An optimized protocol for plant extracellular vesicles isolation from Ophiopogon japonicus root: a comparative evaluation based on miRNA cargo Authors: Yang Xiao, Liqi Feng, Xin Zhao, Siyu Chen, Fengqi Lv, Zihan Li, Qi Zheng, Tao Zhou, Yuntong Ma, Binjie Xu Published: 2025 ABSTRACT AbstractBackgroundPlant extracellular vesicles (EVs), hold significant therapeutic potential due to their roles in intercellular communication and cross-kingdom regulation, primarily mediated by their microRNA (miRNA) cargo. However, isolating high-purity plant EVs from complex plant tissues, such as the tuberous roots of Ophiopogon japonicus, is challenging due to the dense cell wall matrix and high content of contaminants like polysaccharides. Existing isolation methods, including differential ultracentrifugation (DUC) and density gradient ultracentrifugation (DGUC), involve trade-offs between yield, purity, and vesicle integrity, necessitating the development of optimized protocols.ResultsWe developed and systematically optimized an integrated protocol for isolating high-purity EVs from O. japonicus roots. Key optimizations included: (1) refining the DUC protocol by incorporating a double ultracentrifugation step; (2) implementing a modified DGUC approach with a pre-clearing step for superior debris removal; and (3) evaluating enzymatic pre-treatment with cellulase and pectinase to enhance EVs release. Comparative analysis demonstrated that the optimized method, particularly utilizing enzymatic pre-processing and double ultracentrifugation, significantly improved plant EVs yield and purity. Small RNA (sRNA) sequencing of the resulting high-purity EVs successfully characterized their functional miRNA cargo profile, validating the efficacy of the isolation strategy.ConclusionsThis study establishes a robust and adaptable pipeline for isolating high-quality, functionally intact plant EVs from challenging plant root tissues. The optimized protocol effectively addresses the critical methodological challenges of yield and purity, enabling reliable downstream functional characterization and advancing therapeutic investigations of plant-derived EVs.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13007-025-01481-7.Keywords: Plant extracellular vesicles, Medicinal plant, miRNA, Ophiopogon japonicus Background Extracellular vesicles (EVs) are lipid bilayer-enclosed particles with sizes ranging from tens of nanometers to several micrometers, which serve as crucial mediators of intercellular communication. They transport a diverse cargo of bioactive molecules—including proteins, lipids, nucleic acids (DNA, mRNA, miRNA), and metabolites—reflective of their cellular origin [1]. EVs are classified into subtypes such as exosomes (30–150 nm, derived from multivesicular bodies), microvesicles (50–1000 nm, budding from the plasma membrane), and apoptotic bodies (500–4000 nm, released during apoptosis) based on their biogenesis pathways and size [2–4]. While initially viewed as mechanisms for cellular waste disposal [5], EVs are now recognized as sophisticated signaling entities with pivotal roles in mammalian physiology and pathology, including immune regulation, tumor progression, and tissue repair [6]. Their significant diagnostic and therapeutic potential as disease biomarkers and drug delivery vehicles is well-established [7–10], driving intense research into their fundamental biology and functions [11]. Plant EVs, first observed in carrot cell cultures in 1967 [12], have garnered substantial research interest only in recent years. Plant EVs play essential roles in fundamental plant processes such as growth and development, defense against pathogens, and intercellular signaling [13–15]. Particularly, note are plant-derived exosome-like nanovesicles (PDENs), which exhibit remarkable therapeutic potential, demonstrating effects including anti-inflammation [16], antioxidant activity [17], antitumor properties [18], gut microbiota modulation [19], and neuroprotection [20]. The natural biocompatibility, stability, and inherent capacity of PDENs to carry both hydrophilic and lipophilic molecules make them highly promising candidates as nanocarriers for targeted drug delivery applications [21]. A key functional component underpinning the biological activity of plant EVs is their miRNA cargo. miRNAs are small (~ 18–24 nucleotide (nt)) non-coding RNAs that silence target gene expression post-transcriptionally [22]. Importantly, plant EVs can transport these regulatory plant miRNAs across kingdoms, influencing the physiology of recipient cells. For instance, Arabidopsis thaliana EVs deliver miRNAs capable of suppressing virulence genes in the fungal pathogen Botrytis cinerea, thereby inhibiting infection [23–25]. This cross-kingdom regulatory potential highlights the significant functional relevance of plant EVs and their cargo. Critically, the therapeutic efficacy and functional relevance of isolated plant EVs are intrinsically linked to their purity and yield. However, isolating high-quality plant EVs presents unique and substantial challenges due to the complex composition of plant tissues, which are rich in polysaccharides, secondary metabolites, and structural components like cellulose and pectin [1, 26]. Current isolation strategies involve significant trade-offs. Differential ultracentrifugation (DUC), often considered a standard method, frequently suffers from low resolution, significant sample loss, and the co-isolation of non-EV contaminants such as protein aggregates and organelle fragments, resulting in suboptimal purity [1, 26]. Density gradient ultracentrifugation (DGUC), utilizing media like sucrose or iodixanol, significantly improves purity by separating plant EVs based on their characteristic buoyant density (typically 1.10–1.18 g/mL) [27, 28], but it can be time-consuming and may reduce overall yield. Alternative techniques such as ultrafiltration, immunoaffinity capture, polymer-based precipitation, and commercial kits offer different advantages but may introduce biases, risk damaging plant EVs, or lack scalability [29, 30]. Furthermore, optimal isolation protocols must be carefully tailored to specific plant organs (e.g., roots vs. leaves), considering factors like sap content, fiber abundance, and cell wall composition [26, 31]. A particularly persistent challenge is efficiently liberating plant EVs trapped within the dense plant cell wall matrix without introducing contaminants or degrading the vesicles themselves. Enzymatic pre-treatment (e.g., with cellulase and pectinase) offers a potential solution by degrading structural polysaccharides [3], but its impact on final EVs yield and the integrity of their molecular cargo requires careful evaluation. In this study, we address these critical methodological gaps by developing and systematically optimizing an integrated protocol for isolating high-purity EVs from the tuberous roots of O. japonicus, a valuable medicinal plant. We specifically evaluated and optimized enzymatic