3D Nanofibrillar Matrix Stiffness Modulates Extracellular Vesicle Cargo and Pro‐Tumour Functions

ABSTRACT

Extracellular matrix (ECM) stiffness and extracellular vesicles (EVs) are critical regulators of tumour progression, yet their interaction in three‐dimensional (3D) microenvironments remains poorly understood. Most studies on ECM stiffness and EV biology rely on 2D cultures, which do not capture the complexity of the tumour microenvironment. Here, a biomimetic 3D nanofibrillar ECM model based on a cellulose nanofibril hydrogel was established to assess stiffness‐dependent changes in EV properties and functions. EVs derived from stiff matrices (StEVs) exhibited distinct physicochemical characteristics and carried unique protein and microRNA cargo compared with those from soft matrices (SoEVs). Functionally, StEVs more potently promoted tumour cell proliferation and migration, while in vivo mouse models further demonstrated that StEVs enhanced tumour growth. Multi‐omics analyses and pharmacological inhibition studies revealed that StEVs activate the mitogen‐activated protein kinase/extracellular signal‐regulated kinase 1/2 (MAPK/ERK1/2) signalling pathway in recipient cells. These findings highlight the mechanobiological regulation of EV‐mediated intercellular communication within 3D ECM environments and demonstrate how matrix stiffness shapes EV cargo and pro‐tumour activity.

1. Introduction

Mechanical cues within the extracellular matrix (ECM) show a critical influence on cancer progression through mechanotransduction (Agrawal et al. 2025; Massey et al. 2024; Mohammadi and Sahai 2018; Saraswathibhatla et al. 2023; Chen et al. 2024; Zhu et al. 2025). Notably, ECM stiffness is a prominent physical hallmark of cancer, activating signalling pathways that promote tumour cell proliferation, invasion and metastasis (Liang and Song 2023; Nia et al. 2020, 2025; Peng et al. 2025; Zhang et al. 2021). Given its pivotal role in tumour progression, targeting ECM stiffening has gained increasing attention as a potential therapeutic strategy (Feng et al. 2024; Shen et al. 2020; Sleeboom et al. 2024; Zhang et al. 2025). Simultaneously, extracellular vesicles (EVs) have emerged as key mediators of intercellular communication by transferring bioactive molecules such as proteins and microRNA to regulate cellular behaviours and tumour microenvironment (Lopez et al. 2023; Ortiz 2021; Kalluri and McAndrews 2023; Liu and Wang 2023; van Niel et al. 2022). Despite the increasing body of evidence highlighting the crucial roles of ECM mechanics and EVs in cancer biology, the interplay between ECM stiffness and EV‐mediated signalling remains poorly understood, particularly within the context of three‐dimensional (3D) ECM‐like networks.

Emerging studies have examined the influence of matrix stiffness on EV biogenesis and function, most of which have been performed in two‐dimensional (2D) culture systems (Patwardhan et al. 2021; Liu et al. 2022, 2024; Wu et al. 2023; Senigagliesi et al. 2024; Sneider et al. 2024). For example, Patwardhan et al. demonstrated that breast cancer cells cultured on collagen‐coated polyacrylamide hydrogels with increased stiffness (0.5 vs. 5 kPa) exhibited elevated exosome secretion, and that stiffness‐induced exosomes enriched in thrombospondin‐1 (THBS1) promoted cancer cell motility and invasion (Patwardhan et al. 2021). Similarly, Guo et al. employed 2D collagen‐coated substrates with defined stiffness (0.5 vs. 10 kPa) and revealed that exosomes derived from stiff ECMs are enriched in Jagged1 and enhance tumour growth via Notch signalling (Wu et al. 2023). More recently, Denis Wirtz et al. cultured breast cancer cells on stiffness‐defined 2D collagen‐coated matrices (0.5 vs. 25 kPa) and found that EVs from stiff matrices are enriched in integrins, adhesion molecules and immune evasion‐related proteins compared to those from soft matrices (Sneider et al. 2024). While informative, these 2D studies fail to recapitulate the spatial complexity and mechanical heterogeneity of the in vivo tumour microenvironment (Benelli et al. 2023; Acharekar et al. 2023; Gil‐Redondo et al. 2022).

In contrast, 3D culture systems provide a more physiologically relevant platform for modelling the complex cell‐matrix and cell‐cell interactions present within tumour microenvironments (Hartung 2014; Micalet et al. 2021; Sun et al. 2021). To support 3D cell culture, various hydrogel‐based materials have been developed (Abuwatfa et al. 2024; Vernerey et al. 2021; Sievers et al. 2023), among which, Matrigel is widely used but suffers from lot‐to‐lot variability, limited stiffness range and high cost, which together restrict its capacity to model the biomechanical diversity of tumour ECM (Hughes et al. 2010; Reed et al. 2009; Curvello et al. 2019). In comparison, cellulose nanofibril (CNF)‐based hydrogels offer distinguishing advantages due to their fibrillar architecture, low cost, sustainability, low toxicity, excellent water permeability, mechanical properties and biocompatibility (Hossen et al. 2018; Torresan et al. 2025; Curvello et al. 2024; Bhattacharya et al. 2012; Wang et al. 2020). Among these, 2,2,6,6‐tetramethylpiperidine‐1‐oxyl (TEMPO)‐oxidized cellulose nanofibrils (TOCNF) display rigid, collagen fibril‐like morphologies and surface‐anchored carboxyl groups arranged in a polyuronate‐like pattern, closely mimicking key structural and chemical features of native ECM components such as collagen and hyaluronan (Curvello et al. 2021; Hatakeyama and Kitaoka 2021). To complement this structural mimicry, gelatin methacryloyl (GelMA), a methacrylated derivative of collagen, provides biochemical functionality through abundant arginine‐glycine‐aspartic acid (RGD) motifs that facilitate cell adhesion (Piao et al. 2021; Chin et al. 2023). The combination of TOCNF and GelMA yields a hybrid hydrogel that synergistically integrates nanofibrillar structural fidelity with biochemical cues essential for cellular engagement. This hybrid matrix provides a versatile and highly tunable platform for constructing 3D tumour models with architectural and mechanobiological properties resembling those of the ECM.

In this study, a biomimetic TOCNF/GelMA hybrid hydrogel system was developed to examine the effects of 3D ECM mechanics on EV cargo, function and potential mechanisms. The hybrid hydrogel was engineered to simulate the stiffness of both normal and pathological ECM conditions. It was then characterized for its physicochemical properties and biocompatibility. Importantly, EVs derived from stiff 3D matrices (StEVs) exhibited distinct physicochemical properties, including altered Young's modulus and zeta potential, as well as significantly different protein and microRNA cargo profiles compared to those from soft matrices (SoEVs). Functionally, phenotypic assays demonstrated that StEVs exerted a stronger influence on tumour cell proliferation and migration, with subsequent in vivo experiments demonstrating a significant enhancement of tumour growth. Proteomic and small RNA sequencing analyses further revealed that StEVs may activate the mitogen‐activated protein kinase/extracellular signal‐regulated kinase 1/2 (MAPK/ERK1/2) signalling pathway in recipient cells, a finding further validated through pharmacological inhibition experiments. Together, these results identify ECM stiffness as a critical regulator of EV cargo and function in a 3D context, offering new insights into the mechanobiological control of EV‐mediated signalling in cancer.

