Magnetic‐Assisted Aptamer Selection and Isolation of Migrasomes for Application in Lung Adenocarcinoma Proteomic Analysis

ABSTRACT

Migrasomes, newly discovered vesicular organelles, hold promise as diagnostic biomarkers and therapeutic targets in various diseases. However, the exploration of their clinical value remains hindered by the complexity of enriching and analyzing low concentrations of migrasomes in body fluids. To address this issue, a magnetic‐assisted strategy was devised for screening aptamers specific to migrasomes, with the identified aptamer then being utilized for the specific isolation of migrasomes derived from clinical plasma. Initially, lipid‐affinity magnetic nanoparticles were prepared and employed in a Magnetic‐Systematic Evolution of Ligands by Exponential Enrichment (Mag‐SELEX) process to identify aptamers that specifically target migrasomes. An optimal aptamer, Apt_B3, with a dissociation constant (Kd) of 251.9 nM, was successfully identified. This aptamer was subsequently utilized to construct the magnetic aptamer probe system, enabling the precise and rapid capture of migrasomes from plasma within 15 min. Our strategy exhibited exceptional separation efficiency, confirming its reliability and enhanced performance compared to traditional methods such as density gradient centrifugation. Clinical samples were then analyzed to validate the potential role of migrasome‐derived tumor biomarkers in lung adenocarcinoma. These findings underscore the promising applicability of our strategy for studying migrasomes in clinical disease diagnosis.

Mag‐SELEX identified the first aptamer targeting migrasome, offering a potential ligand for their identification and detection in clinical samples. The magnetic aptamer probe, characterised by high specificity and rapid recovery, provides a simple, rapid, and economical method for elucidating migrasome mechanisms in disease and facilitating translational research in clinical diagnostics.

1. Introduction

Migrasomes are newly discovered cellular organelles, first reported in 2015. They originate from the retraction fibers that extend outward from the trailing edge of the migrating cells and are subsequently released into the extracellular environment, providing valuable insights into intercellular communication and signal exchange (da Rocha‐Azevedo and Schmid 2015; Ma et al. 2015). It has been reported that migrasomes are abundant in various human solid tissues and biological fluids, playing significant roles in the progression of numerous diseases by transporting cargos such as mRNA, proteins and metabolites (Jiang et al. 2019; Jiang et al. 2024; Jiao et al. 2021; Zhang et al. 2023). Migrasomes have been proposed as potential therapeutic targets of diseases like proliferative vitreoretinopathy (Wu et al. 2022), cerebral amyloid angiopathy (Hu et al. 2023), and chronic kidney disease(Liu et al. 2020), in which migrasome may promote disease progression and cause cytotoxicity (Liu et al. 2023; Lv and Zhang 2023; Schmidt‐Pogoda et al. 2018; Zhang et al. 2022). Migrasomes can also serve as biomarkers for early podocyte injury (Yang et al. 2024), neuron damage (Iorio et al. 2024), tumors (Deng et al. 2024; Gu et al. 2024; Qin et al. 2022; Wang et al. 2024; Zhang et al. 2024a) and other diseases (Deniz et al. 2023; Li et al. 2023; Liu et al. 2024; Sun et al. 2024; Yu and Yu 2022), as they are laden with materials transmitting signals. However, the isolation of the migrasomes is challenging. Currently, the commonly used migrasome isolation methods are density gradient centrifugation (DGC) and immunoaffinity strategies (Jiang et al. 2023). DGC approach is limited for clinical applications due to demands of high‐grade equipment, large volumes, and low yields. Immunoaffinity approach is limited for no specific target protein for migrasome at present (Liga et al. 2015). The absence of adequate methods for isolating migrasomes from clinical samples underscores the urgent need for convenient and efficient separation techniques to facilitate future research on migrasomes.

Aptamers are short single‐stranded RNA (ssRNA) or DNA (ssDNA) oligonucleotides selected from a combinatorial oligonucleotide library via the SELEX methodology (Fang et al. 2024; Kohlberger and Gadermaier 2022; Zhang et al. 2024b). They exhibit desirable features including high editability, strong stability, low immunogenicity, and cost‐effectiveness, making them suitable for utilization in detection and analysis systems. Aptamers can form unique secondary or tertiary structures and are commonly designated as synthetic antibodies due to their target recognition mechanism that parallels antigen‐antibody interactions. They can chemically bond with a wide variety of targets such as proteins, nucleic acids, small molecules, cells, membrane structure and so on (Duan et al. 2022; Ji et al. 2023; Liu et al. 2022). These oligonucleotide‐based recognition elements demonstrate superior target specificity and enhanced binding affinities compared to conventional antibodies, typically exhibiting dissociation constants (Kd values) in the nanomolar to low micromolar range. Apart from that, aptamers have several appealing characteristics over antibodies, including reduced molecular weight, improved thermal/chemical stability, negligible immunogenicity, facile chemical synthesis and functional modification and increased target discrimination capability. Scientists have tested a large number of extracellular vesicle (EVs) aptamers (Esposito et al. 2021; Yamashita et al. 2017; Zhao et al. 2024), however, there are currently no nucleic acid aptamers against migrasome.

To address this issue, a magnetic‐assisted strategy was devised for screening aptamers specific to intact migrasomes, with the identified aptamer then being utilized for the specific isolation of migrasomes derived from clinical plasma sample. Initially, lipid‐affinity magnetic nanoparticles were prepared and employed in a Magnetic‐Systematic Evolution of Ligands by Exponential Enrichment (Mag‐SELEX) process to identify aptamers that specifically target migrasomes. An optimal aptamer, Apt_B3, with a Kd of 251.9 nM, was successfully identified. We have shown that the aptamer exhibits good stability, high specificity recognizing characteristics towards migrasome. We demonstrated Apt_B3 has significant disparities between migrasome binding and small extracellular vesicle (sEVs) and microvesicles (MVs) binding. Subsequently, we employed the selected aptamer to construct a magnetic aptamer probe (MAP) system for the purification of migrasomes from plasma samples. By using this system, the operation process effectively reducing the procedure time to just 15 min. Thus, the use of MAP system in enriching migrasome holds significant promise in clinical application. The fundamental principle of the Mag‐SELEX approach and MAP system is shown in Scheme 1.

SCHEME 1.

Schematic representation and application of the magnetic‐Systematic Evolution of Ligands by Exponential Enrichment (Mag‐SELEX) strategy and the MAP system.

2. Materials and Methods

2.1. Cell Culture

L929 cells stably expressing Tspan4‐mCherry were gift from Li Yu laboratory at Tsinghua University. L929 cells were cultured in DMEM medium (Gibco) supplemented with 10% fetal bovine serum (Gibco), 100 U/mL penicillin (Hyclone), and 100 µg/mL streptomycin (Hyclone) in a 5% CO2 incubator at 37°C with saturated humidity.

2.2. Live‐Cell Imaging

Cells were cultured in 35 mm glass‐bottom dishes pre‐coated with fibronectin (1 µg/mL), and images were subsequently acquired using an Olympus (SpinSR) confocal microscope.

