, and 2.16-fold for DOTAP-LNP2 when comparing measurements in PBS versus SNF (Figure 4C). Consistent with the increase in particle size, a reduction of at least 2-fold in the apparent diffusion coefficient (D) was observed for all formulations when characterized in SNF (Figure 4B). No differences in apparent D were observed among the LNPs in SNF, and no alterations in particle concentration were found for any of the LNPs (Figure 4D). Zeta potential measurements by ELS revealed a shift toward more negative values for all the LNPs in the SNF (Figure 4E).

When characterized in DMEM cell culture medium supplemented with 10% FBS, NTA revealed a reduction in particle size most pronounced for DSPC-LNP6 (Figure 4F). This apparent decrease in size may reflect the high sensitivity of the instrument and the potential detection of FBS proteins as particles (Figure 4G and S7A–C). Significant changes in zeta potential were observed for DOPE-LNP5 and DOTAP-LNP2 in the supplemented medium (Figure 4H), with both formulations becoming more negatively charged.

Next, we evaluated the stability of LNPs in the complex dispersants at 37°C under shaking, over 4 h in SNF, and 24 h in DMEM, as these conditions could potentially represent the physical state in which LNPs may ultimately interact with the biological environment. During stability testing in SNF, the only parameter that changed was the zeta potential, which shifted toward more neutral values for the DSPC- and DOTAP-based formulations. This shift likely reflects the protonation of the ionizable lipid in the acidic SNF environment (Figure S8A–C). DSPC-LNP6 was the most stable formulation in DMEM-supplemented medium, exhibiting a ~740-fold reduction in particle concentration (Figure S8E) and a 1.75-fold increase in mean size, indicating that smaller particles lost stability first (Figure S8D). Interestingly, no changes in zeta potential were observed for DSPC-LNP6, whereas DOPE-LNP5 and DOTAP-LNP2 became more negative over time (Figure S8F). Stability in PBS over a 24 h period was also carried out (Figure S8G and I). The effect on DSPC-LNP6 was similar to that observed in supplemented DMEM but less pronounced, with a smaller reduction in particle concentration of 3.8-fold (Figure S8H) and a slight increase in size (Figure S8G). The only divergent effect was observed for zeta potential, which became more neutral after 24 h (Figure S8I).

Finally, the membrane disruption ability of LNPs was evaluated using a hemolysis assay. No toxicity of the LNPs in RBCs was found at pH 7.4, as indicated by the comparable results between LNP-treated samples and PBS. In contrast, at pH 5.5, pronounced membrane disruption was observed for DOPE-LNP5 and DOTAP-LNP2 (Figure 4I).

Biological Validation of DSPC-LNP6 Using TRAIL-mRNA in Experimental Lung Metastasis

The top-performing DSPC-LNP6 was then selected for the delivery of a functional mRNA encoding TRAIL immune cytokine, aiming to induce tumor cell death in the lungs following IN administration. The in vivo and in vitro results indicated that this formulation could transfect and engage lung resident cells, thereby potentially contributing to the death of tumor cells that metastasize to the lungs.

Colo 205 cells, both GFP and luciferase positive, were intravenously injected into NSG mice, and 26 days post-injection (DPI), a consistent luminescent signal was detected in the lung region. Mice were randomly divided into two groups, and treatment was immediately initiated with TRAIL mRNA formulated in DSPC-LNP6 (Figure 5A). IN administration of TRAIL mRNA encapsulated in DSPC-LNP6 effectively targeted lung tumor cells. Longitudinal IVIS imaging showed a trend toward reduced lung metastasis progression in the treated group, although the difference was not statistically significant (p = 0.0852) (Figure S9A and B). Ex vivo imaging of the lungs corroborated these findings, showing a reduction in metastatic burden in the treated group compared with the controls (Figure 5B and C). At the end of the treatment, flow cytometry analysis of lung tissues revealed an increased frequency of TRAIL + cells (Figure 5D, E and S10) and higher numbers of GFP⁺ dead cells in the TRAIL-mRNA-treated group than in the controls, indicating TRAIL expression and enhanced cancer cell death (Figure 5D–F). Caspase 3 expression, a marker of death receptor activation by TRAIL, was also elevated in TRAIL mRNA-treated mice (Figure 5D–G). Additionally, within the treated group, TRAIL production was increased in the Colo 205 cells (GFP+) compared to lung resident (GFP-) cells (Figure S9C). These findings indicate that DSPC-LNP6–mediated delivery of TRAIL mRNA induces tumor cell death via TRAIL receptor activation.