digestion using cellulase and pectinase to maximize EVs release while minimizing co-extracted impurities. We refined DUC protocols, incorporating a crucial post-ultracentrifugation wash step to enhance purity. Furthermore, we implemented a modified DGUC approach that included a pre-clearing centrifugation step to maximize debris removal prior to gradient separation. We rigorously compared the yield, purity, and integrity of EVs isolated via single versus double ultracentrifugation and enzymatic versus non-enzymatic pre-processing methods. Finally, to validate the functionality of the isolated vesicles, we utilized the high-purity EVs obtained through our optimized density gradient method for comprehensive sRNA sequencing to characterize their miRNA cargo profile. This work establishes a robust and adaptable pipeline for obtaining high-quality, functionally relevant plant EVs from challenging root tissues, enabling reliable downstream functional characterization and therapeutic investigations. Materials and methods Plant materials and growth condition Fresh O. japonicus samples used in this experiment were all collected from a standardized cultivation base located in Santai County, Sichuan Province, China. According to the differences in cultivation management practices, the samples were divided into two groups, including control group (CK) and treatment group (TK). All seedlings were transplanted in April 2023 and harvested in April 2024, corresponding to a 1-year cultivation cycle. Throughout the growth period, standard management protocols of the cultivation base were strictly implemented, including regular soil tillage (e.g., deep plowing and harrowing to improve soil structure), scientific water-fertilizer management (irrigation and fertilization adjusted based on the growth stage of O. japonicus), and integrated pest control. TK was the application of paclobutrazol during the specific growth stage, with a 15% paclobutrazol wettable powder solution sprayed on the foliage. The application rate was 0.6 kg active ingredient per hectare, and the spray was evenly distributed on both the adaxial and abaxial surfaces of leaves to ensure absorption. Isolation of plant EVs by differential ultracentrifugation (DUC) EVs were isolated from fresh roots of O. japonicus using a DUC protocol. Briefly, roots were weighed, thoroughly rinsed with deionized water, and cut into 0.5 cm segments with sterile scissors. The tissue sections were then homogenized in ice-cold 1 × phosphate-buffered saline (PBS, pH 7.4) using a mechanical homogenizer (Model MJ-BL8068, Midea Group Co., Ltd., Foshan, China). The resulting homogenate was sequentially filtered through sterile gauze to remove large debris. The clarified filtrate was subjected to a series of centrifugation steps at 4 °C: first at 1000 g for 10 min to remove large cellular fragments, followed by 5000 g for 20 min to pellet smaller debris and organelles, yielding the crude homogenate. This was subsequently centrifuged at 10,000 g for 40 min to remove larger vesicles and cellular bodies. The supernatant collected after centrifugation at 10,000 g was subjected to filtration. This was performed using a sterile, individually packed 0.22 μm polyethersulfone syringe filter (Cat. No. BS-PES-22; Biosharp Life Sciences Co., Ltd., Wuhan China), resulting in purified homogenate. Then carefully transferred to ultracentrifuge tubes (Cat. No. 355642, Beckman Coulter, Inc., Brea, CA, USA). Ultracentrifugation was performed at 150,000 g for 90 min at 4 °C using a Beckman Optima XE ultracentrifuge equipped with a SW32TI rotor (Beckman Coulter, Inc., Brea, CA, USA). The resulting pellet, containing crude EVs (designated as the initial pellet), was resuspended in 1.5 mL of ice-cold PBS to obtain the DUC1 EV fraction (Fig. 1A). Workflow scheme for the isolation and purification of O. japonicus roots by different methods. A Key steps in the differential ultracentrifugation (DUC) process. B Key steps in the density gradient ultracentrifugation (DGUC) process. C Key steps in the enzymatic digestion treatment process. For detailed methodology, please refer to the “Materials and methods” section in the main text To further purify the EVs and remove potential soluble contaminants or aggregates, the DUC1 EV fraction was purified by a second ultracentrifugation step under identical conditions (150,000 g, 90 min, 4 °C) and resuspended to yield the DUC2 EV fraction, which was used for subsequent analyses (Fig. 1A). Density gradient ultracentrifugation (DGUC) separates plant EVs The sample was first subjected to differential centrifugation followed by one round of ultracentrifugation. The resulting pellet was resuspended to obtain crude EVs (DUC1), which were then purified using discontinuous sucrose density gradient ultracentrifugation. Sucrose working solutions (8%, 30%, 45%, and 60% w/v) were prepared in sterile 1 × PBS (pH 7.4) and filtered (0.22 μm) [32, 33]. A discontinuous gradient was meticulously constructed in ultracentrifuge tubes (Cat. No. 343664, Beckman Coulter, Inc., Brea, CA, USA) by sequentially layering 2 mL of each sucrose solution, beginning with 60% at the bottom, followed by 45%, 30%, and finally 8% at the top, ensuring minimal disruption of the interfaces. The DUC1 EVs fraction (1.5 mL) was carefully loaded onto the pre-formed gradient. Gradients were subjected to ultracentrifugation at 150,000 g for 120 min at 4 °C using a Beckman Optima XE ultracentrifuge with a fixed-angle rotor (Model Type 70 Ti, Beckman Coulter, Inc., Brea, CA, USA). Following centrifugation, distinct bands were visible. The fraction corresponding to the buoyant density interface between the 30% and 45% sucrose layers was collected by careful pipetting. This fraction was diluted >10-fold with ice-cold PBS and pelleted by ultracentrifugation at 150,000 g for 90 min at 4 °C. The resulting pellet was resuspended in 1.5 mL of ice-cold PBS to yield the sucrose density gradient-purified EV fraction (designated DGUC1, prepared via method DGUC_Method1) (Fig. 1B). To obtain an ultra-purified EVs fraction (designated as DGUC2), the crude homogenate was subjected to repeated low-speed centrifugation before ultracentrifugation. Specifically, the sample underwent multiple sequential pre-clearing steps at 10,000 g for 40 min (4 °C) to exhaustively remove residual large debris. The resulting supernatant was then processed identically to the DGUC1 fraction through sucrose density gradient ultracentrifugation, as described previously. The final pellet was resuspended in 1.5 mL of ice-cold PBS to yield the DGUC2 EVs fraction. This fraction was prepared according to method DGUC_Method2 (Fig. 1B). Enzymatic digestion isolation of plant EVs EVs, designated here as O. japonicus root-derived extracellular nanovesicles (ORDENs), were isolated from fresh O. japonicus roots using an enzymatic digestion protocol. Briefly, roots were washed thoroughly with deionized water to remove soil contaminants, blotted dry, and weighed to