A rapid and reliable strategy for identification of LNP and protein corona in vitro and in vivo

Abstract

This study develops an integrated analytical method for visual verification and subsequent analysis of lipids and protein coronas of lipid nanoparticles (LNPs) in vitro and in vivo. In this strategy, SDS-PAGE combined with Coomassie brilliant blue (CBB) staining enabled rapid and reliable visual verification and semi-quantification of LNPs/lipids in solutions, yielding a linear standard curve of R² = 0.992. LC-MS/MS and GC–MS provided specific detection and quantification of key lipids (SM-102, DSPC, DMG-PEG2000, cholesterol), with excellent linearity for all four lipids (R² > 0.90). DMG-PEG2000 was quantifiable in the original LNP formulation but could not be reliably quantified after gel extraction due to the sample processing workflow. LNPs and associated protein corona were separated by SDS-PAGE and visualized by CBB after isolation from bulk solutions by size-exclusion chromatography (SEC) or sucrose density gradient centrifugation (S-DGC). LC-MS/MS and GC–MS detected SM-102, DSPC and cholesterol respectively in the forefront of SDS-PAGE gel loaded with factions eluted in the first peak of SEC chromatogram or the pellet layer of S-DGC. TEM confirmed the presence of intact LNPs in these fractions. We found that samples collected from both SEC and S-DGC yielded highly consistent protein coronas as measured by protein MS, enriched in apolipoproteins, immune-related proteins. In addition, pathways such as complement/coagulation and cholesterol metabolism were enriched. Importantly, SEC/SDS-PAGE/CBB combined with protein MS were successfully used to analyze in vivo LNP/protein corona, which revealed stark differences in composition between in vivo and in vitro protein corona, with opsonins such as immunoglobulins enriched in the former. Furthermore, we demonstrated that SDS-PAGE/CBB could be effectively used to indirectly assess LNP uptake by cells. In summary, a rapid and reliable analytical metheod for simultaneous detection of LNP and protein corona with visual verification is validated, which will improve quality control in study of LNP protein corona.

Graphical abstract

1. Introduction

Lipid nanoparticles (LNPs) serve as a pivotal platform for the delivery of nucleic acid therapeutics and have demonstrated translational potential in vaccine development and gene therapy (Safford et al., 2025; Chappell et al., 2024). The efficacy and in vivo trajectory of LNPs are influenced not only by their intrinsic lipid composition and physical attributes but also, upon introduction into biological systems, by the rapid formation of a “protein corona” on their surface (Rademacker et al., 2025; Wang et al., 2024; Straten et al., 2024). This spontaneously assembled layer of adsorbed biomolecules masks the original particle interface and subsequently dictates critical biological outcomes, including cellular recognition (Neal et al., 2023), endocytic routing (Bresinskya and Goepfericha, 2025), immune responses (Guo et al., 2025), and tissue distribution (Jean-Michel et al., 2021). Consequently, the precise identification and separation of LNPs and related protein corona from biological solutions for subsequent characterization of the LNP and protein corona-encompassing their composition, architecture, and biological functions—is essential for elucidating the in vivo behavior of LNPs, refining their delivery performance, and evaluating their safety profile. Despite its importance, the field continues to grapple with several methodological limitations. One prominent challenge is the absence of an integrated analytical workflow that effectively combines rapid lipid detection with robust quantification.

LNP typically lack ultraviolet (UV) chromophores, and their quantitative analysis relies on a variety of techniques tailored to specific needs (Ru et al., 2025). For instance, Yang et al. employed the high-performance liquid chromatography-charged aerosol detector (HPLC-CAD) method to determine the contents of four lipids in the LNP-encapsulated Omicron XBB.1.5 COVID-19 mRNA vaccine (Yang et al., 2024). Similarly, Zhang et al. used HPLC coupled with evaporative light scattering detection (HPLC–ELSD) to determine four lipid components in an LNP-based RNA delivery system (Zheng et al., 2025). Lei et al. applied liquid chromatography–tandem mass spectrometry (LC–MS/MS) to study the biodistribution of cationic lipids and their metabolites in rats (Lei et al., 2023). In a complementary approach, Voke et al. performed lipid quantification using fluorescence-based tracking with fluorescently labeled lipids (Voke et al., 2025). Despite the utility of these analytical techniques, each method used for LNP quantification presents notable limitations. When chromatographic methods are coupled with universal detectors such as CAD or ELSD, they often require time-consuming method development for complex lipid mixtures and exhibit a narrow linear dynamic range, which may compromise quantification accuracy. Although mass spectrometry (MS) offers higher sensitivity and specificity, its application is constrained by high operational costs, significant matrix effects in biological samples requiring careful calibration, and cumbersome sample preparation procedures. Alternatively, fluorescence-based strategies, while highly effective for tracking, introduce a fundamental uncertainty: the incorporation of fluorescent labels may alter the native physicochemical properties and biological behavior of LNPs, making quantification indirect and potentially unrepresentative of the unmodified formulation.

Furthermore, current understanding of the LNP protein corona is predominantly derived from in vitro plasma incubation models. However, whether these models can accurately reflect the “biological identity” formed by LNPs within the complex physiological microenvironment in vivo remains lacking in direct and systematic experimental comparison and validation. At the same time, existing methods for assessing LNP cellular uptake efficiency largely rely on fluorescence labeling or molecular detection techniques, which are often cumbersome and time-consuming. Therefore, there is an urgent need to establish a new assessment strategy that is simple, reliable, and capable of rapid implementation.