2.3. Preparation of the Model Migrasomes, sEVs and MVs

The purification of migrasomes typically necessitates 30–80 dishes, each measuring 150 mm in diameter. Initially, the cells were cultured in dishes coated with 1 µg/ml fibronectin (Sigma‐Aldrich, F0895) in complete DMEM medium for 16 h. Upon reaching 50% confluence, cells were detached with trypsin (Gibco), and the suspension containing both cells and migrasomes was collected in 50 mL tubes. The suspension was subjected to sequential centrifugation at 4°C: 1000 g for 10 min, 4000 g for 20 min, and 20,000 g for 30 min. The resulting pellet was collected as crude migrasomes for subsequent DGC using a top‐down density gradient (Sigma‐Aldrich, D1556). Briefly, the crude migrasomes were resuspended with 137.5 µL of dilution buffer, mixed with 400 µL of 1× extraction buffer and 252.5 µL of 60% Optiprep to prepare a 19% sample gradient solution. A continuous Optiprep gradient was then prepared by layering 500 µL each of 30%, 25%, 19%, 15%, 12%, 10%, 8%, 5%, and 2% Optiprep solution from bottom to top. The gradient was centrifuged at 150,000 g for 4 h at 4°C in an MLS‐50 rotor (Beckman, Optima MAX‐XP). Fractions of 480 µL were collected subsequently from the top, and each fraction was centrifuged at 20,000 g for 30 min. The pellets were washed with PBS and centrifuged again to obtain purified migrasomes. Based on characterisation results, the sixth fraction was collected and used for subsequent experiments.

The isolation of sEVs and MVs from L929 cell culture supernatants was performed following a differential ultracentrifugation protocol adapted from previous descriptions (Jeppesen et al. 2019). Briefly, cells were grown to 70–80% confluency and then conditioned in serum‐free medium for 48 h. The conditioned medium was collected and clarified by sequential centrifugation: first at 2000 g for 20 min at 4°C to remove cells, followed by 10,000 g for 30 min at 4°C to pellet MVs. The resulting MV pellet was resuspended in phosphate‐buffered saline (PBS) for downstream analysis. The supernatant was subsequently filtered through a 0.22 µm pore‐size polyethersulfone filter and subjected to ultracentrifugation at 100,000 g for 70 min at 4°C to pellet sEVs. The sEV pellet was washed once in PBS under the same ultracentrifugation conditions and finally resuspended in 100 µL PBS.

2.4. Western Blot

Protein samples from migrasomes were extracted and quantified using the Micro BCA Protein Assay Kit (Thermo Fisher Scientific). A total of 30 µg of proteins from each sample was electrophoresed using 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) followed by enhanced chemiluminescence detection reagents (Beyotime, P0018FS). The primary antibodies used for detection included PIGK (Abcam, ab201693, 1: 2000), NDST1 (Sigma‐Aldrich, SAB1307040, 1: 1000), ITGA5 (Cell Signaling Technology, #4705, 1:1000), TSG101 (Abcam, ab125011, 1:1000), ALIX (Abcam, ab186429, 1:1000), CD63 (Abcam, ab68418;1: 800), ARF6 (Proteintech, 20225‐1‐AP, 1:1000), ANXA1 (Proteintech, 21990‐1‐AP, 1:5000), and secondary antibodies goat anti‐rabbit IgG‐HRP (Affinity, S0001, 1: 5000) and goat anti‐mouse IgG‐HRP (Affinity, S0002, 1: 5000).

2.5. Preparation of Magnetic Nanomaterials

Firstly, a conventional hydrothermal method was utilized for the synthesis of Fe₃O₄ nanoparticles. In summary, FeCl₃·6H₂O (2.43 g) was dissolved in 60 mL of ethylene glycol and subsequently transferred into a Teflon‐lined stainless‐steel autoclave. Following this, sodium acetate (3.60 g) and sodium citrate (0.45 g) were then added into the solution successively while stirring. After one hour of vigorous magnetic stirring, the mixture was transferred to an oven and heated at 200°C for 16 h. Upon cooling to room temperature, the black sediment was collected through magnetic separation and washed sequentially with ethanol and ultrapure water. The resultant particles were re‐dispersed in approximately 15 mL of deionized water for further applications.

Subsequently, we synthesized magnetic Fe₃O₄@SiO₂ core‐shell submicrometer particles using a slightly modified method as previously developed (Wan et al. 2017). Briefly, the re‐prepared Fe₃O₄ particles (100 mg) were accurately weighed and ultrasonically dispersed in 40 mL of 0.1 M HCl for 30 min, followed by washing three times with ultrapure water. Subsequently, the particles were ultrasonically dispersed in a solution containing 32 mL of ethanol, 8 mL of water, and 0.5 mL of concentrated ammonia (28% w/w). Tetraethyl orthosilicate (0.5 mL) was then added dropwise under magnetic stirring, followed by stirring for 4 h at room temperature. The resulting particles were thoroughly washed with ethanol and ultrapure water, then dispersed in ultrapure water for further modification.

To functionalize the Fe₃O₄@SiO₂ composite with amino group, 100 mg of Fe₃O₄@SiO₂ and 0.4 mL of 3‐aminopropyltriethoxysilane were ultrasonically dispersed in a mixture of 36 mL of ethanol and 4 mL of water. The mixture was stirred at room temperature for 6 h; subsequently, 0.4 mL of triethylamine was added, and stirring was continued for an additional 18 h. Finally, the products were collected and washed three times with ethanol and ultrapure water for subsequent use.

To construct NeutrAvidin (NA)‐modified magnetic particles, 5 mg of Fe₃O₄@SiO₂‐NH₂ nanospheres were added to a dimethylformanide solution containing 10% pyridine and 1 mM phenyldiisothiocyanate for 2 h. Subsequently, the resulting particles were thoroughly washed in succession with dimethylformanide, ethanol and ultrapure water. The isothiocyanate‐grafted magnetic microparticles (MMPs) were further combined with approximately 625 µg of NA protein (Thermo Fisher Scientific, #31000) at 37°C for 1 hour, followed by treatment with 1% BSA for 30 min. The product was then washed with three times with PBS.

The freshly prepared NA‐coated MMPs were immediately utilized for the synthesis of magnetic capture probe (MCP) and MAP. The labelling probe (LP), biotin‐tagged DSPE–PEG powder, was dissolved in pure anhydrous ethanol and stored at −20°C. 1 mg of NA‐coated MMPs was dispersed in DSPE‐PEG‐Biotin at a final concentration of 1 µM and incubated at room temperature for 30 min. Upon completion, the assembled particulates were washed three times with deionized water; the collected MCP was then immediately isolated for the purification of migrasomes in the SELEX procedure. For the assembly of MAP, biotin‐modified Apt_B3 was initially heated to 95°C and then cooled on ice to induce a conformational change, followed by the same procedures as described for the MCP. The prepared MAP was utilized for the direct enrichment of migrasomes in biological samples.

The purified model migrasomes were immobilized onto magnetic beads through a lipid bilayer anchored by DSPE on the MCP. The MCP (1 × 10⁵) was resuspended in 1 mL of PBS containing the model migrasomes and incubated at 4°C for 30 min with gentle rotation. Following three washes with 1 mL of PBS, the beads were stored at 4°C in PBS. The association of MCP with model sEVs was achieved using a similar procedure.

2.6. Characterisation

The morphology of the magnetic nanomaterials obtained at each step, along with the EVs, was examined using electron microscopies. For transmission electron microscopy (TEM) tests, samples were deposited onto double‐sided copper grids and dried at room temperature prior to observation (FEI, Tecnai G2 F20 S‐TWIN). For scanning electron microscopy (SEM) tests, 5 µL of each sample was seeded onto the surface of a clean silicon wafer and thoroughly dried overnight. The morphology of the magnetic components was directly examined using the instrument (HITACHI, S‐4800), while EVs required pre‐sputter coating with gold at room temperature. The particle size and Zeta‐potential of the MMPs before and after chemical modification were measured using a Zetasizer (Malvern, ZEN 3690). For the characterisation of sEVs, nanoparticle tracking analysis (NTA, Particle Metrix Zeta View) was employed to determine particle concentration and size distribution. The capture performance of the materials was evaluated using laser confocal microscopy (CLSM, OLYMPUS, SpinSR). The effective recovery rate of magnetic nanomaterials was quantified by measuring the ultraviolet absorption intensity of the supernatant at different magnetic separation intervals using an ultraviolet spectrophotometer (HITACHI, U‐2910).