2. Experimental Section

2.1. Materials and Reagents

Dulbecco's Modified Eagle Medium (DMEM), RIPM‐1640 medium, Foetal Bovine Serum (FBS), Phosphate‐buffered saline (1X PBS, pH 7.4), Propidium iodide solution (P3566), Trypsin and Collagenase (#17101015, Type II) were obtained from Life Technologies (USA). Protease Inhibitor Cocktail (HY‐K0010), Phosphatase Inhibitor Cocktail I (HY‐K0021, 100X in DMSO), Calcein‐AM (HY‐D0041) and Inhibitor U0126 (HY‐12031) were bought from MCE. Phalloidin‐iFluor 647 Reagent (ab176759) was provided by Abcam. BeyoClick EdU‐555 Cell Proliferation Kit (C0075S) was purchased from Beyotime (China), and the Annexin V Apoptosis Detection Kit FITC (#88‐8005‐72) was obtained from Invitrogen (USA). Cellulase from Trichoderma reesei (C2730) was bought from Sigma. TEMPO‐oxidized cellulose nanofibre (TOCNF) was supplied by Tianjin Mujingling Technology Co. Ltd. (China), while GelMA (Gelatin Methacryloyl) and the photoinitiator LAP (Lithium phenyl (2,4,6‐trimethylbenzoyl) phosphinate) were sourced from Engineering for Life of Yongqinquan (Suzhou, China). The miRNeasy Serum/Plasma Kit (#217184) and miRNeasy Tissue/Cells Advanced Mini Kit (#217604) were acquired from Qiagen (Germany). For detailed information on all antibodies, please refer to Table S1.

2.2. Preparation and Characterization of TOCNF/GelMA Hybrid Hydrogel

The GelMA solution was prepared by dissolving solid GelMA in a 0.25% (w/v) LAP photoinitiator solution at 60°C for 40 min under constant stirring. After complete dissolution, the solution was filtered through a 0.22 µm syringe filter to ensure sterility and remove any undissolved particles. To prepare the TOCNF/GelMA hybrid hydrogel, 1.2% (w/v) TOCNF was added to the GelMA solution in a 1:2 volume ratio (TOCNF: GelMA). The mixture was homogenized using a vortex mixer until a uniform dispersion was achieved. The hybrid solution was then crosslinked using a UV curing system (EFL‐LS‐1602, Suzhou, China) with a light intensity of 25 mW/cm². Soft hydrogels were crosslinked for 30 s, while stiff hydrogels were crosslinked for 90 s. Dry samples of the TOCNF/GelMA hybrid hydrogel were prepared by freeze‐drying the dispersion to remove water. These dried samples were subjected to Fourier transform infrared spectroscopy (FT‐IR) for chemical characterization and scanning electron microscopy (SEM) for morphological analysis.

2.3. Swelling Ratio Test

To analyze the swelling behaviour of the hybrid hydrogel, the freeze‐dried TOCNF/GelMA samples were first weighed and recorded as W1. The dried samples were then immersed in PBS (pH 7.4, 0.01 M) and incubated at 37°C for 24 h. After incubation, the hybrid hydrogels were removed, and any excess PBS was carefully removed by Kimwipes. The swollen hydrogel samples were then weighed and recorded as W2. All experiments were independently carried out in triplicate. The swelling ratio of TOCNF/GelMA hybrid hydrogel was calculated using the following formula:

$$\mathrm{Swell}\mathrm{Ratio}\left(\%\right)=\left[\left(\mathrm{W}2-\mathrm{W}1\right)/\mathrm{W}1\right]\times 100\%.$$

2.4. Degradation Study

To characterize the enzymatic degradation properties of the hybrid hydrogel, the samples were first weighed and recorded as W3. The hydrogels were then immersed in a PBS solution containing 28 U/g of cellulase and 2 U/mL of collagenase. At 3, 9 and 30 h, the samples were removed, washed with deionized water, and gently removed excess liquid with Kimwipes. The samples were then weighed and recorded as W4. Experiments were conducted independently three times. The enzymatic degradation rate was calculated using the following formula: Enzymatic Degradation Rate (%) = [(W3 – W4)/W3] × 100%.

2.5. Mechanical Testing

Hybrid hydrogels designated for mechanical testing were crosslinked under 405 nm UV light within a 10 mm polytetrafluoroethylene (PTFE) mould. Following crosslinking, the mould was carefully removed, and the hydrogels were immersed in PBS for 24 h to ensure complete hydration. Prior to testing, excess surface liquid was gently removed using Kimwipes. Compression testing was conducted at a strain rate of 0.5 mm/min with a preloading force of 0.01 N. During formal testing, a maximum applied load of 50 N was used. The compressive modulus was determined from the slope of the linear region (5%–10% strain) of the stress‐strain curve, providing a quantitative assessment of the mechanical properties of the hydrogels. Triplicates were carried out for each test.

2.6. EV‐Depleted Medium Preparation

The EV‐depleted DMEM and RIPM‐1640 were prepared using 10% EV‐depleted FBS and 1% penicillin‐streptomycin. The preparation of EV‐depleted FBS followed the protocol established in our previous work (Wang et al. 2024). Briefly, commercially purchased FBS was centrifuged at 120,000 × g for 18 h at 4°C, resulting in a dark‐coloured precipitate at the bottom of the tube. The supernatant was carefully aspirated to avoid disturbing the precipitate and subsequently filtered through a 0.2 µm syringe filter for decontamination. The filtered supernatant was either used immediately for experiments or stored at 4°C for future use.

2.7. Culture of Cells in 3D Hybrid Hydrogel

Before 3D culture, all necessary experimental materials were placed in a biosafety cabinet for UV sterilization for at least 30 min. H1299/A549 cells, cultured in 15 cm diameter dishes, were digested with 3 mL of trypsin and incubated at 37°C for 3 min. The digestion was then terminated by adding medium. After centrifugation, the cell pellet was collected. The TOCNF/GelMA hybrid hydrogels were then resuspended using an electronic pipette, ensuring gentle handling to avoid bubble formation, and thoroughly mixed to obtain a homogeneous cell suspension. The hydrogel‐cell mixture (800 µL per well) was added to low‐adhesion 6‐well plates (#3471 Corning), and light crosslinking was performed using a UV curing system (EFL‐LS‐1602, Suzhou, China) with an intensity of 25 mW/cm². The curing time for the soft 3D hydrogel system was 30 s, while the stiff 3D hydrogel system was cured for 90 s. After curing, 3 mL of medium was added to each well, and the plates were placed in a 37°C incubator for 5 min. The medium was then replaced with 4 mL of fresh medium, and the plates were returned to the incubator for further culture. The cell culture medium was changed every 2 days using fresh EV‐depleted medium. During the medium change, the hydrogels were washed twice with PBS, taking care to avoid direct contact with the hydrogel. Throughout the experiment, it was ensured that cells in both the soft and stiff conditions were derived from the same cell passage and maintained at a cell density of 1.3 × 10⁶ cells/mL.

2.8. EV Collection From 3D Culture System

Following the 3D culture procedures described above, the medium was replaced with EV‐depleted medium after overnight incubation, marking Day 1 of 3D culture. Culture medium was collected on Days 3, 5 and 7, sequentially centrifuged at 400 × g for 10 min, followed by 3000 × g for 20 min at 4°C. The resulting supernatant was filtered through a 0.22 µm membrane and stored at –80°C for future purification. On Day 7, after collecting the culture medium, enzymatic digestion was performed to recover EVs trapped within the hydrogel matrix. The hydrogel was digested in EV‐depleted medium using cellulose and collagenase at 37°C for 1–2 h, with pipette aspiration facilitating digestion. Once the hydrogel was fully digested, the medium was collected, filtered through a 40 µm membrane, and centrifuged at 3000 × g for 10 min. The resulting supernatant was combined with the previously collected medium for subsequent EV isolation. The remaining pelleted cells were resuspended in 4 mL of PBS, aliquoted into 1.5 mL centrifuge tubes, and centrifuged at 150,000 × g for 5 min. The pellet was washed twice, and after removing the supernatant, it was stored at –80°C for future analysis.

