Generalizability of Enzyme‐Based Isolation Approach for Extracellular Vesicles From Traditional Medicinal Plants

ABSTRACT

Extracellular vesicles (EVs) from plants, particularly traditional medicinal plants, have garnered therapeutic interest. However, therapeutic utility requires establishing effective extraction and purification protocols for manufacturing. Here, we systemically optimized an enzyme‐based EV extraction method and demonstrated the generalizability with different parts from seventeen medicinal plants, fresh, frozen or dried. The enzymatic pulping process (henceforth denoted as Phyto‐EVpure) of fresh ginger, Morus alba leaves and Isatis indigotica Fort. roots improved EV yield and purity over grinding. Ginger EVs processed via Phyto‐EVpure preserved bioactivities and conferred hepatoprotection in a non‐alcoholic fatty liver disease model in vitro and in vivo. Phyto‐EVpure efficiently isolated EVs from leaf, rhizome and root stored fresh, frozen or dried. Phyto‐EVpure enabled us to demonstrate that EVs from phloem of previously intractable dried vine samples exhibited greater antitumour activities than xylem in vitro. Collectively, we provided a manufacturing‐friendly protocol for medicinal plant EV isolation and demonstrated the generalizability of Phyto‐EVpure for efficient isolation of EVs from a wide spectrum of traditional medicinal plants under different conditions.

1. Introduction

Plant extracellular vesicles (EVs) play vital roles in physiological processes and defence against invaders in plants (Cai et al. 2018; Lian et al. 2022; Pinedo et al. 2021) and have also been exploited as therapeutics and drug delivery carriers because of the abundance of active biomolecules, biocompatibility, and stability in the human body, even when ingested (Dad et al. 2021; Zhang et al. 2016b). In particular, EVs from traditional medicinal plants focus on their intrinsic pharmacological activities and are scalable and sustainable resources (Cao et al. 2023). Notably, ginger, aloe, and grape EVs have been tested in clinical trials for oral mucositis and gastrointestinal diseases (Feng et al. 2023).

The commercial development of plant EVs is still in its infancy, and affordable, scalable protocols for high‐quality EV isolation are lacking. Unlike mammalian EVs (Janouskova et al. 2022), plant EVs require two‐step isolation, pulping and purification (Cong et al. 2022). Currently, grinding is extensively employed for pulping prior to purification by differential and sucrose gradient ultracentrifugation (Lee et al. 2020; You et al. 2021). However, low EV yield and purity (Lian et al. 2022) and lack of applicability to dried medicinal plants indicate that grinding is not ideal for manufacturing diverse types of EVs. Alternatives such as utilizing apoplast washing fluid (Chen, He et al. 2022; Rutter and Innes 2017) or cellulase and pectinase digestion of the cell wall (Ou et al. 2023; Zhao et al. 2023) can increase purity and yield, but are either technically demanding and not universal or require optimization for different plants, respectively. A unified pulping and purification protocol enables commercial manufacturers to invest in a single manufacturing workflow for scalable high‐yield and high‐purity production of multiple‐plant EVs, enabling economies of scale.

Here, we systematically optimized and developed a robust enzyme‐based pulping method for a wide spectrum of traditional medicinal plants spanning seventeen species and different plant parts and denoted this approach as Phyto‐EVpure. This approach includes EV depletion from enzymes, usage of optimal antibacterial reagents, and optimized reaction conditions for different medicinal plants under different conditions and verified its generalizability in 17 different medicinal plants. Importantly, EVs obtained from fresh ginger via Phyto‐EVpure under optimized conditions preserved membrane protein activity, showing greater yield and purity than those obtained by grinding. Ginger EVs containing unsaturated fatty acids mitigated fatty acid deposition in a non‐alcoholic fatty liver disease (NAFLD) model in vitro and in vivo. Phyto‐EVpure enabled the robust and efficient isolation of EVs from 17 traditional medicinal plants and their fresh, frozen, and dried parts. Isolation of EVs from dried vine samples, such as Marsdenia tenacissima (Roxb.) Wight et Arn., and Tripterygium wilfordii Hook. F. was achieved using Phyto‐EVpure. The ability to harvest high‐quality EVs from different plant parts enabled us to determine that EVs from the phloem showed higher antitumour activity than those from the xylem in vitro. Our study established a robust general pulping protocol suitable for scalable manufacturing, thus providing a path for commercial production of high‐yield, high‐purity plant EV products.

2. Materials and Methods

2.1. Animals

Adult C57BL/6 wild‐type (6–8‐week‐old) mice were used in all experiments (the number used is specified in the figure legend). Mice were housed under specific pathogen‐free (SPF) conditions in a temperature‐controlled room. All experimental procedures were conducted at the Animal Unit of Tianjin Medical University (Tianjin, China) in compliance with protocols approved by the Institutional Ethics Committee (Approval No. SYXK2023‐0004). Mice were sacrificed by cervical dislocation at the desired time points.

2.2. High‐Fat Diet‐Induced Non‐Alcoholic Fatty Liver Disease (NAFLD) Mouse Models

Mice with NAFLD were established as described previously (Zhuang et al. 2015). Briefly, C57BL/6 mice were randomized into two dietary regimens for 12 weeks: normal chow (10% fat) or high‐fat diet (HFD, 60% fat; D12492, Research Diets). NAFLD model mice received daily oral gavage of ginger‐derived EVs (2 × 10¹¹ particles dissolved in PBS) preprocessed with Phyto‐EVpure; the amount of ginger EVs was adopted from previous studies (Deng et al. 2017; Liu et al. 2020; Zhang et al. 2016a), followed by ultracentrifugation purification, and were administered consecutively for 4 weeks. Twenty‐four hours post‐final treatment, mice were euthanized. Tissues were either snap‐frozen in liquid nitrogen (−80°C storage) or fixed in 4% paraformaldehyde (4°C overnight).

2.3. Cell Culture

The human bronchial epithelial cell BEAS‐2B cell line was kindly provided by Prof. Jun Chen (Tianjin Medical University General Hospital, Tianjin, China) and was cultured as previously described (Lee and Ryu 2023). Human hepatocyte HL7702 and HepG2 cells were purchased from the Shanghai Institute for Biological Sciences (Chinese Academy of Sciences, China), and the human non‐small lung carcinoma cell line A549 and foetal kidney cell HEK293T were purchased from ATCC Biobank and cultured according to the manufacturer's instructions (Bounda et al. 2015; Han et al. 2016; Jurgielewicz et al. 2020). Briefly, cells were cultured in Dulbecco's Modified Eagle's medium (DMEM) or RPMI 1640 supplemented with 10% foetal bovine serum (FBS) and 1% penicillin and streptomycin (P/S) at 37°C in a humidified atmosphere with 5% CO₂.

2.4. Preparation of Enzyme Solutions

All enzymes were purchased from the same company to avoid variability, and the activity was tested prior to use. Cellulase (M29335, Lot Nos. M10065594, 230531100014, MERYE, China) and pectinase (S10007, Lot Nos. JS334544, N27IS233437, Yuanye, China) were dissolved with 20 mmol/L MES (M7670, Solarbio, China), 2 mmol/L CaCl₂, and 0.1 mol/L NaCl and adjusted to pH5.5 as per manufacturer's instructions at the concentration of 20 g/L (2%). EV depletion from cellulase and pectinase was adopted from fungal EV isolation as described previously (Logan et al. 2024). Briefly, cellulase and pectinase stock solutions (1:1 ratio) were centrifuged at 15,000 × g for 15 min at 4°C, and the recovered supernatant was filtered through 0.22 µm filters (SLGP033RB, Millipore, USA), followed by ultracentrifugation (Beckman Optimal‐100 XP, Beckman Coulter, Germany) at 100,000 × g for 1 h at 4°C. To confirm EV removal, the supernatant was subjected to an additional round of ultracentrifugation at 100,000 × g for 1 h at 4°C. The activity of cellulase and pectinase was assessed with the Cellulase Activity Assay Kit (BC2540) and Pectinase Activity Assay Kit (BC2630; Solarbio, China) as per manufacturer's instructions.

