Meso-macroporous hydrogel for direct litre-scale isolation of extracellular vesicles

Abstract

Extracellular vesicles are cell-originated lipid bilayer membrane vesicles that play vital roles in cell-to-cell communications. While extracellular vesicles hold substantial biomedical potential, conventional methodologies for isolating extracellular vesicles require elaborate preprocessing and, therefore, remain labour intensive and limited by throughput. To overcome these challenges, we present a facile fabrication route for generating a meso-macroporous hydrogel matrix with pores of ~400 nm for customizable extracellular vesicle isolation. By combining surface charge-selective capture of extracellular vesicles within the hydrogel matrix and their recovery by high ionic strength, we report direct extracellular vesicle isolation with a throughput range from microlitre to litre scales, without preprocessing, for various biofluids, including whole blood, plasma, ascites, saliva, urine, bovine milk and cell culture media. Furthermore, we demonstrate that the meso-macroporous hydrogel also serves as a solid-phase matrix for preserving extracellular vesicles for on-demand downstream analyses, making it applicable for therapeutics, cosmeceuticals and disease diagnostics.

Fig. 1. Overview of hydrogel-based direct EV isolation and fabrication and porosity characterization of meso–macroporous hydrogel particles.

a, Schematic overview depicting direct EV isolation with meso–macroporous hydrogel from various biofluids and its versatile applications in diagnostics, prognostics, therapeutics and cosmeceutics, including litre-scale isolation and preservation of EVs within the hydrogel. b, Schematic diagram displaying cryo-photocrosslinking process to fabricate meso–macroporous PEG700DA particles. DI, deionized. c, Fluorescence images of meso–macroporous (top) and microporous (bottom) particles during one-dimensional (1D) radial diffusion of various fluorescent indicators at 1 h: fluorescein (~1 nm), 10 kDa dextran labelled with fluorescein isothiocyanate (FITC; 4.72 nm), green-fluorescent silica nanobeads (100 and 200 nm) and FITC-silica nanobeads (450 nm). Intensity scale bars are identical to each indicator. d, Photographs showing an 11 × 11 array snapshot of meso–macroporous hydrogel particles (left) and lyophilized meso–macroporous particles in a glass vial for storage (right). a created with BioRender.com.

Fig. 2. Procedure and underlying principles of meso–macroporous-hydrogel-based direct EV isolation.

a, Schematic diagram displaying EV isolation procedure with meso–macroporous hydrogel particles: all-in-one tube isolation (pink-shaded box) or tube-changeable isolation (green-shaded box). Labels 1, 2 and 3 represent successive transfers of the sample into fresh tubes at each step of the procedure. Each SEM image represents a cross-section of meso–macroporous particles before EV isolation (i), after in-gel capture of EVs (ii) or after off-gel recovery of EVs (iii). b, Yield and protein concentration of EVs with meso–macroporous (blue) and microporous (grey) hydrogel particles (technical replicates, n = 5 for yield and 3 for protein concentration). Statistical significance, ****P < 0.0001. c, Schematic illustrations depicting the underlying principles of active in-gel capture of EVs mediated by meso–macropores of hydrogel and a salt (NaCl), rendering the additional exclusion of permeable but repelled impurities (<400 nm). Dashed boxes (i–iii) illustrate dynamic snapshots of the proposed in-gel EV capture mechanism. d, Cryogenic TEM images of EVs isolated with meso–macroporous hydrogel, showing lipid bilayer structure. Zoomed-in views are indicated by the green dashed boxes. Source data

Fig. 3. Litre-scale direct EV isolation with meso–macroporous hydrogel particles.

a,e, Size distribution, yield and purity of gastric cancer patients’ ascites EVs (a) and bovine milk EVs (e) isolated by hydrogel. Statistical significance, not significant (NS) P = 0.1119 (yield in e), P = 0.3422 (purity in e). Error bars indicate mean ± s.d. (technical replicates, n = 5 for a and 3 for e; biological replicates in a, n = 5). b, Western blot images showing the expression of EV-positive (CD63 and PDCD6IP) and tumour metastasis (CDH2, CLDN1 and ANG) markers from the gastric cancer patients’ ascites EVs. c, Photographs showing lyophilized hydrogel particles (left) in a glass bottle and the bottle with 1 l of milk poured in (right). d, Amount of milk EVs isolated from 1 l (purple) and 1 ml (blue). Statistical significance, ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 3). Source data