Figure 5.

TRAIL-mRNA loaded into DSPC-LNP6 for the treatment of Colo 205 lung metastasis. (A) Experimental design of Colo 205 cells lung metastasis and treatment scheme. (B) ex vivo lungs IVIS images. (C) Quantification of luminescent signal in lungs. Data are presented as mean ± SD analyzed by T test. N=5 animals/group (D) Representative density plot with gates used for TRAIL+ production, GFP+ dead cells (Colo 205) and Caspase 3 determination. (E) Quantification of cells expressing TRAIL in the lungs. (F) Quantification of Colo 205 dead cells in the lungs. (G) Quantification of Caspase 3 in Colo 205 cells. Data are presented as mean ± SD analyzed by T test. N=5 animals/group with one technical replicate per animal. *p<0.05, **p<0.005.

Discussion

Barcode-based screening enables simultaneous comparison of multiple LNP formulations in vivo. Using this technology, the biodistribution of hundreds of LNP formulations for therapeutic applications has been explored.^16,34,35 However, most studies have focused on the IV route of administration to investigate how LNP composition influences their fate in the body, primarily aiming for extrahepatic accumulation.^18,27 Another route of administration that has been investigated is the intramuscular route, which has been used to understand how different LNPs compositions recruit and interact with immune cell populations for different applications, ranging from cancer immunotherapy to infectious diseases.^16,17

Herein, we used the b-DNA approach to investigate the biodistribution of an LNP library administered via the IN route and evaluated how helper lipid identity and changes in the relative helper lipid/cholesterol composition influence the lung-targeting efficiency of LNPs. We chose to focus on the IN route because it provides direct exposure of nanoparticles to the pulmonary tissue once the respiratory tract barriers are overcome. In addition, recent studies have shown that physicochemical targeting of LNPs to the lungs after systemic administration often fails to engage the epithelial compartment, but rather accumulates in the pulmonary vasculature leading to capillary entrapment, resulting in clotting.³⁶ These findings highlight the importance of investigating different administration routes and evaluating the role of helper lipids. While selective organ targeting (SORT) LNPs identified DOTAP as a helper lipid that enhances lung targeting following IV administration,^13,37 its targeting properties may differ depending on the route of administration. This difference could potentially be influenced by variations in the biological environment encountered by LNPs after IV versus IN delivery, including differences in the biomolecular interactions at the LNP interface.

The Biodistribution assessment revealed superior b-DNA delivery to the lungs for LNP formulation containing 50% of DSPC. The other organs exhibited higher b-DNA delivery with DOPE-based LNPs. Lung targeting was enhanced in formulations containing higher amounts of DSPC, followed by formulations containing lower amounts of DOTAP. Importantly, because the total lipid composition was maintained at 100 mol%, increasing the helper-lipid molar fraction necessarily resulted in a corresponding decrease in cholesterol content. Thus, the effects observed across these formulations should instead be interpreted as arising from changes in helper lipid identity together with the relative helper lipid/cholesterol composition. The widespread b-DNA delivery observed for DOPE-based formulations may be attributed to DOPE’s conical geometry, which leads to the formation of an inverted hexagonal HII phase in the LNP membrane. This membrane organization exhibits a higher capacity to destabilize biological membranes, resulting in broader biodistribution of DOPE-LNPs.^14,38