2.5. Isolation of EVs From Ginger by Grinding

EVs were isolated from ginger by the grinding method, which was conducted as described previously (Zhang et al. 2016a). Briefly, fresh ginger was purchased from the local market (Tianjin, China) and washed thoroughly under tap water for 0.5 h to remove traces of any dirt particles and weighed in which ginger was peeled. A total of 500 g of ginger was ground with a mixer (12‐Speed Countertop Blender, Oster, USA) at the maximum speed for 10 min, with every 1 min blending followed by a 1 min pause to prevent overheating. The juice obtained was filtered through gauze and centrifuged at 1000 × g for 10 min, 3000 × g for 20 min, and 10,000 × g for 40 min. The supernatant was ultracentrifuged (Beckman Optima XP‐100, Beckman Coulter, Germany) at 100,000 × g for 90 min, and the EV pellets were resuspended in PBS and passed through a 0.22 µm filter. The resuspended ginger EVs were layered on a linear sucrose gradient (8%, 30%, 45%, and 60%, V900116, Sigma‐Aldrich, USA), and the gradient was centrifuged at 150,000 × g for 2 h at 4°C (Zhang, Xiao, et al. 2016; Zhuang et al. 2015). Four fractions from sucrose gradients were collected and ultracentrifuged at 100,000 × g for 1 h at 4°C to pellet the ginger EVs. The recovered ginger EVs from the 30%/45% fractions were incubated at 37°C and 50°C in a water bath for 2 h, followed by visualization using high‐resolution transmission electron microscopy (TEM) (Hitachi HT7700, Japan).

2.6. Isolation of EVs From Different Medicinal Plants Pulped With Phyto‐EVpure

The plants were sterilized as described previously (Tewelde et al. 2020). Briefly, fresh medical plants were washed thoroughly with tap water for 0.5 h to remove traces of dirt. For example, ginger was cut into 1–2 mm³ pieces, placed in sterile conical flasks, and submerged in 70% ethanol (M34056‐500ML, MERYER, China) for 30 s with gentle shaking, followed by rinsing with cold ddH₂O five times briefly (the entire procedure was conducted in a sterile environment). To test the effect of different detergents and antibiotics, we submerged the ginger samples into (1) in 70% ethanol for 5 min, (2) in 0.5% NaClO (sodium hypochlorite, M98688‐100G, MERYER, China) for 20 min, (3) 3% H₂O₂ (hydrogen peroxide (88597, Sigma‐Aldrich, USA) for 20 min; in the enzyme digestion buffer containing, (4) 1% penicillin‐streptomycin (PS; 03.12001A, EallBio, China), (5) 100 U/mL Bacitracin (B8181, Solarbio, China) or (6) 0.25 µg/mL Amphotericin B (1397‐89‐3, Solarbio, China) for 30 min, respectively. Additionally, 1% PS, 100 U/mL Bacitracin and 0.25 µg/mL Amphotericin B were added to the buffer during the enzymatic digestion. To examine bacterial growth, the reaction solution (0.4 mL) was aspirated, spread onto Luria–Bertani (LB) agar plates, and incubated at 37°C for 24 h. To detect fungal contamination, the reaction solution (1 mL) was mixed with LB liquid medium (5 mL) and incubated overnight at 37°C, followed by glycogen staining on a slide (100 µL) with periodic acid‐Schiff (PAS) staining (G1366, Solarbio, China) according to the manufacturer's instructions.

To optimize the enzyme concentration, PS‐sterilized plant pieces were submerged in EV‐depleted enzyme solutions containing 1% PS and different concentrations of cellulase (C) and pectinase (P) with EV depleted including 1.0% C+0.5% P, 1.0% C+ 0.25% P, 0.75% C+0.25% P, 0.5% C+0.5% P, 0.5% C+0.25% P, 0.25% C+0.25% P, 0.25% C+0.125% P and 0.16% C+0.08% P, and incubated at 37°C for 6 h in the dark. Subsequently, the digested mixture was filtered through gauze to remove large plant tissue debris, and the supernatant was centrifuged at 3000 × g for 20 min and 10,000 × g for 40 min at 4°C. The supernatant was passed through a 0.22 µm filter and ultracentrifuged at 100,000 × g for 70 min to pellet the EVs. Processed EV suspensions were maintained at 4°C for immediate analysis or cryopreserved at −80°C for long‐term storage. To optimize the reaction time, PS‐sterilized ginger (160 g), Morus alba leaf (60 g), and Isatis indigotica Fort root (160 g) were cut into pieces and digested in enzyme buffer containing 1% cellulase, 0.5% pectinase, 20 mmol/L MES, 2 mmol/L CaCl₂, and 0.1 mol/L NaCl and 1% PS (with adjusted pH5.5) at 37°C for different durations (0.5, 1, 2, and 4 h) in the dark, followed by the isolation of EVs as described above. Roots were minced into 1–2 mm³ pieces using a garlic masher (SPOUNR, Germany). Optimized conditions were applied to other rhizomes, leaves, and roots (kindly provided by Tianjin University of Traditional Chinese Medicine, Tianjin, China) under different conditions. The enzyme solution was used throughout the study unless otherwise specified.

EVs isolated from dried Marsdenia tenacissima (Roxb.) Wight et Arn. (MT) (Kaifutang Pharmacy, Hebei province, China), dried MTs were milled (Rongshida, China) and sieved with a stainless‐steel mesh screen (2–4 mm), then soaked in the enzyme buffer containing 20 mmol/L MES, 2 mmol/L CaCl₂, and 0.1 mol/L NaCl (pH 5.5) with or without 1% PS at 4°C for different durations for softening, followed by the detection of microbial contamination with LB culture and EV morphologies with TEM. After determining the softening time, the softened MT pieces were digested with the enzyme solution at 37°C for 0.5 h in the dark, followed by isolation of EVs by ultracentrifugation, as described above. To optimize the reaction time, PS‐sterilized and softened (8 h) MT samples (150 g) were digested in an enzyme solution at 37°C for different durations (0.5, 1, 2, and 4 h) in the dark, and EVs were isolated as described above. Processed EV suspensions were maintained at 4°C for immediate analysis or cryopreserved at −80°C for long‐term storage. The optimized softening and enzyme reaction conditions were used for EV isolation from other dried vine samples, phloem, and xylem. The phloem and xylem were manually separated as described previously (Kumar, Bhushan et al. 2022; Mei et al. 2019), and the rest of the steps were the same as those for the MT samples. EVs derived from human HEK293T cells were isolated as described previously (Gao et al. 2018).

2.7. Characterization of Plant EVs

Plant EVs were characterized according to guidelines for mammalian EVs (Pinedo et al. 2021; Welsh et al. 2024). Nanoparticle tracking analysis (NTA) was performed using a Nanosight NS300 system (Malvern, UK). Five sequential 60s captures were acquired per sample under standardized parameters (cell temperature: 25°C; syringe flow rate: 30 µL/s). EVs yield (particles/g) was calculated by normalizing the particle concentration (particles/mL) to the initial plant biomass. The EV's ultrastructure was analyzed by transmission electron microscopy (TEM; Hitachi HT7700, Japan) following established protocols (Gao et al. 2018). Briefly, EVs were fixed in 4% paraformaldehyde (PBS‐based) and adsorbed onto carbon‐coated copper grids (20 min, RT). The grids were sequentially washed with PBS and ddH₂O (5 min each) and then fixed with 1% glutaraldehyde (5 min), followed by eight ddH₂O washes (2 min each). Negative staining was achieved by sequential incubation with uranyl oxalate (pH 7.0, 5 min) and ice‐cold methylcellulose‐uranyl acetate (4% uranyl acetate/2% methylcellulose, 1:9 v/v, 10 min). The grids were air‐dried prior to TEM imaging.

