Structural and Functional Characterization of EXPO‐Derived Extracellular Vesicles in Plants

ABSTRACT

Extracellular vesicles (EVs) are increasingly being recognized as important mediators of intercellular communication in plants, but their biogenesis, heterogeneity, and membrane origins remain poorly understood. Here, the structural diversity, formation mechanisms and potential functions of EVs from various plant cell types under normal physiological conditions are demonstrated using 3D electron tomography (ET), cryo‐ET, and immunogold transmission electron microscopy (TEM). The EVs are classified into three distinct categories based on their size, content, and molecular marker profiles. These are TET8‐positive small EVs (∼50–200 nm in diameter, lacking organelle content), Exo70E2‐positive medium EVs (∼200–500 nm in diameter, containing ribosomes), and PEN1‐positive large EVs/Extracellular tubules (∼500–2000 nm in diameter, containing ribosomes and small vesicles). The Exo70E2‐positive medium EVs originate from the plant‐specific exocyst‐positive organelle (EXPO), and isolated EXPOs carry cargoes associated with stress responses. Subsequent transcriptomic profiling and pathogen‐resistance assays in exo70e2 mutants indicate that EXPO‐derived EVs contribute to plant defense, potentially by delivering defense‐related proteins during pathogen infection. Collectively, these findings provide a framework for understanding EV heterogeneity in plants and highlight EXPO‐derived EVs as a potential key component of plant defense.

In this study, 3D electron tomography (ET), cryo‐ET, and immunogold transmission electron microscopy (TEM) are employed to characterize plant extracellular vesicles (EVs) under physiological conditions. EVs are classified into three distinct categories according to their size, content, and molecular‐marker profiles. Furthermore, Exo70E2‐positive medium EVs are shown to originate from the plant‐specific exocyst‐positive organelle (EXPO), potentially contributing to plant stress responses.

1. Introduction

Exocytosis is a critical secretory pathway in all eukaryotic cells, involving multiple membrane‐bound organelles that coordinate the transport and delivery of proteins and lipids. In general, there are two main exocytic pathways; namely, the conventional protein secretion (CPS) and unconventional protein secretion (UPS) pathways. In the CPS pathway, proteins are secreted through the endoplasmic reticulum (ER) and then transported to the Golgi apparatus before being delivered to the trans‐Golgi network (TGN). At the TGN, the proteins are packaged into secretory vesicles, which then fuse with the plasma membrane (PM). In this way, the protein cargoes are either released into the cell periphery or, in the case of membrane proteins, delivered to the PM [1, 2]. Entry into the CPS pathway typically requires N‐terminal signal peptides or transmembrane domains. However, recent studies have revealed that many secretory proteins lack both these features, indicating the presence of non‐classical secretion pathways that bypass the CPS in eukaryotic cells [1, 2, 3, 4]. Unlike the CPS, the UPS transports proteins lacking signal peptides, along with other cellular components to the extracellular space. One prominent UPS pathway involves the transport of proteins, RNAs, and lipids via extracellular vesicles (EVs) [5, 6, 7].

EVs are membrane‐bound nano‐vesicles, which are released by cells into the extracellular space and EV‐mediated secretion has been identified in almost all eukaryotic cells [8]. In mammals and yeast, three major types of EV have been well characterized. These are: 1) Exosomes, which are intraluminal vesicles (ILVs) released when multivesicular bodies (MVBs) fuse with the PM; 2) Microvesicles, also known as micro‐shedding vesicles which originate when specific sites of the PM undergo outward budding; and 3) Apoptotic bodies, which are made up of cell debris generated during programmed cell death [9, 10].

The existence and functional roles of EVs in plant cells are relatively less well explored due to the presence of the rigid cell wall, which may impede their formations and functions [6, 11], although in the last few decades, accumulating evidence suggests that plants also secrete functional EVs. Indeed, the analysis of images acquired by 2D transmission electron microscopy (TEM) indicates EV‐like structures outside the cell and the isolation of EVs from the apoplastic wash of leaves under pathological conditions [12, 13, 14, 15, 16]. However, the origin and nature of such EVs in plants remain largely unknown. Plant EVs were first identified in the 1960s when TEM studies on chemically fixed carrot cells revealed the presence of EVs of various sizes and demonstrated the fusion of MVB‐like structures with the PM [17]. Furthermore, in Arabidopsis leaf tissue, upon pathogen infection, some MVB‐like compartments have been observed to accumulate and fuse with the PM, forming paramural bodies and releasing exosome‐like EVs into the apoplast [18]. Moreover, recent study also provided TEM evidence of exosome‐like EVs being released at fungal infection sites in plants [19]. Notably, EV structures in these studies are primarily observed in chemically fixed samples upon biotic stresses, the origin and nature of these EVs under normal physiological condition in intact plant cells remains to be illustrated.

Using the Exo70 family protein Exo70E2 as a marker, a plant‐unique organelle has recently been identified [20, 21]. Through 2D TEM analysis, double‐membrane structures within the cytosol were observed, which were shown to fuse their outer membrane with the PM and simultaneously releasing a single‐membrane‐bound EV into the apoplast. This newly identified organelle was termed the Exocyst‐Positive Organelle (EXPO) [21]. However, the reliance on 2D TEM for studying EV structures presents challenges with regards to accurately characterizing their size, shape, and cargo content [22].

To gain a deeper insight into these nanovesicles, numerous methods have been developed to isolate EVs from various plant species, followed by proteomic and morphological analyses of the isolated vesicle mixtures. For example, a novel procedure was devised to isolate EVs from the apoplastic wash of Arabidopsis leaves by employing vacuum infiltration followed by ultracentrifugation to better preserve the native physiological conditions [23, 24, 25]. This optimized method subsequently enabled RNA profiling and other molecular analyses of EV mixtures across different plant species, significantly advancing our understanding of plant EVs components [26, 27, 28]. Although multiple studies have identified different types of plant EVs and their distinct functional roles, there remains a significant gap in the rigorous and systematic characterization of their nature and identity, which limits our understanding of their biogenesis and trafficking within intact plant cells.

In this study, we employed a combination of 3D electron tomography (ET), Cryo‐ET, correlated light and electron microscope (CLEM) and immunogold‐TEM, to investigate plant EVs across various cell types in different plant species, with a particular focus on root border cells, leaf guard cells and pollen grains and tubes, which were found to be enriched in EV structures [12, 13, 29]. Based on size, content, and molecular marker profiles, we characterized EV populations under normal physiological conditions and classified them into three distinct categories: TET8‐positive small EVs (50–200 nm in diameter and lacking organelle content), Exo70E2‐positive medium EVs (200–500 nm in diameter and containing ribosomes), and PEN1‐positive large EVs/Extracellular tubules (500–2000 nm in diameter and containing ribosomes and small vesicles). We also showed that Exo70E2‐positive medium EVs originate from the plant‐specific EXPO and demonstrated their trafficking in vivo. Furthermore, we performed proteomic analysis on isolated EXPOs and identified novel cargoes that are potentially involved in plant stress responses. Given the critical role of Exo70E2 in EXPO formation and trafficking [20], we conducted RNA sequencing and pathogen defense assays on mutant lines of exo70e2. Our results revealed that EXPO‐derived EVs play a positive role in plant defense against Pst DC3000 infection, potentially by delivering defense‐related proteins during pathogen infection.

2. Results

2.1. Identification of EVs in Different Plant Cell Types

To identify plant EVs, we initially focused on the root border cells. Extensive screening using traditional 2D transmission electron microscopy (TEM) on the wild‐type root tip revealed a significant population of membrane‐bound vesicles within the apoplast that resembled EV structures and exhibited distinct morphology. However, recognizing the limitations of 2D TEM for characterizing stereoscopic vesicular structures [30], we employed 3D electron tomography (ET) that allows precise estimation of vesicle sizes and contents [31, 32] while providing a 3D perspective to determine whether the apoplastic vesicles are completely detached from the PM, a key criterion for identifying these structures as authentic plant EVs.