2.7. Intact‐Migrasome SELEX

A Mag‐SELEX system was developed for aptamer selection against intact migrasome. The initial ssDNA library contained a 40‐nt random region flanked by two 20‐nt constant regions, which served as primer binding site for Polymerase Chain Reaction (PCR) amplification. The forward primer was labeled with FAM at the 5' end to monitor library enrichment, and the reverse primer was labeled with biotin at the 5' end to enable capture of double‐stranded PCR products onto NA‐coated MMPs for generation of the sub‐libraries.

Prior to each selection round, the ssDNA library dissolved in PBS, denatured at 95°C for 10 min and renatured on ice for 15 min. In the first round, 10 nmol of the primary random was incubated with migrasomes (50 µg, BCA assay) pre‐captured onto MCP in 1 mL of PBS, with gentle shaking at 4°C for 90 min. After washing the beads with 0.5 mL of PBS to remove unbound ssDNA, the migrasome‐aptamer complex was heated at 95°C for 5 min, and the eluted ssDNA was collected using a magnetic separator. The harvested oligonucleotides were amplified by PCR (95°C for 5 min; 20 cycles of 95°C for 30 s, 62°C for 30 s, 72°Cfor 30 s; and 72°C for 5 min). The biotinylated PCR products were then immobilized on NA‐coated MMPs at room temperature for 30 min and washed with PBS. The FAM‑labeled sense strand was eluted with 0.1 M NaOH (5 min), neutralized with 0.1 M HCl, and desalted to obtain the sub‐library for subsequent rounds.

To improve specificity, counter‐selection was introduced from rounds 4 to 6. The ssDNA eluted from migrasomes was incubated with sEVs (20 µg) pre‐captured onto MCP (600 µg) in 1 mL of PBS, and the unbound oligonucleotides were collected by magnetic separation for PCR amplification. Selection stringency was progressively increased by adjusting library concentration, selection time, counter‐selection time, washing intensity, washing times and washing solution volume (Table S2).

Enrichment of migrasome‐binding sequences was monitored after each round by flow cytometric using FAM‐labeled sub‐library incubated with migrasome‐loaded MCP, with the initial random library and blank beads as controls. After eight rounds, the enriched pool was amplified, purified by gel extraction (TSINGKE, China), and subjected to high‐throughput sequencing (Sangon Biotech, Shanghai, China). The secondary structures of candidate aptamer were predicted using the M‐fold algorithm (http://www.unafold.org).

2.8. ssDNA Library and Primers

An 80‐nt ssDNA library was utilized for the selection of aptamers. All random oligonucleotide libraries and primers were synthesized by Sangon (Shanghai, China) and purified using HPLC. The sequence information for the random oligonucleotide libraries and PCR primer sequence is outlined as follows:

Randomized library: 5′‐ TCAAGTCACAGGTTCCAGGT‐N40‐ATAGGCACTGA

CACGACACT‐3′; where “40N” denotes a random arrangement of 40 nucleotides comprised of A, T, G, and C.

Forward primer: 5′‐FAM‐TCAAGTCACAGGTTCCAGGT‐3′. The fluorescein modification facilitates a more intuitive detection of ssDNA binding to the target during the screening process.

Reverse primer: 5′‐Biotin‐AGTGTCGTGTCAGTGCCTAT‐3′. The biotin modification enables the rapid and convenient extraction of the PCR‐amplified sub‐library from double‐stranded DNA using NA‐coated MMPs.

2.9. Optimisation of the PCR Conditions

The screened oligonucleotide from each round of SELEX were amplified using the PCR procedure. To maximize the amplification efficiency, the key elements of PCR process were optimized prior to the commencement of formal screening. First, the annealing temperature was optimized at 56, 58, 60, 62, 64, 66, 68,70 and 72°C. Subsequently, the number of PCR cycles was optimized to 20, 22, 24, 26 and 28 cycles. Finally, the volume of templates within the 10 µL system was optimized to 0.5, 1.0, 1.5, 2.0, 2.5 and 3.0 µL. The amplified products were subjected to agarose gel electrophoresis, using the 20 bp DNA ladder marker and nucleic acid stain purchased from Takara. The optimal experimental parameters were selected for subsequent experiments.

2.10. Affinity, Specificity and Selectivity Validation of the Selected Aptamer

To estimate the binding affinity of the candidate aptamer, a series of concentration gradients of FAM‐labeled aptamer (0, 10, 50, 150, 300, 450, 600, 750 nM) were incubated with the MCP‐anchored migrasomes at 4°C for 30 min. After washing the sample three times, the average fluorescence intensity at each concentration was analyzed using flow cytometry. Graphpad Prism software (version 8.4.0) was employed for data plotting and affinity curve fitting. The equilibrium Kd was calculated using the equation $\mathrm{Y}=\frac{\mathrm{Bmax}·\mathrm{X}}{\mathrm{Kd}+\mathrm{X}}$, where X, Y, and Bmax represented the aptamer concentration (mol/L), fluorescence intensity at each concentration, and maximum fluorescence intensity, respectively. All experiments were conducted independently in triplicate.

The specificity and selectivity of Apt_B3 for migrasomes was evaluated through three complementary assays: Western blot, flow cytometry, and CLSM. In each assay, migrasomes were used as the positive control, while pre‑isolated sEVs and MVs served as negative controls to assess cross‑reactivity with other vesicle subtypes. In addition to vesicle‑based controls, the following FAM‐labeled oligonucleotide controls and magnetic nanomaterials‑based blank control were included in all experiments:

The selected aptamer Apt_B3: 5’‐ TCAAGTCACAGGTTCCAGGTGGCGGCAC

GGTTTCCCGGATTTCCGCGTTCGAGCTCCGGCATAGGCACTGACACGACACT‐3’.

Bare beads (Blank, NA‐coated MMPs without aptamer) to assess non‑specific adsorption to the bead surface.

Randomized library (RL) to evaluate background binding of unselected sequences: 5′‐ TCAAGTCACAGGTTCCAGGT‐N40‐ATAGGCACTGACACGACACT‐3′; where “40N” denotes a random arrangement of 40 nucleotides comprised of A, T, G, and C.

Scrambled‑sequence aptamer (scrApt_B3): 5′‑ TCAAGTCACAGGTTCCAGGTC

CGCCGTGCCAAAGGGCCTAAAGGCGCAAGCTCGAGGCCGATAGGCACTGACACGACACT‑3′.

CD63‑specific aptamer (Apt_CD63): 5′‑AAAAACACCCCACCTCGCTCCCGTG

ACACTAATGCT‑3′.(Permpoka et al. 2025)

2.11. Effect of Temperature on Aptamer Binding Ability

To investigate the impact of temperature shift on aptamer affinity, we conducted experiments at varying reaction temperatures. FAM‐labeled aptamers, with a final concentration of 500 nM, were incubated with MCP‐anchored migrasomes at temperatures of 4, 25 and 37°C. After washing and resuspending the migrasomes in PBS, they were subjected to flow cytometry. FAM‐labeled primary random ssDNA libraries were used as a control to assess nonspecific adsorption, while magnetic nanomaterials served as a blank control.