2.8. Purity Measurement of Different EVs

EV purity was measured as described previously (Gao et al. 2018; Tian et al. 2020). Briefly, EVs isolated from fresh ginger via Phyto‐EVpure and grinding methods or EVs derived from the enzyme were incubated with 1% Triton X‐100 (ab286840; Abcam, Cambridge, UK) for 1 h on ice. Flow cytometric analysis (FACSVerse, BD Biosciences) was performed to assess the FSC/SSC profiles of plant EVs before and after Triton X‐100 treatment. Particles resistant to detergent lysis were operationally defined as non‐vesicular contaminants.

2.9. Labelling of Plant EVs

Plant EVs were labelled with DiR (APC‐Cy7) (D12731; Invitrogen, Carlsbad, CA, USA) following standardized protocols. Briefly, 200 µL of the EV suspension was incubated with 10 µM DiR (APC‐Cy7) (37°C, 20 min, dark conditions), followed by the removal of unbound dye via centrifugal ultrafiltration (100 kDa cutoff, UFC510096, Millipore, USA).

2.10. Cellular Uptake

Cellular uptake of plant EVs was measured as described previously (Zhao et al. 2023). To measure the cellular uptake of plant EVs, human HL‐7702, BEAS‐2B, and A549 cells (1 × 10⁵ per well) were seeded and cultured in 24‐well plates (Corning, USA) for 12 h. DiR (APC‐Cy7)‐labelled fresh ginger EVs (1 × 10⁹ particles/mL in PBS) were added to HL‐7702 and BEAS‐2B cells, and DiR (APC‐Cy7)‐labelled EVs from Marsdenia tenacissima (Roxb.) Wight et Arn. (MT) or Tripterygium wilfordii Hook. F. (TW) phloem and xylem (1 × 10⁹ particles/mL in PBS) were added to human A549 cells, followed by incubation for another 24 h. Cellular uptake and localization of different plant EVs were analyzed by flow cytometry (FACSVerse, BD, USA) and confocal fluorescence microscopy (FV1000, Olympus, Japan), respectively.

2.11. Water Permeability Assay

To measure the water permeability of fresh ginger EVs, their osmotic responses were monitored, as described previously (Eto et al. 2010; Wang et al. 2009). Ginger EVs were equilibrated in 20 mM HEPES buffer (pH 7.0, H917470, Macklin, China) containing 1 mM DTT (AC10995, Acmec, China) and subjected to osmotic challenge by rapid mixing (1:1 v/v) with 500 mM sucrose, using a stopped‐flow spectrometer (SX20, Applied Photophysics) at 5°C. The light‐scattering dynamics at 440 nm were monitored to quantify the vesicular volume changes induced by osmotically driven efflux. EVs pretreated with 100 µM AgNO₃ (S197266; Aladdin, China) were analyzed under identical conditions to assess aquaporin‐mediated water permeability (Niemietz and Tyerman 2002). The scattering kinetics were fitted to first‐order decay models (Y = ae^−kt+b) using Origin, where k represents the rate constant, and Y is the normalized scattering intensity.

2.12. Tissue Distribution

Tissue distribution of ginger‐EVs was measured as described previously (Gupta et al. 2020; Liu et al. 2023). To analyze the biodistribution of ginger EVs in vivo, C57BL/6 mice (n = 3 per group) were intravenously injected with DiR (APC‐Cy7)‐labelled ginger EVs processed by grinding or Phyto‐EVpure (5×10¹² particles/kg in PBS). Mice were euthanized, and body‐wide tissues were collected 2, 12, and 24 h after injection. DiR signal intensity was measured using an IVIS spectrum (PerkinElmer, USA) and analyzed with a Living Image in vivo imaging analysis software (PerkinElmer, Waltham, MA, USA).

2.13. Pharmacokinetic Assay

Pharmacokinetic assays were performed as described previously (Hwang et al. 2019). Briefly, C57BL/6 mice were intravenously injected with DiR (APC‐Cy7)‐labelled fresh ginger EVs (5×10¹² particles/kg in PBS) isolated by grinding or Phyto‐EVpure. Blood was drawn from the tip of the tail vein of injected mice using heparinized capillary tubes (Fisher Scientific, Pittsburgh, PA, USA) at 5, 15, 30, 60, and 360 min after injection. Blood samples were centrifuged at 1500 × g for 20 min to obtain serum, which was measured using flow cytometry (FACSVerse, BD, USA) to detect fluorescence signals.

2.14. In Vitro Assay for Hepatic Steatosis

An in vitro assay for liver cell steatosis in vitro was performed as described previously (Xia et al. 2019). Briefly, HepG2 cells (8 × 10⁵ cells/well) were seeded on glass coverslips in 6‐well plates for 12 h. To establish lipid overload conditions, cells were incubated with 0.2 mM oleic acid and 0.1 mM palmitic acid (24 h). Ginger‐derived EVs (1 × 10¹⁰ particles/mL in PBS) were then administered for 24 h. Cellular lipid accumulation was quantified via Oil Red O staining (Beyotime Biotechnology, China) and imaged using an Olympus IX71 microscope (40× objective; Olympus, Japan).

2.15. Histological Analysis

For hematoxylin and eosin (H&E) staining, liver tissues fixed in 4% PFA were embedded in paraffin, cut into 4 µm serial sections, and subjected to H&E staining according to the manufacturer's instructions (Beyotime, China). To examine lipid droplets, liver tissues were dehydrated in 15% sucrose solution, cryosectioned to 8–10 µm, and stained with oil red. The tissue sections were observed under a bright‐field microscope (Olympus BX51, Japan).

2.16. Measurement of Hepatic TG Levels

Liver triglyceride (TG) content was determined using chloroform‐methanol extraction as described previously (Tsai et al. 2022). Briefly, 100 mg of hepatic tissue was homogenized in chloroform: methanol (2:1, v/v; 2 mL) and incubated overnight at 4°C. After centrifugation (3000 × g, 5 min), the supernatants were mixed with 1 mL saline and centrifuged (500 × g, 10 min). The chloroform phase was collected, evaporated under nitrogen, and reconstituted in 500 µL chloroform‐1% Triton X‐100 solution. Following solvent evaporation, the lipid residues were resuspended in 500 µL ddH₂O. TG levels were quantified using a commercial ELISA kit (K622‐100, BioVision), with 1% Triton X‐100 as a blank control.

2.17. Biochemical Analysis

Serum was obtained from the treated or control mice, and aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total cholesterol (TC) levels were measured in the clinical pathology laboratory (Tianjin Medical University, Tianjin, China).

2.18. Simulated Gastrointestinal Digestion

Artificial digestive fluids were formulated as previously (Minekus et al. 2014): (1) Gastric fluid: Pepsin (2000 U/mL, Boster Biological Technology) dissolved in electrolyte solution (pH 2.0, adjusted with 1 M HCl) containing 7.8 mM K⁺, 72.2 mM Na⁺, 70.2 mM Cl⁻, 0.9 mM H₂PO₄ ⁻, 25.5 mM HCO₃ ⁻, 0.1 mM Mg²⁺, 1 mM NH₄ ⁺, 0.15 mM Ca²⁺. (2) Intestinal fluid: Pancreatin (800 U/mL, MedChemExpress) and bile juice (24 mg/mL, Hwrk Chem) in electrolyte solution (pH 7.0, adjusted with 1 M NaOH) containing 7.6 mM K⁺, 123.4 mM Na⁺, 55.5 mM Cl⁻, 0.8 mM H₂PO₄ ⁻, 85 mM HCO₃ ⁻, 0.33 mM Mg²⁺ and 0.6 mM Ca²⁺. Ginger‐derived EVs or HEK293T‐EVs (1 × 10¹¹ particles in PBS) (Li et al. 2025)were sequentially incubated with 10 mL gastric fluid (1 h, 37°C) and 10 mL intestinal fluid (1 h, 37°C) under gentle agitation (50 rpm). EV integrity was determined by TEM imaging after digestion. Quantitative analysis of intact vesicles per group was performed as previously reported (Welsh et al. 2024) with the continuous lipid bilayer defined as intact or ruptured (irregular membrane discontinuity as damaged.