The total collected medium was then concentrated using a Centrate‐70 Plus concentrator (#UFC710008, Millipore), and EVs were isolated via size‐exclusion chromatography (SEC). For SEC‐based EV isolation, the qEV original Gen2 column was used according to the manufacturer's protocol. The concentrated culture medium was loaded onto the column, and EV‐enriched fractions (fractions 4–7) were collected. These fractions were further concentrated to a final volume of 100 µL using a 10 kDa molecular weight cut‐off (MWCO) ultrafiltration filter (Millipore). The final EV samples were either immediately used for experiments or stored at –80°C for future analysis. The protocols for NTA (Nanoparticle Tracking Analysis), TEM (Transmission Electron Microscopy) and AFM (Atomic Force Microscopy) characterization of EVs are detailed in the Supporting Information.

2.9. Live/Dead Staining Assays in Hybrid Hydrogel

Cell viability in 3D hydrogels was evaluated using Live/Dead cell staining. After culturing for 1, 7 and 15 days, the cell‐laden hydrogels were washed three times with PBS, then incubated with a solution of Calcein‐AM (2 µM) and propidium iodide (3 µM) at 37°C for 30 min. The stained samples were imaged using a confocal laser scanning microscope (Nikon A1HD25), with dead cells stained red and live cells stained green. Fluorescence intensity analysis was subsequently performed using ImageJ software.

2.10. Cell Morphology Staining

After 7 days of 3D culture, cells were fixed with 4% paraformaldehyde at room temperature for 30 min, followed by three washes with PBS. The cells were then permeabilized with 0.2% Triton X‐100 at room temperature for 20 min and washed three times with PBS. Subsequently, the cells were blocked with 2% BSA for 20 min. Actin filaments were stained using Alexa‐Fluor 647‐conjugated phalloidin (ab176759, Abcam) for 60 min. After staining, the cells were washed three times with PBS, and then counterstained with DAPI (Thermo Fisher) for 10 min. Following this, the cells were washed three times with PBS, and fluorescence images were captured using a confocal laser scanning microscope (Nikon A1HD25).

2.11. EdU Cell Proliferation Assay

Cell proliferation within hydrogels was assessed on Day 7 of culture using the BeyoClick EdU‐555 Cell Proliferation Kit (#C0075S, Beyotime, China) according to the manufacturer's instructions, with minor modifications for 3D culture. Briefly, encapsulated cells were incubated with 10 µM EdU at 37°C overnight to ensure adequate labelling in the 3D hydrogel environment. Following incubation, samples were fixed with 4% paraformaldehyde for 15 min at room temperature, washed three times with phosphate‐buffered saline (PBS), and permeabilized with 0.5% Triton X‐100 for 15 min. After three additional PBS washes, the click reaction solution was added, and samples were incubated in the dark for 30 min at room temperature. Nuclei were counterstained with DAPI (1:1000 dilution) for 20 min, followed by three PBS washes. Images were acquired using a Nikon A1HD25 confocal laser scanning microscope, and EdU‐positive cells were quantified by fluorescence intensity analysis using ImageJ software.

2.12. Annexin V/PI Apoptosis Assay

Cell apoptosis within hydrogels was assessed on Day 7 of culture using a FITC Annexin V/PI apoptosis kit (#88‐8005‐72, Invitrogen, USA) according to the manufacturer's instructions, with slight modifications for 3D culture. Briefly, hydrogels containing cells were gently washed with 1× binding buffer to remove residual culture medium. Annexin V‐FITC (20 µL) and propidium iodide (5 µL) were then added directly to the culture dish and incubated at room temperature in the dark for 30 min. Following incubation, samples were imaged using a Nikon A1HD25 confocal laser scanning microscope.

2.13. Western Blotting

Cells were lysed with RIPA on ice for 15 min, and then the whole lysate was centrifuged at 12,000 × g for 15 min at 4°C. Protease and phosphatase inhibitors were added during the lysis of cell samples. The BCA (bicinchoninic acid) kit (#23225, ThermoFisher, USA) was used to determine the protein content of EVs and Cell samples. Next, the protein was evenly mixed with 4X Laemmli Sample Buffer (#1610747, Bio‐Rad) and denatured by heating at 95°C for 5 min. The resulting samples were separated at 80–100 V in a precast gel and then transferred onto a 0.45 µm PVDF membrane. The membrane was blocked with Rapid Blocking Buffer from Beyotime (China), followed by incubation with the primary antibodies at 4°C overnight. Subsequently, the membrane was washed by soaking in TBST (Tris Buffered Saline with Tween 20, Beyotime, China) for 15 min and three repetitions, followed by incubation with secondary antibodies for 2 h at room temperature. Finally, the resulting membrane was washed in TBST three times before imaging using an enhanced chemiluminescence (ECL) reagent (Beyotime, China).

2.14. Wound Healing Assay

The wound healing assay was employed to investigate the effects of StEVs and SoEVs on cell migration. The assay was conducted using ibidi Culture‐Inserts (ibidi) following the manufacturer's instructions. A cell suspension of 5 × 10⁵ cells (100 µL) was added to the ibidi Culture‐Inserts placed in 24‐well plates. Once a confluent cell monolayer was formed, the ibidi Culture‐Inserts were removed, creating a defined cell‐free gap of 500 µm in width. The wells were washed with PBS to remove cell debris, and the medium was replaced with EV‐depleted culture medium. StEVs or SoEVs were then added at a concentration of 20 µg/mL, with PBS serving as the negative control. Wound closure was monitored and imaged at 0, 18 and 24 h using a real‐time tracking microscope (LS720, Etaluma, USA). The wound area at each time point was compared to the initial wound area (0 h), and the resulting area was normalized to the initial wound area to assess cell migration ability. The wound area measurement was performed using ImageJ software.

2.15. Cell Proliferation Assay

H1299 or A549 cells (2 × 10³ cells/well) were seeded into 96‐well plates and cultured in standard medium. Following cell attachment, the medium was replaced with EV‐depleted medium, and the cells were treated with either SoEVs or StEVs, with PBS serving as the control. EVs were added at a concentration of 20 µg/mL, and the experiment was conducted over four days. Cell proliferation was assessed on Days 1, 2, 3 and 4 using the Cell Counting Kit‐8 (CCK‐8; Dojindo Molecular Technologies, Japan). Absorbance at 450 nm was measured using a BioTek Synergy H1 Microplate Reader. The proliferation rate was evaluated by comparing absorbance values across treatment groups at each time point, allowing for the assessment of EV‐induced effects on cell proliferation under different matrix conditions.