This study established a rapid and reliable quantitative analytical strategy for comprehensive analysis of LNP lipids and protein corona both in vitro and in vivo. By extending the classic CBB staining technique to the semi-quantitative (based on gray value) analysis of total LNP lipids, we developed a method based on SDS-PAGE combined with grayscale analysis for rapid screening of LNP lipids in biological solutions. This was further integrated with LC-MS/MS and GC–MS technologies to achieve accurate identification and quantitative analysis of key lipid components (SM-102, DSPC, and cholesterol). Meanwhile, proteins of LNP protein corona were simultaneously collected for subsequent protein mass spectrometry analysis. This method effectively eliminates interference from endogenous plasma particles during protein corona analysis. Additionally, using this SDS-PAGE/CBB staining strategy, we systematically evaluated the transfection efficiency of LNPs and endocytosis under different cell treatment conditions. The results showed a significant negative correlation between LNP content in the culture supernatant and both cellular uptake efficiency and target gene expression levels. Thus, our strategy is applicable for a wide range of circumstance requiring analysis of LNP and/or protein corona.

2. Materials and methods

2.1. Chemicals, reagents and materials

HeLa cells (ATCC, Virginia, USA). Water, methanol, acetonitrile, isopropanol (Fisher Scientific, New Hampshire, USA); BeyoBlue™ Plus (Beyotime Biotechnology, Shanghai, China); EEA1 antibody (Cell Signaling Technology, Boston, Massachusetts, USA); Alexa Fluor 555 conjugate (Molecular Probes, Silicon Valley, California, USA); DOPE-Atto 647 (Atto-Tec, Berlin, Germany); Streptavidin-FITC (APE*BIO, Houston, Texas, USA); dUTP-11-biotin (Thermo Fisher Scientific, Waltham, Massachusetts, USA); Cholesterol (Sigma-Aldrich, St. Louis, Missouri, USA); SM102, DSPC, and DMG-PEG (Sinopeg, Xiamen, China). PBS (Servicebio, Wuhan, China); Shim-pack GIST-HP C18 (3 μm, 2.1 × 100 mm)(SHIMADZU, Kyoto, Japan).

2.2. Preparation and characterization of LNPs

The LNP formulation used consists of ionizable cationic lipids, phospholipids, cholesterol, and polyethylene glycol (PEG)-lipids, with a molar ratio of 50:10:38.5:1.5, which is referred to as the classic formulation. The four lipids used were SM-102, DSPC, Cholesterol, and DMG-PEG2000. The N/P ratio was 6. The LNP formulation was prepared via the ethanol injection method, using biotin-labeled and purified linear DNA (Biotin-11-dUTP, ThermoFisher) as the nucleic acid, with the organic phase and aqueous phase mixed at a volume ratio of 1:3 (Luo et al., 2025; Cheng et al., 2024). Ethanol was removed by dialysis and replaced with PBS buffer (155 mM NaCl, 3 mM Na₂HPO₄, 1 mM KH₂PO₄, pH 7.4); Particle size and zeta potential were measured using a Malvern Zetasizer Advance Pro((Malvern, UK)), with 10 μL of LNP sample diluted in 990 μL of deionized water at room temperature (25 °C); the encapsulation efficiency of DNA/mRNA was verified by agarose gel electrophoresis retardation assay: an aliquot of LNP solution was mixed with or without an equal volume of 10% Triton-X 100 aqueous solution, vortexed thoroughly, and incubated at 37 °C for 5 min. The morphological characteristics of LNPs were observed using a transmission electron microscope (TEM, Tecnai G2 F20 S-Twin, FEI, USA).

2.3. Coomassie brilliant blue staining

The collected samples were separated by 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (80 V for 1 h, then switched to 120 V until the bromophenol blue front reached approximately 1.5 cm from the bottom of the stacking gel). After electrophoresis, the gel was removed and stained with 0.1% Coomassie Brilliant Blue R-250 at room temperature for 30 min. Following staining, the gel was de-stained by shaking overnight in ultrapure water with frequent change until the background became transparent and the bands were clearly visible. Finally, the gel was imaged and analyzed using a gel documentation system.

2.4. LC/GC–MS determination and analysis

LC-MS condition:The high-performance liquid chromatography (HPLC) method for lipid detection is as follows: mobile phase A consists of a 10 mM ammonium acetate +0.1% formic acid aqueous solution, while mobile phase B is composed of 10 mM ammonium acetate +0.1% formic acid in a water/isopropyl alcohol mixture (5/95, v/v). The flow rate is set at 0.3 mL/min, and the needle wash solution is an isopropyl alcohol/water mixture (7/3, v/v). The autosampler temperature is maintained at 4 °C, and a Shim-pack GISTHP C18 column (3 μm, 2.1 × 100 mm) is used with the column oven temperature set at 30 °C. A Quadrupole-Time of Flight-Mass Spectrometry instrument was used in positive ion detection mode. Instrument parameters: Gas1 30 psi; Gas2 30 psi; Spray Voltage 5500 V; Curtain Gas 30 psi; Ion Source Temperature 500 °C. MS scanning mode was employed with a scan range of m/z 200–1500, De-clustering Potential (DP) 50 V, and Collision Energy (CE) 40 eV.

GC–MS condition:The chromatographic conditions were set as follows: Agilent J&W DB-5MS capillary column (30 m × 0.25 mm × 0.25 μm) was used; the injection port temperature was 250 °C; high-purity helium (purity ≥99%) served as the carrier gas with an initial pressure of 152.3 kPa, a carrier gas flow rate of 16.2 mL/min, and a column flow rate of 1.20 mL/min; split injection was adopted with a split ratio of 10∶1 and an injection volume of 1 μL; the temperature program was initialized at 250 °C, ramped to 300 °C at 20 °C /min, then further ramped to 320 °C at 5 °C/min, and held for 3 min. For the mass spectrometric conditions, an electron impact (EI) ion source was employed with an electron energy of 70 eV; the ion source temperature was 250 °C, the MS interface temperature was 260 °C, and the solvent delay time was 3 min; the scan modes included Full Scan (scan range m/z 50.0–500.0) and Selected Ion Monitoring (SIM).