2.19. Annexin V and Propidium Iodide (PI) Staining

Apoptosis was assessed using an Annexin V/Propidium Iodide (PI) dual‐staining kit (Sungene Biotech, Tianjin, China) in accordance with the manufacturer's standardized protocol. Briefly, human A549 cells (1 × 10⁵ per well) were cultured in 12‐well plates overnight and treated with EVs from phloem and xylem of MT or TW (2 × 10¹⁰ particles/mL in PBS) at 37°C for 48 h. Following two washes with ice‐cold PBS, cells were detached using EDTA‐free trypsin and collected. The cell suspension was then reconstituted in 100 µL of binding buffer and dual‐stained with Annexin V‐FITC/PI (5 µL each) in the dark at 25°C for 15 min. Prior to flow cytometric analysis (BD FACS Calibur; BD Biosciences, USA), 400 µL binding buffer was added to stabilize the staining system through gentle vortexing.

2.20. 3‐(4,5‐Dimethylthiazol‐2‐yl)‐2,5‐Diphenyltetrazolium Bromide (MTT) Assay

MTT assay was used to determine the cytotoxicity of plant EVs derived from the phloem and xylem in vitro (Fu et al. 2019). Briefly, human A549 cells were cultured in 96‐well plates (5000 cells per well) for 12 h prior to treatment with EVs from phloem and xylem of MT or TW (1 × 10¹⁰ particles/mL in PBS) for 48 h. After the incubation period, each well received 10 µL of MTT solution (5 mg/mL in phosphate‐buffered saline) and underwent 4‐h maintenance at 37°C. Subsequently, the supernatant was carefully replaced with 100 µL dimethyl sulfoxide for optical density determination. Quantitative analysis was performed at a wavelength of 570 nm using a Sunrise microplate reader (Tecan Group AG, Männedorf, Switzerland) following standard ELISA measurement protocols.

2.21. Transwell Assay

Cell migration capacity was quantitatively evaluated using polycarbonate membrane inserts (8 µm porosity, Merck Millipore, Burlington, MA, USA) in 12‐well culture systems (Corning, MA, USA). Human A549 cells (2 × 10⁵ per well) were preconditioned with EVs from phloem and xylem of MT or TW (1 × 10¹⁰ particles/mL in PBS) from designated sources for 24 h before being suspended in 500 µL of serum‐depleted medium for upper‐chamber loading. The basal compartments contained 1 mL of complete growth medium, and the assembly was maintained at 37°C in 5% CO₂ for 24 h. Following migration, non‐transmigrated cells were mechanically cleared using sterile cotton applicators, whereas adherent cells on the inferior membrane surface were fixed and stained with 1% crystal violet histochemical staining (Solarbio Life Sciences, Beijing, China) for 15 min at ambient temperature. Cellular migration patterns were documented using phase‐contrast microscopy (Olympus BX51 Imaging System, Olympus Corporation, Tokyo, Japan) and quantitatively analyzed using the ImageJ software (National Institutes of Health, Bethesda, USA).

2.22. Liquid Chromatography Tandem Mass Spectrometry (LC‐MS/MS) Analysis

Ginger‐derived extracellular vesicles (30 µg) were subjected to tryptic digestion using an optimized filter‐aided sample preparation (FASP) method (Wang et al. 2022). Sequential protein reduction (10 mM dithiothreitol for 50 min at room temperature) and alkylation (40 mM iodoacetamide for 40 min in the dark) were performed. Excess iodoacetamide was neutralized by treatment with 40 mM dithiothreitol treatment (30 min). Buffer exchange with 50 mM ammonium bicarbonate was achieved using 10 kDa molecular weight cutoff filters (Millipore, USA), followed by tryptic digestion (1:30 enzyme‐to‐protein ratio) at 37°C for 16 h. The peptides were desalted, vacuum‐concentrated (SpeedVac; Thermo Fisher Scientific), and reconstituted in 0.1% formic acid. The concentration was determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific) prior to LC‐MS/MS. Chromatographic separation utilized a self‐packed C18 column (150 µm × 29 cm, 1.9 µm particles; Dr. Maisch GmbH) on an Easy‐nLC 1000 system (Thermo Fisher Scientific, Waltham, MA, USA) with a 120‐min linear gradient (8%–95% mobile phase B; 300 nL/min). The mobile phase consisted of 0.1% formic acid in H₂O (A) and 0.1% formic acid in 80% acetonitrile (B). Mass spectrometry was performed on an Orbitrap Fusion Lumos platform (Thermo Fisher Scientific) in data‐independent acquisition (DIA) mode. Full MS scans (350–1000 m/z) were acquired at 60,000 resolutions (AGC target 1 × 10⁶), followed by 38 MS2 scans (200–2000 m/z) with 1.6 m/z isolation windows at 30,000 resolutions.

2.23. Proteomic Data Analysis

Label‐free quantification was performed using DIA‐NN (v1.8) with the following parameters: precursor FDR≤1% (target decoy approach), tryptic digestion allowance (one missed cleavage), mass tolerance of 10 ppm (MS/MS), fixed carbamidomethylation (Cys), and variable oxidation (Met). Swiss‐Prot and TrEMBL FASTA databases for Zingiber officinale (Ginger) (Amomum zingiber) (library contains 71856 proteins) were downloaded from UniPort (https://www.uniprot.org/). Chromatographic alignment and cross‐run normalization were performed using the DIA‐NN high‐accuracy quantification workflow. Peptide identification thresholds included: length of 7–30 residues, mass range of 300–1800 Da, and fragment m/z of 200–1800. Parallel processing was enabled using multithreading (eight cores). Quantitative matrices were generated in R (v4.2.2) using the ‘diann’ package, employing precursor‐level intensities. Data curation involved: (1) Exclusion of single‐peptide protein identifications, (2) MaxLFQ‐based intensity normalization (DIA‐NN implementation), (3) Quantile normalization ('preprocessCore' package), (4) Minimum‐value imputation for missing data, (5) log2 transformation prior to statistical analysis. The complete dataset was deposited in ProteomeXchange (PXD057378) through the iProX infrastructure (Chen, Ma et al. 2022; Ma et al. 2019).

2.24. Lipid Extraction and LC‐MS/MS Analysis

Lipidomics analysis was conducted by Beijing Allwegene Technology Company Ltd. (Beijing, China). Lipids were extracted from ginger EVs (5×10¹⁰ particles) using the methyl tert‐butyl ether (MTBE) method (Matyash et al. 2008). Samples were initially homogenized with 200 µL ddH2O (vortexed for 5 s), followed by sequential addition of 240 µL ice‐cold methanol (vortexed for 30 s) and 800 µL of MTBE. The mixture was ultrasonicated for 20 min at 4°C (Sonics, USA) and incubated at room temperature for 30 min to facilitate phase separation. After centrifugation at 14,000 × g (15 min at 10°C), the upper organic layer was collected and evaporated under nitrogen gas. Lipid separation was performed on a Waters CSH C18 column (1.7 µm, 2.1 mm×100 mm, Waters) using reverse‐phase chromatography. Reconstituted lipid extracts (200 µL 90% isopropanol/acetonitrile, centrifuged at 14,000 × g for 15 min) were injected at 3 µL. The mobile phase comprised Solvent A (acetonitrile/water, 6:4, v/v) and Solvent B (acetonitrile/isopropanol, 1:9, v/v), both containing 0.1% formic acid and 0.1 mM ammonium formate. Gradient elution was programmed as follows: 30% B (0–2 min), linearly increased to 100% B over 23 min, followed by 10‐min of re‐equilibration at 5% B. The flow rate was maintained at 300 µL/min throughout the experiments. Lipid profiling was conducted using a Q‐Exactive Plus mass spectrometer (Thermo Scientific) operating in positive/negative ion‐switching mode. The electrospray ionization (ESI) parameters were optimized as follows: ion source temperature, 300°C; capillary temperature, 350°C; spray voltage: 3.0 kV, S‐Lens RF level, 50%; and full‐scan range: m/z 200–1800. Lipid identification was performed using Lipid Search software (v4.2, Thermo Scientific), which incorporates a database of >30 lipid classes and 1.5 million characteristic fragment ions. The mass tolerance for both precursor and product ions was set to 5 ppm.