2.16. Phenotypic Assays to Evaluate MAPK Pathway Inhibition in StEVs‐Treated Cells

To investigate whether StEVs influence cell function through the MAPK pathway and whether U0126 effectively inhibits the activation of this pathway, thereby reversing or attenuating the biological effects of StEVs, the following three phenotypic assays were conducted:

CCK‐8 Assay: H1299 or A549 cells (2 × 10³ cells/well) were seeded in 96‐well plates and cultured in standard medium. After cell attachment, the medium was replaced with EV‐depleted medium. The cells were then treated under the following conditions: StEVs (20 µg/mL), U0126+StEVs (10 µM U0126 combined with 20 µg/mL StEVs) and U0126 alone (10 µM). The experiment was conducted over 4 days, with cell proliferation assessed on Days 1, 2, 3 and 4 using the Cell Counting Kit‐8 (CCK‐8; Dojindo Molecular Technologies, Japan). Absorbance at 450 nm was measured using a BioTek Synergy H1 Microplate Reader.

Wound Healing Assay: H1299 or A549 cell suspensions (5 × 10⁵ cells in 100 µL) were seeded into ibidi Culture‐Inserts (ibidi, Germany) placed in 24‐well plates. Once the cells reached confluence, the inserts were removed, creating a defined cell‐free gap (500 µm in width). The wells were then washed with PBS to remove cell debris, and the medium was replaced with EV‐depleted culture medium. The experiment was conducted under three conditions: StEVs (20 µg/mL), StEVs+U0126 (20 µg/mL StEVs+10 µM U0126) and U0126 alone (10 µM). Wound closure was monitored and imaged at 0, 18 and 24 h using a real‐time tracking microscope (LS720, Etaluma, USA). The wound area at each time point was compared to the initial wound area (0 h), and the percentage of wound closure was quantified. The wound area measurements were analyzed using ImageJ software.

Transwell Assay: H1299 cells (1 × 10⁴ cells/well) were seeded into the upper chambers of 24‐well Transwell plates (8 µm pore size, #3422, Corning, USA) in EV‐depleted medium. The lower chambers were filled with 600 µL of RPMI 1640 medium supplemented with 10% FBS to serve as a chemoattractant. After cell attachment, cells were treated under three conditions: StEVs (20 µg/mL), StEVs+U0126 (20 µg/mL StEVs+10 µM U0126) and U0126 alone (10 µM). Following 24 h of incubation, Transwell chambers were washed three times with PBS, fixed with 4% paraformaldehyde at room temperature for 20 min, and washed again with PBS. Cells that migrated to the bottom side of the membrane were stained with 0.5% crystal violet for 10 min, followed by multiple PBS washes. Non‐migrated cells remaining in the upper chamber were carefully removed using a cotton swab. Migrated cells were observed and imaged using an optical microscope, and migration was quantitatively assessed using ImageJ software.

2.17. Preparation of Multi‐Omics Grade EVs by Iodixanol Density Gradient Separation

One day prior to the experiment, EV‐depleted culture medium collected from 3D cell cultures was retrieved from –80°C storage and thawed at 4°C. The medium was then concentrated using Centricon Plus‐70 centrifugal filter units (100 kDa MWCO, Millipore). Concurrently, a pre‐prepared 5%–40% density gradient buffer (containing 0.25 M sucrose, 1 mM EDTA, 10 mM Tris‐HCl, and OptiPrep solution) was taken out from the 4°C refrigerator for further use. To establish the density gradient, 1 mL of the EV sample and 2 mL of 60% OptiPrep solution were carefully added to the bottom of Ultra‐Clear tubes (#344059, Beckman). The tube was gently tilted, and the gradient was layered as follows: 3 mL of 20% density buffer, followed by 3 mL of 10% density buffer and finally, 3 mL of 5% density buffer to complete the gradient. The samples were then subjected to ultracentrifugation at 120,000 × g for 16 h at 4°C using an SW 41 Ti rotor. Following ultracentrifugation, the EV solution was fractionated into 10 fractions using an automated Piston Gradient Fractionator. Based on NTA and WB validation, fractions 6–9 were collected. These fractions were further concentrated using a 10 kDa MWCO ultrafiltration filter (Millipore, USA) and subsequently purified via SEC as previously described to obtain the final EV preparations for downstream proteomic and small RNA sequencing analyses.

2.18. Proteomics

LC‐MS/MS analysis was performed on a Q Exactive mass spectrometer (Thermo Scientific) that was coupled to Easy nLC (Proxeon Biosystems, now Thermo Fisher Scientific). The peptides were loaded onto a reverse phase trap column (Thermo Scientific Acclaim PepMap100, 100 µm × 2 cm, nanoViper C18) connected to the C18‐reversed phase analytical column (Thermo Scientific Easy Column, 10 cm long, 75 µm inner diameter, 3 µm resin) in buffer A (0.1% Formic acid in water) and separated with a linear gradient of buffer B (84% acetonitrile and 0.1% Formic acid) at a flow rate of 300 nL/min. The mass spectrometer was operated in positive ion mode. MS data was acquired using a data‐dependent top20 method dynamically choosing the most abundant precursor ions from the survey scan (300–1800 m/z) for HCD fragmentation. Automatic gain control (AGC) target was set to 1e6, and maximum inject time to 50 ms. Dynamic exclusion duration was 30 s. Survey scans were acquired at a resolution of 60,000 at m/z 200 and resolution for HCD spectra was set to 15,000 at m/z 200, and the isolation width was 1.5 m/z. Normalized collision energy was 30 eV and the underfill ratio, which specifies the minimum percentage of the target value likely to be reached at maximum fill time, was defined as 0.1%. The instrument was run with peptide recognition mode enabled. The MS raw data for each sample were combined and searched using the MaxQuant 1.6.14 software for identification and quantitation analysis. Proteomic analysis was performed by Shanghai Applied Protein Technology Co. Ltd. using 4D‐label‐free technology. The criteria for differential proteomic analysis were set as |fold change| > 2 and p < 0.05. For more detailed information, please refer to the Supporting Information.

2.19. Small RNA Sequencing

EV small RNA: Total RNA was extracted and purified from the samples using the miRNeasy Serum/Plasma Advanced Kit (Qiagen, #217204) according to the manufacturer's instructions. RNA concentration and purity were assessed using the RNA Nano 6000 Assay Kit on an Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). For small RNA library preparation, 1–500 ng of total RNA per sample was used as input material. The sequencing library was generated using the QIAseq miRNA Library Kit (Qiagen, Frederick, MD) following the manufacturer's protocol, with index codes added to attribute sequences to each sample. Reverse transcription (RT) primers with unique molecular indices (UMIs) were introduced during cDNA synthesis and PCR amplification to enable accurate quantification of miRNA expression. Finally, library quality was assessed using the Agilent Bioanalyzer 2100 and qPCR analysis. Clustering of the index‐coded samples was performed on the cBot Cluster Generation System using the TruSeq PE Cluster Kit v3‐cBot‐HS (Illumina, San Diego, CA, USA) according to the manufacturer's instructions. After cluster generation, libraries were sequenced on the Illumina NovaSeq platform (Illumina, San Diego, CA, USA), and raw sequencing reads were generated. Library construction and sequencing were carried out by Biomarker Technologies Co. Ltd. (Beijing, China). The detailed steps for cellular small RNA sequencing can be found in the Supporting Information.

2.20. Post‐Small RNA‐Seq Analysis

The raw small RNA‐Seq data were processed by quality control using FastQC and were subjected to adapter and quality trimming using Cutadapt. The cleaned reads were aligned to the reference human genome (hg38) sourced from Ensembl by utilizing the mapper.pl script of miRDeep2. The expression levels of known miRNAs from miRBase were assessed using the quantifier function of miRDeep2. Differentially expressed miRNAs were identified using DESeq2 by comparing SoEVs with StEVs and SoCells with StCells, applying the criteria set at a log₂ fold change (FC) ≥0.58 or ≤–0.58 and an adjusted p value ≤ 0.05. KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was conducted utilizing the clusterProfiler package. Target genes of differentially expressed miRNAs were obtained from Targetscan, miRDB and the miRTarbase databases. The differentially expressed miRNAs were visualized using the ggplot2 package for volcano plots and the ComplexHeatmap package for heatmaps. Additionally, Venn diagrams were created using Evenn software.