To ensure that the observed membrane structures in root cells were retained as close to their native states as possible while minimizing fixation artifacts, we fixed samples using the high‐pressure freezing/freeze substitution (HPF/FS) method [33, 34, 35]. Figure 1A presents a low‐magnification TEM micrograph of a wild‐type Arabidopsis root tip. Higher magnification images of an area encompassing the root border cells show abundant small EVs with uniform sizes ranging from ∼50–200 nm in diameter (mean ± SD = 105 ± 32 nm; n = 66). These round‐shaped vesicles lack visible content (e.g., ribosomes) within their lumen (Figure 1B,C,N and Video S1, Supporting Information) and are clearly separated from the PM.

FIGURE 1.

Identification and characterization of distinct extracellular vesicles (EVs) in Arabidopsis root border cells using 3D electron tomography. (A) A low‐magnification electron microscopy image of a wild‐type Arabidopsis root tip. Scale bars, 10 µm. The boxes highlighted the locations of the interested EV structures with numbers indicating the corresponding areas in (B–G). (B–G) Tomographic slices (B, D and F) and corresponding 3D models (C,E,G) as indicated in (A) reveal small (B,C), medium (F‐G) and large (D,E) EVs are separated from the plasma membrane (PM, membrane model in green color). The arrows in matching color in tomographic slices and models indicate the same structures. Please note that a ribosome (indicated by the red arrow) was found outside the cell (F). Scale bars, 200 nm. (H,I) A double membrane organelle is observed adjacent to the plasma membrane. Scale bars, 200 nm. (J–M) Large EVs with ribosomes inside are connected to the plasma membrane. Scale bars, 200 nm. (N) Statistical analysis on maximum diameters of small, medium and large EVs in wild‐type Arabidopsis root cells based on 3D electron tomography.

The medium EVs are larger than the small EVs, ranging from ∼200–500 nm in diameter (mean ± SD = 302 ± 91 nm; n = 54) (Figure 1F,G,N and Video S2, Supporting Information) and ribosomes are visible within their lumens. Nearby, there was an organelle with a double‐membrane envelope situated adjacent to the PM (Figure 1H,I and Video S3, Supporting Information). This morphologically resembled the EXPOs previously imaged by 2D TEM in Arabidopsis suspension cells [20, 21]. Moreover, similar to the medium EVs, these double‐membrane structures also appeared to contain ribosomes. Based on these observations, which align with the working model for EXPOs, we suggest that the medium EVs may originate from the fusion of EXPOs with the PM [21, 36].

As well as small and medium EVs, we also observed apoplastic, irregularly shaped, large EVs measuring ∼500–2000 nm (mean ± SD = 1001 ± 478 nm; n = 7) in size which contain ribosomes and tiny vesicles within their structures (Figure 1D,E,N and Video S4, Supporting Information). Interestingly, we also noted a heterogeneous distribution of cargo within some of these large vesicles. For example, in the large EV shown (Figure 1D,E and Figure S1, Supporting Information), the lumen of the portion on the right contained ribosomes and small vesicles, whereas that on the left appeared to be empty. Detailed tomograms revealed a membrane connection between the two portions, which suggests that large EVs might result from the fusion of EVs of different origins (Figure 1D,E and Figure S1, Supporting Information). Regarding the biogenesis of these large EVs, we observed outward budding from the PM with a narrow membrane bridge connecting the two, which are ready to be pinched off (n = 3) (Figure 1J,K). This configuration closely resembles microvesicle formation in mammalian cells [37], which suggests that the large EVs in plants might be initiated from comparable shedding events at the PM (Figure 1L,M) [38]. However, although the root border cells appeared to be morphologically normal, we cannot exclude that they might be undergoing programmed cell death, potentially compromising their representation of typical root cell characteristics. Therefore, we redirected our focus to more stable root cell types—specifically the quiescent center and adjacent initial cells—to investigate the EVs associated with these populations. Our ET analysis of these cells consistently identified three distinct EV types, along with their spatial distributions within the root cells (see Figure S2A,B and Video S5, Supporting Information). Furthermore, we present 23 examples of small, medium, and large EVs derived from individual root cell tomograms (Figure S2C, Supporting Information). These results suggest that EV structures are widely distributed across various stages and populations of root cells.

Consistent with our results in the root cells, we also performed whole‐cell ET analysis in leaf tissues [31, 32] and identified analogous EV structures (yellow) in the guard cells which accumulated predominantly around the stomatal boundary (cyan), as clearly depicted in the 3D model (Figure S3A–D and Video S6, Supporting Information). Both types of EVs (Small and Medium) identified in the guard cells exhibited consistent morphological features with those in the root border cells (Figure S3E–H and Videos S7 and S8, Supporting Information), such morphological consistency reinforces the rationale behind classification of plant EVs in different cell types.

Interestingly, we observed a special type of EVs on further investigation of the hypocotyl cells of Arabidopsis seedlings. Previous research has reported the presence of extracellular vesicular‐tubular structures in suberizing cells [39], [14, 40]. For our study, we collected sample sections from the hypocotyl region of 5‐day‐old Arabidopsis seedlings and conducted fixation with HPF/FS (Figure S3I, Supporting Information). Upon close inspection of 3D tomograms, we identified EVs in the apoplast outside suberizing and suberized cells, exhibiting diverse morphologies, including membrane‐bound long tubular structures (Figure S3J–L and Video S9, Supporting Information). These tubular EVs are clearly separated from the PM. In surrounding, we also observed some small EVs. Both the small and the tubular EVs have no visible ribosomes within their lumens, fusions between the small EVs were identified to form short tubular structures (Figure S3M–P and Video S10, Supporting Information). These results suggest that the tubular EVs might be generated from the fusions of the small EVs.

2.2. Cryo‐ET Provided Near‐Native Evidence of EVs in Plant Cells

Although HPF offers higher‐fidelity structural preservation than chemical fixation, the potential impact of secondary fixatives such as osmium tetroxide, along with the lengthy resin substitution and embedding processes, on the morphology of EV membranes should be considered [41]. To address this concern, we employed cryo‐electron tomography (Cryo‐ET), which is well‐suited for capturing the EV structures in native states [42]. However, applying cryo‐ET to plunge frozen higher plant cells/tissues remains challenging due to the large size, the presence of a cell wall, and the complexity of cell types [30, 43]. Consequently, we selected pollen tubes and pollen grains for our initial trials. Pollen tubes, with diameters not exceeding 15 µm, are relatively small and can be readily fixed by plunge freezing, thereby facilitating subsequent processing with cryo‐focused ion beam (FIB) milling [44, 45]. For samples thicker than 20 µm, such as pollen grains, we employed the newly developed serialized on‐grid lift‐in sectioning for tomography (SOLIST) lift‐out technique to dissect and extract sample chunks from high‐pressure frozen carriers for subsequent FIB milling and cryo‐ET imaging [46].

To compare the structural integrity of EVs using these two techniques, we first conducted room temperature ET applying the conventional HPF/FS fixation and resin embedding on germinated pollen tubes and pollen grains. In Arabidopsis pollen tubes, small EVs (∼30–100 nm in diameter; n = 76) were found accumulated outside the tip area (Figure 2A and Video S11, Supporting Information). We also performed ET analysis on Tobacco pollen tubes (Figure S4 and Video S12, Supporting Information) and showed consistent results that small EVs of identical sizes had accumulated near the tip (n = 21) (Figure S4, Supporting Information). Interestingly, we also found that medium EVs were actively released from the PM of the tricellular stage pollen grains (Figure S5A–D and Video S13, Supporting Information) and consistent with our previous findings in roots and leaves, they were enclosed by a membrane invagination on the PM and contained ribosomes within their lumens (Figure S5A–D, Supporting Information). Furthermore, numerous extracellular tubules were observed within the apoplast of tricellular‐stage pollen grains (Figure 2D and Figure S5A–D and Video S13, Supporting Information). Given that pollen grains are ideal single‐cell systems with well‐documented developmental stages, we were able to trace the formation of these extracellular tubules back to earlier stages of pollen grain development. We found that these extracellular tubules bud from the PM during the early bicellular stage and migrate toward the cell boundary as the cell transitions to the late bicellular stage (Figure 2D and Videos S14 and S15, Supporting Information). Ultimately, these tubules degrade and become integrated into the cell wall, supporting the hypothesis that plant EVs might originated from PM budding events (Figure S5E–J, Supporting Information).

FIGURE 2.