2.12. Human Materials

Peripheral blood samples were obtained from 40 healthy donors and 53 patients diagnosed with lung adenocarcinoma (LUAD) at the West China Hospital of Sichuan University. All patients included in this study were newly diagnosed, treatment‑naïve, and confirmed as LUAD (T1N0M0) by pathological biopsy, with no prior history of chemotherapy, radiotherapy, or surgical resection before sample collection. Patients with a history of other malignancies, acute or chronic inflammatory diseases, autoimmune disorders, or infectious diseases were excluded. The healthy controls (HCs) group consisted of individuals who underwent routine health examination at the same institution and were confirmed to be free of malignancy, acute or chronic infection, inflammatory conditions, hematological disorders, or any history of cancer. All HCs were age‑ and sex‑matched to the LUAD patient cohort to the extent possible. Informed consent was obtained from all participants, and approval for the clinical study was granted by the Ethics Committee of West China Hospital of Sichuan University (Approval number 2023 Review, nos. 4 and 1546). Peripheral blood was collected into EDTA‑anticoagulant tubes, and plasma was separated by centrifugation at 3500 rpm for 10 min at 4°C within 2 h of collection, and subsequently used for migrasome extraction.

2.13. Feasibility Verification of MAP for Migrasome Capture

First, the purified migrasome model was collected from cell suspension through DGC. The assembled MAP was resuspended in 1 mL PBS containing the model migrasomes and then incubated at 4°C for 1 hour with gentle rotation. After washing twice with PBS, the beads were resuspended in PBS and analyzed for zeta potential shift of the MAP before and after capture. TEM and SEM were utilized to investigate the morphology of the captured substances, while western blot was employed to assess the expression of marker proteins. Furthermore, MAP was incubated directly with fresh plasma samples, followed by protein extraction after two washes. The applicability of MAP in biological fluid samples containing complex components was verified using western blot.

2.14. Optimisation of Experimental Parameters of MAP

First, the quantity of MAP was optimized for each 2 mL of plasma, corresponding to 50, 150, 300, 450, 600, and 750 µg. Subsequently, the incubation time was optimized based on 5, 10, 15, 30, 45 and 60 min under the optimal feeding ratio. The total protein content of the captured migrasomes was quantified following the instructions provided by the Micro BCA Protein Assay Kit. The capture efficiency was calculated using the formula: $\mathrm{Capture}\mathrm{efficiency}(\%)=\frac{\mathrm{C}1}{\mathrm{C}0}\times 100\%$, where C1 denotes the concentration of captured migrasome protein and C0 denotes the concentration of input migrasome protein. Ultimately, the optimal experimental parameters were selected for subsequent experiments.

2.15. Platelet Activation in Vitro

Platelet‐rich plasma (PRP) was prepared by centrifuging PPACK‐supplemented EDTA‐anticoagulated whole blood at 150 g for 10 min at room temperature. The upper PRP fraction was collected and divided into two aliquots. Thrombin (T4648; Sigma–Aldrich; 1 U ml⁻¹) or PBS was added to each aliquot and incubated with gentle mixing for 30 min at room temperature. Following activation, each sample was subjected to three parallel analyses. For morphological assessment, samples were centrifuged at 10,000 g for 15 min, and the resulting pellets were fixed with 2.5% glutaraldehyde and subsequently examined by TEM. The remaining portions were used for migrasome isolation via MAP or DGC. Isolated migrasomes were stained with PE–anti‐CD62P, APC–anti‐CD41, and wheat germ agglutinin (WGA) for 30 min at room temperature in the dark, washed with PBS, and analyzed by flow cytometry and CLSM.

2.16. Quantitative Proteomic and Bioinformatic Analysis

Label‐free quantitative proteomics combined with bioinformatics analysis was performed on two independent cohorts. In the first, four independent plasma samples were each divided into two aliquots for migrasome enrichment by DGC and the MAP system to compare methodological performance. In the second, migrasomes were isolated using the MAP system from 16 early‑stage LUAD patients and 10 HCs to identify potential biomarkers.

For proteomic sample preparation, frozen samples were thawed on ice and subjected to three cycles of sonication using a high‐intensity ultrasonic processor (Scientz) in lysis buffer. A 2 µL aliquot of each sample was taken for SDS‑PAGE quality control. Protein concentration was determined, and equal amount of protein from each sample were adjusted to a consistent volume with lysis buffer. Proteins were reduced with 5 mM dithiothreitol at 56°C for 30 min, alkylated with 11 mM iodoacetamide at 25°C in the dark for 15 min, and then diluted with 200 mM tetraethyl ammonium bromide (TEAB) to a final volume of 100 µL. Trypsin was added at a 1:50 (trypsin: protein, w/w) ratio for the first digestion at 37°C overnight, followed by a second digestion with trypsin at 1:100 (w/w) ratio for an additional 4 h. The resulting peptides were desalted using a C18 solid‐phase extraction column. For each sample, an equal amount of peptide (4 µg) was loaded for mass spectrometry analysis. Peptides were dissolved in solvent A (0.1% formic acid and 2% acetonitrile in water) and separated using an EASY‐nLC 1200 UPLC system (ThermoFisher Scientific). Mass spectrometry analysis was performed on an Orbitrap Exploris 480 instrument (ThermoFisher Scientific) operating in data‐dependent acquisition (DDA) mode. The raw date was processed using Proteome Discoverer (v2.4.1.15) with a decoy database for false discovery rate (FDR) estimation. Search parameters included: trypsin digestion allowing up to two missed cleavages; minimum peptide length of six residues; precursor mass tolerance of 10 ppm; fragment mass tolerance of 0.02 Da. Carbamidomethylation (C) was set as a fixed modification, and oxidation (M), acetylation (N‐terminus), and methionine loss variants as variable modifications. FDR thresholds for protein, peptide, and PSM identifications were set to <1%. Differential expression was defined as fold change >1.5 with adjusted p < 0.05.

Gene Ontology (GO) annotation of the identified proteins was derived from the UniProt‐GOA database (www.ebi.ac.uk/GOA/), and pathways annotation was performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (www.kegg.jp/kegg/). Based on the proteomic findings, the expression levels of two candidate proteins, LAMC1 and ANKFY1, were further validated in an independent validation cohort using enzyme‐linked immunosorbent assay (ELISA).

2.17. Statistical Analysis

Statistical analyses were conducted using GraphPad Prism 8.4.0 software (GraphPad Software, Inc., CA) and SPSS 22.0. Flowjo version 10.6.2 (FlowJo, LLC., USA) was employed for the analysis of flow cytometry data. Data were obtained from a minimum of three independent experiments. Continuous variables that conformed to a normal distribution were presented as mean (X) ± standard deviation (SD). Statistical significance was assessed using an unpaired two‐tailed Student's t test.

3. Results

3.1. Preparation and Characterisation of MCP and Purified EVs

To enable efficient migrasome isolation, we designed and synthesized a MCP based on lipid insertion‐mediated surface functionalization. The probe was constructed by conjugating a biotinylated lipid–PEG complex onto NA‐coated MMPs. The LP structure comprises three key elements: a DSPE lipid tail for insertion into the migrasome membrane, a PEG spacer to enhance solubility and minimize non‐specific interactions, and a terminal biotin group for stable linkage to the NA‐coated MMPs.

The magnetic core was first prepared via hydrothermal synthesis, yielding Fe₃O₄ nanoparticles with an average diameter of approximately 400 nm (Figure S1A) and exhibited strong superparamagnetic behavior (≈ 70 emu g⁻¹, Figure S2). Sequential surface modifications were then performed to introduce the NA coating. SEM (Figure 1A and Figure S1A–C) and TEM (Figure 1B and Figure S1D–F) confirmed that the spherical morphology and monodispersity were well maintained throughout the coating steps (Fe₃O₄, Fe₃O₄@SiO₂ nanospheres, Fe₃O₄@SiO₂‐NH₂ nanospheres and Fe₃O₄@SiO₂‐NA nanospheres) A uniform SiO₂ shell of about 50 nm thickness was observed around the magnetic core after silica coating (Figure S1E). Surface charge analysis via zeta potential measurements further validated the successful stepwise functionalization: the potential shifted from –9.9 mV (Fe₃O₄) to –22.53 mV (Fe₃O₄@SiO₂ nanospheres), then to +36.27 mV after amine modification (Fe₃O₄@SiO₂‐NH₂ nanospheres), and finally to –11.87 mV following NA and LP conjugation, confirming the assembly of the complete MCP (Figure 1C).