2.21. EV MicroRNA Extraction and qRT‐PCR

The isolated EVs were subjected to lysis to extract and purify microRNA (miRNA) using the miRNeasy Serum/Plasma Kit (#217184, Qiagen) following the manufacturer's manual. Nanodrop One (ThermoFisher, USA) was used to determine the miRNA concentration. For RT‐qPCR, reverse transcription was performed using the miRCURY LNA RT kit (#339340, Qiagen), and the corresponding miRNA levels were calculated by Ct values after synthesis from the miRCURY LNA SYBR Green PCR kit (#339345, Qiagen). Data were analyzed using the 2^−ΔΔCt method. U6 was used as the reference gene, and primer sequences are shown in Table S2.

2.22. In Vivo Tumour Growth Assay

All animal experiments were approved by the Animal Research Ethics Sub‐Committee of City University of Hong Kong and conducted in accordance with institutional guidelines for the care and use of laboratory animals. Female NOD/SCID mice (4–6 weeks old; n = 5 per group) were subcutaneously injected in the right flank with 3 × 10⁶ luciferase‐labelled A549 cells suspended in 100 µL sterile PBS. Mice were randomly assigned to three treatment groups (PBS, SoEVs or StEVs) prior to the first EV administration. EVs were isolated from A549 cells cultured in 3D soft (SoEVs) or stiff (StEVs) TOCNF/GelMA hydrogels as described above, quantified by protein concentration using the BCA assay, and resuspended in sterile PBS. Mice received tail vein injections of 20 µg EV protein in 100 µL PBS twice per week for 5 weeks, resulting in a total of nine injections. The PBS control group received an equivalent volume of sterile PBS on the same schedule. Tumour length (L) and width (W) were measured with digital calipers every 2 days, and tumour volume was calculated as (L × W²)/2. Bioluminescence imaging was performed on Day 33 as a terminal assessment prior to euthanasia using the IVIS Spectrum system (PerkinElmer) following intraperitoneal (i.p.) injection of D‐luciferin (150 mg/kg) 15 min before imaging. Luminescence intensity was quantified using Living Image software (PerkinElmer). On Day 34, all mice were euthanized, and tumours were excised, photographed and weighed.

2.23. Statistical Analysis

All statistical analyses were conducted using GraphPad Prism 9.5.1 and Origin 2021. Data are presented as the mean ± standard error of the mean (SEM) from at least three independent experiments. Statistical analysis was carried out using an unpaired two‐tailed Student's t test when two groups were compared, and by one‐way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was performed to compare multiple groups, with statistical significance indicated as follows: (*p < 0.05, **p < 0.01, *p < 0.001, ****p < 0.0001).

3. Results and Discussion

3.1. Design and Characterization of 3D Nanofibrillar Hydrogels With Varying Stiffness

The ECM in the human body is a 3D fibrous network that plays a pivotal role in tissue homeostasis and cellular dynamics. In cancer, pathological ECM remodelling‐characterized by excessive collagen deposition‐leads to increased stiffness in the tumour microenvironment, profoundly influencing disease progression (Nia et al. 2020, 2025). To more accurately replicate this 3D architecture in vivo, this study introduced an innovative hybrid hydrogel composed of TEMPO‐oxidized cellulose nanofibrils (TOCNF), a plant‐derived biomaterial, integrated with gelatin methacryloyl (GelMA), a widely used biomaterial in tissue engineering (Tamo 2024; Zhang et al. 2023). As depicted in Figure 1a, the hybrid hydrogel was synthesized via UV crosslinking (405 nm), yielding a stable macroscopic structure, confirming the method's efficacy. FT‐IR analysis (Figure 1b) revealed significant changes in the O–H stretching vibration peak (3500 cm), broadening in the hybrid material compared to pure TOCNF, indicating hydrogen bonding between the components. Furthermore, a red shift in the C═O stretching vibration (from 1734 to 1631 cm) suggests hydrogen bond interactions between the carboxyl groups of TOCNF and the amide groups of GelMA (Lohrasbi et al. 2019; Tao et al. 2019). In mechanical assessments, the Young's modulus is an important indicator of hydrogel stiffness, directly affecting cell‐matrix interactions, particularly in the tumour microenvironment, where stiffness variations have a significant impact on cell behaviour and EV functions (Guimarães et al. 2020). After optimization (Figure S1), the Young's modulus of the soft matrix hydrogel is 2 kPa, while the stiff matrix hydrogel has a modulus of 20 kPa (Figure 1c). This design simulates the stiffness difference between healthy lung tissue and pathological lung tissue, providing physiological relevance for subsequent studies on the tumour microenvironment (Guo et al. 2022; Booth et al. 2012; Butcher et al. 2009; Shukla et al. 2016). Swelling behaviour, an indicator of hydrogel's water absorption and nutrient exchange capacity, is shown in Figure 1d, where both hydrogels demonstrated favourable absorption, supporting cell growth and nutrient transfer in a 3D culture context, akin to the in vivo ECM. The pore size distributions of stiff and soft hydrogels, both centred around 20–30 µm, indicated minimal pore size variation between the two matrices, as evidenced by SEM images (Figure 1e,f). Additionally, biodegradability under enzyme treatment (Figure 1g) confirms that the hydrogel can mimic the dynamic in vivo environment, with a notable increase in degradation rate under enzymatic conditions. These findings confirm the successful construction of ECM‐mimicking hydrogels with tunable mechanical properties and favourable physicochemical characteristics for 3D culture.

FIGURE 1.

Preparation and characterization of TOCNF/GelMA hybrid hydrogels. (a) A schematic showing TOCNF and GelMA are mixed and crosslinked under 405 nm UV light to form a 3D hydrogel network. (b) FT‐IR spectra showing distinct characteristic peaks for TOCNF, GelMA and their mixture. (c) Young's modulus analysis for soft and stiff matrix hydrogels. (d) Swelling ratio of soft and stiff matrix hydrogels. (e, f) Pore size distribution of the stiff and soft hydrogels, with statistical analysis showing that the pore sizes are primarily concentrated between 20 and 30 µm; inset shows SEM images of the hydrogels. Scale bar: 20 µm. (g) Degradation rate of the hydrogels under enzyme treatment across different time points.