2.5. LNP and protein corona separation by size exclusion chromatography

Size Exclusion Chromatography (SEC) Analysis: 2 mL of lipid nanoparticle (LNP) solution was mixed with 200 μL of human plasma (resulting in a final volume fraction of approximately 10%). Place the mixture in a thermomixer and incubate at 37 °C with rotation for 30 min (to ensure uniform incubation). After incubation, the protein corona-coated LNP solution was injected onto the size exclusion chromatography (SEC) column (HiScale 16/20 size exclusion column (Cytiva, 28964424), containing 30 mL Sepharose CL-4B (Cytiva, 17015001) on a Akta Pure (Cytiva).

Subsequently, the column was eluted with PBS at a flow rate of approximately 1 mL/min, and collect 36 consecutive chromatographic fractions (∼ 1 mL per fraction). 20 μL from each fraction were subjected to SDS-PAGE/CBB analysis to locate LNP lipids. The protein bands and lipid bands were excised and subjected to proteomic or lipid mass spectrometry analysis.

2.6. Sucrose density gradient centrifugation (S-DGC) analysis

The sucrose density gradient column was constructed by sequentially layering sucrose solutions with concentrations of 40%, 30%, 20%, 10%, and 0% from the bottom to the top. 2 mL of LNP solution was thoroughly mixed 200 μL of human plasma, then incubate at 37 °C (physiologically relevant temperature) for 30 min to ensure sufficient formation of the protein corona. Immediately after incubation, the mixture was slowly loaded onto the top of sucrose density gradient column, and subsequently centrifuged at 110,000 g, and 4 °C for 16 h. After centrifugation, 25 consecutive fractions (0.4 mL per fraction) were collected from the top to the bottom of the centrifuge column by careful pipetting. 20 μL from each fraction were subjected to SDS-PAGE/CBB analysis to locate LNP lipids. The protein bands and lipid bands were excised and subjected to proteomic or lipid mass spectrometry analysis.

2.7. LNP cell uptake assay

LNP cell uptake assay utilized HeLa cells, which were routinely cultured under standard conditions using Dulbecco's Modified Eagle Medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. To investigate the effects of different nutritional states on cellular behavior, a serum deprivation treatment group was included, in which cells were pretreated in serum-free medium for 4 h. Subsequently, the cells were transfected with LNPs loaded with biotin-labeled DNA for a duration of 4 h. Cellular uptake and subcellular localization of the nanoparticles were examined using immunofluorescence. Following fixation and permeabilization, cells were stained with an anti-EEA1 antibody to label early endosomes, streptavidin-FITC to detect the DNA carried by the nanoparticles, and DAPI to counterstain the nuclei. Finally, multichannel fluorescence images were acquired using a high-content imaging system (Operetta CLS, PE), and professional image analysis software was employed to evaluate the co-localization of the nanoparticles with early endosomes, thereby elucidating their intracellular trafficking pathway.

2.8. In vivo LNP administration and plasma collection

All animal experiments were approved by the Animal Care and Use Committee of Dali University (Protocol Number: [2022-P2–24]) and conducted in accordance with the Dali University Guide for the Care and Use of Laboratory Animals. Male Balb/c mice (8 weeks old, weighing 20–25 g) were purchased from Hunan SJA Laboratory Animal Co., Ltd. (License No.: SCXK(Xiang)2021–0002). The mice were housed in individually ventilated cages (5 per cage) under a 12-h light/dark cycle at 22 ± 2 °C and 50–60% relative humidity, with free access to standard chow and water. Before the experiment, the animals were allowed to acclimatize for at least 7 days.Mice were randomly assigned to two groups (n = 4 per group) using a random number table. No blinding was performed due to the nature of the study. Each mouse received a single intravenous injection via the tail vein of 0.5 mg/kg of LNP-encapsulated plasmid DNA (N/P ratio = 6; lipid molar composition: SM-102: DSPC: cholesterol: DMG-PEG = 50: 10: 38.5: 1.5). The injection volume was 200 μL per mouse, and no anesthesia was used during the injection procedure.At 15 and 30 min post-injection, mice were deeply anesthetized with isoflurane, and blood was collected via cardiac puncture. Immediately after blood collection, the mice were euthanized by cervical dislocation. Blood samples were centrifuged at 2000 rpm (approximately 400 ×g) for 10 min at 4 to obtain plasma. Plasma was obtained by centrifugation at 2000 rpm for 10 min, removing red blood cells and blood clots. Plasma from 4 mice were pooled together, which was diluted by a factor of 2, then subjected to SEC separation. Fractions of 1 mL/tube were collected.No animals were excluded from the study, and no adverse events were observed. The sample size (n = 4 per group) was chosen based on previous experience with similar LNP studies; no formal power analysis was performed.

2.9. Statistical analysis

Densitometry analysis of immunoblot bands and fluorescence signal intensity were quantified using ImageJ software. Experimental data are expressed as mean ± s.e.m. Comparisons between two groups were performed using a two-tailed Student's t-test. A p-value ≤0.05 was considered statistically significant.

3. Results

3.1. Establishment of SDS-PAGE/CBB for visual verification and semi-quantification of LNP lipids

To develop a rapid, facile and reliable method suitable for detection of LNP lipids in complex biological solutions, an integrated strategy based on SDS-PAGE separation, CBB staining, coupled with mass spectrometry was developed and validated. The rapid and facile detection of total lipids was established using SDS-PAGE followed by CBB staining. As shown in Fig. 1A, lipid bands appeared at the gel forefront of SDS-PAGE after CBB staining, indicating LNP lipids could be stained by CBB and migrated faster than loading dye during SDS-PAGE. The intensity of lipid bands increased as the concentration of total lipids (2.26–31.59 μg/μl) increased. Quantification of densitometry of the bands using ImageJ yielded a linear curve of total lipid mass versus densitometry value (Fig. 1B), which exhibited excellent linearity in a wide range of lipid concentration. The fitted equation was y = 726.29× + 263.83, with a coefficient of determination (R²) of 0.990, confirming the accuracy and reliability of the method.