FIGURE 1.

Characterisation of the materials and the model EVs derived from L929 cells. (A, B) SEM (A) and TEM (B) images of NA‐coated magnetic microparticles (MMPs). (C) Zeta potentials of synthesized materials. Data are presented as mean ± SD (n = 3). (D) TEM images of small EVs. (E) Size distribution of small EVs measured by NTA. (F) CLSM image showing L929 cells stably expressing TSPAN4‐mCherry cultured on fibronectin‐coated dishes, along with the migrasomes produced by these cells. (G) Western blot analysis of density gradient fractions from L929 cells using antibodies against integrin α5, NDST1, PIGK, Alix, and Tsg101. (H, I) TEM (H) and SEM (I) images of migrasomes isolated from fraction 6.

To evaluate the capture capability of MCP, high‐purity migrasomes were required. Given the established role of tetraspanins in migrasome biogenesis (Huang et al. 2019), we employed DGC to isolate migrasomes from L929 cells stably expressing TSPAN4‐mCherry (Figure 1F). Western blotting analysis of gradient fractions revealed identified migrasome markers and integrin α5 predominantly concentrated in fraction 5–7 (Figure 1G) (Zhao et al. 2019). Morphological assessment by SEM showed intact migrasomes with associated retraction fibers (Figure 1I), while TEM images displayed the characteristic “pomegranate‐like” ultrastructure, with diameters around 500 nm (Figure 1H). In parallel, sEVs were isolated by ultracentrifugation and exhibited the expected saucer‐shape structure morphology under TEM (Figure 1D), with a size distribution of 30–150 nm as quantified by NTA (Figure 1E).

Together, these results confirm the successful fabrication of MCP with well‐defined physicochemical properties and the preparation of morphologically and biochemically validated migrasome and sEVs, providing a solid foundation for subsequent aptamer screening.

3.2. Functional Validation and Optimisation of MCP for Migrasome Capture

To validate the specific migrasome‐binding capability of the MCP, we compared its capture performance against control (NA‐coated MMPs). Purified migrasomes expressing TSPAN4‑mCherry were incubated separately with MCP or control, followed by fluorescence imaging. As shown in Figure S3, MCP‐treated samples displayed distinct red fluorescence signals colocalized with the MCP, whereas no such signal was detected for NA‑coated MMPs, indicating selective migrasome capture mediated by the lipid‑anchoring design.

We further confirmed the lipid insertion mechanism by employing FITC‐labeled NA (FITC‐NA). When migrasomes pre‑labeled with the lipid component were incubated with FITC‑NA, clear colocalization of red (mCherry) and green (FITC) fluorescence was observed, confirming successful biotin–NA linkage via the inserted LP (Figure S4). In contrast, control groups without lipid labeling showed no fluorescence overlap. These results collectively support that MCP captures migrasomes through specific insertion of the DSPE lipid tail into the vesicle membrane, rather than through nonspecific adsorption.

To achieve efficient and reproducible capture, we systematically optimized key experimental parameters. First, the amount of LP conjugated to the magnetic beads was evaluated. Capture efficiency increased with LP concentration up to 1 µmol, beyond which no significant improvement was observed (Figure 2A). Subsequently, the conjugation time between LP and NA‑coated MMPs was examined. The capture efficiency rose steadily over 30 min and then plateaued, indicating complete coupling within this period (Figure 2B).

FIGURE 2.

Optimisation of experimental conditions. (A) Amount of labelling probe (LP). (B) Incubation time of LP with Fe₃O₄@SiO₂‐NA nanospheres. (C) Amount of magnetic capture probe (MCP). (D) Incubation time of MCP with migrasomes. Error bars represent mean ± s.e.m. (n = 3). (E–F) SEM images of migrasomes labeled with 100 µg (E) or 600 µg (F) MCP. Red and black arrows indicate migrasome structures and MCP on their surface, respectively.

The amount of MCP used per capture reaction was also critical. As shown in Figure 2C, capture efficiency peaked at 87.8% when 600 µg of MCP was applied, with a decline observed at higher amounts. To visualize this effect, SEM imaging was performed on migrasomes captured with either 100 µg or 600 µg of MCP. At the lower amount, MCP particles were sparsely distributed on the migrasome surface (Figure 2E) whereas at 600 µg, the vesicle surface was uniformly and densely coated with MCP (Figure 2F). The reduction in efficiency beyond this point likely results from particle aggregation and sedimentation during magnetic separation, which reduces accessible binding sites.

Finally, we optimized the incubation time between MCP and migrasome. Capture efficiency increased from approximately 60% at 5 min to about 80% at 30 min, after which it stabilized (Figure 2D). Based on these findings, the following optimized conditions were established for all subsequent assays: 1 µmol LP, 30 min conjugation time, 600 µg MCP er reaction, and 30 min MCP–migrasome incubation.

3.3. Selection and Characterisation of Migrasome‑Specific Aptamers via Mag‑SELEX

To obtain aptamers that selectively recognize migrasomes, we employed a Mag‑SELEX strategy. In this approach, purified migrasomes were used as the positive, while sEVs served as counter‑selection agents to deplete sequences with cross‑reactivity. Each round of selection involved PCR amplification of the enriched libraries using FAM‑labeled forward primers and biotin‑labeled reverse primers. Prior to initiating the screening process, key PCR parameters—including annealing temperature, template amount, and cycles number—were systematically optimized to ensure efficient and specific amplification (Figures S5, S6 and Table S1).

Selection stringency was progressively increased over successive rounds to drive the enrichment of high‑affinity binders. This was achieved by gradually reducing the migrasome concentration, shortening the incubation time with library, extending the counter‐selection step, and intensifying the washing intensity (Table S2). The enrichment process was monitored by flow cytometry using FAM‐labeled sub‐libraries. As shown in Figure 3A, the fluorescence relative intensity (FRI) increased steadily over the first six rounds, indicating successful enrichment of migrasome‑binding sequences. The FRI plateaued after the eighth round, suggesting that the library had converged; therefore, the eighth‑round products were subjected to further high‐throughput sequencing.

FIGURE 3.

Screening and characterisation of candidate aptamer Apt_B3. (A) Flow cytometry analysis of enriched ssDNA pools from Rounds 6 and 8 showing binding to migrasomes during SELEX progression. Bare beads (Blank) and the primary random library (RL) were served as negative controls. (B) Predicted secondary structure of Apt_B3. (C) Western blot analysis of migrasomes, sEVs and MVs following Apt_B3 pull‑down using antibodies against integrin α5, PIGK, CPQ, Alix, CD63, and ARF6. Controls included Blank, RL, a scrambled sequence (scrApt_B3), and a CD63‑specific aptamer (Apt_CD63). (D) Flow cytometry‑based binding specificity of Apt_B3 to migrasomes, sEVs and MVs, with the same negative controls as in (C). (E) CLSM images of Apt_B3 binding to migrasomes, sEVs and MVs. Negative controls were identical to those in (C). (F) Temperature‑dependent binding of Apt_B3 to migrasomes assessed by flow cytometry. (G) Determination of the binding affinity of Apt_B3 for migrasomes.