Fig. 4. Meso–macroporous hydrogel as EV-preserving carrier.

a, Photographs of meso–macroporous hydrogel particles in wet (left), dehydrated (middle) and rehydrated (right) states. White dashed circles indicate the particle size in the wet state before dehydration through lyophilization. b, Yield of EVs isolated with meso–macroporous particles in the aged state (red; 1 year equivalent acceleration), rehydrated-after-dehydration state (yellow) and wet state (blue). Statistical significance, NS P = 0.8578. c, Schematic diagram displaying the procedure to isolate EVs from lyophilized EV-captured meso–macroporous hydrogel particles after storage at room temperature. d, Size distribution and purity of EVs recovered immediately (control, blue) and from lyophilized EV-captured hydrogel particles after storage for 5 days (red). Statistical significance, NS P = 0.4445. e, Relative yield (the yield divided by the yield on day 0) of isolated EVs over storage time after lyophilization. Error bars indicate mean ± s.d. (technical replicates, n = 3 for b and 5 for d and e). Source data

Fig. 5. Customizability and versatility of hydrogel-based EV isolation.

a, Yield (blue) and amount (grey) of EVs by varying off-gel recovery volume. Statistical significance to 200 µl, ***P = 0.0002 and ****P < 0.0001 for yield; NS P = 0.3949 (40 µl) and 0.2104 (400 µl) for amount. b,c, Amount of EVs isolated with 10 µl hydrogel particles by varying off-gel recovery volume (b), and size distribution, yield and purity by varying isolation time (c). Statistical significance, NS P = 0.1779 (b) and 0.0894 (yield in c) and *P = 0.0403 (purity in c). d, Size of EVs by varying freezing temperatures during cryo-photocrosslinking of hydrogel precursor. Statistical significance, **P = 0.0011. e, Size distribution, yield and purity of EVs isolated by hydrogel particles frozen at –80 °C (blue) and –195 °C (red) during cryo-photocrosslinking. Statistical significance, *P = 0.0204 (yield) and NS P = 0.8866 (purity). f, Comparative schematic diagrams depicting the procedure of EV isolation from human whole blood with timelines. g, Size distribution, yield and purity of whole blood EVs isolated by hydrogel (blue) and UC (grey). Statistical significance, **P = 0.0035 (yield) and NS P = 0.7986 (purity). Error bars indicate mean ± s.d. (technical replicates, n = 3 for a–c and g and 5 for d and e). f created with BioRender.com. Source data

Fig. 6. Downstream analyses for therapy and diagnosis.

a, Comparative schematic diagram depicting the procedure of EV isolation from bovine milk with timelines. b, Size distribution, yield and purity of milk EVs isolated by hydrogel (blue) and UC (grey). Statistical significance, ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 3). c,d, Identical-milk-volume (c) and equal-particle-number (d) analyses for the proliferation of hDFs at 3 days in vitro (DIV 3) after treating milk EVs isolated by hydrogel (blue) and UC (grey) for 24 h, relative to proliferation without milk EVs (white). Statistical significance, *P = 0.0110 (gel–UC in d) and 0.0348 (gel–control in d), ***P = 0.0003 and ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 4 for c and 7 for d). e, Fluorescence images showing hDFs’ nuclei (blue) and expression of collagen II (green), depending on the treatment with milk EVs—isolated by hydrogel (left), isolated by UC (middle) and no treatment (right)—as part of an equal-particle-number analysis. f, Equal-particle-number analysis for the proliferation of HaCaT cells at DIV 3 after treating with milk EVs isolated by hydrogel (blue) and UC (grey) for 24 h, relative to proliferation without milk EVs (white). Statistical significance, **P = 0.0014 and ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 4). g, Western blot image showing glutathionylated proteins and GAPDH protein expression by HaCaT cells as part of an equal-particle-number analysis. h, Size distribution and purity of human urine EVs isolated by hydrogel (blue) and UC (grey). Statistical significance, NS P = 0.7522. Error bars indicate mean ± s.d. (technical replicates, n = 3). i, Schematic diagram of profiling urinary EV miRNAs as a diagnostic downstream analysis, and a SuperPlot displaying ratiometric fluorescence signals (intensity I of miR-6090 divided by I of miR-3665) from 12 healthy controls and 12 prostate cancer patients. Statistical significance, ****P < 0.0001. Error bars indicate mean ± s.d. (biological replicates pooled from 12 controls or patients, n = 2; technical replicates, the number of conventional hydrogel particles (Supplementary Fig. 10), n = 13). a created with BioRender.com. Source data