Fig. 1.

Establishment of SDS-PAGE/CBB staining for visualization and semi-quantification of LNP Lipids. (A), LNPs lipids visualized on SDS-PAGE after CBB staining. (B), Linear curve of lipid concentration versus densitometry in (A). (C), Representative LC-MS total ion chromatogram (TIC) of key LNP lipid components (SM-102, DSPC, and DMG-PEG2000) under full-scan mode. Red circles mark [M + H]⁺ and [M + Na]⁺ ions, green circles mark diagnostic ions. (D), LC-MS chromatogram of three lipids (DMG-PEG2000, SM-102, DSPC) in MRM mode. (E). GC/MS chromatogram of Cholesterol. (F-H), Linear regression of peak area vs concentration of SM-102 (F), DSPC (G) and cholesterol (H) derived from quantitative MS. Results are representative of three independent experiments. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Next, methods for detection and quantification of LNP lipids using LC-MS/MS or GC/MS were established. Under LC-MS full-scan mode, effective separation of SMC-102, DSPC, and DMG-PEG2000 was obtained (Fig. 1C). Under positive-ion mode electrospray ionization mass spectrometry (ESI-MS), SM-102 primarily exhibits the protonated molecular ion [M + H]⁺ (m/z 710.8) and its sodium adduct [M + Na]⁺ (m/z 732.8). In addition, the characteristic fragmentation in its primary mass spectrum originates from alpha-cleavage near the ionizable tertiary amine headgroup. Specifically, the protonated molecular ion undergoes C—N bond cleavage, resulting in the neutral loss of a structural unit comprising the morpholine ring, N,N-dimethyl group, and an adjacent carbonyl moiety (inferred formula C₁₂H₂₃NO₃, neutral loss of 398.5 Da). Concurrently, the positive charge remains on the hydrophobic alkyl chain fragment at the opposite end, generating the characteristic diagnostic ion at m/z 312.3. For DSPC, both [M + H]⁺ (m/z 790.6) and [M + Na]⁺ (m/z 813.0) were detected. The latter further underwent loss of trimethylamine to form the diagnostic fragment [M + Na-59]⁺ (m/z 754.7), confirming its phosphatidylcholine structure. In the mass spectrum of DMG-PEG2000, the dominant charged ion distribution lies within the m/z range of 100–1300. Within the m/z region of 700–1200, a series of single-charged fragment ions can be observed, such as m/z 590.2, 634.2, 663.5, 707.5, 766.2, 810.3, 869.3, 913.2, 1008.5, 1052.4, 1228.2 and 1272.7. The consistent 44 Da mass difference between these adjacent fragment ions corresponds exactly to the molecular weight of an ethylene glycol unit, confirming that they all originate from the PEG chain. Based on these identifications, a multiple reaction monitoring (MRM) method was established. The resulting chromatograms showed that well-resolved baseline separation of all three components was achieved (Fig. 1D), providing a foundation for accurate quantification. To address the low ionization efficiency of cholesterol in LC-MS, GC/MS analysis of cholesterol was performed (Fig. 1E). A clear fragmentation pattern was observed, with characteristic ions including the molecular ion [M + H]⁺ at m/z 386.3, the dehydration product [M-H₂O]⁻ (m/z 368.8), and fragment ions at m/z 353.3, 301.3 and 255.2. The result indicates a method suitable for cholesterol quantification was established.

The LNP lipid containing bands were excised from the SDS-PAGE gel, extracted, and quantified separately by LC-MS/MS (MRM) or GC–MS. SM-102, DSPC and cholesterol were readily detected and quantified. DMG-PEG2000 was detected but not readily quantifiable in our experiment setting. As shown in Figs. 1F–1H, the calibration curves for SM-102, DSPC, and cholesterol all exhibited good linearity. The regression equations were y = 2736.2× + 462.96 (R² = 0.9431), y = 10,423× +10,512 (R² = 0.9741), and y = 862,353× - 2E+06 (R² = 0.9635), for SM-102, DSPC and cholesterol respectively. R² value of all regression curves exceeded 0.94, demonstrating that SDS-PAGE coupled with CBB is a reliable method for visual verification and semi-quantification of LNP lipids across a wide concentration range.

3.2. Visual verification and detection of LNP and protein corona in SEC fractions by SDS-PAGE/CBB

SEC achieves separation based on molecular size differences: smaller molecules can enter the pores of the stationary phase, while larger molecules are excluded and thus pass through the chromatography column more quickly (Kasper et al., 2021). A workflow including SEC separation of protein-LNP complexes from bulk solution, followed by SDS-PAGE/CBB staining, gel excision, and subsequent lipid and protein mass spectrometry analysis was established (Fig. 2A). As shown in Fig. 2B-C, the SEC chromatogram of LNP/plasma mixture could be divided into three discrete peaks corresponding to large, medium, and small size particles/molecular complexes, respectively. A total of 36 factions were collected. Fractions No. 1–6, No. 7–24, and No. 25–36 corresponding to peak one, two and three respectively. SDS-PAGE and CBB staining revealed that the majority of LNP lipids concentrated in fractions 1–6, i.e. peak one (Fig. 2C). The lipid-containing gel bands were excised for mass spectrometry analysis. Specifically, SM-102 and DSPC were detected by LC-MS/MS in MRM mode, and cholesterol was identified by GC–MS (Fig. 2D, E). Next, we employed TEM to investigate whether or not intact LNPs were present in SEC chromatogram peak one (fractions No.1–6). The results confirmed the presence of intact LNPs presence in peak one (Fig. 2F, G). Under TEM, both naïve LNPs (Fig. 2F) and protein corona coated LNPs (Fig. 2G) displayed multi-lamellar structure. The corona-coated LNPs exhibited a distinct, fluffy shell-like layer (i.e., the protein corona), resulting in an overall larger structure. In contrast, the native LNPs showed a smooth particle surface, which is consistent with previously reported observations (Imiołek et al., 2025).