Among the candidate sequences identified, aptamer_B3 (Apt_B3) exhibited the strongest binding signal. Its predicted secondary structure, analyzed by Mfold, formed stable stem‐loop motifs with a free energy of −5.186 kcal·mol⁻¹ (Figure 3B). To comprehensively evaluate the specificity of Apt_B3, we performed a series of orthogonal assays. Western blot analysis following aptamer pull‑down revealed that Apt_B3 selectively enriched migrasome‑specific markers (integrin α5, PIGK, CPQ), while showing no detectable pull‑down of canonical sEVs markers (Alix, CD63) or the MVs marker ARF6 (Figure 3C). Importantly, control experiments using blank, the RL, scrApt_B3, and Apt_CD63 all yielded negative results, confirming that the observed binding is specific to Apt_B3 and not attributable to nonspecific adsorption or sequence‑independent interactions. Flow cytometry‑based binding assays further demonstrated that Apt_B3 exhibited strong fluorescence signal exclusively with migrasomes, with minimal binding to sEVs or MVs (Figure 3D). This high selectivity was visually corroborated by CLSM imaging, which showed clear colocalization of FAM‑labeled Apt_B3 with migrasomes, while no appreciable signal was observed on sEVs or MVs under identical conditions (Figure 3E). Together, these multi‑methodological validations consistently demonstrate that Apt_B3 possesses outstanding specificity and selectivity for migrasomes over other EVs subtypes.

The binding affinity of Apt_B3 was quantitatively assessed by incubating bead‑captured migrasomes with increasing concentrations of Apt_B3 (10–750 nM) Nonlinear regression analysis yielded a Kd of 251.9 nM, confirming high affinity interaction (Figure 3F). Temperature‑dependent binding assays revealed that Apt_B3 maintained robust recognition at 4°C and 25°C, with a moderate decrease in fluorescence intensity at 37°C, indicating preserved but slightly reduced binding under physiological temperature (Figure 3G).

These results collectively establish Apt_B3 as a promising, specific, and affinity‑matured aptamer for migrasome recognition under both experimental and biologically relevant conditions.

3.4. Development and Analytical Validation of the MAP System

We next constructed a MAP system by conjugating biotin‐labeled Apt_B3 onto the surface of NA‐coated MMPs for specific migrasome isolation. Successful functionalization showed that was verified through physicochemical characterisation. Dynamic light scattering Apt_B3 conjugation did not significantly alter nanoparticle size (Figure 4A). In contrast, the zeta potential shifted from −8.78 ± 0.52 mV (NA‐MMPs) to −20.1 ± 0.56 mV (MAP), consistent with the addition of negatively charged oligonucleotides (Figure 4B). Furthermore, CLSM imaging of FAM‑labeled Apt_B3 confirmed specific fluorescence signals on MAP particles, whereas control particles (Fe₃O₄@SiO₂‐NH₂ and NA‐MMPs alone) showed no detectable fluorescence after washing (Figure S7), confirming successful aptamer immobilization.

FIGURE 4.

Construction and validation of the MAP system. (A) Size distribution and (B) Zeta potential of Fe₃O₄@SiO₂‐NA nanospheres and MAP. All bars represent means ± SD (n = 3). (C) Western blot analysis of migrasomes purified by MAP compared with DGC, using antibodies against indicated markers. (D) Representative (a) SEM and (b) TEM images of migrasomes captured by the MAP system from cellular samples, with red arrows indicating migrasomes and black arrows indicating surface‐bound MAP. (E) TEM images of plasma migrasomes isolated by (a) MAP and (b) DGC; red and black arrows mark migrasomes and surface MAP, respectively. (F) Western blot of migrasomes isolated from plasma of nine individuals using MAP. The pellet was analyzed by western blot using antibodies against indicated proteins. (G) Recovery efficiency of the MAP system over different separation time with a commercial magnetic scaffold (inset: MAP suspension before and after 60 s of magnetic isolation). (H) Capture efficiency of different amounts of MAP incubated with migrasomes. (I) Capture efficiency of 300 µg MAP incubated with migrasomes for varying durations. (J) Flow cytometric analysis of migrasome levels enriched by MAP or DGC from resting or thrombin‑activated PRP. Blank and RL were included as negative controls. (K) Flow cytometric analysis of surface expression of the platelet marker CD41 and the activated platelet marker CD62P on migrasomes enriched by MAP or DGC from resting versus thrombin‑activated PRP. (L) CLSM images showing migrasome level and surface expression of CD41 and CD62P following enrichment by MAP or DGC from resting versus thrombin‑activated PRP.

We then assessed the capture specificity of the MAP system using purified cellular migrasomes. Western blot assay revealed that migrasome markers (NDST1, PIGK and integrin α5) were clearly enriched by both MAP and DGC, while markers of sEVs (ALIX and CD63) and MVs (ARF6 andANXA1) were prominent in DGC isolates but were absent or barely detectable in MAP‑enriched samples (Figure 4C). Morphological examination by SEM revealed that MAP‑captured migrasomes retained spherical morphology with diameters around 1 µm, and magnetic nanoparticles were uniformly distributed on the vesicle surface (Figure 4 Da). TEM further confirmed the characteristic ultrastructure of intact migrasomes isolated by MAP (Figure 4Db).

The applicability of the MAP system to complex biofluids was evaluated using clinical plasma samples (n = 9). Western blot analysis of MAP‑enriched fractions from individual donors consistently showed strong signals for migrasome markers, with minimal detection of sEVs or MVs contaminants (Figure 4F), confirming robust specificity in a clinically relevant matrix. Morphologically, plasma‑derived migrasomes isolated by MAP exhibited well‑preserved ultrastructure comparable to those obtained by DGC (Figure 4Ea, Eb).

We further optimized the operational parameters of the MAP system for plasma samples. Capture efficiency reached a maximum at 300 µg of MAP per 2 mL of plasma (Figure 4H) and plateaued within 15 min of incubation (Figure 4I). The system also demonstrated excellent magnetic responsiveness, with over 90% recovery achieved within 60 s using a commercial magnetic scaffold (Figure 4G).

To systematically evaluate whether the MAP platform co‑isolates platelet‑derived particles—particularly under conditions of platelet activation that elevate vesicle release—we performed a controlled comparison using PRP under resting and thrombin‑activated states, with DGC as a methodological control. Platelet activation was confirmed by TEM, which showed the expected morphological shift from discoid resting platelets to pseudopod‑spreading activated platelets (Figure S8). We first examined the impact of platelet activation on the capture of glycoprotein‑positive vesicles using WGA, a general glycoprotein stain. As shown in Figure 4J, the WGA‑positive population captured by MAP remained stable regardless of activation state. In contrast, upon platelet activation, DGC‑isolated samples showed a substantial increase in WGA signal. This indicates that DGC, unlike MAP, co‑isolates abundant glycoprotein‑carrying particles, likely platelet‑derived vesicles, which MAP dose not selectively capture. We next directly analyzed the expression of platelet‑specific markers on vesicles isolated by both methods. Flow cytometry revealed that the percentages of CD41‑ and CD62P‑positive populations in MAP isolates did not increase significantly following activation, whereas both markers were markedly elevated in DGC isolates from activated PRP (Figure 4K). CLSM corroborated these findings. In platelet‐activated samples, Apt_B3‑captured migrasomes showed minimal co‑localization with CD41 or CD62P, whereas DGC‐enriched fractions exhibited extensive overlap of both markers with the isolates (Figure 4L).