Investigation of extracellular vesicles in developing Arabidopsis pollen tubes and pollen grains using (cryo‐)electron tomography. (A) 2D TEM micrograph and tomographic slice of the tip region of a wild‐type Arabidopsis pollen tube, showing small EVs localized at the tube tip, along with the corresponding 3D model. Scale bars, 500 nm. (B) Workflow for sample fixation, FIB milling, Cryo‐ET, and 3D reconstruction. (C) Workflow of sample preparation for Cryo‐ET: (a) germinated pollen tubes were transferred onto carbon‐coated grids and fixed by plunge‐freezing; Scale bars, 50 µm. (b) the shank region was thinned by Cryo‐FIB (focused ion beam) milling to generate lamellae for subsequent Cryo‐ET; Scale bars, 5 µm. (c) tomographic slice of a reconstructed pollen tube, with boxes indicating membrane‐bounded vesicles in the extracellular space; enlarged views (d, f) showing detailed EV features, with corresponding 3D models (e, j). Scale bars, 100 nm. E, extracellular space; I, intracellular space. (D) A 2D TEM micrograph and tomographic slice of an Arabidopsis pollen grain at the early bicellular stage, showing extracellular tubules within the intine region, together with the corresponding 3D model. (E) Workflow of the SOLIST lift‐out technique (F) Anther samples from wild‐type Arabidopsis were high‐pressure frozen, after which sample chunks were lifted from the carrier (a,b) and mounted onto new grids for FIB milling, producing lamellae (c) for subsequent Cryo‐ET imaging. Scale bars, 5 µm. (d,e) Zoom‐in tomographic slices showing EVs in pollen grains. Scale bars, 500 nm.

As cryo‐ET offers a high spatial resolution and the ability to capture EVs preserved in a near‐native state [47], we next employed cryo‐FIB milling to prepare lamellae from germinating Arabidopsis pollen tubes after plunge freezing (Figure 2B,C) [30]. Cryo‐ET analysis indicated that within the apoplast area of the pollen tube, oval‐shaped EV structures exhibited a high level of similarity (with regards to size and the lack of luminal contents) to the small EVs observed using room temperature ET analysis (Figure 2C d–j and Video S16, Supporting Information). For pollen grains, the lifted‐out sample chunks were processed using FIB to acquire lamellae for cryo‐ET imaging, revealing consistent extracellular tubular structures within the apoplast (Figure 2E,F), confirming that room temperature ET effectively preserves the EV and extracellular tubule structures [48].

In summary, by conducting ET analysis of cells from five different tissues across two species, we have identified and classified a total of three major types of EVs based on their morphological characteristics. (Table S1, Supporting Information)

2.3. EVs of Different Sizes are Labeled by Distinct Putative EV Markers

ET analysis revealed diverse populations of plant EVs with distinct morphological features. However, classification based solely on morphology is challenging, and their biogenesis and secretion mechanisms remain poorly understood [49]. Thus, it is essential to identify molecular markers that can be used to distinguish between the individual EV subtypes, three putative EV marker proteins have been described: TETRASPANIN 8 (TET8), PENETRATION 1 (PEN1, also known as SYP121), and Exo70E2 [19, 20, 21, 36, 50]. To elucidate their roles in plant EV identity and trafficking, we constructed transgenic Arabidopsis plants expressing TET8‐GFP, GFP‐PEN1, and Exo70E2‐GFP. Using laser scanning confocal microscope, we observed the distribution of their signals in the root meristem, leaf cells, pollen grains, and pollen tubes of these transgenic plants. TET8‐GFP and GFP‐PEN1 exhibited typical PM patterns with few punctate signals inside the cell, whereas Exo70E2‐GFP was localized as discrete puncta both within the PM and in the cytosol (Figure S6A, Supporting Information).

To directly observe whether the TET8, PEN1, and Exo70E2 positive puncta might be secreted out of plant cells, we treated transgenic plant roots with 10% [w:v] sodium chloride solution to induce plasmolysis. After treatment, the plant cells exhibited cell wall separation that facilitates visualization of the enlarged apoplastic regions. We observed punctate GFP signals in the apoplast, which indicated that TET8, PEN1, and Exo70E2 positive organelles can be secreted into the extracellular spaces (Figure S6B, Supporting Information).

Next, we wanted to explore the differences in membrane origin and trafficking of the organelles marked by TET8, PEN1, and Exo70E2. In Arabidopsis, the Rab5 family protein ARA7 serves as standard marker for MVBs and the plant‐specific ARA6 shows partial co‐localization with ARA7 and represents a special MVB population [51]. Furthermore, SYP61 was used as the marker indicating trans‐Golgi network [52]. Upon transient expression of these proteins in Arabidopsis protoplasts, we observed complete colocalization of ARA6‐GFP with TET8‐mRFP, whereas ARA6 showed partial colocalization with ARA7 but did not overlap with SYP61 (Figure S7A, Supporting Information). These findings align with previous research, confirming the localization of TET8 in MVBs marked by ARA6 [19]. We also found that there was complete spatial separation between the YFP‐PEN1 and RFP‐ARA7 signals, but there was overlap between YFP‐PEN1 and mRFP‐SYP61, indicating that cytosolic PEN1 signals are localized to the PM and the trans‐Golgi network but not to MVBs (Figure S7B, Supporting Information). Complete spatial separation was also found between Exo70E2‐GFP and either mRFP‐ARA7 or mRFP‐SYP61 (Figure S7C, Supporting Information). This finding supports previous reports identifying Exo70E2 as being exclusively localized to the EXPO [21, 36, 53].

To further investigate whether EVs marked by TET8, PEN1 or Exo70E2 are derived from different membrane origins, we transiently co‐expressed Exo70E2‐GFP, ARA6‐mRFP, and CFP‐PEN1, and showed that they are localized distinctly from each other in the cytosol (Figure S7D, Supporting Information).

Next, we performed immunogold‐TEM analysis to examine whether Exo70E2, TET8, and PEN1 label distinct populations of EVs. Interestingly, our data showed that gold particles conjugated with antibodies targeting Exo70E2‐GFP preferentially labeled EVs between 200 to 500 nm, which corresponds to medium EVs in our classification system. Indeed, we found that 76% of medium EVs were Exo70E2‐positive (i.e., defined as having more than one gold particle on the membrane, n = 30) (Figure 3A,B,E). In contrast, TET8 predominantly labeled small EVs (50–200 nm), with 87% of these testing positive (n = 33) (Figure 3C,F). Moreover, PEN1 was detected on large EVs in root cells and extracellular tubules in pollen grains, as revealed by immunogold‐TEM labeling with anti‐GFP antibodies in GFP‐PEN1 transgenic plants (Figure 3D,E; Figure S8, Supporting Information). Interestingly, some PEN1‐positive large EVs were observed to remain connected to the PM, suggesting that they may originate from PM budding. Taken together, these findings indicate that at least three distinct EV populations and pathways exist in Arabidopsis.

FIGURE 3.

Immunogold‐TEM showing distinct EV subpopulations individually labeled by Exo70E2, TET8, or PEN1, respectively. (A,B) Immunogold‐TEM of guard and pavement cells from Exo70E2‐GFP seedling leaves, showing gold‐labeled Exo70E2‐positive EXPO structures. Scale Bars, 2µm. (C) Immunogold‐TEM of root cells from TET8‐GFP plants showing TET8‐positive EVs, predominantly small EVs. Scale bars, 500 nm. (D) Immunogold‐TEM analysis of root cells from GFP‐PEN1 plants showing PEN1‐positive EVs, predominantly large EVs. Scale bars, 200 nm. (E–G) Quantitative analysis of Exo70E2‐positive/negative medium EVs, TET8‐positive/negative small EVs, and PEN1‐positive/negative large EVs.

Next, we focused on the plant‐specific EXPO‐derived EV organelle, using Exo70E2 as a marker [20], to further elucidate its nature and identity in vivo. Exo70E2‐GFP is localized as puncta on the PM and inside the cytosol when expressed in plant cells. To gain deeper insights into these punctate structures at the ultrastructural level, we performed immunogold‐TEM analysis on ultrathin sections prepared from transgenic Exo70E2‐GFP plants, which were fixed by HPF/FS (Figure 3A,B). In leaf guard cells and pavement cells, Exo70E2‐positive compartments, indicated by gold particle labeling, exhibited a double‐membrane structure in the cytosol (Figure 3A‐2,B‐4,5), appearing to fuse with the PM with its outer membrane (Figure 3A‐1,B‐3) and expelled single‐membrane vesicles out of the cell (Figure 3A‐3,4,B‐1,2), consistent with the model previously described in Arabidopsis suspension cells [20, 21]. Taken together, our data suggests that the medium EVs observed by ET represent the Exo70E2‐positive EXPO‐derived organelles in plants.