record fresh biomass. The cleaned roots were homogenized in ice-cold 1 × PBS (pH 7.4) using a mechanical homogenizer at a tissue-to-buffer ratio of 1:2 (w/v). The resultant homogenate was filtered through sterile muslin cloth to remove large particulate matter. To digest the plant cell wall matrix and facilitate EVs release, the filtrate was supplemented with the following enzymes (final concentrations): 30 mg/mL cellulase (CAS 9012-54-8, Yakult Pharmaceutical Industry Co., Ltd., Tokyo, Japan) and 2 mg/mL pectinase (CAS 9033-35-6, Yakult Pharmaceutical Industry Co., Ltd., Tokyo, Japan) [28, 34]. The enzymatic digestion proceeded for 12 h at room temperature (approximately 25 °C) under gentle agitation. Following incubation, the digest was subjected to differential centrifugation at 4 °C to pellet cellular debris, including residual cell wall fragments and protoplasts. The centrifugation steps were as follows: 1000 g for 10 min, 5000 g for 20 min, and 10,000 g for 40 min. The supernatant enriched with ORDENs was passed through a 0.22 μm filter and subjected to ultracentrifugation to isolate ORDENs. Specifically, it was transferred to ultracentrifuge tubes (Cat. No. 355642, Beckman, USA) and centrifuged at 150,000 g for 90 min at 4 °C (Model Optima XE; Beckman Coulter, Inc., Brea, CA, USA). The resulting pellet was resuspended in PBS to obtain the purified ORDEN (Fig. 1C). Storage of isolated plant EVs The isolated EVs were resuspended in ice-cold 1 × PBS (pH 7.4) and immediately aliquoted. The aliquots intended for characterization analysis were stored at 4 °C post-isolation, with all characterizations completed within 3 days. The aliquots designated for sRNA sequencing were transferred into RNase-free tubes, snap-frozen in liquid nitrogen, and stored in a − 80 °C ultra-low temperature freezer until RNA extraction. Transmission electron microscope (TEM) analysis of plant EVs To confirm plant EVs morphology, 10 µL of sample suspension was dropped on the copper mesh with a liquid gun, timed for 10 min, and the excess liquid was sucked away with a small piece of filter paper. Each sample was stained with 3% uranyl acetate for 1–3 min and used a filter paper to wick away excess staining liquid before being imaged at 100 KV using a Transmission Electron Microscope (Model FEI Tecnai G2 12, Thermo Fisher Scientific Inc., Waltham, MA, USA). Analysis of plant EVs concentration and size distribution by nano flow cytometry (NanoFCM) The particle size distribution and concentration of the isolated plant EVs were determined using nanoparticle flow cytometry (Model Flow NanoAnalyzer U30E, NanoFCM biotech Co., Ltd., Xiamen, China). Briefly, the EVs sample was appropriately diluted with 1 × PBS to a concentration within the instrument's optimal detection range (typically 5 × 108 to 5 × 109 particles/mL) to minimize coincidence events. A minimum of 10,000 events were recorded for each sample at a flow rate of 100 µL/min. The volumetric absolute count method was applied for direct concentration measurement without the need for external standards. The derived particle size was calculated based on the scattered light intensity. Methodology for post-cryopreservation plant EVs characterization For long-term storage, EVs were aliquoted in PBS and preserved at − 80 °C. The frozen EV samples were maintained under these conditions for 18 months and did not undergo any freeze-thaw cycles prior to subsequent characterization. Post-thawed EVs were characterized for particle concentration and size distribution using NanoFCM (Model CytoFLEX nano, Beckman Coulter, Inc. Brea, CA, USA). Morphological assessment was performed by TEM as previously described in the Methods section. The total protein concentration of the isolated EVs was quantified with a bicinchoninic acid (BCA) assay kit (Cat. No. P0010S, Beyotime Biotechnology Co., Ltd., Shanghai, China). Analysis of EVs purity by membrane staining To confirm the lipid membrane integrity of the isolated vesicles and distinguish them from non-lipid particles, the samples were stained with a lipophilic fluorescent dye. A 10 µM working solution of POMV Membrane Green Stains (Cat. No. NHAPOMV-20, NanoFCM biotech Co., Ltd., Xiamen, China) was freshly prepared by a 10-fold dilution of the stock solution with the provided Dilution C. Then, 10 µL of EVs (approximately 2–5 × 1010 particles/mL, pre-diluted with Diluent C if necessary) was incubated with 10 µL of the working solution at 37 °C for 10 min in the dark. Following incubation, the stained sample was diluted 100-fold with 1 × PBS and analyzed immediately on the Flow NanoAnalyzer U30E (NanoFCM biotech Co., Ltd., Xiamen, China). An unstained EVs sample, processed identically but without the dye, was used as a negative control to set the fluorescence threshold. The percentage of membrane-positive particles was defined as the proportion of events exhibiting fluorescence intensity above the 99.99th percentile of the unstained control. Small RNA (sRNA) library construction and sequencing The isolated EVs from each sample were immediately snap-frozen in liquid nitrogen and stored at − 80 °C until RNA extraction. Total RNA was extracted from approximately 1 × 107 EV particles using TRNzol Universal reagent (Cat. No. DP424, Tiangen Biochemical Technology Co., Ltd., Beijing, China) according to the manufacturer’s protocol. Briefly, the EVs were lysed in 1 mL of reagent by vortexing, followed by a 5 min incubation at room temperature. Then, 0.2 mL of chloroform was added, the mixture was vigorously shaken for 15 s and incubated for 3 min at room temperature. Phase separation was achieved by centrifugation at 13,400 g for 15 min at 4 °C. The resulting aqueous phase (approximately 500 µL) was transferred to a new tube, and the RNA was precipitated with an equal volume of isopropanol during a 10 min incubation at room temperature. The RNA pellet was recovered by centrifugation at 13,400 g for 10 min at 4 °C, washed once with 1 mL of 75% ethanol, and centrifuged again at 10,000 g for 5 min at 4 °C. Finally, the pellet was air-dried and dissolved in 50 µL of RNase-free water. The RNA amount and purity of each sample was quantified using NanoDrop ND-1000 (Thermo Fisher Scientific Inc., Waltham, MA, USA). The RNA integrity was assessed by Bioanalyzer 2100 (Agilent Technologies, Inc., CA, USA) with RIN number > 7.0, and confirmed by electrophoresis with denaturing agarose gel. The sRNA libraries were constructed following the manufacturer’s instructions of TruSeq sRNA library preparation kits (Cat. No. RS-122-2001, Illumina Inc., San Diego, United States). All libraries were sequenced by Hiseq 2500 Sequencing & Pipeline filter in Lianchuan biotech Co., Ltd. (Hangzhou, China). Bioinformatics analysis of sRNAs and identification of miRNA Bioinformatics analysis of sRNA and miRNA identification were performed using the ACGT101-miR (v4.2) platform (LC Sciences, LLC, Houston, TX, USA). Raw reads were processed by first removing adapter sequences, contaminants, and