2.25. Statistical Analysis

Data are expressed as mean ± SEM. Intergroup comparisons were analyzed using SigmaStat 3.5 (Systat Software Inc., USA), with statistical significance defined at p < 0.05. Parametric or nonparametric tests were selected based on data distribution, as annotated in the figure legends. A priori power analysis for sample size estimation was conducted the PASS 11.0 (NCSS, UT, USA), and ANOVA power calculations were conducted using Sigmastat (SystatSoftware, Inc. USA), with the observed mean difference, standard deviation, and sample size. The analysis (Student–Newman–Keuls test, α = 0.05) revealed all statistical powers above 80%, suggesting that the study had sufficient power to detect the observed difference.

3. Results

3.1. Optimization of Phyto‐EVpure Workflow for Plant EV Isolation

To develop Phyto‐EVpure, we selected ginger, a commonly used Chinese medicinal plant (Zhu and He 2023), as a prototypical plant. As cellulase and pectinase used for cell wall digestion are primarily extracted from fungi, to exclude potential contamination, we ultracentrifuged the enzyme solution to deplete enzyme‐derived EVs. As shown by the sauce‐cup shape and double‐layered vesicles via transmission electron microscopy (TEM) (Figure 1a) and significantly reduced number of lipid vesicles after Triton X‐100 disruption (Figure 1b), a method to differentiate vesicular and non‐vesicular structures (Osteikoetxea et al. 2015), ultracentrifugation lowered the EV concentration in the enzyme solution. Comparable digestion enzyme activities were observed across different concentrations before and after depletion of EVs (Figure 1c), indicating that depletion of EVs did not affect the activities of cellulase and pectinase. As the optimal reaction temperature for the enzymes was 50°C (Figure S1a), we examined the morphology of ginger EVs at 50°C and 37°C, temperatures used in previous studies (Ou et al. 2023; Zhao et al. 2023). Consistent with previous reports (Ou et al. 2023; Zhao et al. 2023), EVs maintained membrane integrity at 37°C, whereas aggregation of EVs was observed at 50°C (Figure 1d). Thus, we chose 37°C for subsequent tests.

FIGURE 1.

Stepwise optimization for Phyto‐EVpure. (a) Representative transmission electron microscopic (TEM) images for EVs derived from cellulase and pectinase solution via ultracentrifugation (scale bar = 200 nm). (b) Flow cytometry to assess lipid vesicles and non‐lipid vesicles (forward scatter versus side scatter (FSC/SSC)) of enzyme‐derived EVs before or after treatment with 1% Triton X‐100 (n = 3) (**p < 0.01, two‐tailed t‐test). (c) Measurement of cellulase and pectinase activities before or after EV depletion via ultracentrifugation. (d) Representative TEM images for ginger EVs incubated at different temperatures (scale bar = 200 nm). Purified EVs were derived from ginger processed by grinding. (e) Representative TEM images for ginger EVs (the upper panel) and bacterial colonies (the lower panel) under different pre‐treatments (scale bar = 200 nm). Ginger was pretreated with 0.5% NaClO, 3% H₂O_2, or 70% alcohol for 20 min prior to enzyme digestion, or pretreated with 70% alcohol for 30 s followed by enzyme digestion containing 1% PS or 100 U/mL bacitracin. NC means negative control, referring to the enzyme reaction buffer containing ginger. Buffer represents the enzyme buffer alone without ginger. After 4 h of reaction in 1% cellulase and 0.5% pectinase, the reaction buffer without or with ginger was spread on agarose for examining bacterial colonies. (f) Examination of bacterial contamination in the ginger‐containing reaction buffer with or without 1% PS or 0.25 µg/mL amphotericin B (B) or the combination of both reagents (PSB), in which ginger was pretreated with 70% alcohol for 30 s. The reaction time was 4 or 6 h.

Minimizing any possible contamination without disrupting the EVs is a prerequisite for the isolation of high‐quality EVs. To this end, we tested different detergents and antibacterial reagents for ginger sample processing using Phyto‐EVpure. As expected, no contamination was found in the enzyme buffer; however, bacterial colonies grew in the buffer after addition of ginger and incubation for 4 h (Figure 1e). Compared to alcohol and bacitracin (Zhao et al. 2023), there was no bacteria were found in the presence of 1% penicillin and streptomycin (PS) or ginger pretreated with 0.5% NaClO or 3% H₂O_2, although EVs with much smaller sizes or compromised membrane integrity were present in the NaClO‐ or H₂O₂‐treated samples (Figure 1e). Furthermore, to examine the presence of fungal contamination, we added an antifungal agent, amphotericin B (Kapil et al. 2018) to the reaction solution and incubated it for 4 and 6 h, respectively. Compared to 1% PS, bacterial colonies were formed in the reaction solution containing 0.25 µg/mL amphotericin B, but not in 1% PS or the combination of PS and amphotericin B (Figure 1f). Corroborating this observation, no fungi were found in the reaction solution containing ginger for 6 h, as shown by periodic acid‐Schiff staining (Figure S1b), a method for fungi staining (Dadaci et al. 2015), demonstrating that 1% PS is sufficient to eliminate bacterial contamination throughout the procedure, without affecting the integrity of plant EVs.

3.2. Phyto‐EVpure Enables Efficient Ginger EV Isolation Without Altering Intrinsic Properties

Next, we optimized the reaction conditions for ginger by starting with different concentrations of cellulase and pectinase and digestion for 6 h. More complete digestion was observed with 1.0% cellulase and 0.5% pectinase than with other concentrations, as reflected by the lower amounts of filamentous materials (Figure S2a). Further evaluation revealed that more uniform EVs with higher purity were obtained from ginger treated with 1% cellulase and 0.5% pectinase for 1 h compared to other conditions (Figure 2a). Compared with grinding, significantly higher purity was achieved for ginger EVs via Phyto‐EVpure, as shown by TEM (Figure 2b), nanoparticle tracking analysis (NTA) (Figure 2c), and lipid vesicle counts after Triton X‐100 disruption (Figure 2d). Consistent with previous reports (Zhang, Xiao, et al. 2016; Zhuang et al. 2015), the majority of particles were distributed in 30%/45% sucrose with buoyant densities between 1.09 and 1.17 g/mL after sucrose gradient ultracentrifugation, with up to a 60‐fold higher yield for Phyto‐EVpure than for grinding (Figure 2e–g). Furthermore, Phyto‐EVpure optimized for ginger was also applicable to other rhizomes, such as yam, carrot, and potato, with carrot showing the highest yield (Figures 2h and S2b), indicating the generalizability of this protocol.

FIGURE 2.

Effect of enzyme reaction time on and characterization of ginger EVs. (a) Representative TEM images for EVs purified by ultracentrifugation after ginger was digested in 1% cellulase and 0.5% pectinase for different timepoints (scale bar = 200 nm). Representative TEM images (b) and size distribution with NTA (c) of ginger EVs purified by ultracentrifugation with ginger pulped either with grinding or Phyto‐EVpure (scale bar = 200 nm). (d) Flow cytometry to assess the morphological parameters (forward scatter versus side scatter [FSC/SSC]) of ginger EVs before or after treatment with 1% Triton X‐100 (n = 3) (**p < 0.01, two‐tailed t‐test). (e) Representative TEM images of ginger EVs in different buoyant densities after sucrose gradient centrifugation (scale bar = 200 nm). NTA analysis for size distribution (f) and yield (g) of ginger EVs distributed in 30%/45% of sucrose after sucrose gradient centrifugation, in which ginger was pulped with Phyto‐EVpure or grinding (n = 3) (**p < 0.01, two‐tailed t‐test). (h) Representative TEM images and NTA of purified EVs from yam, carrot and potato after processed with Phyto‐EVpure.