However, the 2D TEM projection on individual sections is limited in determining whether these extracellular EXPO‐derived EVs were completely separated from the PM. Thereby, we conducted correlated electron and light microscopy (CLEM) studies on roots treated with 10% [w:v] sodium chloride to induce plasmolysis [54]. Fluorescent micrographs showed Exo70E2‐GFP signals as discrete puncta in the apoplast and on the PM (Figure S9A, Supporting Information). At the ultra‐structural level, these GFP signals correlated with two populations: apoplastic medium EVs with cytosolic content (Figure S9D, Supporting Information) and electron‐dense patches on the PM (Figure S9E,F, Supporting Information). Comparable structures labeled by antibodies targeting Exo70E2‐GFP were also observed by immunogold‐TEM in root cells (Figure S9G–I, Supporting Information).

In summary, our findings illustrate the EXPO pathway in vivo, aligning with the previous model reported in plant suspension cells [21, 55]. The EXPO, which appears as a double‐membrane organelle in the cytosol, is destined to fuse with the PM and release a medium‐sized EV outside the cell.

2.4. EXPO‐Derived EVs are Potentially Involved in the Plant Response to Stress

Recognizing EXPO‐derived EVs as a distinct subtype prompted further investigation into their unique functions. To this end, we aimed to isolate and characterize their cargoes; However, as these vesicles are embedded within the cell wall, structurally fragile, and easily disrupted during cell wall digestion and lysis, making them unsuitable for consistent, high‐quality proteomic analysis. We therefore applied our immune‐isolation strategy on intact intracellular EXPO organelles from transgenic Arabidopsis suspension cells expressing the EXPO marker Exo70E2‐GFP to characterize their cargoes [20, 21, 55]. Protoplasts were obtained by digesting the cell wall, and then we lysed the cell membrane followed by discontinuous subcellular fractionation (Figure 4A). The Exo70E2 signal, appearing as punctate, accumulated mainly in the 41%–46% sucrose fraction (Figure 4B) as confirmed by immuno‐blot detection of the Exo70E2‐GFP protein (Figure 4C). These vesicles were then confirmed to be EXPO organelles by a combination of negative staining and immunogold‐TEM, and they were shown to be labeled by both anti‐Exo70E2 and anti‐GFP antibodies (Figure 4D). Furthermore, 2D TEM imaging of the resin embedded EXPO‐enriched fraction revealed an accumulation of membrane‐bound structures that resembled EXPOs (Figure 4D).

FIGURE 4.

Immuno‐isolation of EXPOs for cargo identification. (A) Workflow for EXPO purification from Exo70E2‐GFP transgenic Arabidopsis suspension cells. (B) Protoplasts were isolated from transgenic Arabidopsis suspension cells expressing the EXPO marker Exo70E2‐GFP, which displayed the typical EXPO localization pattern. Enrichment of EXPOs was observed in sucrose fractions 41%–46%. Scale Bars, 10µm. (C) Immunoblot analysis of transgenic cells expressing Exo70E2‐GFP using Exo70E2 and GFP antibodies shows enrichment of Exo70E2‐GFP in the 41%–46% sucrose fractions, which is absent in wild‐type cells. In contrast, the Golgi apparatus, detected with ManI antibodies, is enriched in the <36% sucrose fractions. (D) Immunogold‐TEM analysis of the EXPO‐enriched fraction with anti‐Exo70E2 and anti‐GFP antibodies. Resin embedding followed by ultrathin sectioning produced 2D TEM micrographs, revealing EXPO‐like structures. Scale bars: 100 nm. (E) EXPO vesicles are successfully isolated using Exo70E2 or GFP antibodies in conjunction with protein A agarose beads, or GFP antibody‐conjugated Sepharose beads. Control groups show no GFP signals. Scale bars: 10 µm. (F) Western blot analysis indicates that Exo70E2‐GFP is detectable in elutions from Exo70E2 antibody (E2 Ab) Protein A and GFP antibody (GFP Ab) Protein A groups but absent in the control (empty protein A beads). (G) Silver staining of immuno‐isolated EXPOs using different beads as indicated. Blank protein A agarose or Sepharose beads are served as negative controls. (H) Multiple rounds of LC/MS/MS analysis are conducted on immuno‐isolated vesicles. The table presents a subset of the full list of putative EXPO‐related component proteins, verified through co‐localization analysis via co‐expression with Exo70E2.

Next, we used protein A‐agarose beads conjugated with anti‐GFP or anti‐Exo70E2 antibodies as well as anti‐GFP Sepharose beads for the immuno‐isolation of EXPOs from the enriched sucrose fraction. Dispersed Exo70E2‐GFP fluorescent puncta, indicative of EXPOs, were found adhering to anti‐GFP Protein A‐agarose, anti‐Exo70E2 Protein A‐agarose, and anti‐GFP Sepharose beads, but were absent from Protein A‐agarose and Sepharose beads (Figure 4E). The presence of Exo70E2‐GFP was further confirmed by western blot analysis (Figure 4F). Proteins were extracted from the purified EXPO organelles for proteomic analysis using LC‐MS/MS (Figure 4G). As expected, in addition to Exo70E2, numerous ribosomal proteins from both the 40S and 60S subunits were identified (Table S2, Supporting Information), consistent with TEM observations showing ribosomes within EXPOs (Figure 4H). Furthermore, 217 EXPO‐related proteins were identified by LC‐MS/MS, 10 cargo candidates out of which were further verified by subcellular colocalization studies with Exo70E2 (Figure 4H and Figure S10, Supporting Information). For example, we identified prohibitin 2 (PHB2), involved in salt stress [56]; SLY1, involved in gibberellin acid signaling [57]; and a plant basic secretion protein family protein (BSP) AT2G15130, previously reported to be potentially involved in pathogen defense response [58, 59, 60, 61] (Figure S11, Supporting Information), as EXPO‐related protein cargos. Interestingly, the AT2G15130 has two splice variants AT2G15130.1 and AT2G15130.2, and of the two, the former has a signal peptide at C terminal, as predicted by SignalP4.0 whereas the latter does not. We found that colocalization only occurred between AT2G15130.2 and Exo70E2 (Figure S11, Supporting Information), which suggests that BSP proteins encoded by different mRNA splice variants might undergo distinct secretion pathways. These preliminary findings suggest that as a distinct subtype of plant EVs, EXPO‐derived EVs might mediate the unconventional secretion of their protein cargoes.

It was suggested that Exo70E2 is crucial for EXPO formation and trafficking in plant cells [20, 21, 36, 53]. To address the possible roles of the Exo70E2/EXPO pathways in cellular processes, we next focused on the Exo70E2‐related mutants exo70e2‐2 and exo70e2‐3. The exo70e2‐2 mutant is a knock‐out mutant with a T‐DNA insertion in the exon region of the Exo70E2 gene, while exo70e2‐3 is also a T‐DNA mutant but with an insertion in the promoter region (Figure 5A). The exo70e2‐3 mutant displayed enhanced transcription levels of the Exo70E2 gene which were verified with RT‐PCR, and homozygous lines of these mutants were identified by genotyping (Figure 5B). Under normal growth conditions, exo70e2‐2 and exo70e2‐3 mutants, along with the Exo70E2‐GFP overexpression line, exhibited mildly retarded growth phenotypes (Figure 5C).

FIGURE 5.

