Hydrogel-Based Extracellular Vesicle Isolation Method from Various Biological Solutions

Abstract

Extracellular vesicles (EVs), which are secreted by almost all living cells and play crucial roles in various physiological processes, have attracted increasing research interest. Numerous protocols for EV isolation, such as ultracentrifugation and polymer precipitation, have been developed; however, each conventional method has inherent strengths and limitations, particularly regarding purity, yield, and bias toward specific EV subpopulations. Herein, we present an EV isolation method, hydrogel adsorption separation (HAS), which uses a hydrogel scaffold for the selective and high-purity isolation of EVs from diverse biofluids. HAS leverages the affinity between membrane vesicles and palmitoyl groups on the hydrogel surface, enabling the isolation of less-biased EV subpopulations. EVs isolated using the HAS method were compatible with downstream analyses, including proteomics and flow cytometry, demonstrating their suitability for both basic and applied research. This approach is a practical and versatile tool for advancing the EV-based research and applications.

Introduction

Extracellular vesicles (EVs) are nanometer-scale lipid bilayer vesicles secreted by almost all living cells, including prokaryotes and eukaryotes. − These vesicles carry a diverse array of bioactive molecules, such as proteins, nucleic acids, sugars, and small metabolites. Although EVs were once considered cellular waste, they play pivotal roles as mediators of intercellular communication. Once released from donor cells, EVs travel through body fluids to reach local and distant recipient cells. Upon uptake, their molecular cargo modulates a wide range of physiological and pathological responses, including immune responses, tumor progression and metastasis, and neurodegenerative diseases. Elucidating the physiological functions of EVs is essential to advance our understanding of complex biological systems.

The clinical potential of EVs has attracted significant substantial attention. − Mesenchymal stem cell (MSC)-derived EVs have are effective in tissue regeneration. In addition to native EVs, engineered EVs are being developed to enhance cellular uptake and modifiability, either by direct drug loading or genetic engineering of parent cells. , Furthermore, the use of EVs as biomarkers for liquid biopsy is a promising diagnostic application. , The molecular heterogeneity of EVs in body fluids reflects the physiological or pathological state of their origin cells. Tumor-derived EVs display specific molecular signatures that can facilitate cancer diagnosis. ,− Therefore, the detection of these disease-specific EV markers in small volumes of blood or urine has potential as noninvasive diagnostics.

Basic and translational EV research require robust, simple, and efficient methods for isolating EVs from biological fluids. ,− Although numerous methods for EV isolation and purification such as ultracentrifugation (UC), polymer precipitation (PP), and affinity-based isolation methods have been developed, achieving high yield and purity while preserving EV integrity and minimizing processing time remains a substantial challenge. Affinity-based isolation methods targeting EV-specific biomarkers, such as tetraspanins can provide high EV purities; however, these approaches are inherently biased toward specific EV subpopulations defined by their target molecules. In contrast, less-biased methods that exploit the physical or physicochemical properties of EVs, such as density, particle size, surface charge, and hydration, often result in lower purity owing to the coisolation of contaminants. Therefore, there is a clear need for new protocols that enable unbiased isolation of EVs with improved purity.

In our preliminary studies, hydrogels could serve as effective scaffolds for EV adsorption. Hydrogels are widely used in biomedical applications, such as wound dressings, drug delivery systems, and tissue engineering. Polyacrylate hydrogels have marked water absorption capacity, enabling them to retain several hundred times their own weight. In our experiments, we used polyacrylate hydrogels to concentrate urinary EVs via water adsorption and found that EVs were readily adsorbed onto the hydrogel surface. These findings prompted us to develop a novel EV isolation strategy, called hydrogel adsorption separation (HAS).

Herein, we describe the design and preparation of a modified polyacrylate hydrogel for EV isolation using the HAS method and evaluate its performance in purifying EVs from various biological fluids. We demonstrated that the HAS method using a polyacrylate hydrogel yields intact EV solutions with minimal impurities and comprehensively characterized the properties of EVs isolated using this approach.

Results

Molecular Design and Preparation of Hydrogel for Selective EV Adsorption

Commercially available cross-linked polyacrylates were used to establish a proof-of-concept for EV isolation. When these polyacrylate beads were incubated with urine and subsequently treated with sodium chloride, EV molecular markers were detected using Western blotting. However, the resulting solution contained high levels of albumin contamination.

Selective adsorption of EVs onto a hydrogel surface is essential for high-purity EV isolation using the HAS method. The nonspecific adsorption of protein impurities, particularly albumin, one of the most abundant proteins in human body fluids, can significantly compromise the purity of EVs, as confirmed by our initial findings. ,

To minimize albumin binding, a neutralized cross-linked polyacrylate hydrogel was selected as the adsorption scaffold. At neutral pH (7.0), both polyacrylate and albumin are negatively charged (albumin pI: 5.0; polyacrylate pK _a: 4.5–5.0), , which should promote electrostatic repulsion and reduce nonspecific albumin adsorption.

However, because EVs also possess a negative surface charge, their adsorption onto the polyacrylate hydrogel can be hindered by electrostatic repulsion. To overcome this challenge, we focused on anchoring lipids to the EV membrane. − We hypothesized that the introduction of a hydrophobic palmitoyl group would facilitate the anchoring of EVs to the hydrogel surface. This modification enabled the reversible introduction of the palmitoyl group, thereby rendering the hydrogel surface suitable for EV-selective adsorption and dissociation.

Adsorption and Subsequent Dissociation of Vesicles onto the Hydrogel Surface

To validate the adsorption capacity of the polyacrylate hydrogel, liposomes were used as model particles (Figure ). Specifically, we used fluorescein-labeled liposomes, which closely mimic their size (∼100 nm) and negative surface potential. , The liposomes were suspended in phosphate-buffered saline (PBS) and incubated with a polyacrylate hydrogel. After the hydrogel swelled and was washed with saline to remove the unbound material, we examined the hydrogel using fluorescence microscopy (Figure A). Strong green fluorescence was observed on the hydrogel surface, indicating efficient adsorption of the liposomes. Correspondingly, the fluorescence intensity of the residual solution markedly decreased after swelling, further confirming the adsorption process (Figure B).