In contrast, lung delivery was favored for DSPC-LNP6, which is expected to adopt a lamellar phase owing to the cylindrical geometry of the DSPC. This geometry led to a more tightly packed LNP membrane enriched in helper lipids, consistent with the higher particle concentration observed for DSPC-LNP6. Owing to this lamellar organization and increased membrane packing, DSPC-LNP6 is likely less susceptible to destabilization by surfactants present in mucus, allowing it to reach the lungs and maintain more integrity. DSPC-LNP6 exhibited enhanced lung delivery and transfection efficiency.³⁹

Although fewer DOPE-LNPs remained in the lungs, resulting in lower transfection, their stronger membrane destabilization capacity likely promoted widespread biodistribution, making them top performers in other organs. The performance of the DOTAP-based LNPs may be related to the permanent positive charge of DOTAP. This feature may enhance the interaction with negatively charged mucus, limiting the diffusibility of these LNPs within the respiratory tract, particularly when higher DOTAP molar ratios are used.

In summary, the application of a nucleic acid barcode strategy enabled a comprehensive assessment of whole-body biodistribution of the LNP library following IN administration. While our primary objective was to evaluate the lung-targeting efficiency, this approach also provided broader insights into the systemic fate of LNPs beyond the respiratory tract. Barcode-based screening strategies have inherent limitations, as higher nucleic acid delivery does not necessarily translate into greater functional transfection efficiency. In addition, changes in the nucleic acid cargo may affect LNP physicochemical properties, cellular uptake, intracellular trafficking, and ultimately transfection performance. Therefore, screening results obtained with b-DNA should not be interpreted as a direct surrogate for mRNA delivery and require subsequent validation and characterization after reformulation with the intended RNA cargo. For this reason, the lead formulations identified in our b-DNA screening were further evaluated using luciferase mRNA both in vitro and in vivo, allowing us to assess the correlation between barcode-based delivery and functional mRNA expression. Despite these limitations, barcode-based strategies provide a valuable high-throughput approach for narrowing large libraries to a smaller number of leads suitable for detailed functional validation. Our results demonstrated that DSPC-containing formulations, particularly those with higher DSPC-to-cholesterol ratios, exhibit an enhanced capacity to overcome respiratory tract barriers, resulting in more efficient lung delivery and transfection. This observation is consistent with those of recent studies,^40,41 reinforcing the importance of overall lipid composition, including helper lipid identity and the relative helper lipid/cholesterol balance, as a determinant of pulmonary delivery.

The helper lipid geometry and resulting biophysical properties of the LNP membrane provide a useful framework for interpreting the transfection profiles observed in this study. Our in vitro results are consistent with previous reports demonstrating that DOPE-based formulations exhibit superior transfection efficiency.¹⁸ This behavior is likely associated with the DOPE geometry discussed above, facilitating membrane fusion and endosomal escape. In contrast, our in vivo findings indicate that DSPC-containing formulations enhance transfection in the lungs following IN administration.

Because the in vitro transfection profile differs from the in vivo outcomes, simple 2D cultures using pulmonary cells appear to be limited for screening LNP formulations aimed at lung targeting. These findings also highlight the importance of high-throughput in vivo approaches for identifying and optimizing lung-targeting LNPs following intranasal administration. Nevertheless, the physicochemical characterization of LNPs in complex media, such as SNF- and FBS-supplemented culture media, offers new opportunities to better understand LNP performance and stability over time. Incubation in SNF affected all formulations in a comparable manner, with no significant differences observed in particle size or apparent diffusion coefficients across the selected LNPs. These results suggest that under in vitro conditions, the surrounding medium may mask composition-dependent differences, leading to convergence in measurable physicochemical properties. Revisiting the role of helper lipid composition helps to contextualize these findings.