Fig. 2.

Visual verification and detection of LNP and protein corona in SEC fractions by SDS-PAGE/CBB. (A), Workflow for visual verification and detection of LNP lipids and protein corona in SEC fractions and subsequent mass spectrometry analysis. (B), SEC chromatogram of LNP/plasma mixture. LNP was incubated with human plasma at 37 °C for 30 min, followed by separation via SEC. (C), Collected fractions were then subjected to SDS-PAGE/CBB staining. Red rectangle marks the lipid bands in fractions No.1–6, which mobilized faster than loading dye. (D), LC-MS/MS (MRM mode) detection of SM-102 and DSPC in excised gel band. (E) GC–MS detection of cholesterol in in excised gel band. (F), TEM images of LNPs in PBS. (G), TEM images of LNPs in SEC chromatogram peak one (fractions No. 1–6). (H), Classification of proteins in protein corona of LNPs incubated with human plasma in vitro. (I), Pathway enrichment Sankey plot. (J), Characterization of enriched lipoproteins. Results are representative of three independent experiments. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

We conducted LC-MS analysis of the protein corona adsorbed to LNPs (fractions No. 1–6) to characterize its composition. As shown in Fig. 2H, apolipoproteins (28%) and immune-related proteins (25%) accounted for half of the protein corona, consistent with previous reports. Complement proteins (8.8%), endocytosis/exocytosis-related proteins (2.2%), and coagulation factors (1%) were present at relatively lower proportions (Fig. 2H). Subsequent signaling pathway analysis of the protein corona revealed that the enriched proteins were implicated in complement and coagulation cascades, cholesterol metabolism, and phagosome pathways (Fig. 2I), consistent with our prior report (Suzuki et al., 2025). Further analysis of the cholesterol metabolism pathway notably identified apolipoprotein E (ApoE), apolipoprotein A-I (ApoA-I) and apolipoprotein A-II (ApoA-II) as significantly enriched components within the corona, which is in line with established literature findings (Santhni et al., 2023).

3.3. Visual verification and detection of LNP and protein corona after S-DGC by SDS-PAGE/CBB

S-DGC is another technique commonly employed for LNP protein corona separation (Vaidya et al., 2024). LNPs were incubated with human plasma at 37 °C for 30 min, followed by ultracentrifugation on a discrete sucrose gradient (top to bottom, 10%, 20%, 30%, 40%) at 110,000 ×g for 16 h. After centrifugation, fractions (No. 1–24) were collected from top to bottom, which were subjected to SDS-PAGE/CBB staining to localize lipid-enriched regions, followed by gel excision and subsequent lipid identification and proteomic analysis (Fig. 3A). After centrifugation, the LNPs/plasma mixture initially loaded on top of the sucrose gradient was pelleted to a layer between 30% and 40% sucrose gradient (Fig. 3B). SDS-PAGE and CBB revealed that LNPs were highly enriched in this pellet (fractions No.19–22) (Fig. 3C). Consistently, agarose gel electrophoresis showed that the encapsulated DNA was enriched in the pellet (fractions No. 19–22) (Fig. 3D). The lipid-containing SDS-PAGE gel bands were excised for mass spectrometry analysis. Specifically, SM-102 and DSPC were identified by LC-MS/MS in MRM mode (Fig. 3E), while cholesterol was identified by GC–MS (Fig. 3F). Quantitative analysis showed that LNP lipids were highly enriched in the pellet (fractions No.19–22) (Fig. 3G), which correlated very well with the SDS-PAGE/CBB staining result (Fig. 3H). Dynamic light scattering (DLS) analysis of S-DGC fractions No.19–22 of LNP/plasma mixture showed a polydispersity index (PDI) close to 1 (Table S1), suggesting a high degree of LNP aggregation in these fractions. TEM examination showed LNPs in these fractions, although relatively intact, were highly aggregative (Fig. 3I). The LNPs in these fractions were deformative compared to that in PBS (Fig. 3J).

Fig. 3.

Visual verification and detection of LNP and protein corona after S-DGC by SDS-PAGE/CBB. (A), Workflow for visual verification and detection of LNP lipids and protein corona in S-DGC fractions and subsequent mass spectrometry analysis. (The schematic diagram of the experimental workflow was generated using BioRender). (B), Image of LNP/plasma mixture loaded sucrose gradient density column after S-DGC. LNP was incubated with human plasma at 37 °C for 30 min, followed by separation via S-DGC, and fractions were collected. (C), Collected S-DGC fractions were examined by SDS-PAGE/CBB staining. (D) Collected S-DGC fractions were examined by agarose gel electrophoresis to locate LNP encapsulated plasmid DNA. (E-F) Lipid containing gel bands from (C) were excised and subjected to LC-MS (MRM mode) analysis of SM-102 and DSPC (E) and GC–MS analysis of cholesterol (F). (G), Quantification of lipid in S-DGC fractions No.1–25 based on MS results. (H), Pearson's correlation between MS quantification of lipids extracted from SDS-PAGE and lipid densitometry on SDS-PAGE. Fractions No.19–25 were analyzed. (I), TEM images of LNPs in PBS. (J), TEM images of LNPs in S-DGC fractions No. 19–22). (K), Volcano plot showing enriched/reduced proteins in LNP protein corona (fractions No. 19–22) vs plasma control. (L), Functional classification of proteins in LNP protein corona. Results are representative of three independent experiments.