Extended Data Fig. 1. Optimization of meso–macroporous hydrogel-based direct EV isolation.

a, Yield of human plasma EVs by varying the volumetric composition of PEG700DA in precursor solutions. Statistical significance, ***P = 0.0002, *P = 0.0443 and ****P < 0.0001. b, Yield of EVs by varying input volume per batch while fixing the volume of the meso–macroporous hydrogel as 40 µL (1 particle). Statistical significance to 300 µL, NS P > 0.9999 (350 µL), P = 0.9998 (400 µL) and P = 0.3997 (800 µL). c, Yield of EVs by varying the volume of meso–macroporous hydrogel (that is, the number of hydrogel particles; 40 µL each) per batch while fixing the input plasma volume as 800 µL. Statistical significance to 120 µL (3 particles), ****P < 0.0001, *P = 0.0343 and NS P = 0.2161. d, Yield (blue) and purity (grey) of EVs depending on ionic strength (that is, NaCl concentration) during the in-gel capture. Statistical significance to 1.5 M, ***P = 0.0003 (0.5 M) and 0.0006 (2.5 M) for yield; ***P = 0.0002 (0.5 M) and 0.0003 (2.5 M) for purity. e, Yield of EVs depending on in-gel capture time. Statistical significance to 1 h, NS P = 0.9998 (2 h) and 0.0626 (24 h). f, Purity (blue) and yield (grey) of EVs depending on the number of washes. Statistical significance to 3 washes, ****P < 0.0001 (1 wash), ***P = 0.0001 (2 washes) and NS P = 0.6999 (4 washes) and 0.9896 (5 washes) for purity; ***P = 0.0006 (1 wash), ****P < 0.0001 (2 washes), NS P = 0.2559 (4 washes) and ***P = 0.0001 (5 washes) for yield. g, Yield of EVs depending on off-gel recovery time. Statistical significance to 5 min, NS P = 0.9961 (10 min), 0.7515 (30 min) and 0.2854 (60 min). For all these plots, error bars indicate mean ± s.d. (technical replicates, n = 3 for a, c, e, f and g and 5 for b and d). Source data

Extended Data Fig. 2. Reproducibility of meso–macroporous hydrogel-based isolation across batches and individuals.

a,b, Size distributions of human plasma EVs isolated by hydrogel across five batches (a) and three individuals (b). c, Size of EVs isolated by different individuals. Statistical significance, NS P = 0.3908 (A-B), 0.5503 (A-C) and 0.9771 (B-C). Box plots display the interquartile range (box), with the horizontal line inside each box representing the median and the whiskers indicating the minimum and maximum values (technical replicates, n = 5). Source data

Extended Data Fig. 3. Characterizations for human plasma EVs isolated with meso–macroporous hydrogel.

a, Western blot images showing the expression of EV-positive (CD63, CD9 and TSG101) and EV-negative (CNX and APOB) markers from the isolated nanoparticles by meso–macroporous hydrogel and other conventional methods: density gradient ultracentrifugation (UC), UC, size exclusion chromatography and ExoQuick. b,c, Yield (b) and purity (c) of nanoparticles isolated by the four methodologies. Statistical significance of the conventional methods to hydrogel, ****P < 0.0001, ***P = 0.0002 and *P = 0.0122 for yield (b); **P = 0.0045, NS P = 0.6296, ****P < 0.0001 and *P = 0.0164 for purity (c). Error bars indicate mean ± s.d. (technical replicates, n = 3). d, Median particle size and its range. Statistical significance of the conventional methods to hydrogel, *P = 0.0116, NS P > 0.9999 (UC), = 0.9444 (SEC) and 0.9985 (ExoQuick). Plots display the median size (circle) and the whiskers indicating the minimum and maximum values (technical replicates, n = 3). e, Western blot images showing the expression of EV-positive (CD63 and TSG101) and EV-negative (CNX, APOB, APOA1, GOLGA2 and CSN1S1) markers in nanoparticles isolated from human plasma (left), hiPSC culture media (middle) and bovine milk (right) by meso–macroporous hydrogel, UC, size exclusion chromatography and positive controls. Source data

Extended Data Fig. 4. Negative-stain TEM to visually identify contaminants.