EXPO‐derived EVs play a key role in plant resistance to pathogen infection. (A) Schematic representation of T‐DNA insertions in exo70e2 mutant lines. (B) Genotyping analysis and RT‐PCR analysis confirming exo70e2‐2 as a knockout mutant line, while exo70e2‐3 exhibiting increased Exo70E2 transcription levels compared to the wild‐type (WT). Primers used are indicated in (A). (C) Phenotypes of WT, exo70e2‐2, exo70e2‐3 and Exo70E2 at 3 (upper panel) and 6 (lower panel) weeks under normal growth conditions. (D) RT‐qPCR analysis on gene transcriptions of ATG8a, YLS9, CPB60G and PR1 in WT, exo70e2‐3, UBQ10‐driven Exo70E2‐GFP and exo70e2‐3/e2‐cas9 lines. The results demonstrate enhanced expression of these genes in the Exo70E2 overexpression background. Differing letters represent significant differences (p value ≤ 0.05; statistical analysis was conducted using two‐way ANOVA analysis). (E) 4‐week‐old wild‐type, exo70e2‐2 and exo70e2‐3 plants were spray‐inoculated with Pst DC3000, respectively. The lg‐transformed CFU/cm² values of the pathogen in the leaves were counted at 0, 24, and 48 h after inoculation. (*p value ≤ 0.05; **p value ≤ 0.01; statistical analysis was conducted using student's t‐test) (F) TEM imaging and quantitative analysis of EXPO‐derived EVs in root border cells from exo70e2‐2 and exo70e2‐3 mutants. Scale bars: 500 nm. (*p value ≤ 0.05; **p value ≤ 0.01; statistical analysis was conducted using student's t‐test).

Next, we performed RNA sequencing (RNA‐seq) on leaves of wild‐type (Col‐0), exo70e2‐2, and exo70e2‐3 plants infected with the virulent bacterial strain Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) at 0h and 5h post‐infection to analyze the effect of the mutations on gene expression in response to the infection (Figure S12A, Supporting Information). The genes that were differentially expressed in the Exo70E2 overexpression line (H5_e23) and WT (H5) (padj≤ 0.05) were annotated into 20 functional groups (Figure S12B, Supporting Information). We found that these differentially expressed genes were mostly enriched in three functional groups: plant‐pathogen interaction, protein processing in the ER, and plant hormone signal transduction.

With a specific interest in plant pathogen defense, we selected four genes from the group “plant‐pathogen interaction,” encoding the Pathogen‐related protein 1 (PR1), NDR1/HIN1‐like protein 10 (NHL10), Calmodulin‐binding protein 60‐like G (CBP60g), and the autophagy‐related protein 8a (ATG8a), for further analysis of their expression patterns using RT‐qPCR. Our results indicated that in the Exo70E2 overexpression lines (exo70e2‐3) and the Exo70E2‐GFP transgenic plant, all four genes displayed enhanced transcription levels compared to the wild type. However, when Exo70E2 was knocked out in exo70e2‐3 with CRISPR/Cas9, the transcription levels of these genes returned to normal (Figure 5D). These findings hint of the potential role of Exo70E2 in regulating pathogen response pathways.

We next performed pathogen inoculation assays to compare the disease resistance phenotypes of Exo70E2‐related mutant plants against the bacterial pathogen Pst DC3000 [62]. For this analysis, WT, exo70e2‐2, and exo70e2‐3 plants were spray‐inoculated with a Pst DC3000 suspension. The bacterial load, measured as colony‐forming units (CFUs), was quantified on leaves at 0, 24, and 48 h post‐inoculation. The exo70e2‐2 knockout mutant exhibited reduced resistance to Pst DC3000 compared with the WT, whereas the exo70e2‐3 overexpression line showed enhanced resistance (Figure 5E). We next wondered how Exo70E2 overexpression could enhance plant resistance to pathogen infections. Therefore, we applied 2D TEM imaging to the exo70e2‐3 overexpression mutant lines. Interestingly, we observed an accumulation of EXPO‐derived EVs outside the root border cells where EXPOs are seldom observed in exo70e2‐2 knock‐out mutant lines (Figure 5F). This finding is consistent with previous studies suggesting that Exo70E2 expression induces EXPO formation [20] and subsequent release of EXPO‐derived EVs, thereby potentially mediating the secretion of defense‐related materials into the apoplast during pathogen infection and enhancing plant resistance.

In conclusion, these results indicate that EXPO‐derived EVs contribute positively to plant immunity, likely by mediating the delivery of defense‐related proteins during pathogen infection

3. Discussions

The study of EVs in the field of plant biology research is still in its infancy. To date, researchers have identified at least three distinct subclasses of EVs in plants: TET8‐positive EVs, PEN1‐positive EVs, and Exo70E2‐positive EVs [10, 49], demonstrating the EV heterogeneity in the plant cells. However, there is still a lack of rigorous and systematic characterization of the nature and identity of these EVs in vivo under normal physiological conditions [3, 4, 5, 22]. In this study, we selected Arabidopsis thaliana and Nicotiana tabacum as model systems because of their well‐characterized genetics, the availability of EV marker lines, and their suitability for advanced imaging. These species also align with prior EV studies, allowing for meaningful comparisons. Using 3D ET, cryo‐ET, and immunogold‐TEM with antibodies against established EV markers (TET8, PEN1, and Exo70E2), we characterized plant EVs, revealing their structural diversity in near‐native states and their associations with specific molecular markers. Based on morphological features, cellular contents, and marker profiles, we preliminarily classified these EVs into three distinct populations and further identified the plant‐specific EXPO‐derived EVs, elucidating their trafficking in vivo, analyzing their cargoes, and investigating their functions in response to stress.

Beyond these established sources of EV formation, alternative origins cannot be excluded. For example, the small vesicle structures observed outside root border cells might represent vacuole‐derived EVs. In plants, root border cells undergo programmed cell death (PCD) as part of their normal root cap development. This process is critical for root cap turnover, root growth, and defense against pathogens. Previous studies have demonstrated that during vacuole‐mediated PCD, vacuoles can fuse with the plasma membrane to release hydrolytic contents, and in some cases, membrane‐bounded vesicles may also be discharged into the extracellular space, contributing to antimicrobial defense [63, 64, 65]. This represents an important aspect of the plant cell‐autonomous immune system. Nevertheless, the precise biogenesis and release mechanisms of vacuole‐derived EVs remain poorly understood. Future research using advanced live‐cell imaging, immunogold‐TEM, CLEM, and vacuole‐specific molecular markers will be necessary to clarify their origin, cargo composition, and functional significance in plant development and immunity.

3.1. How Plant EVs Cross the Cell Wall

There has been a long‐standing debate regarding the existence and functionality of plant EVs, particularly in the context of the plant cell wall. However, accumulating evidence now indicates that multiple types of EVs exist across different plant cell types. For example, TEM studies have documented the release of exosome‐like EVs at fungal infection sites [19]. Moreover, these EVs have been implicated in plant–microbe interactions, particularly during arbuscular mycorrhizal symbiosis, where they are secreted into the peri‐arbuscular space to engage with fungal hyphae [15]. These findings suggest that EVs are capable of traversing both plant and fungal cell walls to carry out their functions. Our data also suggest that, in tobacco pollen tubes, small EVs (30–100 nm) can traverse the cell wall and localize within its outer regions, as illustrated in the model (Figure S5D, Supporting Information). Several factors possibly account for the ability of EVs to traverse the cell wall: (1) During plant–microbe interactions (e.g. symbiosis), plants may secrete cell wall–modifying enzymes that soften and remodel the wall, thereby facilitating the passage of EVs; and (2) The plant cell wall is not a static barrier but rather a porous and dynamic structure, allowing the passage of EVs. An important direction for future research is to elucidate the mechanisms by which plant EVs traverse the cell wall during active secretion.

3.2. Tissue‐Specific Localization and Functions of Plant EVs

The abundance of EVs varies across plant tissues, thereby contributing to their tissue‐specific functional roles, such as cell wall thickening, development, and plant–microbe interaction. Based on our TEM analysis, root border cells exhibit a high abundance of EVs spanning multiple subtypes (Figure 1). We propose that these EVs may contribute to cell‐wall assembly and remodeling, promoting the development of the characteristically thick cell walls of root border cells and thereby conferring mechanical resilience in the soil environment. We also identified EVs in the apoplast outside suberizing and suberized cells, exhibiting diverse morphologies, including membrane‐bounded long tubular structures (Figure S3I–P, Supporting Information), consistent with a previous study suggesting their role in suberin deposition in hypocotyl cells [14]. In bicellular pollen grains, EVs within the cell wall rupture and degrade (Figure 2D, Figure S5E–J, Supporting Information), with their remnants seemingly incorporated into the intine, potentially contributing to pollen development [66, 67]. In leaf guard cells and pavement cells, EXPO‐derived medium EVs were identified (Figure 3A,B). Proteomic analysis of their cargo revealed numerous stress‐related proteins (Figure 4H), suggesting that EXPO‐derived medium EVs play a role in plant–microbe interactions [68, 69].