Comparative LC-MS analysis of LNP protein corona in S-DGC fractions No. 19–22 of LNP/plasma mixture versus human plasma controls was performed. The volcano plot (Fig. 3K) revealed significant difference between the experimental group and control: proteins such as complement C1s (C1S), IGHV3–43, C3 and ApoE were markedly up-regulated, whereas IGHV3–64, alpha-2 macroglobulin (A2M), CCDC110 and others were significantly down-regulated in the former. Proteomic analysis of the protein corona revealed its major composition as follows: apolipoproteins (32.3%) and immunoproteins (13.9%) (Fig. 3L). Complements (4.9%), endocytosis/exocytosis-related proteins (2.5%), and coagulation factors (1.2%) were present at relatively lower proportions. The protein corona composition obtained by analyzing S-DGC fractions No. 19–22 of LNP/plasma mixture was largely consistent with that obtained by analyzing elutes in peak one of SEC chromatogram, albeit with some differences.

3.4. Visual verification and detection of in vivo LNP and protein corona in SEC fractions by SDS-PAGE/CBB

Thus far, we have established a rapid and reliable strategy for visual verification and detection of LNP lipids and protein corona simultaneously in in vitro samples. We asked whether this strategy could be used to visual verify lipids and protein corona in in vivo samples. LNP (N/P = 6) of class formulation (SM-102: DSPC: Chol: DMG-PEG = 50: 10: 38.5: 1.5, molar ratio) was administered to mice via tail vein injection. Fifteen to thirty minutes later, blood was collected via cardiac puncture and plasma was harvested. First, we attempted to separate LNP/protein complexes from mouse plasma by S-DGC. However, S-DGC did not pellet LNP/protein complexes from in vivo samples into a narrow layer (Fig. S1), indicating these LNP/protein complexes were highly heterogeneous. This suggests that S-DGC may not be suitable for separation of LNP/protein corona complexes in in vivo samples. Next, we employed SEC for this task (Fig. 4A). Fig. 4B and C showed the SEC chromatograms of in vivo samples at 15 min (Fig. 4B) and 30 min (Fig. 4C) respectively. Fractions were collected and subjected to SDS-PAGE/CBB staining. For samples collected 15 min post administration, LNP lipids were found predominantly in SEC chromatogram peak one (fractions No. 1–5) (Fig. 4D). However, lipids were found present in SEC chromatogram peak two (Fig. 4D), albeit at low levels, which is different from separation of in vitro samples by SEC. For samples collected 30 min post administration, LNP lipids were not found in SEC chromatogram peak one (Fig. 4D), indicating LNPs in peak one were subjected to fast clearance. The lipid-containing SDS-PAGE gel bands from peak one of 15-min samples were excised and further analyzed by LC-MS/MS and GC–MS, which successfully identified SM-102, DSPC (Fig. 4F) and cholesterol (Fig. 4G). TEM was employed to investigate the presence of LNPs in peak one of 15-min samples, which not only confirmed the presence of LNPs but also revealed that LNPs were intact structurally in the blood of mice 15 min post injection (Fig. 4H). Notably, the size of the LNPs in in vivo samples were highly heterogeneous (Fig. 4H).

Fig. 4.

Visual verification and detection of in vivo LNP and protein corona in SEC fractions by SDS-PAGE/CBB. (A), Schematic diagram of verification and detection of LNP and protein corona in in vivo samples (The schematic diagram of the experimental workflow was generated using BioRender). (B-C), SEC chromatograms of in vivo LNP plasma mixtures 15 min (B) and 30 min (C) post LNP i.v. administration into mice. (D-E), SDS-PAGE/CBB analysis of SEC fractions of in vivo LNP plasma mixtures 15 min (D) and 30 min (E) post LNP administration. Red rectangles mark lipid bands or not in peak one. Yellow rectangle marks the faint lipid bands in peak two. (F), LC-MS/MS (MRM mode) detection of SM-102 and DSPC in lipid bands (fractions No.1–5) extracted from SDS-PAGE gel of 15 min samples. (G) GC MS detection of cholesterol in lipid bands in samples as in (F). (G), TEM image of LNPs in fractions No. 1–5. Arrows point to LNPs of heterogeneous size. (H), Functional classification of in vivo LNP protein corona identified by MS analysis. (I), KEGG pathway analysis of in vivo LNP protein corona. (J), Go enrichment analysis of in vivo LNP protein corona. Results are representative of three independent experiments. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

To elucidate the composition of the in vivo LNP protein corona, we performed mass spectrometry to analyze the proteins in corresponding fractions No.1–5 (peak one). The results showed that immunoglobulins and related proteins constituted the largest proportion (46.1%), followed by metabolic enzymes (23.1%). Other components included complement system proteins (6.8%), coagulation and fibrinolysis system proteins (4.9%), protease inhibitors (4.9%), apolipoproteins (4.2%), cytoskeletal and structural proteins (4.5%) (Fig. 4I). This protein composition profile indicates that the in vivo LNP protein corona is predominantly composed of immune-related proteins while encompassing a diverse range of functional proteins. Subsequently, KEGG pathway analysis revealed that the enriched proteins in the LNP protein corona involved in biological pathways such complement and coagulation cascades, cholesterol metabolism (Fig. 4J). Further analysis indicated that apolipoproteins such as ApoE, ApoA-I and ApoA-II were significantly enriched in the protein corona (Data not shown), a finding consistent with previous literature reports [20]. In the GO enrichment analysis (Fig. 4K), the enriched biological process (BP) included proteins involved in processes such as hemostasis, coagulation, complement activation, and cell adhesion. These functions are closely associated with immune regulation and hemocompatibility, suggesting that the LNP protein corona may activates immune responses and coagulation regulation in vivo. The cellular component (CC) showed that the proteins in the protein corona are predominantly localized to the secretory pathway, with the “secreted” category exhibiting the most significant enrichment (highest -log₁₀(P-value). Additionally, notable enrichment is observed in the cytoplasm and extracellular matrix. This distribution pattern aligns with the biological context in which the protein corona forms. The enriched molecular function (MF) revealed proteins such as actin binding, protease inhibitor activity, and heparin binding, indicating their ability to mediate substrate recognition, enzyme activity regulation, and other biological activities during interactions. Notably, a discrepancy in the proportion of apolipoproteins and immune-proteins adsorbed to LNPs in vitro and in vivo was observed. This may reflect the complex interaction between LNPs and plasma within the physiological environment. Together, these results indicate that SDS-PAGE/CBB coupled with MS can be used to visual verify lipids and protein corona in in vivo samples.