To investigate whether the degradation enzymes in Phyto‐EVpure affected the surface properties and, consequently, the tropism of plant EVs, we labeled equal particles of ginger EVs processed via Phyto‐EVpure or grinding and purified by sucrose gradient ultracentrifugation. As hepatocytes and bronchial epithelial cells have been extensively employed for testing systemic (hepatic) and local (pulmonary) EV delivery (Cao et al. 2019; Swindle et al. 2009; Zhang et al. 2024; Zhang et al. 2025), thus, we examined cellular uptake of EVs in H7702 hepatocytes and BEAS‐2B bronchial epithelial cells as previously reported (Teng et al. 2021; Wang et al. 2025). Comparable labeling efficiency and cellular uptake in human hepatocytes and bronchial epithelial cells (Han et al. 2020) were obtained with ginger EVs processed via Phyto‐EVpure or grinding (Figure 3a,b), with fluorescence signals mainly localized to the cytoplasm of human hepatocytes and bronchial epithelial cells (Figure 3c). As expected, fluorescence signals primarily accumulated in the liver and spleen of wild‐type C57BL/6 mice after a single intravenous administration of equal particle numbers (5 × 10¹² particles/kg) of ginger EVs, similar to EVs derived from Ginseng and grapefruits (Cao et al. 2019; Wang et al. 2013), with signals gradually declining 24 h after injection (Figure 3d,e), although a slightly longer circulatory half‐life was observed in EVs processed via Phyto‐EVpure than with grinding (Figures 3f and S3a). Consistent with previous studies (Zhang, Xiao, et al. 2016; Zhuang et al. 2015), the integrity of ginger EVs was preserved after digestion with simulated gastrointestinal fluid (Figure 3g), whereas no intact vesicles were found in human HEK293T cell‐derived EVs under identical conditions (Figure S3b), confirming the resistance of ginger EVs to harsh gastrointestinal environments. Altogether, these findings demonstrate that ginger EVs processed via Phyto‐EVpure show a similar biodistribution to EVs processed by grinding.

FIGURE 3.

In vitro and in vivo evaluation of EVs isolated from ginger processed with Phyto‐EVpure. (a) Flow cytometry to examine the labelling efficiency of EVs obtained from ginger processed with Phyto‐EVpure or grinding. (b) Cellular uptake of DiR (APC‐Cy7)‐labelled ginger EVs in human hepatocytes and bronchial epithelial cells (n = 3). (c) Representative confocal fluorescence microscopic images of DiR (APC‐Cy7)‐labelled ginger EVs in human hepatocytes and bronchial epithelial cells (scale bar = 20 µm). Nuclei were counterstained with DAPI. Tissue distribution (d) and quantitative analysis (e) of DiR (APC‐Cy7)‐labelled ginger EVs in body‐wide tissues at different timepoints after single intravenous injection (n = 3). MFI means mean fluorescence intensity. Br‐brain; H‐heart; Sp‐spleen; K‐kidney; Lu‐lung; Li‐liver. (f) Measurement of serum half‐life of EVs from ginger processed with Phyto‐EVpure or grinding. C_max refers to the maximal concentration. T1/2α means the distribution half‐life. (g) Representative TEM images of ginger EVs after incubation in the simulated gastrointestinal fluid in vitro, in which ginger was processed with Phyto‐EVpure or grinding (scale bar = 200 nm).

3.3. Ginger EVs Processed via Phyto‐EVpure Preserve Biological Activities

To determine the effect of Phyto‐EVpure on the proteomic and lipidomic profiles of ginger EVs, we performed Liquid Chromatography with tandem mass spectrometry (LC‐MS/MS) and lipidomic analysis. Consistent with previous observations (Kalarikkal et al. 2020; Zhang et al. 2016a), more membrane proteins and fewer ribosomal proteins were found in ginger EVs processed via Phyto‐EVpure than by grinding, in which ribosomal proteins are considered contaminants (Théry et al. 2018), as revealed by functional classification related to cellular components with Gene Ontology (GO) annotation (Figure S4), with a maximum of 2245 or 2012 proteins identified for ginger EVs obtained via grinding or Phyto‐EVpure, respectively. Consistent with previous findings (Zhang, Xiao, et al. 2016), aquaporins, lipoxygenase, and ATP synthase proteins were detected in ginger EVs processed using both methods (Figure 4a). To examine whether the involvement of enzymes affects the biological activity of membrane proteins in EVs, we tested the water permeability of ginger EVs mediated by aquaporins, which are key membrane channels responsible for the transport of water and small neutral molecules in plants (Maurel et al. 2015). The osmotic responses and water efflux of ginger EVs were measured using stopped‐flow analysis (Wang et al. 2009), in which the increase in light scattering reflects vesicle shrinkage caused by osmotic water efflux driven by the osmotic gradient (Eto et al. 2010). Increases in scattering light intensities were observed with ginger EVs processed with Phyto‐EVpure and ground in a similar manner after mixing with sucrose, whereas significantly decreased scattering intensities were found in the presence of AgNO₃ (Figure 4b), a specific inhibitor of aquaporins (Wang et al. 2024), indicating that aquaporins preserve biological functions and mediate water transport in ginger EVs prepared via both Phyto‐EVpure and grinding.

FIGURE 4.

Proteomic and lipidomic profiling and functional validation of EVs from ginger processed with Phyto‐EVpure or grinding. (a) Quantitative analysis of proteins identified in ginger EVs via proteomic analysis, in which ginger was processed with Phyto‐EVpure or grinding (n = 8; *p < 0.05, **p < 0.01; two‐tailed t‐test). (b) Stopped‐flow spectrometric analysis of ginger EVs’ water permeability (n = 10). Ginger was processed with Phyto‐EVpure or grinding prior to ultracentrifugation. The shrinkage of EVs was monitored by measuring the increase in scattered light. (c) Lipidomic profiling of ginger EVs (n = 3). PA: phosphatidic acids; lysoPG: lysophosphatidylglycerol; PE: Phosphatidylethanolamine; lysoPC: lysoPhosphatidylholine; DGDG: digalactosyldiacylglycerol; PS: Phosphatidylserine; PG: Phosphatidylglycerol; PI: phosphatidylinositol; MGDG: monogalactosyl monoacylglycerol; lysoPE: lysoPhosphatidylethanolamine; PC: Phosphatidylholine. (d) Oil Red staining of lipid droplets in OA‐induced human HepG2 cells (scale bar = 50 µm). NC means HepG2 cells without any treatment. (e) Histological analysis of liver of NAFLD mouse treated with ginger EVs. Ginger was processed with Phyto‐EVpure (scale bar = 50 µm). Measurement of serum total cholesterol (TC) and hepatic triglyceride (TG) (f) and serum alanine transaminase (ALT) and aspartate transferase (AST) (g) of mice with NAFLD after treatment with ginger EVs (n = 6). **p < 0.01; One way‐ANOVA post hoc Student–Newman–Keuls test was used for statistical analysis.