Another interesting question is lipid composition of individual EV subtypes. In the current work we classify EV subtypes by markers—TET8‐positive small EVs (originating from ILVs of MVBs), Exo70E2‐positive medium EVs (EXPO; likely ER‐derived), and PEN1‐positive large EVs (PM‐shed). Accordingly, we expect their basal lipidomes to reflect distinct biogenesis routes: ILVs of MVB‐derived EVs enriched in phosphoinositides (such as PI3P) and phosphatidic acid/diacylglycerol [70, 71]; EXPO‐derived EVs enriched in ER‐type phospholipids (phosphatidylcholine, phosphatidylethanolamine, and phosphatidylinositol) with curvature‐promoting phosphatidic acid and diacylglycerol; and PM‐shed EVs enriched in sterols (sitosterol and stigmasterol) and sphingolipids (glycosylinositol phosphorylceramides and glycosylceramide) with PI(4,5)P₂ microdomains [72, 73, 74]. However, technical challenges and the limited yields of EVs from individual cell types prevented us from generating a robust, reproducible dataset for lipidomic profiling. Future investigations using targeted lipidomic approaches will be essential to resolve the lipid signatures of these EV subtypes. Such studies will provide an important complement to our current proteomic analysis and help elucidate how lipids contribute to EV biogenesis, cargo sorting, and their roles in plant–microbe interactions.

3.3. Plant Small EVs Versus Mammalian Exosomes

Exosomes, as a major type of EV in mammalian cells, originate from intraluminal vesicles (ILVs) of the MVBs and are released into extracellular spaces during the fusions of MVBs with the PM [75]. In plants, previous study demonstrated the enrichment of plant TET8 on the exosome‐like EVs which predominantly accumulated and were released at the plant‐fungi interaction sites on the PM. Co‐localization of MVB marker ARA6 and TET8 has been observed within these PM interaction sites [19, 76], suggests that TET8‐positive EVs may be derived from MVB‐PM fusion. Several studies in plants have demonstrated that some vesicle‐containing bodies, resembling MVBs, do fuse with the PM and release vesicles into the apoplast, such as paramural bodies (PMBs) and EV‐containing bodies (EVBs) [14, 40, 77, 78, 79]. Recent research also demonstrated upon invasion of Arbuscular Mycorrhizal Fungi on the rice root cells, exosome‐like EVs are released from the plant MVBs adjacent to the fungal hypha [15, 80, 81]. These results provided evidence supporting that MVB‐PM fusions contribute to exosome‐like EV formation in plants. Nevertheless, whether TET8‐positive EVs in Arabidopsis are functionally equivalent to mammalian exosomes remains to be further investigated. It should also be noted that although TET8 serves as a useful EV marker in Arabidopsis, its universality across plant species is limited, and validation in each species is necessary before assuming functional equivalence.

In our data, the small EVs are predominantly positive for TET8 labeling. However, under normal physiological conditions, as examined in our work, we seldom detected MVB–PM fusion events in TEM sections. We believe this discrepancy may be explained by two major factors: (1) Subcellular distribution of MVBs. Under normal growth conditions, MVBs are generally not enriched at the cortical regions of plant cells but are instead located deeper within the cytoplasm. In our study, the TEM imaging strategy was specifically optimized to capture EV release events occurring at or near the plasma membrane. Consequently, our sections focused on cortical regions several micrometers away from the main MVB population, thereby reducing the probability of capturing intact MVB structures within the imaged field; and (2) Transient nature of MVB–PM fusion. MVB–PM fusion represents a highly dynamic and short‐lived intermediate step in EV release. Even when MVBs are present near the cell periphery, the actual fusion process is rapid and therefore difficult to capture by static TEM imaging. This inherent limitation makes visualization of fusion intermediates a rare event in conventional ultrastructural analyses. Taken together, these technical and biological constraints likely account for the limited visualization of MVB–PM fusion in our dataset. To overcome these limitations, complementary approaches such as live‐cell imaging with fluorescent markers and correlative light and electron microscopy (CLEM) would be required to more reliably capture dynamic membrane fusion events and delineate EV biogenesis pathways at higher temporal and spatial resolution.

3.4. Plant Large EVs Versus Mammalian Ectosomes

Ectosomes, also known as microvesicles (MVs), are another major route of EV secretion. During MV formation, the PM undergoes outward budding, resulting in the release of vesicles into apoplast [37, 38, 82]. In mammalian cells, the machinery of MV formation and releasing are well studied, including the reorganization of the cytoskeleton, ESCRT complex which help in the direct budding process and scission of vesicles from the PM and also the tetraspanin proteins defining MVs size, composition, and the specific cargo sorting [38, 82]. However, a similar pathway has not yet been characterized in plant cells.

In this study, we identified large EVs, approximately 500–2000 nm in diameter, in intact plant cells under normal condition, characterized by the presence of ribosomes and numerous small intraluminal vesicles (Figure 1D,E). ET data revealed that some of these large EVs budded from specific regions of the PM (Figure 1J‐N), resembling microvesicle formation in mammalian cells. Furthermore, immunogold‐TEM labeling experiments demonstrated that PEN1 preferentially associates with these large EVs, with some remaining connected to the PM (Figure 3D and Figure S8, Supporting Information). Together, these results support the hypothesis that large EVs may originate from PM shedding events. In contrast, previous studies reported that PEN1‐positive EVs isolated from the leaves of 5–7‐week‐old plants ranged from approximately 50 to 300 nm in diameter [23]. The discrepancy likely reflects differences in both the biological materials used (roots and pollen grains vs. leaves) and the methodologies applied (imaging of cryo‐fixed intact cells in vivo vs. EV isolation by differential ultracentrifugation). Our data in intact cells in this study suggest that the natural shape of large EVs under ET is predominantly long and tubular, exhibiting a greater maximum length compared to spherical structures in isolation buffer of equivalent volume. Additionally, we also note that some EVs with sizes larger than 450 nm may encounter challenges during filtering processes with 0.45µm filters, which could potentially lead to blockage or fragmentation. Consequently, these larger EVs might be less detectable in previous studies that employed filtering steps. Nevertheless, we cannot exclude the possibility that PEN1 also labels smaller EVs (50–300 nm in diameter) and that the existence of distinct populations (based on size) of PEN1‐positive EVs in other tissue types. In future work, additional PEN1‐associated markers, including specific cargo proteins, will be employed to more comprehensively characterize PEN1‐positive EVs in plants. Similarly, although PEN1‐positive EVs were rarely detected in guard cells in the whole‐cell observations (Figure 3A–D), we cannot rule out their presence in other leaf cell types, such as mesophyll cells, which could be examined in future studies.

Another intriguing aspect is that the formation of large EVs appears to be more complex. In our study, we observed a fusion event between a large EV and an electron‐transparent EV, characterized by a membrane tubule bridging the two vesicles. Given the differences in their content, we propose that large EVs may undergo fusion events with EVs from distinct origins in the apoplast. As additional evidence of EV fusion, the unique tubular shaped EVs observed in suberizing/suberized cells exhibit a high divergence in morphology compared to other types of EVs. Inspired by the hetero‐fusion observed in large EVs, it is suggested that these tubular EVs could result from the homo‐fusion of smaller EVs, which show merging with each other in surroundings and similarly lack cytosolic content.

Furthermore, the PM shedding observed in pollen grains is an intriguing example where the budded tubules fuse with one another to form a tubular network. As the pollen grains mature, these extracellular tubules are relocated to the boundary of the cell wall (Nexine) for degradation, ultimately becoming part of its composition.

These findings highlight the complexity of plant EV formation and indicate a crosstalk between different populations of EVs in the extracellular spaces, where they occasionally encounter both homo‐ and hetero‐fusions with each other. This could help revise our classification and rigorous nomenclature of EVs in plant cells.