3.5. Indirect measurement of LNP internalization by cells via SDS-PAGE/CBB

We extended our method to indirectly measure LNP internalization by cells (Fig. 5A). First, we examined LNPs internalization by cells via fluorescence microscope. LNP (N/P = 6) of class formulation was spiked with 0.1 mol% DOPE-Atto647 for lipid tracing. The encapsulated nucleic acid cargo was linear DNA labeled with biotin-11-dUTP. Following overnight incubation with LNPs, cells in complete medium (Fed cells) accumulated significant higher levels of cargo DNA (DNA-FITC) and lipids (DOPE-Atto647) than cells in medium deprived of FBS (-FBS cells) (Fig. 5B-D), indicating the former cells internalized higher levels of LNPs than the latter. Consistently, transgene GFP expression was significantly higher in Fed cells than that in -FBS cells transfected with LNP-pGFP (GFP plasmid) (Fig. 5E-F).

Fig. 5.

Indirect measurement of LNP internalization by cells via SDS-PAGE/CBB. (A), Schematic diagram of indirect quantitative analysis of LNP endocytosis by cells via SDS-PAGE/CBB. (The schematic diagram of the experimental workflow was generated using BioRender). (B), Endocytosis of DOEP-Atto647 incorporated, biotin-DNA encapsulated LNP by Fed cells and -FBS cells for overnight. (C), Quantification of LNP internalization by Fed and -FBS cells as measured by cargo DNA (DNA-FITC) fluorescence intensity. (D), Quantification of LNP internalization by Fed and -FBS cells as measured by DOPE-Atto647 intensity. (E), Images of GFP expression in Fed and -FBS cells transfected with LNP-pGFP. (F), Quantification of transfection efficiency in (E). (G), SDS-PAGE/CBB analysis of supernatants of Fed and -FBS cells transfected with LNPs for overnight. (H), Lipid densitometry in (G). (I), LC-MS/MS (MRM mode) detection of SM-102 and DSPC in lipid bands in (G). (J), GC–MS detection of cholesterol in lipid bands in (G). (K), Pearson's correlation between lipid densitometry and DNA-FITC fluorescence intensity. Data are mean ± s.e.m. **, p < 0.01 by t-test. Results are representative of three independent experiments.

Next, we collected supernatants from both cells, which were subjected to SDS-PAGE followed by CBB staining. The residual LNPs/lipids in the supernatant were visualized by SDS-PAGE/CBB (Fig. 5G), which showed that supernatant of -FBS cells contained significantly higher level of residual LNPs/lipids than supernatant of Fed cells (Fig. 5H). Subsequently, LC-MS and GC–MS clearly detected three lipids: SM-102, DSPC, and cholesterol in the excised bands (Fig. 5I and Fig. 5J). Moreover, LNP endocytosis inversely correlated with residual LNPs/lipids in the supernatant (Fig. 5K). These results indicate SDS-PAGE/CBB staining could be used to indirectly measure LNP internalization.

4. Discussion

4.1. Mechanism of CBB staining for LNP Lipids

The reported method successfully applies traditional protein staining technology to the semi-quantitative analysis of LNP lipids. SDS-PAGE/CBB staining allows for rapid visual verification of presence of LNPs in biological solutions. The binding of Coomassie Brilliant Blue to proteins primarily relies on positively charged amino acid residues (Wang et al., 2012); however, its staining mechanism for lipids has not yet been fully elucidated. Based on the results of this study and with reference to previous reports on the interactions of CBB with amino acids, peptides, proteins, and other compounds, we speculate that the dye may preferentially bind to the positively charged moieties in the headgroups of zwitterionic lipids, while also potentially interacting with aromatic structures and hydrophobic regions within the lipid molecules. Notably, it was observed experimentally that cholesterol and DMG-PEG2000 could also be stained, despite neither containing charged groups themselves (Data not shown). This phenomenon indicates that electrostatic interactions alone cannot fully explain the staining process. As reported in existing literature, the binding of CBB to lipids may be a combination of interaction with charged head group and hydrophobic regions (Tommy et al., 2021; Claudia et al., 2018). To further validate the specificity of this method for LNP lipids, we conducted a series of control experiments. First, positive staining results with individual cationic lipids (SM-102, DOTAP) confirmed that CBB can effectively stain the lipid components within LNPs. Second, to assess potential interference from endogenous lipids in complex biological matrices, we incubated LNPs with fetal bovine serum, human plasma, mouse plasma, and rat plasma, respectively. The results showed no significant interfering bands from endogenous lipids in any case. Additionally, no co-migrating bands were observed in any of the blank matrix controls (fetal bovine serum, human/mouse/rat plasma), confirming the matrix tolerance of the method. To ensure comparability of results, all comparative samples were run on the same gel, and consistent results were obtained across multiple plasma types, effectively ensuring the reproducibility of the method.These results demonstrate that this method possesses good functional selectivity for LNP lipids over endogenous components. Although DMG-PEG2000 could be detected in excised lipid bands, its response signal in mass spectrometry analysis was weak and did not allow for reliable quantification. Therefore, the quantitative analysis of lipids in this study focused primarily on the three components: SM-102, DSPC, and cholesterol. The observation that endogenous lipids did not interfere with SDS-PAGE/CBB detection is noteworthy, and the underlying mechanism warrants further systematic investigation if the method is to be extended to more complex biological matrices in the future.