low-quality reads to obtain clean data, after which sRNAs within the plant-specific length of 18–25 nt were retained. These filtered sequences were then aligned against mRNA, Rfam, and Repbase databases to exclude non-miRNA sequences such as rRNAs, tRNAs, snRNAs, snoRNAs, and repetitive elements. The remaining sequences were mapped to known miRNA precursors and the reference genome to identify known miRNAs and novel miRNAs derived from either the 3′ or 5′ arm of hairpin structures. The expression levels of miRNAs were normalized, and differential expression between comparison groups was evaluated using statistical methods such as Student's t-test or ANOVA, with significance thresholds set at p < 0.05 or 0.01. Finally, putative target genes of differentially expressed miRNAs were predicted and subjected to functional enrichment analysis to interpret their biological roles. To systematically analyze and identify potential biological pathways, the omics platform was used for GO and KEGG pathway enrichment analysis with the target genes of different miRNAs. GO enrichment analysis include three categories: biological process (BP), molecular functions (MF), and cell components. The GO enriched terms and KEGG pathway terms were identified with p < 0.05. The Top 25 of KEGG enrichment were selected for further analysis. Results DUC enhances EVs purity at the expense of yield DUC, a standard method for EVs isolation, was systematically evaluated by comparing single (DUC1) versus double (DUC2) ultracentrifugation wash steps for yield, purity, and integrity of EVs derived from O. japonicus roots. Visual inspection revealed significantly reduced turbidity in the DUC2 fraction compared to DUC1 (Fig. 1A). TEM confirmed higher contamination in DUC1, with abundant protein aggregates and fibrils, while DUC2 exhibited cleaner vesicle morphology (Fig. 2B). NanoFCM quantified this improvement: DUC2 EVs displayed a smaller mean diameter (67.5 nm vs. DUC1: 61.5 nm) and reduced concentration (1.27 × 1012 vs. DUC1: 5.05 × 1012 particles/mL), reflecting the removal of non-vesicular material (Fig. 2C). Fluorescent membrane labeling further validated enhanced purity, with 49.0% of DUC2 particles retaining intact membranes versus 46.4% in DUC1 (Fig. 2D). This demonstrates that the additional ultracentrifugation step effectively depletes co-precipitated contaminants (e.g., apoptotic bodies), albeit at the cost of reduction in particle yield. Double ultracentrifugations increase EVs purity but decrease yield. A Comparison of the color of EVs extracted by two different ultracentrifugation methods. Both methods used the same weight of roots (50 g). B Representative Transmission electron microscope (TEM) images of EVs separated by two DUC methods. Scale bars, 600 and 200 nm. C Particle size distribution graphs of EVs extracted by two DUC methods. D Particle size distribution graphs of EVs extracted by two DUC methods after being labeled with fluorescent dyes (Red: fluorescently labeled membrane-bound particles, i.e., EVs; Blue: particles not labeled with fluorescent dyes; Black: total number of particles) After long-term storage, the structural integrity of the EVs (DUC1 and DUC2) was confirmed by TEM, which showed that most EVs maintained a concave bilayer membrane structure (Fig. S1A). NanoFCM measurement determined that DUC1 had a concentration of 1.61 × 109 particles/mL and a mean size of 151.2 nm, while DUC2 had a concentration of 8 × 107 particles/mL and a mean size of 120.9 nm. Corresponding protein concentrations (BCA assay) were 1.58 mg/mL for DUC1 and 0.32 mg/mL for DUC2. Enhanced purity of EVs through homogenate purification and DGUC Fluid pre-clearing to achieve higher purity EVs suitable for downstream applications, sucrose DGUC was optimized. Sucrose gradients (60% / 45% / 30% / 8%) were layered, followed by loading of EVs pre-isolated via ultracentrifugation. Ultracentrifugation migrated EVs to their characteristic buoyant density (1.10–1.18 g/mL), visible as a discrete band at the 30–45% sucrose interface. We collected the target band, diluted it in PBS, and performed a final ultracentrifugation to pellet the purified material. (DGUC1, DGUC2) (Fig. 1B). Critically, we introduced a key improvement in the homogenate preparation to minimize contaminating debris. Specifically, the crude homogenate was subjected to iterative centrifugation at 10,000 g for 40 min. This cycle was repeated until no visible pellet formed, yielding the purified homogenate. (Fig. 1B). EVs derived from both homogenate sources banded identically during DGUC (Fig. 3A). TEM imaging revealed superior vesicle integrity and reduced fibrillar/protein contamination in the DGUC2 fraction compared to DGUC1 (Fig. 3B). NanoFCM confirmed DGUC2 EVs were smaller (62.4 nm vs. P3: 73.4 nm) and more concentrated (1.16 × 1012 vs. DGUC1: 2.8 × 1011 particles/mL) (Fig. 3C), consistent with effective contaminant removal. Fluorescent labeling demonstrated significantly higher membrane integrity in DGUC2 (75.3% labeled vesicles) versus DGUC1 (65.1%) (Fig. 3D), quantitatively validating the enhanced purity achieved by combining pre-cleared homogenate with DGUC. Purification of EVs by sucrose density gradient ultracentrifugation. (A) Left panel: the purified homogenate obtained using DGUC_Method1 and the EVs band formed after sucrose density gradient purification. Right panel: the purified homogenate obtained using DGUC_Method2 and the corresponding EV band following sucrose density gradient purification. Both methods used the same weight of roots (50 g). (B) TEM images of EVs separated by two different DGUC methods. Scale bars, 600 and 200 nm. (C) Particle size distribution graphs of EVs extracted by two different DGUC methods. (D) Particle size distribution graphs of EVs extracted by two different DGUC methods after being labeled with fluorescent dyes (Red: fluorescently labeled membrane-bound particles, i.e., extracellular vesicles; Blue: particles not labeled with fluorescent dyes; Black: total number of particles). DGUC1 and DGUC2 retained their structural integrity after storage (Fig. S1B). Characterization by nano-flow cytometry revealed a particle concentration of 5.13 × 107 particles/mL and a mean size of 143.3 nm for DGUC1, compared to 2.72 × 108 particles/mL and 126.6 nm for DGUC2. Subsequently, the BCA protein assay measured concentrations of 1.06 mg/mL and 0.52 mg/mL for DGUC1 and DGUC2, respectively. Enzymatic digestion enhances EVs release and purity To address the challenge of efficiently liberating EVs embedded within the robust plant cell wall matrix, we evaluated an enzymatic pre-digestion strategy. This approach utilizes hydrolytic enzymes, including cellulase and pectinase, to degrade the primary structural components of the cell wall, thereby facilitating the release of entrapped vesicles into solution while simultaneously reducing co-isolation of intracellular contaminants [35]. Building on its successful application in species like Catharanthus roseus [36], we adapted this method for O. japonicus roots, yielding the ORDENs. TEM confirmed that ORDENs exhibited classic cup-shaped exosome morphology with minimal non-vesicular debris (Fig. 4A). NanoFCM revealed a homogenous size profile, with the majority of particles ranging from 45 to 150 nm and a peak diameter of 77 ± 27.3 nm (Fig. 4B). Crucially, membrane integrity assays using fluorescent dyes demonstrated that 83% of particles in the ORDENs fraction were intact, labeled vesicles (Fig. 4C), indicating superior vesicle integrity and purity compared to fractions isolated by ultracentrifugation alone (DUC1, DUC2) or density gradient centrifugation (DGUC1, DGUC2). This confirms enzymatic digestion as a highly effective pre-processing step for isolating high-quality EVs from complex plant tissues. Extraction of EVs by enzymatic digestion method. (A) Representative TEM images of EVs isolated by enzymatic method. Scale bars, 600 and 200 nm. (B) Particle size distribution graph of EVs extracted by enzymatic digestion method. (C) Particle size distribution graph of EVs extracted by enzymatic digestion method after being labeled with fluorescent dyes (Red: fluorescently labeled membrane-bound particles, i.e., extracellular vesicles; Blue: particles not labeled with fluorescent dyes; Black: total number of particles). Following cryopreservation, the ORDENs were characterized and found to maintain structural integrity. NanoFCM analysis determined a particle concentration of 1.26 × 109 particles/mL with an average size of 130.4 nm(Fig. S1C). The protein concentration, as measured by a BCA assay, was 1.03 mg/mL. High-quality miRNA sequencing data validates the effectiveness of our isolation method We successfully isolated EVs from the roots of both control and treated O. japonicus using the DGUC (DGUC2, as outlined in Fig. 1B) (Fig. 5A&B). This method was selected over enzymatic digestion to evaluate an extraction protocol that avoids the potential co-isolation of residual enzymatic activities, which, despite yielding ORDENs with high purity, could interfere with subsequent functional assays. To rigorously evaluate the reliability and reproducibility of our method, high-quality miRNA sequencing was performed on EVs derived from two biologically independent replicates for both control and paclobutrazol-treated groups. Overall, 9,074,206–29,418,203 raw reads were obtained from these four sRNA libraries. The GC content of each sample was between 55.19 and 56.17%, and the Q30 value exceeded 95% (Table S1). After eliminating 3' adapters and low-quality sequences, 27.00 M clean reads were obtained. Compared the clean reads sequences with the RFam and Repbase databases, and removed the non-miRNA sequences from the clean data. Phenotype and miRNA profiles between control and treatment groups in O. japonicus. (A) Morphology of O. japonicus plants and roots in the control group (CK). (B) Morphology of O. japonicus plant and roots in the treatment group (TK) (C) Venn diagram of miRNAs identified in control and treatment groups. (D) Differential expression miRNAs (DEMs) between groups. (E) Correlation of miRNA expression between treatment and control groups. (F) Significance and expression of key DEMs. After filtering out the junk reads, repeats, and adapters sequences, 3,944,772–13,379,771 clean reads were obtained (Table S2). Then, the remaining sequences were aligned to mRNA, Rfam and Repbase database to discard ncRNAs (rRNA, tRNA, snoRNA, and snRNA), other Rfam RNA, and repeat sequences, and the filtered sequences were used for miRNAs identification. A total of 228 expressed miRNAs were identified in all samples (Table S3). All these miRNAs were divided into five groups, then 201 pre-miRNAs (1, 13, 79, 33 and 75 in gp1, gp2a, gp2b, gp3 and gp4, respectively) and 174 unique miRNAs (2, 14, 59, 35 and 64 in gp1, gp2a, gp2b, gp3 and gp4, respectively) were obtained (Table S4). Interestingly, the number of unique miRNAs in gp4 (novel miRNA group) was the largest. All mapped sRNAs within the length range of 18–24 nt were counted for the total and unique reads, most of samples showed the highest abundance at 21 nt (Fig. S2A and Table S5). A total of 142 known and 86 novel mature miRNAs were consistently identified across all replicates, confirming that our EVs extraction method effectively captures a diverse miRNA population. The known miRNAs were classified into 32 families, with miR2592 (9 members), miR168 (8 members), and miR156 (7 members) being the most abundant (Table S6). The length distribution of miRNAs showed that those sequences of 21 nt were the most abundant size class of the unique miRNAs followed by 18, 22, 19 and 20 nt (Fig. S2A). Both known and novel miRNAs showed a typical length distribution, further supporting the successful enrichment of EV-derived miRNAs. Further, to reveal the conservation of identified miRNAs with other species in this study, we compared pre-miRNAs with other species in miRbase, and found that 139 miRNAs were highly conservative with their homologue in soybean (Glycine Max) (Fig. S2B). To enhance the reliability and stability of miRNAs identified here, the miRNAs expressed only in a library of two biological repeats were filled out. The Venn analysis showed that 79 known miRNAs were identified both in the control and treatment groups, while 47 and 8 miRNAs were expressed only in treatment group and control group, respectively (Fig. 5C). Expression analysis showed that 78 known miRNAs were differentially expressed in the control and treatment groups. 13 miRNAs were differentially expressed in the treatment group, with 6 up-regulated and 7 down-regulated (p < 0.05; Table S7 and Fig. 5D). Among them, 9 were known Differentially expressed miRNAs (DEMs) from 7 miRNA families. miR156 and miR159 families had 2 members each, and the other 5 families possessed 1 member each (Fig. 5E&F). Functional analysis and validation of known and unknown miRNA in EVs To explore the specific functions of candidate miRNAs mentioned above, their target genes were screened and obtained from degradome data. The GSTAr (v1.0) was used to predict the target genes of the known and novel miRNAs. A total of 232 target genes were predicted. Then, functional classification of target genes was performed by GO analysis under biological processes (BP), cell components (CC) and molecular functions (MF) categories (Table 1). GO enrichment analysis of the target genes of differentially expressed miRNAs revealed that in the biological process category, multiple processes related to leaf development (GO:0048366), root development (GO:0048364), and the cell cycle were significantly enriched. In the cellular component category, terms like extracellular region (GO:0005576) and exosome (GO:0000178) were enriched, indicating that some target gene products function in extracellular vesicles (exosomes). In the molecular function category, functional groups such as sesquiterpene synthase activity (GO:0010334) and P-type calcium transporter