Unsurprisingly, a similar lipidomic profile was observed for ginger EVs processed via both methods, with phosphatidic acid (PA) ranking the highest, followed by monogalactosyl mono‐acylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), phosphatidylinositol (PI), and phosphatidylcholine (PC) (Figure 4c), consistent with previous reports (Zhang, Xiao, et al. 2016; Zhuang et al. 2015). To assess whether the functionality of ginger EVs is preserved, we used a non‐alcoholic fatty liver disease (NAFLD) model in vitro and in vivo as PA in ginger EVs has been reported to prevent liver damage (Kumar, Sundaram et al. 2022; Zhuang et al. 2015). Remarkably, lipid droplets were significantly reduced in oleic acid (OA)‐induced human hepatocytes and the livers of NAFLD mice fed a high‐fat diet, a model commonly used for NAFLD (Grasselli et al. 2017; Kumar, Sundaram et al. 2022), and treated with ginger EVs processed via Phyto‐EVpure (Figure 4d,e). Levels of serum total cholesterol (TC) and hepatic triglyceride (TG) (Figure 4f), biomarkers for NAFLD (Neuman et al. 2014), and alanine aminotransferase (ALT) and aspartate transaminase (AST) (Figure 4g), biochemical parameters for liver functions, were significantly decreased in mice treated with ginger EVs processed via Phyto‐EVpure compared to those in PBS‐treated mice, indicating that ginger EVs processed via Phyto‐EVpure preserve biological activities. Collectively, these data strengthen the conclusion that Phyto‐EVpure does not affect the proteomic and lipidomic profiles or potency of ginger EVs.

3.4. Phyto‐EVpure Is Applicable to Traditional Medicinal Plants Under Different Conditions

To further explore the generalizability of Phyto‐EVpure, we tested fresh Morus alba leaves and Isatis indigotica Fort. roots under conditions identical to those of ginger. Surprisingly, varying incubation time in 1% cellulase and 0.5% pectinase affected purity and morphologies of EVs with 4 and 0.5 h working best for fresh Morus alba leaves and Isatis indigotica Fort. roots (Figure 5a). Phyto‐EVpure could also isolate EVs from fresh leaves of Lonicera macranthoides Hand.‐Mazz., Houttuynia cordata Thunb. and Koelreuteria paniculata Laxm. and the roots of Polygonum multiflorum Thunb., Sophora flavescens and Salvia miltiorrhiza Bge. (Figures 5b and S5a), although the yield varied, with Salvia miltiorrhiza Bunge ranking at the top (Figure 5c). As snap‐freezing is a common practice for the preservation of fresh Chinese herbs or fruits (Liu et al. 2022), we treated snap‐frozen Morus alba leaves with Phyto‐EVpure to test its applicability. EVs were efficiently isolated with intact morphologies, although the yield was lower than that of fresh samples (Figures 5d,e, and S5b). To examine the applicability of Phyto‐EVpure in dried traditional medicinal plants, we tested dried ginger and Isatis indigotica Fort. root. Strikingly, EVs were also isolated from dried samples via Phyto‐EVpure, although with significantly lower yields than the corresponding fresh samples (Figure 5f,g), demonstrating that EVs are relatively stable upon freezing or drying, and roots are generally richer in EVs than in other parts. Traditional medicinal plants such as Marsdenia tenacissima (Roxb.) Wight et Arn., and Tripterygium wilfordii Hook. F. has been clinically used for antitumour and immunomodulation (Astry et al. 2015; Chen et al. 2006; Dai et al. 2017; Pan et al. 2023; Wang et al. 2018; Wang et al. 2018; Wong et al. 2012; Yi et al. 2023; Zhang et al. 2022; Zhang et al. 2021; Zhao et al. 2020), and we examined the feasibility of Phyto‐EVpure to extract EVs from dried vine samples. To enable effective enzyme digestion, we attempted to soften dried vine samples by soaking them in the buffer for different time points, with 8 h showing a better digestion effect combined with a 0.5 h reaction time, as reflected by the presence of more uniform EVs (Figure S5c) without any bacterial contamination (Figure S5d). Further optimization revealed that a 0.5 h reaction time was sufficient for EV isolation from Marsdenia tenacissima (Roxb.) Wight et Arn. in the presence of 1% cellulase and 0. and 5% pectinase (Figure 5h). EVs with sauce‐cup shapes were effectively isolated from Spatholobus suberectus Dunn and Tinospora sinensis (Lour.) Merr., Polygonum multiflorum Thunb. and Tripterygium wilfordii Hook. F. via Phyto‐EVpure under identical conditions (Figure 5i), with Spatholobus suberectus Dunn showing slightly smaller sizes than the others, as reflected by NTA, and Marsdenia tenacissima being richer in EVs than Tripterygium wilfordii and Spatholobi caulis (Figure 5j). Taken together, these findings show that Phyto‐EVpure can be generalized to different traditional medicinal plants under different conditions.

FIGURE 5.

Optimization of reaction conditions for EVs from leaves, roots and dried vine samples via Phyto‐EVpure. Representative TEM images EVs from Morus alba. leaves and Isatis indigotica fort. roots (a) or other leaves and roots (b) after processing with Phyto‐EVpure (scale bar = 200 nm). Insets represent magnified images. LHM‐ Lonicera macranthoides Hand.‐Mazz.; HCT‐Houttuynia cordata Thunb.; KPL‐ Koelreuteria paniculata Laxm.; PMT‐Polygonum multiflorum Thunb.; SF‐Sophora flavescens; SMB‐Salvia miltiorrhiza Bge. (c) Yield comparison of EVs derived from leaves and roots processed with Phyto‐EVpure (n = 3). Representative TEM images (d) and size distribution with NTA (e) of EVs from fresh or snap‐frozen Morus alba. leaves pulped with Phyto‐EVpure (scale bar = 200 nm). Representative TEM images (f) and size distribution with NTA and yield (g) of EVs from fresh or dried ginger and Isatis indigotica Fort. roots pulped with Phyto‐EVpure (scale bar = 200 nm) (n = 3; **p < 0.01; two‐tailed t‐test was used for statistical analysis). Representative TEM images of EVs from dried Marsdenia tenacissima (Roxb.) Wight et Arn. (h) or other dried vine samples under optimal conditions (i) pulped with Phyto‐EVpure (scale bar = 200 nm). SC‐Spatholobus suberectus Dunn.; MT‐ Marsdenia tenacissima (Roxb.) Wight et Arn.; TSM‐ Tinospora sinensis(Lour.)Merr.; PMT‐ Polygonum multiflorum Thunb.; TW ‐Tripterygium wilfordii Hook. F. (i) Size distribution with NTA and yield of EVs from different dried vine samples pulped with Phyto‐EVpure.

3.5. EVs From Phloem of Dried Vine Samples Show Potent Antitumour Activities In Vitro

Considering that xylem and phloem are two major vascular subsystems of vines with different functions (Jyske and Hölttä 2015; Kumar et al. 2022; Mei et al. 2019), we investigated whether EVs from the xylem and phloem would function differently (Figure S6). Interestingly, more EVs were isolated from xylem than from phloem of Marsdenia tenacissima (Roxb.) Wight et Arn. and Tripterygium wilfordii Hook. F. under identical conditions with the same starting mass (Figure 6a–c), with slightly smaller EVs observed in the xylem and phloem of Tripterygium wilfordii Hook. F. than Marsdenia tenacissima (Roxb.) Wight et Arn. (Figure 6a,b), respectively. Although comparable membrane‐labelling efficiency was achieved for EVs from the xylem and phloem (Figure 6d), significantly enhanced cellular uptake was observed in human non‐small cell lung cancer (NSCLC) cells with EVs from the phloem compared to those from the xylem (Figure 6e,f). Given that extracts of Marsdenia tenacissima (Roxb.) Wight et Arn., and Tripterygium wilfordii Hook. F. exhibits antitumour activity against NSCLC (Lv et al. 2024; Wang et al. 2018), and we tested the antitumour effect of EVs in vitro. As expected, potent cytotoxicity (Figure 6g), induction of apoptosis (Figure 6h,i), and tumour cell migration inhibition (Figure 6j,k) were observed in EVs from the phloem and xylem of Marsdenia tenacissima (Roxb.) Wight et Arn., and Tripterygium wilfordii Hook. F. in human NSCLC cells in vitro with EVs from the phloem of Marsdenia tenacissima (Roxb.) Wight et Arn. exhibited the highest potency. These data demonstrated that EVs from dried vine samples, particularly from the phloem, can be used as potent antitumour agents.