3.5. EXPO‐Derived EVs, a Plant‐Unique Organelle

Aligning with previous models, EXPOs are identified as cytosol‐containing EVs with an average size of around 200–500 nm. EXPOs appear as double‐membrane vesicles inside the cytosol that are destined to fuse with the PM and release a single‐membrane medium EV outside the plant cells [20, 21, 36].

EXPO is defined by Exo70E2, previous studies suggest the Exo70E2 function as an agent to recruit other exocyst subunits and facilitate the fusion between EXPO and the PM [20]. In our CLEM results, besides membranous vesicles, we have also identified the Exo70E2‐positive protein patches on the PM and adjacent to the ER/Golgi, the nature and identity of which are still elusive. We are wondering whether Exo70E2 also exhibits such electron opaque patches under endogenous expression conditions and, if so, it would be interesting to further investigate the involvement of these Exo70E2‐positive protein patches in EXPO secretion and formation.

We also successfully isolated EXPO organelles with the identification of novel protein cargoes. Our study revealed that EXPOs transport defense related proteins for modulating cellular homeostasis and reinforcing plant tolerance to unfavored environments. However, it is believed divergent types of EVs are also involved in these processes in plants, what is the difference in cargo content between different type of EVs and how these cargoes delivered by EVs function in apoplast under certain circumstances are pending further investigations.

4. Experimental Section

4.1. Cloning and Plasmid Construction

Green fluorescent proteins (GFP) and monomeric red fluorescent proteins (mRFP) constructs are utilized for transient expression in Arabidopsis protoplasts. The target genes are amplified from the cDNA library and cloned into the pBI221 backbones driven by the Ubiquitin‐10 promoter (UBQ10) by Gateway cloning (Invitrogen). For protein candidates identified in the proteomic analysis, fluorescent protein tags were fused to either the C‐terminus or the N‐terminus of the target proteins. Both fusion orientations were tested to confirm that the fluorescent tag did not affect the native subcellular localization of the target protein. For well‐characterized marker proteins, the fusion orientation followed previous reports, with “‐XFP” denoting a C‐terminal fusion and “XFP‐” denoting an N‐terminal fusion [55].

To generate transgenic plants stably expressing Exo70E2‐GFP, TET8‐GFP, and GFP‐PEN1, the full cDNA sequences of Exo70E2, TET8, and PEN1 are amplified and cloned into either the pCAMBIA1300‐UBQ10pro::gene or pBI121‐UBQ10pro::gene vectors. Exo70E2 and TET8 are tagged with GFP at their C‐terminus, while PEN1 is tagged with GFP at its N‐terminus. Arabidopsis thaliana plants (accession Col‐0) were transformed with these constructs via Agrobacterium tumefaciens strain GV3101 by floral dipping to generate individual lines.

For CRISPR/Cas9 construct generation, psgR‐Cas9‐At backbone was used to generate sgRNA‐containing vector, The forward and reverse DNA oligomers for sgRNA targeting Exo70E2 are E2‐cas9‐F and E2‐cas9‐R. The whole cassette of sgRNA‐Cas9 was amplified by PCR from the sgRNA containing construct and subcloned to pCAMBIA1300 to generate the pCAMBIA‐sgRNA‐Cas9 construct. All primer details are listed in the Table S3 (Supporting Information).

4.2. Plant Material and Growth Conditions

Tobacco (Nicotiana tabacum) and Arabidopsis (Arabidopsis thaliana) plants in soil are grown in green house under normal condition (16 h light/8 h dark, 22°C). For seedlings of Arabidopsis germinated on Murashige and Skoog (MS) agar basement, MS plates were prepared from MS salts (Sigma‐aldrich), 0.8% [w:v] phyto agar (Plant media), and 1% [w:v] sucrose, adjust pH to 5.7 with KOH. The seeds were spread on plates and kept in 4°C for 16 h before place in growth chamber for germination (16 h light/8 h dark, 22°C). For Arabidopsis and Tobacco pollen tube in vitro germination, healthy flowers were collected from adult plants and rinse with germination buffer by vortex. (Arabidopsis: 0.01% [w:v] H₃BO₃, 1mM CaCl₂, 1mM Ca(NO₃)₂, 1mM MgSO₄ with 18% Sucrose and adjust PH to 7; Tobacco:0.01% [w:v] H₃BO₃, 1mM CaCl₂, 1mM Ca(NO₃)₂, 1mM MgSO₄ with 10% [w:v] Sucrose and adjust PH to 6.5) The pollen were then germinated in germination buffer at either 27.5°C for 1–3 h (Tobacco) or 22.5°C 8–10 h (Arabidopsis) [30, 83]. For generating the transgenic plant, the constructs were transformed into GV3101 Agrobacterium tumefaciens strain. The suspension culture of transformed Agrobacteria was applied to the Columbia‐0 (Col‐0) wildtype Arabidopsis plant by floral dipping. The T‐DNA insertion mutant lines of Exo70E2 GK072F12 (exo70e2‐2) and CS827449 (exo70e2‐3), PEN1 CS72343 (pen1), and Tetraspanin 8 SALK_136039C (tet8) were obtained from the Arabidopsis Biological Resource Center (https://abrc.osu.edu). All these mutant lines are in the Col‐0 background. Genotyping PCR analysis was performed to verify the homozygous mutants.

4.3. 3D Electron Tomography

For sample fixation and resin embedding, Roots, leaves of 5‐day‐old seedlings, and pollen grains from soil‐grown Arabidopsis or Tobacco plants were collected. Pollen grains were germinated as mentioned to produce pollen tubes. Samples were fixed with high‐pressure freezing, then immersed in acetone with 2% osmium tetroxide and 0.1% uranyl acetate at −85°C for 48 h and warmed to 0°C at 1°C per hour. After washing thrice to remove osmium, samples underwent stepwise substitution with Epon 812 resin (Sigma‐Aldrich) at concentrations of 5%, 10%, 25%, 50%, 75%, and 100%, and were embedded and heated to 65°C for resin polymerization. Following the previously described general procedure [31], serial sections of the samples, either 250 nm or 300 nm thick, were prepared using an ultramicrotome. To aid in image alignment during the tilt series, 15‐nm gold particles were applied to the grids to adhere to the section surfaces, serving as fiducial markers. These thick sections were then imaged with a Tecnai F20 electron microscope (FEI Company) operated at 200 kV. For each serial section, a single tilt series consisting of 81 images was captured, spanning angles from +60° to −60° with 1.5° increments. A second tilt series was subsequently obtained by rotating the grid at 90°. Dual‐axis tomograms were generated using the etomo tool from the IMOD software package (version 4.9.7). Magnifications of ×3500, ×9600, and ×14 500 were used to different cell samples corresponding to pixel sizes of 6.2, 2.3, and 1.5 nm, respectively. For 3D model creation, contours were manually traced and meshed using the 3dmod tool within IMOD.

4.4. Correlated Light and Electron Microscope

For sample fixation and resin embedding, Seedlings treated with 10% [w:v] sodium chloride for plasmolysis were harvested at the root tip and subjected to high‐pressure freezing. Freeze substitution of the samples are performed in acetone with 0.25% glutaraldehyde and 0.2% uranyl acetate at −85°C for 24 h, followed by a gradual increase to −35°C. LR White Resin (Electron Microscopy Sciences) is then substituted in increments of 33%, 66%, and 100% at −35°C, and the samples were UV exposed for 16 h for embedding The general workflow of Correlative Light and Electron Microscopy (CLEM) has been outlined previously [54, 84]. Briefly, 150 nm ultrathin sections are prepared from cryo‐fixed root blocks expressing Exo70E2‐GFP. These sections were then analyzed using confocal microscopy (Leica Stellaris 8) and Transmission Electron Microscopy (TEM) (Hitachi High‐Technologies). Alexa Fluor 568 fluorescent beads (red) are used as fiducial markers to facilitate the correlation of images. Which give red fluorescent under the light microscope and appear as 15 nm sharp dots within the TEM. Finally, images obtained from light microscope and TEM are digitally combined by aligning with fluorescent beads using Adobe Photoshop.