3.2. The 3D Hybrid Hydrogel Exhibits Excellent Biocompatibility

Following the synthesis and characterization of the physicochemical properties of the TOCNF/GelMA hybrid hydrogel, its biocompatibility was evaluated using multiple complementary assays. Cell viability of H1299 and A549 cells encapsulated in the hydrogels at various time points (Days 1, 7 and 15) was first assessed using Calcein‐AM/PI staining. In the H1299 cell line (Figure 2a,c), viability remained consistently high (>80%) throughout the culture period in both soft and stiff hydrogels, indicating excellent biocompatibility and effective support for cell growth. Similarly, A549 cells (Figure 2b,d) exhibited high viability under both stiffness conditions, confirming the hydrogel's suitability for long‐term culture. Tumour spheroids formed in both hydrogel types after 7 days of culture (Figure 2e,f), with morphological differences observed between stiffness conditions. Spheroids in soft matrices tended to display a larger size and area compared with those in stiff matrices, a phenomenon consistently observed in both cell lines and in agreement with a previous report that spheroids expand more readily in softer matrices (Chin et al. 2023). Additional assays further confirmed the hydrogel's ability to support healthy cell growth (Figures S2 and S3). EdU staining demonstrated active cell proliferation under both stiffness conditions, with EdU‐positive cells showing red fluorescence and nuclei counterstained with DAPI in blue. Annexin V‐FITC/PI staining further revealed low levels of apoptosis, characterized by non‐apoptotic cells with no fluorescence, early apoptotic cells with green fluorescence, late apoptotic cells with both green and red fluorescence, and necrotic cells with red fluorescence only. A direct comparison with GelMA hydrogels (Figure S4) showed that the TOCNF/GelMA hybrid hydrogel maintained higher cell viability, underscoring its robust biocompatibility. Collectively, these results highlight the hybrid hydrogel's suitability as a reliable 3D culture platform for subsequent biological research.

FIGURE 2.

Biocompatibility evaluation of TOCNF/GelMA hybrid hydrogels. (a) Cell viability of H1299 cells in soft and stiff hydrogels at different culture times (Days 1, 7 and 15) determined by Calcein‐AM/PI staining. Green indicates live cells; red indicates dead cells. (b) Cell viability of A549 cells in soft and stiff hydrogels at different culture times (Days 1, 7 and 15) determined by Calcein‐AM/PI staining. (c) Quantitative analysis of H1299 cell viability in soft and stiff hydrogels (n = 3). (d) Quantitative analysis of A549 cell viability in soft and stiff hydrogels (n = 3). (e) Tumour spheroid morphology of H1299 cells after 7 days of culture. (f) Tumour spheroid morphology of A549 cells after 7 days of culture. Scale bars: 100 µm; the scale bar in the first row applies to all images.

3.3. ECM Stiffness Modulates the Physicochemical Properties of EVs

Following the characterization of TOCNF/GelMA hydrogels, a 7‐day 3D culture using H1299 and A549 cells was conducted to investigate ECM mechanics' impact on EVs (Figures 3a and S5). This hybrid hydrogel system mimics lung tumour matrices with varying stiffness, providing a physiologically relevant model for EV studies. To ensure the collection of sufficient EV samples for subsequent analysis, we regularly replaced the EV‐depleted medium and collected the medium on Days 3, 5 and 7 (Figure 3b). The detailed procedures for 3D culture and EV isolation are outlined in the Experimental section (Sections 2.7 and 2.8), with all steps conducted in accordance with the MISEV2018/2023 guidelines to ensure the rigor and reproducibility of the experimental results (Théry et al. 2018; Welsh et al. 2024). The selection of a 7‐day culture period was based on two key considerations: first, a shorter culture period could result in an insufficient yield of EVs, which would be inadequate for downstream experiments; second, an extended culture period might increase the risk of contamination (Rocha et al. 2019). After successfully isolating the EVs, we performed a comprehensive characterization to assess the impact of mechanical stiffness on EV properties. Both EV types exhibited a size distribution predominantly below 200 nm, as determined by NTA, indicating that their sizes fall within the typical biological range (Figure 3c,d). The average EV size for both cell types showed no significant difference between H1299‐ and A549‐derived EVs, regardless of whether they were cultured in soft or stiff matrices (Figure S6a). It is evident that for both StEVs and SoEVs, the proportion of EVs in the 100–200 nm range is significantly higher compared to the 30–100 and 200–500 nm ranges (Figure S6b,c). The zeta potential analysis of EVs derived from H1299 and A549 cells is presented in Figure 3e. The results indicated that SoEVs have a more negative zeta potential compared to StEVs. This suggested the ECM stiffness may influence the surface characteristics or membrane composition of EVs, potentially affecting their stability and interactions with recipient cells. This observation contrasts with results obtained under different stiffness conditions in 2D cultures, which may be attributed to differences in the cell lines used or the culture environment (3D vs. 2D) (Patwardhan et al. 2021). TEM analysis showed that SoEVs and StEVs derived from H1299 and A549 cells exhibited the characteristic concave, saucer‐shaped morphology, consistent with previous reports, thereby validating sample quality and confirming typical EV structure (Figure 3f) (Chuo et al. 2018). WB analysis of H1299 cells (SoCell, StCell) and their EVs (SoEVs, StEVs) showed EV‐specific markers (Flotillin‐1, Syntenin‐1, TSG101, CD81, CD9) in EVs and Calnexin only in cells, confirming EV identity and purity (Figure 3g). A549 cells and their EVs showed the same pattern (Figure 3h). AFM revealed EVs with spherical or ellipsoidal shapes, consistent with their typical morphology in physiological environments (Figure 3i) (Parisse et al. 2017; Sharma et al. 2018). Height measurements showed that most EVs from H1299 cells were within 20 nm (Figure 3j), and those from A549 cells were mainly within 15 nm (Figure S6d), consistent with previous reports (Bairamukov et al. 2022; Gazze et al. 2021). Young's modulus analysis revealed that StEVs had lower values than SoEVs, suggesting that matrix stiffness may modulate EV mechanical properties by altering the expression of membrane rigidity‐related proteins (Figure 3k). To explore this mechanism, WB analysis was performed for ERM (Ezrin/Radixin/Moesin) and Vinculin, two proteins crucial for cytoskeletal and membrane mechanics (Parihar et al. 2022; Sato et al. 1992). Results showed higher ERM and Vinculin levels in HSoEVs and ASoEVs than in their stiff‐matrix counterparts (HStEVs and AStEVs) (Figure S7a), with quantitative data (Figure S7b,c) further supporting that matrix stiffness affects EV biogenesis via cytoskeletal protein regulation and membrane mechanical properties. In summary, EVs secreted by H1299 and A549 cells cultured in TOCNF/GelMA 3D matrices with varying stiffness were characterized using NTA, TEM, WB and AFM. The findings demonstrate that mechanical stiffness influences the physical properties of EVs, which may in turn affect their function and biological impact, underscoring the importance of matrix stiffness in EV‐mediated cellular communication.

FIGURE 3.

3D culture and EV isolation and characterization. (a) Schematic representation of 3D culture in soft and stiff hydrogel matrices. (b) Time points for exosome collection during the culture process. (c) NTA analysis: size distribution of EVs from H1299 cells cultured in soft and stiff hydrogels. (d) Size distribution of EVs from A549 cells cultured in soft and stiff hydrogels, as determined by NTA. (e) Zeta potential analysis of EVs from H1299 and A549 cells. (f) TEM analysis showing the morphological characteristics of exosomes from H1299 and A549 cells. (g) WB validation of EV markers in H1299 cells and their derived EVs. (h) WB analysis of EV markers in A549 cells and their derived EVs. (i) AFM analysis showing the surface morphology of EVs derived from H1299 and A549 cells. (j) Height distribution analysis of H1299‐derived EVs, measured by AFM. (k) Young's modulus analysis of EVs from H1299 and A549 cells. Scale bars: 100 nm (f), 400 nm (i). Data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by unpaired, two‐tailed Student's t‐tests.