activity (GO:0005388) were significantly enriched (Fig. 6A). Interestingly, it was showed that most target genes of identified miRNAs were enriched in sesquiterpene synthase activity and sesquiterpene biosynthetic process. Table 1GO enrichmentGO IDGO TermGO CategoryRich FactorP valueSig. GO:0051762Sesquiterpene biosynthetic processBP0.3757E-18****GO:0010334Sesquiterpene synthase activityMF0.32144E-17****GO:0005388P-type calcium transporter activityMF0.19232E-16****GO:00084083’-5’ exonuclease activityMF0.15256E-14****GO:0004527Exonuclease activityMF0.10472E-12****GO:0010507Negative regulation of autophagyBP0.3753E-12****GO:0033897Ribonuclease T2 activityMF0.33336E-12****GO:0010168ER bodyCC0.27272E-11****GO:0042325Regulation of phosphorylationBP0.38462E-10****GO:0009048Dosage compensation by inactivation of X chromosomeBP0.82E-10****GO:1,904,872Regulation of telomerase RNA localization to Cajal bodyBP0.82E-10****GO:0071044Histone mRNA catabolic processBP0.66677E-10****GO:0090503Obsolete RNA phosphodiester bond hydrolysis, exonucleolyticBP0.66677E-10****GO:0000178Exosome (RNase complex)CC0.57142E-09****GO:0045339Farnesyl diphosphate catabolic processBP0.53E-09****GO:0004532RNA exonuclease activityMF0.53E-09****GO:0071048Nuclear mRNA surveillanceBP0.53E-09****GO:0016075rRNA catabolic processBP0.10711E-08****GO:0031597Cytosolic proteasome complexCC0.10531E-08****GO:0036402Proteasome-activating activityMF0.10531E-08****GO:0045899Positive regulation of RNA polymerase II transcription preinitiation complex assemblyBP0.10531E-08****GO:0006401RNA catabolic processBP0.10531E-08****GO:0045338Farnesyl diphosphate metabolic processBP0.17861E-08****GO:0016838Carbon-oxygen lyase activity, acting on phosphatesMF0.36362E-08****GO:0007155Cell adhesionBP0.16672E-08****GO:0031595Nuclear proteasome complexCC0.09682E-08****GO:0005887Plasma membraneCC0.01595E-08****GO:0006364rRNA processingBP0.03179E-08****GO:0005576Extracellular regionCC0.00931E-07****GO:0004521RNA endonuclease activityMF0.06821E-07****GO:0000325Plant-type vacuoleCC0.03492E-07****GO:0008540Proteasome regulatory particle, base subcomplexCC0.06382E-07****GO:0007568Obsolete agingBP0.06322E-07****GO:0017025TBP-class protein bindingMF0.05714E-07****GO:0032211Negative regulation of telomere maintenance via telomeraseBP0.16675E-07****GO:0005622Intracellular anatomical structureCC0.01325E-07****GO:0000956Nuclear-transcribed mRNA catabolic processBP0.166E-07****GO:0071034CUT catabolic processBP0.15387E-07****GO:0004321Fatty-acyl-CoA synthase activityMF0.11432E-06****GO:0019888Protein phosphatase regulator activityMF0.06253E-06****GO:0035327EuchromatinCC0.11113E-06****GO:00162074-coumarate-CoA ligase activityMF0.10813E-06****GO:0005773VacuoleCC0.01133E-06****GO:0009851Auxin biosynthetic processBP0.10264E-06****GO:0008285Negative regulation of cell population proliferationBP0.05684E-06****GO:0030433Ubiquitin-dependent ERAD pathwayBP0.03775E-06****GO:0071035Nuclear polyadenylation-dependent rRNA catabolic processBP0.08897E-06****GO:0071028Nuclear mRNA surveillanceBP0.08169E-06****GO:0061088Obsolete regulation of sequestering of zinc ionBP0.081E-05****GO:0043231Intracellular membrane-bounded organelleCC0.01231E-05****GO:0004816Asparagine-tRNA ligase activityMF0.06672E-05****GO:0006421Asparaginyl-tRNA aminoacylationBP0.06672E-05****GO:00001753’-5’-RNA exonuclease activityMF0.06562E-05****GO:0000460Maturation of 5.8 S rRNABP0.05485E-05****GO:0005777PeroxisomeCC0.01455E-05****GO:0031408Oxylipin biosynthetic processBP0.05136E-05****GO:0004252Serine-type endopeptidase activityMF0.01916E-05****GO:0008559ABC-type xenobiotic transporter activityMF0.04887E-05****GO:0009850Auxin metabolic processBP0.04887E-05****GO:0009695Jasmonic acid biosynthetic processBP0.04828E-05****GO:0010321Regulation of vegetative phase changeBP0.33330.0001***GO:0000176Nuclear exosome (RNase complex)CC0.04350.0001***GO:0005385Zinc ion transmembrane transporter activityMF0.04210.0001***GO:0048366Leaf developmentBP0.01990.0002***GO:0003916DNA topoisomerase activityMF0.20.0003***GO:0010043Response to zinc ionBP0.03280.0003***GO:0046686Response to cadmium ionBP0.00910.0004***GO:0003917DNA topoisomerase type I (single strand cut, ATP-independent) activityMF0.11760.0009***GO:0016485Protein processingBP0.02420.001**GO:0048658Anther wall tapetum developmentBP0.11110.001**GO:0030527Structural constituent of chromatinMF0.10530.0012**GO:0048364Root developmentBP0.01260.0018**GO:0043682P-type divalent copper transporter activityMF0.08330.0019**GO:0004477Methenyltetrahydrofolate cyclohydrolase activityMF0.07690.0022**GO:0004488Methylenetetrahydrofolate dehydrogenase (NADP+) activityMF0.07690.0022**GO:0010076Maintenance of floral meristem identityBP0.07410.0024**GO:0009554MegasporogenesisBP0.06060.0035**GO:0030163Protein catabolic processBP0.01620.0045**GO:0050105L-gulonolactone oxidase activityMF0.05260.0046**GO:0052694Jasmonoyl-isoleucine-12-hydroxylase activityMF0.50.0053**GO:0016874Ligase activityMF0.01540.0053**GO:0035066Positive regulation of histone acetylationBP0.04760.0057**GO:0003885D-arabinono-1,4-lactone oxidase activityMF0.04760.0057**GO:2,000,012Regulation of auxin polar transportBP0.04650.0059**GO:0016020MembraneCC0.00520.0073**‘**’ indicates p < 0.005, ‘***’ indicates p < 0.001, ‘****’ indicates p < 0.0001 GO enrichment ‘**’ indicates p < 0.005, ‘***’ indicates p < 0.001, ‘****’ indicates p < 0.0001 GO and KEGG enrichment analysis of DEMs target genes. (A) GO enrichment bar plot. (B) GO enrichment scatter plot. (C) KEGG enrichment bar plot. (D) KEGG enrichment scatter plot Subsequently, KEGG analysis was used to explore the metabolic pathways regulated by target genes, and showing that 18 metabolic pathways were significantly enriched (Table S8). Interestingly, most pathways were enriched in Metabolism, especially in metabolism of terpenoids and polyketides and biosynthesis of other secondary metabolites (Fig. 6C&D), the sesquiterpenoid and triterpenoid biosynthesis (map00909) accounts the most (Table 2). Combined with the GO analysis, it was found that target genes were enriched in biosynthesis and activity of sesquiterpenoids and triterpenoids, which are two crucial classes of secondary metabolites in plants. Sesquiterpenoids are usually involved in plant defense, insect attraction, and antifeedant responses, whereas triterpenoids, including saponins (such as ginsenosides and soyasaponins), sterols, play key roles in defense, signaling, and interactions with microorganisms [37, 38]. This indicates that miRNAs in exosomes can target the biosynthetic genes of these pathways, and suggests that O. japonicus plants can remotely and precisely" turn off" or "downregulate" these metabolic processes in recipient cells. Table 2KEGG enrichment analysis of DEMs target genesPathwayKEGG Level 2Pathway NameRich FactorP valueSig.map00909Metabolism of terpenoids and polyketidesSesquiterpenoid and triterpenoid biosynthesis0.05237.54311E-10****map00525Biosynthesis of other secondary metabolitesAcarbose and validamycin