1.

Adsorption of liposomes and EVs onto the hydrogel surface. (A) Fluorescence microscopy images of hydrogels after incubation with a fluorescein-conjugated liposome solution: left panel, after hydrogel swelling, right panel, after addition of a high ionic strength solution to induce liposome dissociation. Emission filter: 525/50 nm, exposure time: 0.4 s, scale bar: 500 μm. (B) Recovery rates of liposomes in the flow-through fraction after hydrogel swelling (Flow Through) and in the elution fraction following addition of a high ionic strength solution (Elution). Recovery rates were determined by measuring the fluorescence intensity of fluorescein (excitation: 488 nm, emission: 520 nm) and accounting for the volume of each fraction, normalized to the fluorescence intensity of the input liposome solution (Input). (C) Stereomicroscope image of the hydrogel after swelling with an EV-containing sample (scale bar: 1 mm). (D) Field emission scanning electron microscopy (FE-SEM) image at 100× magnification, showing the rough surface morphology of the swollen hydrogel (scale bar: 100 μm). (E) FE-SEM image at 30,000× magnification. Arrows indicate particle-like structures (50–200 nm in diameter) identified as EVs adsorbed on the hydrogel surface (scale bar: 500 nm).

To induce dissociation of the adsorbed liposomes, we introduced a high-ionic-strength solution that disrupted the electrostatic interactions between the polyacrylate hydrogel and palmitoyl anchor. Following this treatment, fluorescence microscopy revealed a substantial reduction in fluorescence on the hydrogel surface (Figure A), whereas the elution solution containing the released liposomes exhibited strong fluorescence (Figure B). Collectively, these results demonstrate that the engineered hydrogel enables both efficient adsorption and controlled dissociation of liposomes, supporting its suitability for EV isolation using the HAS method.

The adsorption of EVs onto the hydrogel surface was further confirmed using field-emission scanning electron microscopy (FE-SEM), which allowed the observation of the hydrogel under hydrated conditions with high resolution. The hydrogel, which had absorbed the EV-containing sample and swelled to approximately 500 μm to 1 mm in diameter, exhibited a rough and uneven surface (Figure C,D). At a magnification of 30,000×, numerous particle-like structures with diameters ranging from 50 to 200 nm were adhered to the hydrogel surface (Figure E), consistent with the expected size of the EVs.

Together, these results suggest the potential for reversible adsorption and dissociation of vesicles, including liposomes and EVs.

EV Purification by the HAS Method

HAS consists of three main steps: (1) adsorption of EVs onto the hydrogel, (2) removal of residual impurities by washing, and (3) dissociation of EVs from the hydrogel (Figure S1). Human urine and conditioned medium from HEK293 cells were used for EV isolation using the HAS method. The isolated EVs were characterized by nanoparticle tracking analysis (NTA), which revealed a sharp size distribution ranging from approximately 50 to 200 nm (Figure S2A,B).

The yield of EVs purified from human urine and conditioned medium was evaluated using a CD9/CD63 sandwich ELISA (Table S1), which specifically detects EVs expressing both CD9 and CD63 surface markers. These results confirm that the HAS method enables the efficient and successful isolation of EVs.

Additionally, the presence of EV-specific miRNAs in the purified EV solutions was assessed using RT-qPCR. A 200-μL aliquot of each EV solution was processed using the Exosomal RNA Isolation Kit. Two miRNAs (hsa-miR-15a-5p and hsa-miR-22-3p) previously reported as exosomal or small EV (sEV) markers ,, were analyzed using TaqMan probe assays. Both miRNAs were successfully amplified from urinary and HEK293 cell-derived EVs (Table S2).

Comparison of EV Isolation Efficiency: HAS Method versus UC and Commercial Kits

To evaluate the efficiency of the HAS method relative to UC and other commercially available EV isolation techniques, we purified EVs from 1 mL urine samples collected from patients with chronic kidney disease (CKD) using four different methods: UC, polymer precipitation (PP), Tim4 affinity-based isolation kit (Tim4), and HAS.

Protein impurities and EV markers in the purified EV solutions were analyzed using SDS-PAGE with silver staining and Western blotting (Figure A). EVs isolated using UC and PP exhibited numerous intense protein bands, indicating substantial protein contamination, whereas EVs isolated using HAS exhibited only a few bands. Western blotting revealed clear CD9 bands in EVs isolated by UC, PP, and HAS, but not in those isolated by Tim4.

2.

EVs isolation from urine using several isolation methods. (A) SDS-PAGE (upper panel) and Western blotting (lower panel) analyses of urinary EVs isolated from a patients with CKD (CKD1) using the four different EV isolation methods: ultracentrifugation (UC), polymer precipitation (PP), Tim4 affinity, and the HAS method. The polyacrylamide gel was silver-stained and the CD9 marker was detected by Western blotting. (B) Quantification of CD9⁺CD63⁺ EVs in the purified solutions obtained by each method, as determined using sandwich ELISA. (C) Purity of the EV solution was calculated as the ratio of CD9⁺CD63⁺ EVs (pg) to total protein content (μg) in each sample.

The recovery yield of CD9⁺ CD63⁺ EVs was determined using a CD9/CD63 sandwich ELISA (Figure B and Table S3). Although the absolute amount of CD9⁺ CD63⁺ EVs varied among patients with CKD, the HAS method consistently yielded higher recovery compared with that of the other methods. EV purity was further evaluated by calculating the ratio of CD9⁺ CD63⁺ EVs to total protein content in the EV solutions (Figure C and Table S3). The HAS method produced significantly purer EV preparations than that of the UC and PP methods. The Tim4 method achieved the highest purity; however, this method uses the specific affinity between Tim4 and phosphatidylserine (PS), and thus, has a biased selectivity for PS-positive EV subpopulations.