3.4. StEVs Enhance Tumour Cell Growth In Vitro and In Vivo

To further evaluate the biological impact of stiffness‐modulated EVs, we first assessed their effects on the migration and proliferation of recipient tumour cells. Wound healing assays were performed at 0, 18 and 24 h using H1299 and A549 cells treated with PBS, SoEVs or StEVs (Figure 4a,b). The StEVs treatment group exhibited markedly enhanced wound closure at 24 h compared with both the PBS control and SoEVs groups, indicating a pronounced pro‐migratory effect, consistent with previous reports under 2D conditions (Patwardhan et al. 2021). Quantitative analysis confirmed significantly higher migration rates in the StEVs group at both 18 and 24 h for both cell lines (Figure S8a,b). CCK‐8 assays further demonstrated that StEVs significantly promoted the proliferation of H1299 and A549 cells compared to PBS and SoEVs groups, with consistently higher absorbance values (Figure S8c,d). To validate these findings in vivo, we established a subcutaneous xenograft tumour model, in which mice received tail vein injections of StEVs, SoEVs or PBS. Over the observation period, the StEV‐treated group exhibited markedly stronger bioluminescence signals (Figure 4c), larger excised tumours (Figure 4d) and greater tumour weights (Figure 4e) compared with the other groups. Consistently, longitudinal tumour volume measurements (Figure 4f) revealed the fastest growth rate in the StEV group, whereas the PBS and SoEV groups displayed slower progression. These in vivo observations, together with our in vitro results, consistently demonstrate that EVs derived from stiff 3D matrices possess enhanced tumour‐promoting capacity, underscoring the critical role of mechanical stiffness in modulating EV‐mediated cancer progression.

FIGURE 4.

StEVs promote tumour cell migration and in vivo tumour growth. (a) Wound healing assay of H1299 cells cultured in soft and stiff hydrogels, imaged at 0, 18 and 24 h. (b) Wound healing assay of A549 cells cultured in soft and stiff hydrogels, photographed at 0, 18 and 24 h. (c) Bioluminescence images of subcutaneous xenograft tumours derived from luciferase‐expressing A549 cells following PBS, SoEVs or StEVs treatment (n = 5). (d) Photographs of excised tumours from each treatment group at the endpoint. (e) Tumour weights at the endpoint (n = 5). (f) Tumour volume growth curves over time (n = 5). Scale bars: 100 µm; the scale bar in the first row applies to all images. Data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one‐way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test.

3.5. Proteomics Reveals Distinct EV Protein Cargo Under Varying 3D ECM Stiffness

To investigate how 3D ECM stiffness influences EV protein composition, we performed a 4D label‐free quantitative proteomic analysis comparing SoEVs and StEVs. EVs were purified using a two‐step strategy combining density gradient centrifugation and SEC to ensure high quality for downstream analyses (Figure 5a,b) (Wang et al. 2024; Zhang et al. 2023; Dhondt et al. 2020). NTA of density gradient fractions (Figure 5c) showed low particle concentrations in fractions 1–5, with a marked increase from fraction 6 and a peak at fraction 8, indicating EV enrichment. The ratio of particle number to protein concentration (particles/µg protein) (Figure 5d) confirmed that fractions 6–9 had high purity, with fraction 8 showing the highest purity and concentration (Webber and Clayton 2013). Western blot analysis of EV markers, including Syntenin‐1, Flotillin‐1 and Tsg101 (Figure 5e), further validated EV enrichment in fractions 6–9, with the strongest bands observed in fractions 7 and 8. Fractions 6–9 were therefore pooled and further purified by SEC before proteomic analysis. A total of 3965 proteins were shared between SoEVs and StEVs (Figure 5f), representing proteins potentially involved in fundamental EV structure and function. Differential expression analysis revealed 502 proteins upregulated and 165 proteins downregulated in StEVs relative to SoEVs (Figure 5g), indicating that ECM stiffness has a pronounced regulatory effect on EV cargo composition. Principal component analysis (PCA) showed clear separation between SoEVs and StEVs, with tight clustering of replicates within each group (Figure 5h). Gene Ontology (GO) enrichment analysis (Figure S9) indicated that differentially expressed proteins were predominantly localized to the cytoplasm, extracellular regions and exosomes, and were enriched in processes such as protein transport. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis (Figure 5i) revealed significant enrichment in pathways related to tumour progression and cell communication, including MAPK, PI3K‐Akt and ECM‐receptor interaction. To validate the proteomic results, several metabolism‐related differentially expressed proteins (e.g., ALDOA, ALDOC, LDHA, PGAM1) were selected for targeted Western blot analysis (Figure S10a), and their relative expression levels were quantified (Figure S10b). The trends observed in WB were consistent with the proteomic data, confirming the robustness and reliability of the dataset. Collectively, these findings demonstrate that ECM stiffness profoundly influences the protein cargo of EVs, leading to enrichment of key signalling pathways associated with tumour progression, cell‐matrix interactions and intercellular communication.

FIGURE 5.

Impact of mechanical stiffness on EV proteomic analysis. (a) Diagram illustrating the EV isolation and purification process. The image shows the combination of density gradient centrifugation and size exclusion chromatography for EV purification. (b) EV purification process. Density gradient centrifugation separates the samples into 10 fractions, followed by SEC to obtain high‐quality EV samples. (c) NTA analysis of particle concentration in each fraction. (d) Particle purity analysis in each fraction. (e) WB analysis of EV markers Syntenin‐1, Flotillin‐1 and Tsg101. (f) Venn diagram of protein identification results. (g) Volcano plot of differentially expressed proteins. (h) Principal component analysis (PCA). (i) KEGG pathway enrichment analysis of differentially expressed proteins between StEVs and SoEVs.

3.6. Small RNA Sequencing Reveals EV miRNA Profiles Shaped by ECM Stiffness

Following the proteomic study, we next analyzed the small RNA to further elucidate the molecular mechanisms by which ECM stiffness influences EV‐mediated tumour progression. EVs used for small RNA sequencing were isolated using the same two‐step purification strategy (density gradient centrifugation followed by size exclusion chromatography) as in the proteomic analysis, ensuring high purity and consistency across datasets. Small RNA sequencing revealed that most identified miRNAs clustered within 21–23 nt (Figure 6a). Venn diagram analysis identified 427 miRNAs shared between SoEVs and StEVs, with a substantial proportion of StEV miRNAs overlapping with entries in established EV databases (Vesiclepedia, ExoCarta, EVmiRNA) (Figure 6b,c), supporting the reliability of the dataset (Chitti et al. 2023; Keerthikumar et al. 2016; Liu et al. 2019). Differential expression analysis revealed 37 upregulated and 54 downregulated miRNAs in StEVs compared to SoEVs (Figure 6d), suggesting that matrix stiffness affects selective miRNA loading into EVs, potentially influencing tumour cell behaviour. Heatmap analysis of these differentially expressed miRNAs (Figure 6e) showed distinct clustering between StEVs and SoEVs, with subsets of miRNAs markedly enriched in StEVs. Among the differentially expressed miRNAs, we selected several miRNAs that have been reported in the literature as being relevant to non‐small cell lung cancer research, including miR‐1246, miR‐1290, miR‐210‐3p, miR‐330‐3p and miR‐125a‐3p, and performed qPCR validation (Jiang et al. 2010; Jin et al. 2018; Kim et al. 2016; Mo et al. 2015; Świtlik et al. 2018; Wei et al. 2018; Zhang et al. 2015). The results indicated that miR‐1246 and miR‐1290 were significantly upregulated in StEVs compared to SoEVs (Figure 6f), in agreement with the sequencing data. KEGG pathway enrichment analysis of differentially expressed miRNA targets (Figure 6g) revealed significant enrichment in oncogenic signalling pathways, including MAPK, Ras and focal adhesion. Among these, the mitogen‐activated protein kinase (MAPK) pathway stood out as a central node, in agreement with proteomic results (Figure 5i) and its established role in regulating proliferation, migration, apoptosis and matrix stiffness responses (Guo et al. 2020; Zhang and Liu 2002; Liu et al. 2020; Provenzano et al. 2009). To validate these predictions, recipient cells were treated with StEVs, SoEVs or PBS for 12 h, followed by protein extraction and Western blot analysis of key signalling molecules (Figure 6h). StEV treatment significantly increased ERK1/2 phosphorylation compared with SoEVs and PBS, whereas activation of PI3K‐Akt, TGF‐β and JAK‐STAT pathways was substantially lower, underscoring MAPK/ERK1/2 signalling as the predominant pathway through which StEVs enhance tumour cell migration and proliferation. These findings highlight ECM stiffness as a critical determinant of EV cargo and its downstream functional effects.