4.5. Cryo‐Electron tomography

Follow general workflow of sample preparation with Cryo‐focused ion beam (Cryo‐FIB) and Cryo‐ET imaging as described previously [30, 48]. In brief, in vitro germinated Arabidopsis pollen tubes are transferred to silica coated grids and plunge freezed in liquid ethane and propane with Vitrobot (Thermo Scientific). For sample preparation of the pollen grains, high‐pressure freezing is applied to fix the anthers from 10‐week‐old Arabidopsis plants using 3 mm copper carriers. This is followed by the lift‐out of sample chunks from the high‐pressure frozen sample carriers using the SOLIST (Serialized on‐grid lift‐in sectioning for tomography) method as described previously [46], sample chunks further undergo dissecting and landing onto new grids. Sample grids containing pollen tube/lifted‐out pollen grains were loaded into aquilos 2 (Thermo Scientific) and are milled with ion beam. The final thickness of lamella is less than 300 nm for following Cryo‐ET imagine. Raw tilt serial images were collected from Titan Krios (Thermo Scientific) operating at 300 kV and reconstructed into tomograms. The model generation is further done with 3dmod tool within IMOD.

4.6. Immunogold Labeling TEM

For sample fixation and resin embedding, Root tips and leaves from 5‐day‐old Arabidopsis seedlings were harvested, cut, and rapidly frozen using a Leica EM ICE High‐Pressure Freezer then transferred to the Leica AFS system. The samples are submerged in acetone with 0.25% glutaraldehyde and 0.2% uranyl acetate at −85°C for 48 h, then warmed up gradually to −35°C at a rate of 1.5°C per hour. HM20 resin (Lowicryl) substitution followed stepwise at concentrations of 33%, 66%, and 100% at −35°C, with subsequent UV exposure for embedding. For immunolabeling with established protocols previously [85]. In brief, 100 nm ultrathin sections of leaf samples were obtained from UC7 ultramicrotome (Leica Microsystems). The sections were then incubated with custom‐made anti‐GFP polyclonal antibodies at a concentration of 40 µg/mL followed by 8 or 10 nm gold‐labeled secondary antibodies at 1:50 dilution. The prepared sections were analyzed using the TEM operating at 80 kV (Hitachi High‐Technologies).

4.7. Transient Expression in Arabidopsis Protoplast

Transient expression in Arabidopsis protoplasts was carried out following the method previously described [86]. In brief, 50 mL of a 5‐day‐old Arabidopsis suspension cell culture was centrifuged at 1000 rpm for 2 min at room temperature, and the supernatant was discarded. The cell pellet was then treated with 50 mL of a sterile enzyme solution (1% cellulase, 0.05% pectinase, 0.5% driselase in MS liquid medium) to isolate the protoplasts. This mixture was incubated at 27°C with gentle shaking (65 rpm) for 2 h. After the incubation, the protoplasts were centrifuged at 80 g for 15 min at room temperature. The supernatant was removed, and the protoplasts were gently washed with 30 mL of electroporation buffer. The washed cells were then resuspended in the same buffer to a final concentration of 5×10⁶ protoplasts/ mL. For electroporation, 500 µL of the protoplast suspension was transferred into 0.4 cm gap electroporation cuvettes along with 40 µg of DNA. The mixture was allowed to stand for 10 min at room temperature before electroporation was performed at 130 V and 1000 µF for a single pulse. The protoplasts were then allowed to recover at room temperature for 20 min. Following recovery, the protoplasts were plated in Petri dishes containing 2 mL of TEX buffer and incubated in the dark at 26°C for 16 h. After this period, the protoplasts were ready for imaging using a Leica Stellaris 8 laser scanning confocal microscope (Leica microsystem).

4.8. Subcellular Fractionation and EXPO Isolation

The procedures of subcellular fractionation for plant cells have been described previously [55, 87], in brief, 3‐day‐old cells were pelleted by centrifugation at 500 rpm and washed twice with 0.4 M mannitol. The cells were then incubated in lysis buffer (1% cellulase, 0.05% pectinase, 0.5% driselase in 0.4 M mannitol) at 30°C with shaking at 120 rpm for 1 h to produce protoplasts. After washing, the protoplasts were resuspended in a 13.7% (w:v) sucrose solution in a basic buffer (40 mM HEPES, pH 7.2; 10 mM KCl; 3 mM MgCl2; 0.1 mM EDTA) containing protease inhibitors (Roche). The suspension was then passed through a 25Gx5/8″ needle (Terumo NN2516R) at least four times to lysis the cells.

The cell lysate was layered on top of a sucrose gradient and centrifuged at 110 000 g for 2 h at 4°C. Fractions were collected from the interfaces of the resulting gradients. These fractions were prepared for negative staining for transmission electron microscopy (TEM) analysis, immunoisolation, or membrane protein extraction. GFP antibodies, antibodies against AtExo70E2, or GFP were coupled to CNBr‐activated Sepharose beads (Sigma, USA) or Protein A agarose beads (Invitrogen) for 3 h at 4°C in phosphate‐buffered saline, then equilibrated in the basic buffer. The EXPO‐enriched fraction was diluted 1:2 in basic buffer and incubated with the antibody‐coated beads for 4 h at 4°C. The beads were washed three times with 13.7% (w:v) sucrose in basic buffer and collected for high‐pressure freezing/frozen substitution for TEM analysis, or they were washed in 1% SDS and stored for proteomic analysis. Appropriate negative controls, such as Protein A‐agarose and Sepharose beads alone immune‐isolation, were included to assess background binding, and proteins detected in these control samples were excluded from the final EXPO proteome dataset.

4.9. Pathogen Treatment and Colony Counting

For pathogen inoculation, Pseudomonas syingae pv. tomato (Pst) DC3000 bacterial is grown on KB medium containing 100 mg/L rifampicin, incubated at 28°C for 2–3 days. 20 single colonies were selected and transferred into 100 mL of KB medium containing 100 mg/L rifampicin. Incubate at 28°C with shaking until the OD600 reaches 0.8‐1. Bacteria cultures were centrifuged at room temperature at 3000 rpm for 10 min. Discard the supernatant and resuspend the pellet in 10 mM MgCl₂ to an OD600 of 0.8, adding 0.03% Silwet (300 µL/L). Spray the bacterial suspension onto the leaves until they are thoroughly wet. Cover with a lid or plastic wrap to maintain humidity.

For colony counting, harvest leaves from the test plants at 0h, 24h and 48h post‐inoculation (try to keep the leaf positions consistent across different plants). Soak the leaves in 70% ethanol solution for 1 min, then remove the leaves and discard any residual alcohol. Rinse the leaves in sterile water for 1 min and remove any excess water. Use a 0.6 cm punch to obtain leaf disks, transferring them into a 2 mL centrifuge tube. Add steel beads and 1 mL of 10 mM MgCl₂ to homogenize the leaf disks. Perform a serial dilution and take 5–10 µL of the diluted solution, plating it sequentially onto solid KB plates containing rifampicin. Incubate at 28°C for 2 days. Count the number of colonies and calculate the CFU value of the pathogen.

4.10. RNA Extraction and Bioinformatic Analysis

RNA extraction for Arabidopsis plant has been described previously [88]. In brief, total RNAs were isolated from 4‐week‐old Arabidopsis plant leaves using the Plant RNA Isolation Mini Kit (Agilent Technologies). After removing the genomic DNA by incubating with gDNAse Eraser, cDNA was synthesized using PrimeScript RT Reagent Kit (Takara). For qPCR, three biological replicates were included for each reaction. The UBQ5 or ACTIN2 gene was used as an internal control. The PCR reaction was conducted using the Applied Biosystems StepOne Plus real‐time PCR machine (Thermo). For RNA‐sequencing, RNA samples were isolated from 12‐day‐old Arabidopsis whole seedlings and cDNA libraries were prepared by Novogene (Beijing) using the NEBNext Ultra RNA library prep kit for Illumina sequencing (NEW ENGLAND BIOLABS). Next generation sequencing data was generated by Illumina NovaSeq using 150‐bp paired‐end sequencing. The clean data was generated by removing reads with adaptors or low‐quality sequences and aligned to the Arabidopsis thaliana genome (TAIR10) using HISAT2 [89]. PCA plot was produced using DESeq2 package in R [90]. Differential expressed gene (DEG) analysis was conducted and DEGs with padj value < 0.05 were further analyzed and categorized using KEGG pathway database [91].

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting file 1: advs73546‐sup‐0001‐SuppMat.docx

Supporting file 2: advs73546‐sup‐0002‐Supplemental Videos.zip

Data Availability Statement

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