EV Purification from Blood-Derived Samples

We evaluated the performance of the HAS method for isolating EVs from as little as 100 μL of serum or plasma. The particle characteristics of the recovered serum and plasma EVs were analyzed using NTA, which revealed a sharp size distribution ranging from 50 to 200 nm (Figure S2C,D). The particle concentrations were 1.75 × 10¹¹ particles/mL for serum and 1.48 × 10¹¹ particles/mL for plasma (Table S4), demonstrating that the HAS method enables efficient recovery of EVs from limited blood-derived samples.

The protein content of the purified EV solutions was analyzed using SDS-PAGE (Figure S2E). Notably, EVs isolated using HAS exhibited substantially lower levels of protein impurities than that of serum EVs isolated using UC, as visualized using silver staining (Figure S2F).

These results demonstrate that the HAS method enables the efficient and high-purity isolation of EVs from small volumes of blood-derived samples, highlighting its strong potential for EV-based diagnostics using a minimal sample volume.

Proteomic Analysis of Purified EVs

To assess the impact of different EV purification methods on the proteomic profile, we performed comprehensive proteomic analyses of EV solutions obtained from human serum using HAS, Tim4, and UC. We compared the identification of well-known EV proteins among different isolation methods (Table S5). EVs isolated using HAS and Tim4 showed robust detection of key EV markers listed in MISEV, such as CD9, CD63, TSG101, and HLA, whereas EVs obtained using UC showed lower identification rates for these biomarkers.

The advantages of high-purity isolation methods are reflected in the total number of proteins identified. The total number of proteins identified in the EV solutions purified by the HAS method was 610, which exceeded the number obtained using Tim4 and UC (Tim4: 476, UC: 417, Figure A). To evaluate the specificity of the EV-associated proteins, the identified proteins were cross-referenced using the Vesiclepedia database. Notably, the proportion of EV-specific proteins relative to the total protein content was highest in the HAS-isolated samples (87.2%), compared with that of Tim4 (84.7%) and UC (73.1%).

3.

Comparison of proteins identified in EV solutions obtained by HAS, Tim4, and UC. (A) Number of identified proteins in EV solutions (blue: EV proteins, orange: non-EV proteins). (B) Venn diagrams showing the complete set of detected proteins (left) and the EV protein subset (right) for each isolation method. (C) Correlation heatmap of identified proteins across EV solutions obtained by each method.

Venn diagrams were generated to visualize the overlap and uniqueness of the protein populations identified using each method (Figure B). HAS uniquely detected a greater number of proteins, the vast majority of which were classified as EV proteins (194 of 202 unique proteins). Furthermore, a correlation heatmap of the proteomic profiles (Figure C) revealed that the protein composition of EVs isolated by HAS was more similar to that of Tim4 (Pearson correlation coefficient: 0.68) than to UC (0.48), while the correlation between Tim4 and UC was 0.55. The similarity between HAS and Tim4 may reflect the higher purity of EVs in these samples, because high protein impurities can hinder the detection of low-abundance peptide signals in LC-MS/MS analysis. However, the relatively low correlation between HAS and Tim4 may also be influenced by the differences in the EV capture mechanism of each method.

Together, these findings suggest that the HAS method not only yields a higher number of EV-specific proteins but also enables broader detection of proteins compared with that of the other two methods.

Single EV Analysis Using Flow Cytometry

We performed single-particle flow cytometric analysis to characterize the subpopulations of purified EVs. Mouse serum EVs were isolated using both HAS and UC methods and subsequently stained with fluorophore-conjugated antibodies targeting specific EV markers. Flow cytometry was conducted using a BD FACSymphony A1 Cell Analyzer with BD Small Particle Detector (Becton, Dickinson and Company), using side scatter signal detector for small particles (SP SSC) (Figures and S3). The SP SSC histogram for EVs isolated using HAS showed a main signal population at lower intensities and another population at higher intensities (Figure A). The side-scattering intensity depends on the particle size, therefore, it is possible to estimate the size of the detected particles. We measured a standard polystyrene bead mixture (Figure B) and applied these parameters to Mie theory to generate a calibration curve for EV size determination , (Supporting Text).

4.

Flow cytometric analysis of purified mouse serum EVs isolated using the HAS method. (A) Signal intensity histogram of side scatter for small particles (SP SSC) from mouse serum EVs isolated using the HAS method. (B) Signal intensity histogram of standard polystyrene beads mixture (Megamix-Plus SSC). The signal intensity for 500 nm polystyrene beads reaches the detection maximum threshold of the SP SSC detector. (C) Plot of particle size versus SP SSC signal intensity. The calibration curve for polystyrene beads (blue) and theoretical side scattering intensity for EVs (orange) were analyzed using Mie theory. The red line indicates the signal saturation threshold for the SP SSC detector. Error bar: 2-fold of standard deviation. (D) Size distribution of mouse serum EVs isolated using the HAS method, measured using NTA. No particles with diameters between 300 and 550 nm were observed. (E) Two-dimensional plot of identified events showing CD9 fluorescence intensity versus SP SSC signal intensity. Particles were gated as small EVs (sEV), large EVs (lEV), and aggregates based on the predicted SP SSC intensity calculated using Mie theory. (F) Two-dimensional plots of CD9 and CD81 fluorescence intensity in each gated subpopulation (left panel: sEV, right panel: lEV).