FIGURE 6.

Small RNA analysis of EVs under varying mechanical stiffness. (a) Length distribution of known miRNAs. (b) Venn diagram of expressed miRNAs in SoCell, StCell, SoEVs and StEVs, showing that StEVs harbour more unique miRNAs than SoEVs. (c) Overlap of miRNAs in StEVs and SoEVs with public EV databases (Vesiclepedia, ExoCarta, EVmiRNA). (d) Volcano plot of differentially expressed miRNAs in EVs. (e) Heatmap of differentially expressed miRNAs ranked by log₂FC in EVs. (f) qPCR validation of differentially expressed miRNAs in EVs. (g) KEGG enrichment analysis of target genes of up‐regulated miRNAs in StEVs. (h) WB analysis of signalling pathway activation in recipient cells treated with StEVs or SoEVs. Data are presented as the mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by unpaired, two‐tailed Student's t‐tests.

3.7. StEV‐Induced Migration and Proliferation Depend on MAPK/ERK1/2 Activation

Given the observation that StEVs predominantly activate the MAPK/ERK1/2 pathway, we next examined whether their pro‐migratory and pro‐proliferative effects are dependent on this signalling route. The selective ERK1/2 inhibitor U0126 was used to test the requirement of MAPK/ERK1/2 activation for StEV‐induced phenotypes (Konieczny et al. 2022; Wang et al. 2025; Yuan et al. 2023). Mechanistic validation was first performed in H1299 cells. Cells were divided into three groups: StEVs alone, U0126 alone, and a combination of StEVs and U0126. This design enabled direct evaluation of MAPK dependency by comparing phenotypic outcomes with and without pathway inhibition. The untreated and SoEVs groups, previously characterized (Figure 4), were omitted here to specifically delineate the contribution of MAPK/ERK1/2 signalling to StEV‐induced effects. Transwell migration assays confirmed that StEVs significantly promoted cell migration, an effect that was effectively inhibited by U0126 treatment (Figure 7a,b). Similarly, wound healing assays showed enhanced migration in the StEV group, which was notably suppressed by U0126 (Figure 7c,d). CCK‐8 assays further confirmed that StEVs significantly increased cell proliferation from Days 1 to 4, while U0126 effectively attenuated this effect (Figure 7e). Western blot analysis confirmed that total ERK1/2 levels remained unchanged, while StEV‐induced ERK1/2 phosphorylation was strongly reduced by U0126, with the lowest p‐ERK1/2 levels in the inhibitor‐only group (Figure 7f,g). To confirm the generality of these findings, additional validation was performed in A549 cells, focusing on wound healing and CCK‐8 assays (Figure S11). The results showed the same trend as in H1299 cells, with U0126 treatment effectively reversing the pro‐migratory and pro‐proliferative effects of StEVs. Collectively, these results elucidate a molecular mechanism whereby StEVs enhance tumour cell migration and proliferation via MAPK/ERK1/2 pathway activation, underscoring the pivotal role of stiffness‐modulated EVs in regulating tumour progression through mechanical signalling cues in the extracellular matrix.

FIGURE 7.

StEVs promote cell proliferation and migration via the MAPK/ERK1/2 signalling pathway. (a) Transwell migration assay demonstrates that StEV treatment promotes cell migration, whereas U0126 inhibits this effect. (b) Quantification of migrated cells. (c) Wound healing assay shows that StEV‐treated cells exhibited faster wound closure, while U0126 inhibited StEV‐induced migration enhancement. (d) Quantification of wound healing assay. (e) CCK‐8 proliferation assay. (f) WB analysis of U0126 inhibition on StEV‐induced MAPK pathway activation. (g) Quantification of the p‐ERK/ERK ratio. Scale bars: 50 µm (a), 100 µm (c); the scale bar in the first row applies to all images in (c). Data are presented as the mean ± SEM. Statistical analysis was carried out using unpaired two‐tailed Student's t test when two groups were compared, and one‐way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was performed to compare multiple groups, with statistical significance indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

4. Conclusion

In summary, this study presents a TOCNF/GelMA‐based 3D hydrogel system with excellent biocompatibility, enabling systematic investigation of how matrix stiffness modulates the physicochemical properties and pro‐tumour functions of EVs. Compared to the soft matrix‐derived counterparts (SoEVs), EVs generated in stiff matrices (StEVs) exhibit distinct biophysical features and carry significantly different protein and microRNA cargo profiles. Functionally, StEVs more potently promote tumour cell proliferation and migration in vitro, and the in vivo xenograft model further demonstrated that StEVs accelerate tumour growth compared with SoEVs and PBS controls. Multi‐omics analyses, combined with pharmacological inhibition, reveal that StEVs activate the MAPK/ERK1/2 signalling pathway in recipient cells, positioning stiffness as a critical upstream regulator of EV‐mediated signalling. While this study provides key insights into stiffness‐dependent EV regulation in a 3D context, future work should incorporate lipidomic profiling and studies in diverse tumour models. Such efforts will help clarify the mechanistic complexity and therapeutic potential of mechanically modulated EV communication.

Author Contributions

Zesheng Wang: conceptualization (equal), data curation (lead), formal analysis (lead), methodology (lead), writing‐original draft (lead), writing‐review & editing (lead). Xulin Xie: resources (supporting), methodology (supporting). Yicen Zhou: formal analysis (supporting), writing‐review & editing: (supporting). Huimin He: writing‐review & editing (equal). Zhenjun Guo: conceptualization (supporting), writing‐ review & editing (supporting). Zhengdong Zhou: conceptualization (supporting), investigation (supporting), writing‐review & editing (supporting). Beilei Liu: investigation (supporting), writing‐review & editing: (supporting). Jiayu Sun: conceptualization (supporting), resources (supporting), writing‐review & editing (supporting). Wenxiu Li: conceptualization (supporting), formal analysis (supporting), resources (supporting), writing‐review & editing (supporting). Qichang Nie: methodology (supporting), resources (supporting). Jun Dai: data curation (supporting), formal analysis (supporting). Wenkai Yi: conceptualization (supporting), writing‐review & editing (supporting). Xiaoyu Zhou: investigation (supporting), resources (supporting), writing‐review & editing (supporting). Jian Yan: writing‐review & editing (supporting). Mengsu Yang: conceptualization (lead), project administration (lead), resources (lead), supervision (lead), funding acquisition (lead), writing‐review & editing (supporting).

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supplementary Information

Wang, Z. , Xie X., Zhou Y., et al. 2025. “3D Nanofibrillar Matrix Stiffness Modulates Extracellular Vesicle Cargo and Pro‐Tumour Functions.” Journal of Extracellular Vesicles 14, no. 10: e70165. 10.1002/jev2.70165

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.