FIGURE 6.

Characterization and in vitro evaluation of EVs from phloem and xylem of dried vine samples processed via Phyto‐EVpure. Representative TEM images (a) and size distribution with NTA (b) and yield comparison (c) of EVs from phloem and xylem of dried Marsdenia tenacissima (Roxb.) Wight et Arn. (MT) and Tripterygium wilfordii Hook. F. (TW) pulped with Phyto‐EVpure (scale bar = 200 nm). Flow cytometric analysis of labelling efficiency (d) and cellular uptake (e), and quantitative analysis of cellular uptake (f) of EVs from phloem and xylem of dried vine samples processed with Phyto‐EVpure (n = 3). Measurement of cytotoxicity (g) and apoptosis (h and i), and migration inhibition via transwell (j and k) of EVs from phloem and xylem of dried vine samples processed with Phyto‐EVpure in human NSCLC cells in vitro. *p < 0.05, **p < 0.01; two‐tailed t‐test was used for statistical analysis in (c, f, and g). One way‐ANOVA post hoc Student–Newman–Keuls test was used for statistical analysis in (i and k).

4. Discussion

Although medicinal plant EVs have emerged as a new therapeutic modality, a unified sample processing protocol that enables scalable production in the same manufacturing facility is lacking, particularly for the pulping step unique to plant EV isolation (Nemati et al. 2022). In Phyto‐EVpure, pectinase functions through hydrolyzing pectin‐rich middle lamella and primary cell wall matrices (Anderson and Pelloux 2025), and cellulase enables degradation of cellulose microfibrils within cell walls (Doi and Kosugi 2004). Thus, this enzymatic action disrupts the structural integrity of the plant cell walls, thereby facilitating release of EVs entrapped within the apoplastic space or bound to the cell wall (An et al. 2006; Halperin and Jensen 1967). In this study, we optimized a generalizable enzyme‐based pulping protocol (Phyto‐EVpure) for effective release and isolation of high‐quality EVs from a wide spectrum of traditional medicinal plants under different conditions, particularly from dried vine samples, without detectable contamination. Importantly, no enzyme activities can be detected in the final EV preparations (data not shown), confirming the absence of enzymes. Phyto‐EVpure enables the isolation of EVs with higher yield and greater purity than grinding from fresh leaves, roots, and rhizomes of traditional medicinal plants. Extensive validation of Phyto‐EVpure in 17 different traditional medicinal plants under different conditions, especially frozen and dried samples, not only proves the generalizability of this protocol but also indicates that plant EV isolation might not be subject to the limitation of seasonal planting via Phyto‐EVpure. Most importantly, Phyto‐EVpure enabled EV isolation from previously intractable plants such as dried vine samples and vascular subsystems, including the phloem and xylem, with optimized protocols, thus providing a tool for EV isolation in a much wider spectrum of medicinal plants in a more subtle way and facilitating the identification of conserved biomarker proteins for plant EVs. Worth mentioning, consistent enzyme‐to‐biomass ratios were applied within each plant category, such as dried stems, leaves, rhizomes, and roots. Although we tried to standardize the enzymatic digestion protocol within the same category of medicinal plants, substantial variations were still observed in EVs yield within plant species as reported earlier (Mao et al. 2024; Zhao et al. 2024). This suggests that other factors, such as the biogenesis of EVs, might also have impact on EV yield. More detailed optimization and mechanistic studies are warranted for any specific medicinal plant in future.

Although pectinase and cellulase were previously used for sample processing of fresh Morinda officinalis roots and Catharanthus roseus leaves previously (Ou et al. 2023; Zhao et al. 2023), stepwise optimization, standardization, and systemic validation are key to follow‐up studies and therapeutic use. In the present study, we undertook a comprehensive and stepwise optimization to minimize any potential contamination source to jeopardize the quality of EV for future clinical employment and established a standardized operation protocol without compromising the integrity and activity of medicinal plant EVs. In addition to the depletion of enzyme‐derived EVs, we carefully monitored each step to avoid microbial contamination during incubation and enzyme digestion, thereby eliminating the potential contamination of EVs derived from bacteria and fungi. PS has been extensively used for mammalian cell culture without affecting exosomes derived from mammalian cells (Patel et al. 2019) thus, it is an ideal antibacterial reagent for plant EV isolation. Importantly, the use of pectinase and cellulase in Phyto‐EVpure did not alter the intrinsic properties and biological activities of plant EVs, particularly transmembrane proteins, such as aquaporins, demonstrating the potential and feasibility of Phyto‐EVpure for sample treatment and isolation of EVs from traditional medicinal plants.

The EVs isolated from Marsdenia tenacissima (Roxb.) Wight et Arn., particularly from the phloem, exhibited potent antitumour effects in human NSCLC cells in vitro, indicating therapeutic potential for the antitumour treatment of NSCLC. The toxicity of Marsdenia tenacissima (Roxb.) Wight et Arn. extracts in the clinic (Li et al. 2018), EVs from Marsdenia tenacissima (Roxb.) Wight et Arn. might serve as an alternative, particularly via nebulization, which can reach tumours in the lungs more directly. Also, polysaccharide of Marsdenia tenacissima (Roxb.) Wight et Arn. has been shown to affect the immune regulation of tumour (Jiang et al. 2016), it would be interesting to test the immunomodulatory effect of Marsdenia tenacissima (Roxb.) Wight et Arn. EVs in future. Nevertheless, more extensive in vivo functional studies and mechanistic investigations exploring key elements contributing to the potency of EVs from the phloem are warranted in the future, particularly now that a tool for robust, efficient isolation of plant EVs is available, as demonstrated in the present study.

Overall, we established an optimized and standardized pulping protocol (Phyto‐EVpure) for the isolation of high‐quality plant EVs in a stepwise manner and demonstrated its generalizability to 17 species and different parts under different conditions, thus accelerating the commercial manufacturing and therapeutic use of plant EVs.

Author Contributions

Qian Wang: methodology, validation, data curation, software, writing – original draft, investigation. Renwei Jing: methodology, software, data curation, investigation, validation. Nan Cao: methodology, software, data curation, investigation, validation. Xingjie He：data curation, investigation, validation. Zhongqiu Yang: data curation, investigation, validation. Leijie Zhang: data curation, investigation, validation. Yi Liu: software, visualization, methodology. Ruibing Chen: software, visualization, methodology. Beibei Xiang: resources. Xiaodong Xie: resources. HaiFang Yin: conceptualization, supervision, resources, funding acquisition, writing – original draft, writing – review and editing, project administration.

Ethics Statement

All animal experiments were carried out in the Animal Unit of Tianjin Medical University (Tianjin, China), according to procedures authorized by the Institutional Ethical Committee (Permit Number: SYXK2023‐0004).

Consent

Mice were sacrificed by cervical dislocation at the desired time points.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Materials: jex270090‐sup‐0001‐SuppMat.docx

Supplementary Figure 1: jex270090‐sup‐0002‐FigureS1.tif

Supplementary Figure 2: jex270090‐sup‐0003‐FigureS2.tif

Supplementary Figure 3: jex270090‐sup‐0004‐FigureS3.tif

Supplementary Figure 4: jex270090‐sup‐0005‐FigureS4.tif

Supplementary Figure 5: jex270090‐sup‐0006‐FigureS5.tif

Supplementary Figure 6: jex270090‐sup‐0007‐FigureS6.tif

Wang, Q. , Jing R., Cao N., et al. 2025. “Generalizability of Enzyme‐Based Isolation Approach for Extracellular Vesicles From Traditional Medicinal Plants.” Journal of Extracellular Biology 4, no. 10: e70090. 10.1002/jex2.70090

Data Availability Statement

The Proteomic datasets generated and analyzed in the current study are available in the ProteomeXchange Consortium repository [https://proteomecentral.Proteomexchange.org] dataset identifier: PXD057378. Other data generated or analyzed during this study are included in this published article and its supplementary information files.