Negative-stain TEM images showing EVs isolated with meso–macroporous hydrogel (columns 1 and 2), native biofluid before isolation (column 3) and remains of the sample left over after isolation (column 4) for human plasma (top), urine (middle) and hESC culture media (bottom).

Extended Data Fig. 5. Proteomic analysis of isolates acquired by meso–macroporous hydrogel and conventional methodologies.

a,b, Doughnut charts (a) and abundance rank plots (b) for proteins identified and classified in four categories: EV only, EV & plasma, plasma only and others. c, Number of proteins classified in the four categories over various methodologies, including the meso–macroporous hydrogel-based isolation. d, Venn diagram displaying proteomic composition (that is, the number of EV proteins) of EVs isolated by meso–macroporous hydrogel, density gradient UC, UC and SEC. Source data

Extended Data Fig. 6. RNA sequencing of human plasma EVs isolated by meso–macroporous hydrogel.

a, Doughnut chart displaying the composition of plasma EV RNAs. b, Venn diagram showing the number of miRNAs in EVs isolated by hydrogel (blue) and those listed in two databases of Vesiclepedia and EVmiRNA (red). c, Doughnut chart displaying plasma EV miRNAs with top 10 reads. Source data

Extended Data Fig. 7. Reusability of meso–macroporous hydrogel particles for EV isolation.

a,b, Size distribution (a) and relative yield (Yield / Yield_Fresh; b) of EVs isolated by fresh (blue) and reused hydrogel particles after RIPA treatment (red). Statistical significance, P = 0.2139. Error bars indicate mean ± s.d. (technical replicates, n = 3). Source data

Extended Data Fig. 8. Enhancement of purity for reduced isolation time.

a–c, Size distribution (a), yield (b) and purity (c) of EVs isolated by 10 µL-hydrogel particles for 15 min with extended wash steps (three times each for 5 min; brown) and 80 min (blue). Statistical significance, NS P = 0.1096 (b) and 0.7830 (c). Error bars indicate mean ± s.d. (technical replicates, n = 3). Source data

Extended Data Fig. 9. Direct EV isolation from various biofluids.

a, Size distribution, yield and purity of mouse whole blood EVs isolated by hydrogel (blue) and UC (grey). Statistical significance, ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 5). b,c, EV isolations from human saliva (b) and hESC culture media (c) by hydrogel (blue) and UC (grey): size distribution, percentage of the area under the size distribution curves between 30 and 400 nm, yield and purity. Statistical significance, *P = 0.0304 (yield in b) and *P = 0.0117 (purity in b) for hESC cell culture media; NS P = 0.5364 (yield in c) and *P = 0.0316 (purity in c) for saliva. Error bars indicate mean ± s.d. (technical replicates, n = 3). Source data

Extended Data Fig. 10. Equal-protein-amount analyses for bovine milk EVs’ cosmeceutical (therapeutic) potential.

a, Equal-protein-amount analysis for the proliferation of human dermal fibroblasts (hDFs) at three days in vitro (DIV 3) after treating milk EVs isolated by hydrogel (blue) and UC (grey) for 24 h, relative to proliferation without milk EVs (white). Statistical significance NS P = 0.6465 and ****P < 0.0001. Error bars indicate mean ± s.d. (technical replicates, n = 7). b, Fluorescence images showing hDFs’ nuclei (blue) and expression of collagen II (green), depending on the treatment of milk EVs isolated by hydrogel (left), UC (middle) and no treatment (right) as part of the equal-protein-amount analysis. c, Equal-protein-amount analysis for the proliferation of human keratinocytes (HaCaT cells) at DIV 3 after treating milk EVs isolated by hydrogel (blue), UC (grey) for 24 h, relative to proliferation without milk EVs (white). Statistical significance, NS P > 0.9994 (gel-UC), **P = 0.0031 (gel-control) and 0.0033 (UC-control). Error bars indicate mean ± s.d. (technical replicates, n = 4). d,f, Western blot image showing glutathionylated proteins and GAPDH expression by HaCaT cells. e, Proliferation of HaCaT cells at DIV 3 after treating EV-depleted milk (hatched) for 24 h, relative to proliferation without milk EVs (white). Statistical significance, NS P = 0.0844. Error bars indicate mean ± s.d. (technical replicates, n = 4). Source data