Importantly, the SNF experiments did not reveal significant formulation-dependent differences in particle size or apparent diffusion behavior and therefore do not directly demonstrate greater stability of DSPC-LNP6 in the respiratory environment. SNF represents a simplified and static model that does not fully recapitulate the biochemical complexity and dynamic biological barriers of the respiratory tract, including mucus turnover, mucociliary clearance, interactions with epithelial cells, and other components present in vivo. These limitations may contribute to the absence of measurable physicochemical differences among the LNP formulations under the conditions tested and may partly explain the differences between the in vitro and in vivo outcomes. Thus, although DSPC-containing LNPs are expected to form more tightly packed and rigid membranes, this structural advantage, which is likely critical for maintaining stability in the respiratory tract in vivo, does not appear to confer a measurable distinction in SNF under the conditions tested. In contrast, functional assays provided a clearer differentiation between the formulations. The hemolysis assay demonstrated that DOPE-based LNPs induced pronounced membrane destabilization, consistent with their fusogenic character.^14,18,38 Additionally cryo-TEM revealed the non-classical bleb-like structures in DSPC-LNP6. Recent literature suggests that bleb formation is associated with enhanced transfection in some LNP systems, whereas bleb-rich populations may display reduced transfection efficiency under conditions that limit the availability of ionizable lipids required for endosomal membrane destabilization. Thus, the bleb morphology observed in DSPC-LNP6 may influence mRNA organization and intracellular delivery; however, its specific contribution to endosomal escape remains to be determined.⁴²

As highlighted in this work, DOPE- and DOTAP-based formulations exhibited more negative zeta potential values when characterized in FBS-supplemented cell medium and over time, suggesting a greater propensity for protein corona formation compared to DSPC formulations, further indicating the pronounced susceptibility of these LNPs to the destabilizing effects of the respiratory tract.

Stability studies of DSPC-LNP6 in cell culture medium revealed a decrease in particle concentration, accompanied by an increase in particle size, indicative of particle destabilization. This behavior may be linked to the relatively high phase transition temperature of DSPC, which can limit membrane fluidity under standard in vitro conditions and compromise colloidal stability.⁴³ The limited fluidity of DSPC-LNP6 may also explain its reduced susceptibility to destabilization in mucus, achieving higher b-DNA delivery in the lungs and enhanced transfection.

Regarding the biological application of the screened and selected LNP formulations, we focused on a pulmonary metastasis setting, given that the lungs represent one of the most frequent sites of secondary tumor colonization. Clinically, up to 50% of patients with advanced malignancies develop lung metastases, with colorectal cancer showing a particularly strong propensity for dissemination to the pulmonary microenvironment.⁴ In line with this clinical relevance and building on our group’s previous work demonstrating the therapeutic potential of TRAIL mRNA delivered via LNPs for colorectal tumor regression,^28,44 we employed an experimental model designed to recapitulate key aspects of metastatic spread. Importantly, our therapeutic objective was not to target the primary Colo 205 tumor through intranasal administration. Instead, we aimed to evaluate whether pulmonary delivery of TRAIL mRNA via DSPC-LNP6 could potentially induce apoptosis in tumor cells that migrated to and colonize the lung tissue. This approach provides a framework for assessing the potential of IN LNP-mediated mRNA delivery as a strategy to target disseminated tumor cells within the pulmonary niche, a critical and clinically relevant stage of metastatic disease progression.