biosynthesis0.16131.71603E-08****map00523Metabolism of terpenoids and polyketidesPolyketide sugar unit biosynthesis0.13893.76866E-08****map00521Biosynthesis of other secondary metabolitesStreptomycin biosynthesis0.11111.19904E-07****map03050Folding, sorting and degradationProteasome0.02097.38953E-06****map00515Glycan biosynthesis and metabolismMannose type O-glycan biosynthesis0.08330.00011423***map04392Signal transductionHippo signaling pathway - multiple species0.28570.000139345***map04111Cell growth and deathCell cycle - yeast0.250.000185479***map04113Cell growth and deathMeiosis - yeast0.15380.00051233***map00910Energy metabolismNitrogen metabolism0.02650.000666004***map00220Amino acid metabolismArginine biosynthesis0.02290.001152312**map00592Lipid metabolismAlpha-Linolenic acid metabolism0.01590.00424437**‘**’ indicates p < 0.005, ‘***’ indicates p < 0.001, ‘****’ indicates p < 0.0001 KEGG enrichment analysis of DEMs target genes ‘**’ indicates p < 0.005, ‘***’ indicates p < 0.001, ‘****’ indicates p < 0.0001 Discussion EVs encapsulate bioactive molecules such as proteins, lipids, and nucleic acids (including miRNAs). Their lipid bilayer membrane structure protects the internal miRNAs from degradation in the extracellular environment and enables uptake by recipient cells through endocytosis, thereby facilitating intercellular communication. In plants, the functional study of EVs, similar to exosomes, is emerging as a hot research field. Robust isolation of high-purity EVs from plant tissues remains a significant methodological challenge, hindering the exploration of their diverse roles in intercellular communication, development, and stress responses [26, 35, 39]. This study addresses this gap by presenting a systematic optimization and comparative evaluation of EVs isolation protocols specifically tailored for the fibrous roots of O. japonicus, a homology plant of food and medicine. Our integrated approach, refining both tissue pre-processing and downstream purification strategies, demonstrably enhances EVs yield while critically minimizing co-isolated contaminants, as rigorously validated through multi-modal characterization TEM, NanoFCM, sRNA-Sequencing) and adherence to MISEV (Minimal information for studies of extracellular vesicles) guidelines [36, 40, 41]. Typically, the isolation of EVs from plant material requires the prior acquisition of a tissue homogenate [42]. While sap-rich tissues often allow direct juice expression, low-sap tissues like O. japonicus roots necessitate alternative approaches [43]. We critically compared juice extraction and enzymatic digestion (cellulase-pectinase). Juice extraction offered procedural simplicity and high throughput, yielding larger quantities of crude EVs (Fig. 1A). However, this method inevitably co-isolated substantial amounts of plant-derived contaminants, primarily cellulose and pectin fragments, as evidenced by TEM and impurity analysis (Figs. 2Band 3B). This observation aligns with the reported challenges in plant EVs research, where bulk preparations from complex tissues are often confounded by non-vesicular contaminants [44, 45]. In contrast, enzymatic digestion yielded EVs of markedly higher purity (Fig. 4A–C). In mammalian studies, enzymatic treatments (e.g., collagenase) are routinely applied to tissues to liberate EVs from the extracellular matrix [46, 47]. Our work demonstrates that a parallel principle is applicable to plants, where targeting the unique polysaccharide-based cell wall with specific hydrolases facilitates a more selective release of EVs, thereby minimizing co-isolation of intracellular components and wall debris [28, 41]. However, the enzymatic method resulted in a decreased final yield of EVs (Fig. 4B). This clear trade-off between yield and purity underscores the importance of selecting the pre-processing method based on the specific downstream application requirements (e.g., bulk characterization vs. detailed molecular profiling). DUC is widely employed for plant EVs isolation [15, 17, 32, 48, 49], but standard protocols require optimization for specific tissues. In this study, we successfully isolated EVs from O japonicus roots using DUC. The implementation of a second ultracentrifugation step effectively reduced soluble contaminants and co-precipitated aggregates during the first spin, leading to a substantial improvement in EVs purity (Fig. 2C&D), despite an expected reduction in absolute particle number. This refinement is particularly valuable for applications demanding high sample purity. DUC combined with density gradient centrifugation can isolate EVs with higher purity than using DUC alone [42, 50]. The method has been successfully applied to isolate plant EVs from a variety of plants, including Centella Asiatica, Portulaca oleracea, and Citrus limon, thereby facilitating research into their biochemical composition and therapeutic potential [51–53]. To achieve the higher purity essential for confident molecular characterization and functional studies, we employed and optimized sucrose DGUC, building upon existing methods [32, 54]. Prior to loading the crude EVs sample onto the sucrose gradient, an additional high-speed pre-clearing centrifugation step (10,000 g) was introduced. This pre-clearing step maximized the removal of residual plant debris and large aggregates from the homogenate, resulting in a cleaner input for the density gradient (Fig. 1B). DGUC leverages the characteristic buoyant density of plant EVs (1.10–1.18 g/mL, corresponding to the 30–45% sucrose interface) to effectively separate them from co-sedimenting contaminants with overlapping sedimentation coefficients, such as protein complexes, lipoprotein aggregates, and residual small debris [31, 55, 56]. EVs isolated from the optimal density fraction (DGUC2) exhibited superior purity, provide a more confident assignment of the obtained sRNA cargo to bona fide EVs. To ensure the reproducibility of downstream functional studies and clinical applications, the long-term storage stability of EVs is a critical prerequisite [57]. This study focused on evaluating the effects of − 80 °C freezing on the physicochemical properties of plant EVs. TEM results showed no significant morphological changes in the EVs after one freeze-thaw cycle, indicating relatively good physical structural stability (Fig. S1). This finding is consistent with related studies on EVs derived from various plant and mammalian sources, further supporting that basic freezing conditions can generally preserve vesicle integrity [58]. However, NanoFCM data revealed a decrease in particle concentration and a slight increase in average particle size post-freeze-thaw. Cheng et al. also demonstrated a decrease in EVs concentration after 24 h storage at − 80 °C [59]. Wu et al. observed that the percentage of small particles (30–150 nm) in small EVs isolated from cells showed a decreasing trend with prolonged storage time, while the proportion of large particles (150–500 nm) increased [60]. We