4.2. Comparison and selection of protein corona separation techniques

We systematically compared two mainstream techniques for separation of LNP-protein corona complexes: SEC and S-DGC. SEC achieves separation based on molecular size differences. SEC efficiently separated LNP-protein corona complexes from free plasma proteins (Fig. 2). More importantly, SEC can be used to separate in vivo LNP-protein corona complexes (Fig. 4). After SEC separation, the structural integrity of LNPs was upheld, as confirmed by TEM examination (Fig. 4H). However, the fluid shear forces inherent for SEC elution may dislodge proteins from LNP surface as compared to S-DGC (see discussion below).

The separation process of SEC is gentle and effectively preserves the morphology and size of LNPs. It exhibits strong applicability for in vivo samples, efficiently separating LNP-protein corona complexes from plasma, with standardized operation and good reproducibility. However, its limitations include the fluid shear forces during elution, which may cause weakly bound proteins to detach, potentially biasing the protein corona composition toward the detection of “hard” corona proteins. Additionally, there is a risk of incomplete separation of macromolecular protein complexes with sizes similar to LNPs.In contrast, S-DGC is based on isopycnic separation principles and theoretically offers better retention of weak interactions(Liao et al., 2024; Amy et al., 2021). However, ultracentrifugation compresses particles into a narrow band, leading to significant LNP aggregation. Through blank plasma control experiments, we confirmed that blank plasma subjected to the same S-DGC procedure did not exhibit significant aggregation, indicating that the observed aggregation is primarily related to the intrinsic properties of LNPs rather than being a mere methodological artifact. Nevertheless, aggregation does compromise LNP structural integrity and subsequent functional studies. Furthermore, S-DGC may suffer from co-sedimentation of endogenous proteins in in vivo samples and is operationally complex and time-consuming.Notably, despite the aggregation observed with S-DGC, comparative proteomic analysis revealed that the major functional protein classes (e.g., apolipoproteins, immune-related proteins) obtained by both methods were highly similar, and the enriched biological pathways were largely consistent. This suggests that aggregation does not fundamentally distort the identification of key corona components.Based on these findings, we recommend SEC as the preferred mqethod for in vivo sample analysis, particularly for studies requiring subsequent functional assays. For specific research purposes that require isopycnic separation, S-DGC may serve as a complementary method, provided that appropriate blank matrix controls are included, results are interpreted with caution, and cross-validation with SEC data is performed.

4.3. Differences in protein corona composition in vitro and in vivo and their biological significance

Our comparative analysis conducted under in vivo conditions revealed significant compositional differences compared to the in vitro-formed protein corona. The in vitro protein corona was enriched in apolipoproteins (approximately 28%), whereas the in vivo protein corona was dominated by immunoglobulins (approximately 46.1%) and coagulation factors (Nabar et al., 2024; Mohri et al., 2025; Guzman et al., 2025). These differences observed at early time points suggest that LNPs are rapidly recognized by the humoral immune system and the coagulation cascade upon entering the bloodstream. This rapid recognition phenomenon in a living system, which cannot b eqee fully recapitulated by simple in vitro plasma incubation, may have a profound impact on the subsequent “biological identity” and fate of LNPs. We acknowledge that this represents only a snapshot of a dynamic process. The compositional evolution of the protein corona at different time points and its variability across different tissues remain important questions for future investigation. Nevertheless, our findings highlight that even at these initial critical moments of interaction, the in vivo protein corona exhibits distinct differences. This underscores the necessity of prioritizing in vivo characterization in the rational design of LNPs with predictable pharmacokinetic properties and immunogenicity. It should be noted that our in vivo protein corona analysis was performed on LNPs collected from blood at 15 min or 30 min post-injection. This represents only the subset of LNPs that remained in circulation at this time point, while a significant portion of the injected dose may have already been enriched in organs such as the liver. The corona composition on these circulating particles may differ from that on particles enriched in the liver, potentially introducing a sampling bias. Future studies incorporating earlier time points and tissue-resident LNPs would provide a more complete picture of the dynamic relationship between protein corona and LNP biodistribution.

4.4. Establishment and application of SDS-PAGE/CBB staining method for evaluating LNP cellular uptake

Assessment of LNP cellular uptake usually requires laborious fluorescence labeling of LNP's cargo and/or lipids (Luo et al., 2025; Cheng et al., 2024), followed by fluorescence microscope imaging, which requires high-end equipment and high-quality fluorophore. In this study, we established an indirect measurement method based on SDS-PAGE/CBB staining, which is simple and semi-quantitatively. This method measures the residual lipids in the supernatant of cells, which in turn is used to infer the extent of LNP uptake by cells. Experimental results demonstrated that under Fed conditions, active cellular internalization led to a significant reduction in LNP/lipids in the supernatant, which positively correlated with higher levels of reporter transgene expression. On the other hand, higher levels of LNP/lipids were found in the supernatant of -FBS cells due to suboptimal LNP endocytosis by cells, which inversely correlated with lower levels of reporter transgene expression. The method reported here provides a straightforward operational tool for fast evaluation of LNP endocytosis by cells.

5. Conclusion

This study presents a fast and reliable strategy for identification of LNP and protein corona in vitro solutions and in vivo biological solutions. This strategy provides simultaneous visual verification LNP and protein corona, which is lacking in current literature and will improve quality control in study of LNP protein corona.

CRediT authorship contribution statement

Rui Wang: Writing – original draft, Methodology. Hui Wang: Methodology. Ting Li: Methodology. Guanghui Zi: Methodology. Zhenyu Yang: Methodology. Yuhong Xu: Resources, Conceptualization. Baowei Peng: Writing – review & editing, Visualization, Methodology, Data curation, Conceptualization.

Consent for publication

All authors approved the final manuscript and the submission to this journal.

Funding

Funding information is not available.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

The data and materials supporting the findings of this study are available from the corresponding author upon reasonable request.