Building on the identification of DSPC-LNP6 as a top-performing formulation for pulmonary delivery, we evaluated its functional potential in a disease-relevant context by delivering mRNA encoding the TRAIL immune cytokine. This approach was aimed at determining whether IN administration could achieve sufficient transfection of lung-resident cells to induce apoptosis in metastatic tumor cells colonizing the pulmonary niche. At the cellular level, flow cytometry analysis provided clear evidence of the biological activity. An increased frequency of TRAIL⁺ cells in lung tissues confirmed successful mRNA delivery and expression, whereas a higher proportion of GFP⁺ dead tumor cells in the treated group demonstrated enhanced cancer cell killing. This effect was further supported by elevated caspase-3 expression, indicating enhanced apoptotic activity in the treated group. Although this finding is consistent with the expected biological activity of TRAIL, cleaved caspase-3 is a downstream apoptosis marker and does not, by itself, specifically demonstrate activation of the TRAIL death-receptor pathway. Together, these findings established that DSPC-LNP6 enables functional mRNA delivery to the lungs and can induce tumor cell apoptosis within the metastatic microenvironment. Despite these outcomes, the overall reduction in metastatic progression was not significant. Several factors may have accounted for this discrepancy. The therapeutic regimen, including dosing frequency, mRNA dose, treatment duration, and the transient nature of mRNA expression, may not have been sufficient to achieve a robust antitumor effect at the whole-organ level.

Consistent with this interpretation, although lung bioluminescence showed a downward trend, the reduction did not reach statistical significance. In contrast, flow cytometry demonstrated significant biological effects, including increased TRAIL expression and tumor-cell apoptosis, providing mechanistic evidence that DSPC-LNP6 successfully delivered functional TRAIL mRNA and activated the intended apoptotic pathway. These findings should therefore be interpreted primarily as proof-of-concept for functional intranasal mRNA delivery. Further optimization of the dosing regimen, including higher mRNA doses, increased dosing frequency, or prolonged treatment, may be required to achieve a greater reduction in metastatic burden. Combination strategies with complementary antitumor therapies, including immune checkpoint blockade, may also warrant investigation. Although no evident treatment-related toxicity was observed under the experimental conditions evaluated, further studies are required to assess the potential cumulative and long-term toxicity associated with repeated dosing. Additional studies will be necessary to determine the durability of the response, cumulative toxicity associated with repeated administration, and the therapeutic benefit of optimized monotherapy or combination regimens. Overall, our findings demonstrate that intranasal delivery of TRAIL mRNA using DSPC-LNP6 can induce biologically relevant apoptotic responses in metastatic tumor cells in the lung and support further development of this delivery platform.

Conclusion

This study demonstrates the utility of DNA barcodes (b-DNA) for high-throughput in vivo screening of LNP libraries to identify formulations optimized for lung delivery via intranasal administration. By systematically varying helper lipid types and their molar ratios, we established a robust platform to directly assess biodistribution and transfection efficiency in the lung. Our findings revealed that DSPC-LNP6 successfully overcame the physiological barriers of the respiratory tract, mediating efficient lung transfection in vivo, while DOPE-LNP5 achieved the highest in vitro transfection, correlating with an enhanced membrane-disruption capacity. Importantly, the lack of correlation between in vitro and in vivo performances underscores the limitations of conventional cell-based models in predicting lung delivery, highlighting the value of in vivo high-throughput screening approaches. Finally, application of the top-performing DSPC-LNP6 for the delivery of TRAIL mRNA demonstrated a promising proof-of-concept, inducing tumor cell death in a lung metastasis model. Overall, these findings support b-DNA-based screening as a useful strategy for accelerating the identification of LNPs for pulmonary delivery and provide a foundation for the further development and optimization of RNA therapeutics targeting the respiratory tract.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

Dr Marco Túllio Rodrigues Alves reports Support for the manuscript from CNPq; PRPq-UFMG; CAPES; FAPEMIG; MCTI/FINEP—MS/SCTIE/DGITIS/CGITS; MCTI/CNPq/CAPES/FAPs 2025, during the conduct of the study. Dr Pedro Pires Goulart Guimarães reports Support for the manuscript from MCTI/FINEP—MS/SCTIE/DGITIS/CGITS, FAPEMIG, Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), PRPq-UFMG, National Council for Scientific and Technological Development-CNPq, during the conduct of the study. The authors declare that they have no conflicts of interest in this work.