speculate that this phenomenon may be related to EVs adhesion or aggregation during long-term storage, such as vesicle adsorption to the tube walls or slight fusion between vesicles. Despite the differences in the biological origins of EVs, the patterns of change in physical stability post-cryopreservation demonstrate a certain degree of conservation, which may be attributed to the universal biophysical properties of the vesicle lipid bilayer. Additionally, variations in instrument detection could also contribute to the observed results. Despite the differences in the biological origins of EVs, the patterns of change in physical stability post-cryopreservation demonstrate a certain degree of conservation, which may be attributed to the universal biophysical properties of the vesicle lipid bilayer. sRNA sequencing of DGUC2 EVs served a dual purpose: providing insights into the EV miRNA cargo and establishing a novel, molecular-based quality assessment metric. We identified 228 mature miRNAs (142 known + 86 novel), with the miR2592, miR168, and miR156 families being predominant (Table S3&S6). The remarkable abundance of miR2592, rarely a major player in model plants, highlights potential unique regulatory networks in medicinal species like O. japonicus and serves as a distinctive molecular signature of its root EVs. The prevalence of miR168, a known regulator of AGO1 and thus global RNA silencing efficiency [61, 62], suggests a potential mechanism for selective miRNA sorting into EVs or a role in modulating recipient cell silencing machinery. Among 13 DEMs identified in response to treatment (|log2FC| >1, FDR < 0.05; Fig. 5F), two belonged to the miR156 family. This is mechanistically significant as miR156 targets SPL transcription factors to repress developmental transitions [63–65]. Its altered abundance in EVs provides a compelling molecular link to the observed paclobutrazol-induced root thickening phenotype in O. japonicus, further corroborated by GO term enrichment for "root development" (GO:0048364). Strikingly, target genes of these DEMs were significantly enriched in terms directly associated with EV biology ("exosome" GO:0000178, "extracellular region" GO:0005576; Fig. 6B). This strongly suggests that these DEMs, shuttled within EVs, may orchestrate the intercellular transport and regulation of key enzymes, such as monoterpene synthases, involved in specialized metabolism. Furthermore, KEGG pathway analysis revealed that DEMs target genes were significantly enriched in biosynthesis pathways for pharmacologically active compounds, particularly sesquiterpenoids/triterpenoids (map00909) (Fig. 6D). This positions EVs and their miRNA cargo as key regulators of the valuable secondary metabolites characteristic of O. japonicus. In summary, this study delivers an optimized, integrated pipeline for isolating high-purity EVs from the challenging matrix of O. japonicus roots, establishing a viable source for plant-derived EVs with potential biotechnological and therapeutic applications. However, the field of plant EVs research continues to face substantial methodological hurdles: (1) The inherent complexity of plant extracts, rich in polysaccharides, phenolics, and other secondary metabolites, presents persistent challenges for achieving absolute purity and significantly increases processing complexity and cost. (2) A critical limitation is the current lack of conserved, plant-specific EV protein markers analogous to CD63/TSG101 in animal systems. Identification still heavily relies on physical characteristics (NanoFCM size distribution, TEM morphology) and cargo analysis, necessitating complementary approaches like the miRNA profiling demonstrated here. Our findings that O. japonicus root EVs carry functionally relevant miRNAs, including development-regulating miR156 and those targeting medicinal compound pathways, provide compelling evidence that paclobutrazol may modulate root architecture, in part, by altering the EV-mediated intercellular trafficking of regulatory RNAs. This underscores EVs as active signaling entities in plants. Future research must focus on elucidating the precise loading mechanisms of specific miRNAs into EVs, their uptake and functional impact in recipient cells, and their definitive roles in mediating developmental and metabolic reprogramming within plant tissues. Conclusions We conducted a comparative analysis of three methods for isolating EVs from O. japonicus roots. Our results demonstrate that the isolated EVs exhibit sufficient yield, purity, and structural integrity for downstream analyses. Utilizing this optimized approach, we performed sRNA sequencing on O. japonicus roots-derived EVs. This detailed characterization of EV miRNA profiles will advance our understanding of vesicle-mediated signaling and the mechanistic actions of agrochemicals. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements We extend our sincere gratitude to the Researchers Supporting Project at the State Key Laboratory of Southwestern Chinese Medicine. Abbreviations Bicinchoninic acid Biological process Cell components Control group Differential ultracentrifugation Differentially expressed miRNAs Density gradient ultracentrifugation Extracellular vesicles microRNA molecular functions Nucleotide Minimal information for studies of extracellular vesicles NanoFCM Ophiopogon japonicus root-derived extracellular nanovesicle Phosphate-buffered saline Plant-derived exosome-like nanovesicles Small RNA Transmission electron microscope Treatment group Author contributions Y.X. carried out the design of this research work and writing this manuscript; Y.X., X.Z and S.Y.C. performed the experiments; Z.H.L., Q.Z. and L.Q.F. curated the data; F.Q.L. and Y.X. performed the formal analysis and validated the results; B.J.X. administered the project; Y.T.M., T.Z and B.J.X. supervised the study; Y.X. B.J.X. wrote the main manuscript text; T.Z. B.J.X. reviewed and edited the manuscript. All authors read and approve the final manuscript. Funding This work is supported by Natural Science Foundation of Sichuan province (No. 2025ZNSFSC0165). Data availability Raw Illumina sequence data were deposited in the Short Read Archive of the NCBI database (Biosample accession numbers are SAMN50730936, SAMN50730937, SAMN50730938, SAMN50730939). MiRNA data are available via NCBI with accession (PRJNA1309075). Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Generative AI statement The authors declare that no generative AI was used in the creation of this manuscript. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Tao Zhou, Email: 364462907@qq.com. Yuntong Ma, Email: mayuntong@cdutcm.edu.cn. Binjie Xu, Email: binjiexu@outlook.com. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement Raw Illumina sequence data were deposited in the Short Read Archive of the NCBI database (Biosample accession numbers are SAMN50730936, SAMN50730937, SAMN50730938, SAMN50730939). MiRNA data are available via NCBI with accession (PRJNA1309075).