We next performed a quantitative proteomic comparison to comprehensively assess the performance of MAP relative to DGC using four independent plasma samples (Figure 5A). Both methods demonstrated high inter‐sample consistency, with 93.1% of proteins commonly identified across all four MAP‐isolated samples and 89.7% across all four DGC‐isolated samples (Figure 5B). From the four samples processed by each method, 17,409 and 16,854 proteins were identified by MAP and DGC, respectively. The overlapping proteins between the two methods accounted for 91.9% of the MAP identifications and 94.9% of the DGC identifications (Figure 5C). Collectively, these results demonstrate that MAP achieves high reproducibility and reliability, with protein identification coverage comparable to that of the established DGC method.

FIGURE 5.

Quantitative proteomic analysis of migrasome proteins enriched by MAP and DGC. (A) Workflow for evaluating two migrasome isolation strategies from plasma. (B) Venn diagram showing the numbers of shared and unique proteins identified in migrasomes isolated from four clinical plasma samples using MAP or DGC. (C) Venn diagram showing the number of shared and unique proteins identified in migrasomes enriched by MAP and DGC. and 22 previously reported migrasome marker proteins were detected in samples isolated by both methods. (D) Heatmap showing expression levels of migrasome, sEV, and MVs marker proteins across eight migrasome samples enriched by DGC or MAP. (E) UpSet plot showing the overlap of proteins identified by DGC and MAP methods. Subset sizes (vertical bars) and intersection composition (lower matrix) illustrate the degree of commonality and method‐specific protein profiles across eight plasma‐derived migrasome samples.

Direct comparison of vesicle subtype‐specific markers provided clear evidence for the enhanced selectivity of MAP (Figure 5D). While both methods robustly detected migrasome markers (e.g., NDST1, EOGT, PIGK, CPQ, TSPAN9), canonical sEVs markers (Alix, CD9, TSG101) and MVs markers (ANXA1, ARF6) (Jiao et al. 2024) were consistently lower in MAP isolates, with ARF6 completely absent. Furthermore, 22 previously reported migrasome‐associated proteins, including structural components (e.g., TUBB, KIF5B), membrane receptors (e.g., ITGA5, ITGB1, CSF1R), and vesicle trafficking regulators (e.g., RAB35, RAB10), were successfully identified by both the MAP and DGC methods (Figure 5C). Quantitative comparison of relative expression abundances further demonstrated the specific enrichment of migrasomes by MAP. While the marker composition profiles (Figure 5D) illustrated qualitative differences in vesicle subtype markers between the two methods, quantitative analysis revealed that the levels of key migrasome markers NDST1, CPQ, and PIGK showed no statistically significant differences between MAP and DGC, whereas EOGT was significantly higher in MAP isolates (Figure S9). This lack of statistical difference for most core migrasome markers is likely attributable to multiple factors, including inherent biological variability in migrasome marker expression and the challenges associated with quantifying low‐abundance proteins, where increased inter‐sample variation can reduce statistical power. Importantly, the abundances of highly abundant plasma proteins, including IGHM, FGG, FGB, and FGA, were comparable between the two methods, and albumin was not detected in either MAP or DGC isolates, indicating that both methods effectively exclude this highly abundant contaminant (Table S4). These results confirm that MAP achieves specific enrichment of migrasomes without introducing bias toward non‐specific plasma proteins.

To further assess the reproducibility of the MAP system, we evaluated its performance across diverse clinical sample types. Western blot analysis confirmed consistent detection of established migrasome markers, including NDST1, CPQ, PIGK, and EOGT, in migrasomes isolated by MAP from multiple clinical sample sources such as serum, plasma, hydrothorax, and ascites, but not from cerebrospinal fluid (Figure S10A); the lack of detection likely reflects the low abundance of migrasomes in cerebrospinal fluid. We next analyzed the reproducibility of MAP between two independent migrasome extraction batches from plasma. Batch‑1 consisted of two plasma samples (MAP _a, MAP _b), and Batch‑2 comprised four plasma samples (MAP _1, 2, 3, 4). Over 93% of the identifications in Batch‐1 and more than 91% in Batch‐2 were shared within the intersecting protein set, revealing a substantial common proteome and a high degree of overlap between the two batches, as demonstrated by Venn diagram and UpSet plot analyses (Figure S10B). Key migrasome markers including NDST1, CPQ, PIGK, and EOGT were consistently identified within this high‐overlap core, underscoring the method's high inter‑batch reproducibility in capturing migrasome‑associated proteins.

In addition to its specificity, the MAP system offered a substantial reduction in processing time, approximately 40‑fold faster than DGC, while maintaining comparable or superior selectivity (Table S3). Collectively, these data demonstrate that the MAP system provides a rapid, specific, and efficient method for migrasome isolation from various clinical samples.

3.5. Plasma Migrasomes Serve as Diagnostic Biomarkers for Early‐Stage LUAD

Having established a robust and specific method for migrasome isolation, we next investigated whether plasma migrasomes could reflect disease state and serve as liquid biopsy biomarkers for early‐stage LUAD. To explore this, we performed an unbiased proteomic analysis of migrasomes isolated by MAP from the plasma of a discovery cohort comprising early‐stage LUAD patients (n = 16) and HCs (n = 10). Label‐free LC‑MS/MS analysis quantified 3305 proteins out of 3580 identified. As an initial assessment of global alterations, principal component analysis (PCA) of the migrasome proteome revealed a moderate separation between early‐stage LUAD patients and HCs, indicating that disease‐associated proteomic signatures are detectable even at an early stage (Figure S12A). To dissect these signatures, we next analyzed the composition of the migrasome proteome in detail. A conserved set of 1162 proteins were found to be consistently present across all 26 individuals, suggesting a conserved migrasome architecture (Figure S11A). The complex overlap of protein profiles across all individuals was further detailed by an UpSet plot (Figure S11B).

Comparative analysis of the migrasome proteome identified significant alterations in LUAD. A total of 693 DEPs were identified in LUAD patients compared with HCs, including 404 up‐regulated and 289 down‐regulated proteins (Figure 6A, |fold change| >1.5, p < 0.05). Unsupervised hierarchical clustering of these proteins effectively segregated the samples by disease status, visually reinforcing the distinct molecular signature of LUAD‐derived migrasomes (Figure 6B). Among the most significantly altered proteins, compared with the HC group, the LUAD group exhibited increased levels of LAMC1, ANKFY1 and FN1, and decreased levels of ENO2, PRDX6 and RNF123.

FIGURE 6.

Identification of potential diagnostic biomarkers for lung adenocarcinoma (LUAD). (A) Volcano plot and (B) hierarchical clustering heatmap display proteome‐wide differentially expressed proteins between LUAD patients and healthy controls. (C) GO and (D) KEGG enrichment analyses identify significantly associated biological processes and pathways. (E−F) Validation confirms elevated levels of two migrasome‐derived proteins (LAMC1 and ANKFY1) in LUAD patients (****p < 0.0001). (G) ROC analysis demonstrates superior diagnostic performance of these migrasome markers over conventional CEA, particularly in a combined model.

To elucidate the biological implications of the altered protein profiles, we performed functional enrichment analyses. GO analysis revealed that the DEPs were broadly involved in diverse biological processes, including inflammatory responses, immune activation, and metabolic processes (Figure 6C). Additionally, KEGG pathway analysis not only highlighted their distribution across major functional categories such as signal transduction, transport and the immune system but, more importantly, confirmed a pronounced enrichment in key oncogenic pathways, including the PI3K‐Akt and NF‐kappa B signaling pathway (Figure 6D). To explore the interconnections among these proteins, we constructed a protein‐protein interaction (PPI) network based on key DEPs (e.g., ANKFY1, LAMC1, and AKT1), which revealed functionally interconnected modules aligning with enriched pathways (Figure S12B). Furthermore, an independent analysis restricted to low‐abundance proteins revealed a similar functional landscape, with consistent enrichment in immune‐related processes and oncogenic signaling pathways (Figures S12C, D). Collectively, these results indicate that LUAD‐derived migrasomes carry a unique repertoire of molecular cargo, potentially facilitating intercellular communication and signal transduction that contribute to malignant phenotypic conversion. Based on their significant fold‐change, central hub status in the PPI network, and documented relevance to cancer biology, LAMC1 and ANKFY1 were selected as lead candidates for subsequent diagnostic validation.