Due to signal saturation of the SP SSC detector for 500 nm polystyrene beads, the calibration curve was fitted using three points (160, 200, and 240 nm) (Figures C and S4). To apply this calibration to the light scattering of EVs, the theoretical scattering intensity was calculated using the reported refractive index of EVs (1.427 ± 0.026). Although there was a region of decreasing signal between 350 and 500 nm, NTA analysis of HAS-isolated EVs showed no particles larger than 300 nm (Figure D). We classified the detected particles in the EVs solution into three subpopulations: small EVs (sEV; SP SSC < 1 × 10⁴), large EVs (lEV; SP SSC: 1–9 × 10⁴), and aggregates with a higher scattering intensity than that predicted from the EV refractive index (SP SSC: > 9 × 10⁴) (Figures E and S3A).

The sEV and lEV subpopulations were analyzed for CD9 and CD81 expression (Figures F and S3B). The majority of particles in the sEV and lEV subpopulations of HAS-isolated EVs were CD9⁺ and/or CD81⁺ EVs (sEV: 90.7%, lEV: 98.8%), which was a higher proportion than observed in EVs isolated using UC (sEV: 64.6%, lEV: 77.9%) (Table ).

1. Ratio of CD9/CD81-Positive Particles in Each Subpopulation from EV Solution Obtained by HAS and UC.

Method	EVs	CD9+ CD81+ (%)	CD9+ CD81– (%)	CD9– CD81+ (%)	CD9– CD81– (%)
HAS	sEV	46.6	31.3	12.7	9.3
	lEV	79.8	10.6	8.4	1.2
UC	sEV	10.0	32.4	12.1	45.4
	lEV	36.5	28.9	12.6	22.1

We also measured the ApoA1 marker levels to assess non-EV particle contamination. ApoA1 is a major component of high-density lipoprotein (HDL) and chylomicrons. These lipoproteins are micelle-like globules composed of lipid compounds that may interact with the palmitoyl groups of the polyacrylate hydrogel and potentially contribute to particle contamination. Analysis of the ApoA1⁺ subpopulation by CD9/CD81 expression revealed that about 91% of ApoA1⁺ particles were also CD9⁺ and/or CD81⁺, consistent with previous reports of ApoA1 presence in EVs or exosomes (Figure S5). ,

ApoA1-positive particles without EV markers (CD9 and CD81) were considered lipoprotein particles. However, the size of HDL (∼10 nm) is much smaller than the detection threshold of the SP SSC detector. Therefore, most of these ApoA1⁺ particles without EV markers are considered to be chylomicrons (75–1000 nm). Thus, we defined chylomicron particles as ApoA1-positive particles without EV markers (CD9 and CD81) (Table ). The proportion of chylomicron particles in HAS-purified EV solution (2.1%) was lower than that in UC-purified EV solution (7.0%). The proportion of other contaminating particles without EV markers or ApoA1 in HAS-purified EV solution (7.1%) was also lower than those in UC-purified solution (39.1%). These results indicate that the HAS method yields EV solutions with significantly lower particle contamination from chylomicrons and non-EV particles than that with UC.

2. Proportion of Each Subpopulation by Flow Cytometry Based on Particle Size, CD9 and CD81 Markers, and ApoA1 Marker in EV Solutions Purified Using HAS and UC.

EV marker	CD9+ and/or CD81+	CD9– CD81–
ApoA1	–	ApoA1+	ApoA1–
Subpopulation	EV (%)	Chylomicron (%)	Others (%)
HAS EVs	90.2	2.1	7.1
UC EVs	53.5	5.2	39.1

Cellular Uptake of EVs Isolated by HAS

To assess the functional integrity of EVs isolated using the HAS method, we conducted cellular uptake experiments using EVs derived from human MSCs (hMSCs) from the umbilical cord matrix (hMSC-UC) (Figure ). EVs were purified from hMSC-UC-conditioned medium using the HAS method.

5.

Cellular uptake of isolated MSC-derived EVs isolated using the HAS method in different cell types. (A) Internalization of EVs in dermal papilla cells. EV membranes were labeled with ExoSparkler EX01 (green) and cell nuclei were stained with DAPI (blue). (B) Internalization of EVs in skin cells. EV proteins were labeled with ExoSparkler EX05 (red), and cell nuclei and cytoskeleton were stained with DAPI (blue) and phalloidin (green).

Membranes of isolated MSC-EVs were labeled with ExoSparkler EX01 (Dojindo Laboratories) and incubated with cultured papilla cells. Fluorescence microscopy revealed distinct green fluorescent puncta within the cytoplasm, indicating successful internalization of EVs (Figure A). In a separate experiment, the proteins within the MSC-EVs were labeled with ExoSparkler EX05 (Dojindo Laboratories) and added to cultured skin cells. Red fluorescent dots were observed in the cytoplasm, further confirming EV uptake (Figure B).

These results demonstrate that EVs isolated using HAS retained their ability to be internalized by recipient cells, suggesting that their structural and functional integrity was preserved throughout the isolation and purification processes.

Discussion

Hydrogels have several advantages as EV isolation scaffolds: (1) high hydrophilicity, which reduces the irreversible adsorption of EVs and proteins; (2) chemical tunability of both the scaffold and surface; and (3) water absorption capacity, which enables sample condensation. In particular, the precise control of the hydrogel mesh size to be smaller than the EV diameter is critical for preventing EVs from becoming trapped within the gel matrix. The theoretical average mesh size of the polyacrylate hydrogel used in this study was approximately 30 nm, which is smaller than the minimum diameter of small EVs (50 nm). The surface entrapment of EVs is further enhanced by the incorporation of lipid anchors, such as palmitoyl groups, which are inserted into the lipid bilayer of vesicles via hydrophobic interactions. This strategy enables the HAS method to purify a broad range of vesicles, including liposomes, plant-derived EVs, and bacterial EVs.