We next sought to validate the diagnostic potential of plasma migrasomes proteins as non‐invasive biomarkers for early‐stage LUAD. Using ELISA in an independent validation cohort (37 LUAD patients and 30 HCs), we confirmed that the levels of ANKFY1 and LAMC1 were significantly elevated in LUAD‐derived migrasomes compared to those from HCs (Figure 6E, F). Given the clinical challenge of early detection, we specifically evaluated the diagnostic performance of these migrasomal markers in early‐stage LUAD patients. Carcinoembryonic antigen (CEA) is a standard serological biomarker used in lung cancer management. Receiver operating characteristic (ROC) analysis demonstrated that both migrasomal markers outperformed CEA in discriminating LUAD patients from HCs. LAMC1 achieved an area under the curve (AUC) of 0.734 (sensitivity 90.6%, specificity 50.0%), and ANKFY1 achieved an AUC of 0.855 (sensitivity 96.9%, specificity 64.3%). Remarkably, a combined diagnostic model integrating both markers yielded an AUC of 0.940, with a sensitivity of 90.6% and a specificity of 92.9% (Figure 6G).

Collectively, these results establish migrasomal ANKFY1 and LAMC1 as promising non‐invasive biomarkers for early‐stage LUAD. With the combined model, demonstrating superior diagnostic performance and potential clinical utility for early detection.

4. Discussion

The functional significance of migrasomes has been increasingly recognized across diverse physiological and pathological contexts, including coagulation (Jiang et al. 2024), angiogenesis (Zhang et al. 2023), cellular homeostasis maintenance (Jiao et al. 2021), embryo development (Jiang et al. 2019). Their involvement has also been implicated in various diseases such as cancer (Deniz et al. 2023; Qin et al. 2022), viral infections (Liu et al. 2023; Lv and Zhang 2023; Zhang et al. 2022), cerebral amyloid angiopathy(Hu et al. 2023), and neurological disorders (Li et al. 2023; Schmidt‐Pogoda et al. 2018). Efficient enrichment methods are essential for further elucidation of their clinical role. However, isolating migrasomes from complex biological samples remain a methodological challenge, as existing approaches often lack the requisite specificity and are unsuitable for rapid processing of clinical sample volumes. This study presents an alternative strategy by developing a MAP system designed for the efficient migrasome enrichment from human plasma.

The core of this strategy lies in the selection of a migrasome‐specific DNA aptamer (Apt_B3) via a magnetic bead‐based SELEX (Mag‐SELEX) process. Unlike conventional SELEX targeting purified recombinant proteins, our approach employed intact migrasomes as the target. This introduces a distinct challenge: as organelles derived from the plasma membrane, migrasomes inherently share a substantial portion of their membrane proteome with their parental cells and with other EVs. Achieving high specificity against this background of structurally and compositionally related vesicle populations therefore required a stringent selection strategy. To address this, our Mag‐SELEX protocol implemented rigorous counter‐selection against sEVs, aiming to shift selection pressure towards epitopes or structural motifs that are uniquely presented or enriched on the migrasome surfaces. The resulting aptamer, Apt_B3, which demonstrates preferential binding to migrasomes, may therefore recognize conformational or combinatorial targets characteristic of migrasome organization. A shared limitation of such whole‐target SELEX approaches, however, is that the precise molecular identity of the aptamer's binding partner within the native membrane environment remains unknown—a challenge that future work to deconvolute Apt_B3's target will aim to address.

The integration of Apt_B3 with magnetic materials translates this molecular recognition into a practical tool. The most immediate advantages are operational, offering a contrast to conventional DGC by reducing isolation time from hours to 15 min and requiring only standard laboratory equipment (Ludwig et al. 2018; Patel et al. 2019). Beyond speed, comparative proteomic analysis revealed that the MAP system yielded a migrasome‐enriched fraction with a reduced background of common co‐isolated vesicle proteins compared to DGC. By significantly simplifying operations, the MAP system is more user‐friendly and cost‐effective, thereby enabling its application in clinical scenarios.

Leveraging this platform, we employed the MAP system to profile the proteome of plasma migrasomes in a pilot cohort of early‐stage of LUAD patients and HCs. This exploratory analysis revealed differential protein abundance, and two candidates were further validated in a separate cohort, highlighting the potential of migrasomes to carry disease‐associated molecular information. Beyond this proof‐of‐concept in LUAD, emerging research suggests that migrasomes may serve as promising indicators for other diseases. Realizing this potential, however, requires tools capable of isolating migrasomes from diverse clinical samples. The development of specific and efficient technologies such as the MAP system is therefore critical to advance the study of migrasomes across different sample types (e.g., cerebrospinal fluid, saliva, and tissues). Looking forward, the integration of the MAP system with advances detection platforms, such as electrochemical, optical, or biochip‐based biosensors, may enable the construction of sensitive and specific diagnostic assays for disease‐ associated migrasome markers, thereby contributing to future developments in precise disease diagnosis.

In conclusion, we have developed a proof‐of‐concept aptamer‐based method, MAP, for the rapid and selective isolation of migrasomes from human samples. By combining affinity‐based selectivity with operational simplicity, the MAP system addresses a critical technological gap in migrasome research. This tool is expected to facilitate not only the biological study of migrasomes but also their exploration as a reservoir of circulating biomarkers for liquid biopsy.

Author Contributions

Yixin Xu: methodology, conceptualization, writing – original draft. Lin Jiao: software, writing – review and editing. Qiangying Yi: conceptualization. Yuzuo Chen: visualization. Bei Cai: formal analysis. Junlong Zhang: software. Zhuochun Huang: investigation. Yao Luo: data curation. Yanjun Si: data curation. Yao Wu: resources. Binwu Ying: funding acquisition, supervision. Jie Chen: project administration, funding acquisition. Juan Zhou: conceptualization, validation, project administration.

Conflicts of 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.

Supporting information

Supporting Figure: jev270283‐Sup‐0001‐FigureS1.tiff

Supporting Figure: jev270283‐Sup‐0002‐FigureS2.jpg

Supporting Figure: jev270283‐Sup‐0003‐FigureS3.tiff

Supporting Figure: jev270283‐Sup‐0004‐FigureS4.tiff

Supporting Figure: jev270283‐Sup‐0005‐FigureS5.tiff

Supporting Figure: jev270283‐Sup‐0006‐FigureS6.jpg

Supporting Figure: jev270283‐Sup‐0007‐FigureS8.tif

Supporting Figure: jev270283‐Sup‐0008‐FigureS9.tiff

Supporting Figure: jev270283‐Sup‐0009‐FigureS10.tif

Supporting Figure: jev270283‐Sup‐0010‐FigureS11.tif

Supporting Figure: jev270283‐Sup‐0011‐FigureS12.tif

Supporting Table 4: jev270283‐Sup‐0012‐Tables4.xlsx

Supporting Table 5: jev270283‐Sup‐0013‐Tables5.xlsx

Supporting Information: jev270283‐Sup‐0014‐SuppMat.docx

Data Availability Statement

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD074521 (Proteomic Analysis of the Protein Composition of Migrasomes Enriched by Magnetic Aptamer Probe System and Density Gradient Centrifugation) and PXD074511 (Proteomic Screening Lung Adenocarcinoma Specific Migrasome Proteins). All other data supporting the findings of this study are available from the corresponding author on reasonable request.