HAS offers a novel conceptual framework to address the persistent trade-off between yield, purity, and subpopulation bias. EVs isolated from biological fluids using the HAS method exhibited higher purity compared with that of other less biased methods, such as UC and PP, although affinity-based highly biased methods, such as Tim4, still achieved the highest purity. Notably, the HAS method enables efficient EV isolation from small volumes of serum and plasma, facilitating downstream analyses, such as proteomics and flow cytometry. The ability to work with a limited sample volume is a significant advantage of EV-diagnostic applications.

The relatively lower purity of HAS-isolated EVs compared with that of the Tim4 method was primarily due to the nonspecific adsorption of abundant proteins onto the hydrogel surface. Proteomic analysis revealed that almost all of the top 10 proteins in HAS-isolated serum EVs were highly abundant, including albumin, apolipoproteins, coagulation-related proteins, and other serum-derived proteins (Table S6). The high concentrations of these protein impurities make it difficult to achieve the same level of purity as that achieved by affinity-based methods. Additional limitations of the HAS method include the requirement for a high-salt concentration during EVs elution and potential contamination by cationic or hydrophobic compounds. Although the salt concentration in the eluted solution is reduced compared with that of the input elution buffer owing to hydrogel shrinkage, it remained sufficiently high (∼4%) to interfere with certain immunoassays. Contaminants, such as amines or cell culture medium components (e.g., phenol red), generally do not affect most downstream analyses, but can interfere with specific assays (e.g., amine contamination might disrupt the BCA assay). In such cases, further purification by ultrafiltration and subsequent desalting using desalting columns, or dialysis without ultrafiltration are recommended.

Conclusion

We developed a novel hydrogel-based EV isolation system, HAS, which enables the efficient and high-purity isolation of EVs from a wide range of biofluids, including cell culture medium, urine, serum, and plasma. By incorporating a long alkyl chain as a lipid anchor, HAS demonstrates broad applicability to EVs from diverse sources, including mammalian plants and microorganisms. EVs isolated using HAS are compatible with various downstream analyses, ranging from standard techniques, such as NTA and ELISA to advanced applications, including proteomics and single-EV profiling by flow cytometry. Furthermore, the commercialization of the HAS system as an “Exorption” EV isolation kit provides a practical and versatile tool for both basic and clinical EV research. Future efforts will focus on optimizing the hydrogel matrix and elution conditions to further reduce the salt concentration and minimize small-molecule contamination, thereby expanding the use of the HAS system to an even broader spectrum of downstream applications.

Materials and Methods

Synthesis of Cross-Linked Polyacrylate

Chemicals were purchased from commercial suppliers and used without further purification. Acrylic acid (116.5 g) and deionized water (272.2 g) were added to a 1 L glass beaker and stirred until completely mixed. An aqueous solution of NaOH (48.5%, 96.1 g) was gradually added to the mixture, while cooling in an ice bath to achieve partial neutralization. Subsequently, 1,2-bis­(glycidyloxy)­ethane (3.0 g) was added and mixed thoroughly. The resulting mixture was cooled again to 4 °C, and an aqueous solution of K₂S₂O₈ (2.0 wt %, 9.3 g) was added and mixed well to prepare the starting monomer solution.

Cyclohexane (1120 mL) and sorbitan monostearate (7.1 g) were placed in a 2 L separable flask equipped with a stirrer and condenser, and heated with stirring the flask to 60 °C to dissolve the sorbitan monostearate. The solvent was then degassed by bubbling N₂ to reduce the O₂ below 0.1 ppm and the temperature was raised to 80 °C. The prepared monomer solution was added dropwise to the solvent under stirring for suspension polymerization. After the addition was complete, the reaction mixture was further stirred for 2 h to complete the polymerization reaction.

The resulting hydrogel suspension was filtered through a nylon mesh to remove cyclohexane. The hydrogel was dried in vacuo in a vacuum oven at 130 °C for 1 h to obtain the spherical polymer. The dried polymer was sieved to collect beads of the desired size and other polymers with sizes (yield: 62.5 g).

Preparation of Neutralized Polyacrylate

The synthesized polyacrylate (13.1 g) was placed in a stainless-steel vat and neutralized by spraying it with a K₂CO₃ solution (50 wt %, 6.5 g) in a spray bottle, followed by thorough mixing. The resulting mixture (16.3 g) was then dried at 130 °C for 1 h. The pH of the polyacrylate suspension was measured as follows (pH 7.2):

polyacrylate (0.5 g) was added to 0.9% NaCl solution (100 mL) and stirred for 30 min at room temperature. The pH was measured using a pH meter (Laqua F-74, Horiba, Japan).

Preparation of Lipid Anchor-Coated Polyacrylate

Neutralized polyacrylate (10.0 g) was placed in a stainless-steel vat and hexadecyltrimethylammonium chloride (0.1 g) dissolved in ethanol (2 mL) was added by spraying with a spray bottle. The resulting beads were dried at 130 °C for 30 min to obtain lipid-anchor-coated polyacrylate (polymer 1, yield: 10.0 g).

Adsorption and Dissociation of Fluorescence-Labeled Liposome

Diluted fluorescent liposomes (10 μL; DOPC/CHOL liposomes labeled with Fluorescein DHPE, FormuMax) were diluted with PBS to a final volume of 1 mL (0.5 mM) and added to a spin column containing polymer 1 (55 mg), followed by incubation for 30 min to allow hydrogel swelling. The residual flow-through solution was collected via centrifugation (2000 × g, 1 min). NaCl solution (0.9%, 0.5 mL) was added to the spin column containing the hydrogel three times to wash the hydrogel. Small amounts of the liposome-bound hydrogels were transferred to 24-well microplates for fluorescence microscopy.

After washing the hydrogel, 100 μL of Elution Buffer in Exorption (Sanyo Chemical) was added and mixed using a vortex mixer for 10 s. The mixture was incubated for 30 min and elution solution was collected using centrifugation (2000 × g, 1 min). A small aliquot of the liposome-released hydrogel was transferred into a 24-well microplate. The fluorescence intensities of the input liposome solution, flow-through solution, and elution solution were measured using a microplate reader (excitation: 488/40 nm; emission: 520/40 nm, Synergy LX, BioTek).

The fluorescence microscopy images of each hydrogel bead were obtained using an all-in-one fluorescence microscope (BZ-X810, Keyence, Japan) equipped with a Plan Fluorite 4× PH objective (NA 0.13, BZ-PF04P, Keyence). Green fluorescence was detected using a GFP filter (excitation: 470/40 nm, emission: 525/50 nm, dichroism: 495 nm, OP-87763, Keyence).

Cell Culture and Preparation of Conditioned Medium

The HEK293 cell line was cultured at 37 °C in a humidified atmosphere containing 5% CO₂ in DMEM (4.5 g/L glucose, phenol red, with GlutaMAX supplement; Thermo Fisher Scientific) supplemented with 10% FBS (Biosera) and 0.1 mg/mL of gentamicin (Merck Millipore). The cells were cultured for 2–3 days, and after reaching 70–80% confluence, the medium was discarded and the cells were washed with PBS. DMEM without phenol red and FBS (4.5 g/L glucose and l-glutamine; Thermo Fisher Scientific) was then added to the culture dish, and the cells were incubated for 48 h to allow EV production. The resulting conditioned medium was collected and centrifuged at 2000 × g for 10 min at 4 °C to remove cell debris and large particles. The samples were aliquoted into 1 mL portions and stored at −80 °C.

Biofluid Sample Collection

Human urine from healthy donors were taken from volunteers, obtained right before the EVs isolation experiments, and centrifuged for 10 min (2000 × g, 4 °C). The CKD urine samples analyzed in this study were obtained from individuals with reduced eGFR and positive albuminuria. This study was approved by the Ethics Committee of Tokushima University Hospital (approval no.: 4143–1). Informed consent was obtained from all the participants.

Commercially available pooled sera (human and mouse) and human plasma (K₂EDTA) were purchased from a distributor.

EV Adsorption onto the Hydrogel Surface

The EVs adsorbed onto the hydrogel surface were treated with NanoSuit solution Type III (NanoSuit Co. Ltd., Japan) and subsequently observed using FE-SEM (JEM-7100F, JEOL Ltd., Japan). All SEM imaging was performed at NanoSuit Co. Ltd.

EV Isolation from Urine and Conditioned Medium Samples by HAS Method

EV isolation using the HAS method was performed according to an extraction protocol. The sample solution (1 mL) was added to a spin column containing polymer 1 (55 mg) and incubated for 30 min. The spin column with the hydrogel was centrifuged at 2000 × g for 10 min to remove the flow-through solution. Next, 0.5 mL of 0.9% NaCl was added to the spin column and centrifuged at 2000 × g for 10 min to wash out residual impurities. This washing step was repeated three times. Subsequently, the elution buffer (0.1 mL) for extraction (Sanyo Chemical) was added to the spin column, the top cap was closed, and the column was mixed using a vortex mixer for 10 s. The column was left to stand for 30 min and elution solution was collected using centrifugation at 2000 × g for 1 min.

EV Isolation from Serum and Plasma Samples Using HAS

Before EV isolation, samples were centrifuged at 2000 × g for 10 min at 4 °C to remove debris. Subsequently, 0.1 or 0.2 mL of sample was diluted with PBS (0.2 μm-filtered) to a final volume of 1 mL and subjected to the above procedure.

EV Isolation by UC

Urine samples from patients with CKD (1 mL) and pooled serum (human and mouse, 1 mL) were centrifuged at 10,000 × g for 30 min at 4 °C to remove cells, debris, and large EVs. The supernatant was carefully transferred to ultracentrifuge tubes. The supernatant was then centrifuged at 210,000 × g for 1 h at 4 °C. After UC, the supernatant was carefully discarded and 1 mL of PBS was added to the pellet. The resuspended EV pellet was subjected to a second round of ultracentrifugation (210,000 × g, 1 h, 4 °C), and the supernatant was gently discarded. Purified EVs were resuspended in the residual supernatant.

EV Isolation by PP

EV isolation by polymer precipitation was performed according to the protocol for Total Exosome Isolation (from urine) (Thermo Fisher Scientific). Urine (1 mL) from patients with CKD was centrifuged at 2000 × g for 30 min at 4 °C to remove the cells and debris. The supernatant was transferred to a new microtube and 1 mL of the Total Exosome Isolation reagent was added and mixed thoroughly by vortexing. The mixture was incubated for 1 h at room temperature, and then centrifuged (10,000 × g, 1 h, 4 °C). The supernatant was removed by pipetting and pellet was resuspended in 0.1 mL of PBS.

EV Isolation Using Tim-4 Affinity

EV isolation from 1 mL of urine from patients with CKD was performed according to the MagCapture Exosome Isolation Kit PS Ver.2 (Fujifilm Wako Pure Chemical, Japan).

Diluted human serum samples (serum: 0.2 mL, PBS, 0.8 mL) were also processed using the manufacturer’s protocol to isolate EVs.

NTA

The purified EV solutions were diluted with ultrapure water. The particle concentration and size distribution were measured using NTA with a NanoSight NS300 equipped with a blue laser (488 nm) (Malvern Panalytical). For analysis, the particles were tracked five times under continuous flow for 60 s. The acquired data were analyzed using the NanoSight NTA software version 3.4.4 (Malvern Panalytical).

ELISA

Purified EV solutions were diluted with 0.2% BSA without any added salt or surfactant, and the diluted EV solutions were applied to a commercially available microplate immobilized with anti-CD9 antibody (CD9/CD63 Exosome ELISA Kit, Human, Cosmo Bio). After incubation, each well was washed with wash buffer and anti-CD63 antibody was added according to the manufacturer’s instructions. The absorbance of each well was measured using a microplate reader (absorbance: 450 nm, Synergy LX, BioTek) and the EV concentration was determined using a calibration curve generated from the absorbance of CD9-CD63 epitope-fused standard protein solutions.

RNA Isolation and RT-qPCR

Total RNA was isolated from 200 μL of EV solution using the Exosomal RNA Isolation Kit (Norgen) and eluted in 50 μL. An aliquot of the eluted RNA solution (2 μL) was used for reverse transcription using the TaqMan Advanced miRNA cDNA Synthesis Kit (Thermo Fisher Scientific), according to the manufacturer’s instructions, to generate the cDNA template. miRNA expression analysis was performed using TaqMan probes; 0.5 μL of cDNA solution was diluted with 4.5 μL of 0.1× TE buffer and subsequently mixed with TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific) and the appropriate miRNA primer solution according to the manufacturer’s protocol. cDNA amplification was monitored using the QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific) and cycle threshold (Ct) values were analyzed using QuantStudio Design & Analysis Software v1.5 (Thermo Fisher).

BCA Assay

EV solution (50 μL) obtained from the urine of patients with CKD using HAS was desalted using a HiTrap Desalting column (Cytiva) to eliminate compounds that interfered with the BCA reaction. The BCA assay was performed according to the protocol of the Micro BCA Protein Assay Kit (Thermo Fisher Scientific). The absorbance of each well was measured using a microplate reader (absorbance: 562 nm; Synergy LX, BioTek).

SDS-PAGE and Silver Staining

Purified EV solutions (2 μL), 6× SDS-PAGE Sample Buffer (2 μL; 0.375 M Tris-HCl, 0.03 w/v% bromophenol blue, pH 6.8, Nacalai), and ultrapure water (8 μL) were mixed in a 1.5 mL microtube and heated at 95 °C for 10 min. The resulting denatured samples were loaded onto a 12.5% polyacrylamide gel (Atto) and SDS-PAGE was performed to separate proteins. The gel was subsequently stained with 2D-Silver Stain·II (Cosmo Bio).

SDS-PAGE and Western Blotting

The 6× SDS-PAGE Sample Buffer (2 μL; Nacalai) and purified EV solutions (UC: 2.4 μL, PP: 2.7 μL, Tim4: 2.7 μL, HAS: 10 μL) were mixed in 1.5 mL microtubes, and brought up to 12 μL with ultrapure water. The volume of each EV solution was determined to standardize the concentration factor based on the starting sample volume (e.g., HAS: 0.37 mL EV solution from 1 mL urine, 10 μL loaded; Tim4: 0.10 mL EV solution from 1 mL urine, 2.7 μL loaded). The mixture was heated at 95 °C for 10 min. SDS-PAGE was performed as described above, and the proteins were transferred onto a PVDF membranes (Cytiva). The membrane was blocked with 5% skim milk (Fujifilm Wako) solution for 1 h and incubated with anti-CD9 antibody diluted in PBS containing 0.05% Tween 20 (PBS-T) for 1 h. The membrane was washed five times with PBS-T and incubated with an HRP-conjugated secondary antibody for 1 h. After washing, the membranes were analyzed using ECL Prime (Cytiva) and ImageQuant 800 (Cytiva).

LC-MS/MS

For peptide MS/MS analysis, four volumes of acetone were added to the EV solution, mixed by vortexing, and incubated at −20 °C, overnight to precipitate proteins. After incubation, the mixture was centrifuged (14,000 rpm, 30 min, 4 °C), and the supernatant was removed. Lysis Buffer (7.5 μL; Thermo Fisher Scientific) was added to the protein pellet, which was then sonicated to completely dissolve the pellet. The recovered proteins were subsequently reduced, alkylated, and digested with trypsin/Lys-C using an EasyPep Magnetic MS Sample Prep Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. The peptide concentration was measured using the Pierce Quantitative Colorimetric Peptide Assay (Thermo Fisher Scientific) following the manufacturer’s protocol.

All LC-MS/MS analyses were performed on an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific) coupled to a Vanquish Neo system (Thermo Fisher Scientific). Peptides (200 ng) were separated using reverse-phase liquid chromatography using a C18 column (EASY-Spray PepMap, 150 mm × 75 μm, 3 μm particles, Thermo Fisher Scientific) at a flow rate of 300 nL/min. The separation was achieved using a 61 min linear gradient from 3% to 31% buffer B, followed by a 5 min linear gradient from 31% to 50% buffer B, and a 4 min linear gradient from 50% to 100% buffer B (buffer A: 0.1% formic acid; buffer B: 0.1% formic acid in 80% acetonitrile).

The mass spectrometer was operated in the DIA positive ion mode. A survey scan (30,000 resolution) from 495 to 745 m/z was followed by an MS2 scan (30,000 resolution) with a 4 m/z isolation window and a first mass of 200 m/z. For MS2 scans, peptides were fragmented using higher-energy collisional dissociation with a stepped collision energy of 30%.

DIA-NN version 1.9 was used for the data processing. Human data were searched against the Swiss-Prot Homo sapiens database (release 2024_06). The threshold for protein identification was set to a false discovery rate <1% at both the precursor and protein levels.

EV Protein Identification

The EV protein data set provided by VesiclePedia was downloaded for EV protein identification (version 5.1; September 2023). The UniProtKB accession numbers of the proteins identified in each proteomic result were queried using UniProt Retrieve/ID mapping (https://www.uniprot.org/id-mapping) to obtain a list of Entrez Gene IDs for comparison with the data set. The number of EV proteins in each proteomic analysis was calculated using Python (version 3.10.13) and Pandas (version 2.0.2). Venn diagrams and correlation heat maps were generated using Seaborn (version 0.13.2). The correlation of the identified proteins between each isolation method was assessed using Pearson’s correlation coefficient.

Flow Cytometry

Mouse serum EVs were purified using HAS and UC methods. Fluorophore-conjugated antibody solutions (CD9: RB780 antimouse CD9, BD Biosciences; CD81: BV421 antimouse CD81, BD Biosciences; ApoA1: Allophycocyanin-anti-ApoA1, Novus Biologicals) were centrifuged at 16,000 × g for 2 min to remove the debris. The EVs isolated using HAS were further concentrated using an ultrafiltration spin column (MWCO: 300 kDa; Pharma Foods, Japan). Both HAS- and UC-derived EV solutions were then diluted with filtered PBS to adjust the particle concentrations to 2.0 × 10¹¹ particles/mL. A 100-μL aliquot of each EV solution was mixed with a premixed antibody solution (0.2 μg per antibody) and incubated at 4 °C for 2 h in the dark. The mixture was washed using an ultrafiltration spin column (MWCO: 300 kDa) to remove unbound antibodies.

Multiplex-stained and single-stained EV controls were analyzed using a FACSymphony A1 flow cytometer (Becton Dickinson). A polystyrene bead mixture (Megamix-Plus SSC, BioCytex SARL) was used for particle size calibration based on side-scattering signals. For the particle size calibration, the theoretical scattering intensity was calculated using NumPy (version 1.23.5), scipy (version 1.12.0), and scikit-learn (version 1.2.2), and graphs were generated using matplotlib (version 3.10.1). The acquired raw data were analyzed using FlowJo Software (version 10.10.0; Becton Dickinson) and the results were analyzed and visualized using FlowKit (version 1.2.3) and Bokeh (version 3.6.2).

Cellular Uptake

The hMSC-UC cell line was cultured at 37 °C in a humidified atmosphere containing 5% CO₂ in mesenchymal stem cell growth medium 2 (C-28009; Promo Cell). The cells were cultured for 3–4 days and after reaching 70–80% confluence. The culture medium was collected and centrifuged at 2000 × g for 10 min at 4 °C to remove cell debris and large particles. The supernatant was purified from EVs using HAS. Membranes of the isolated MSC-EVs were labeled with ExoSparkler EX01 (Dojindo Laboratories). Proteins within MSC-EVs were labeled with ExoSparkler EX05 (Dojindo Laboratories).

The normal human keratinocyte cells (PHK16-0b) cell line was cultured at 37 °C in a humidified atmosphere containing 5% CO₂ in Keratinocyte Medium (C-20011; Promo Cell). The cells were cultured for 3–4 days and ExoSparkler EX05 labeled EVs (red) were added to the cultured cells. Following an additional 24-h incubation, the cells were rinsed with PBS, fixed in 4% formaldehyde, and stained with phalloidin-FITC and DAPI for cytoskeletal and nuclear visualization.

The human follicle dermal papilla cells cell line was cultured at 37 °C in a humidified atmosphere containing 5% CO₂ in Follicle Dermal Papilla Cell Growth Medium (C-26501, Promo Cell). The cells were cultured for 3–4 days and ExoSparkler EX01 labeled EVs (green) were added to cultured cells. After an additional 24-h incubation, the cells were rinsed with PBS, fixed in 4% formaldehyde, and stained with DAPI. Stained cells were imaged using a BZ-X800 fluorescence microscope (Keyence).

Detected particles were first divided according to CD9/CD81 fluorescence intensity, with CD9⁺ and/or CD81⁺ particles defined as EVs. The remaining particles (CD9^– CD81^–) were further analyzed by ApoA1 fluorescence intensity. ApoA1⁺ particle in the CD9^– CD81^– subpopulation was defined as chylomicron, while ApoA1^– particles (others) were considered unknown contamination or debris.

Glossary

Abbreviations

HAS

Hydrogel adsorption separation

UC

Ultracentrifuge

PP

Polymer precipitation.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsabm.5c01838.

• Detailed description of the calculation of side scattering intensity by Mie theory; schematic representation of EV isolation by the HAS method (Figure S1); characterization of EVs isolated using the HAS method (Figure S2); flow cytometric analysis of purified mouse serum EVs isolated using the UC methods (Figure S3); curve fitting of side scattering intensity versus observed signal intensity at the SSC detector (Figure S4); two-dimensional plots of identified particles from EV solution obtained by the HAS method (Figure S5); concentration and recovery of CD9⁺CD63⁺ EVs from urine and conditioned medium samples (Table S1); threshold cycles (Ct) values of exosomal miRNAs measured using a TaqMan probe assay (Table S2); quantification of CD9⁺CD63⁺EV yield, amount of proteins in EV solution, and EV purity in urinary EV solution isolated from patients with CKD using four different methods (Table S3); particle concentration and mode diameter of purified EV solutions from serum and plasma (Table S4); list of representative EV-specific proteins identified by proteomic analysis of human serum EVs isolated using each method (Table S5); list of abundant serum proteins identified by proteomic analysis of human serum EVs isolated using each method (Table S6) (PDF)

R.K., K.T., and T.T. developed the concepts and designed the experiments. K.T. and K.U. designed the polymers. R.K., R.T., and K.T. synthesized and prepared the polymers. R.K., R.T., and T.T. performed the experiments. R.K. and T.T. analyzed the data and wrote the manuscript.

R.K., R.T., and K.T. are employees of Sanyo Chemical Industries, Ltd. This study was supported by a research grant from Sanyo Chemical Industries, Ltd.

The authors declare the following competing financial interest(s): Ryota Kajihara, Kyohei Tobinaga, and Riho Tanigaki are employees of Sanyo Chemical Industries,Lt., and this study was supported by a research grant from Sanyo Chemical Industries, Ltd.