Fungal small RNAs ride in extracellular vesicles to enter plant cells through clathrin-mediated endocytosis

Small RNAs (sRNAs) of the fungal pathogen Botrytis cinerea can enter plant cells and hijack host Argonaute protein 1 (AGO1) to silence host immunity genes. However, the mechanism by which these fungal sRNAs are secreted and enter host cells remains unclear. Here, we demonstrate that B. cinerea utilizes extracellular vesicles (EVs) to secrete Bc-sRNAs, which are then internalized by plant cells through clathrin-mediated endocytosis (CME). The B. cinerea tetraspanin protein, Punchless 1 (BcPLS1), serves as an EV biomarker and plays an essential role in fungal pathogenicity. We observe numerous Arabidopsis clathrin-coated vesicles (CCVs) around B. cinerea infection sites and the colocalization of B. cinerea EV marker BcPLS1 and Arabidopsis CLATHRIN LIGHT CHAIN 1, one of the core components of CCV. Meanwhile, BcPLS1 and the B. cinerea-secreted sRNAs are detected in purified CCVs after infection. Arabidopsis knockout mutants and inducible dominant-negative mutants of key components of CME pathway exhibit increased resistance to B. cinerea infection. Furthermore, Bc-sRNA loading into Arabidopsis AGO1 and host target gene suppression are attenuated in those CME mutants. Together, our results demonstrate that fungi secrete sRNAs via EVs, which then enter host plant cells mainly through CME.

Interestingly, cross-kingdom sRNA trafficking has been found to be important even for symbiosis, specifically in mycorrhizal interactions 10 and between prokaryotic Rhizobia and host soybean roots 8 . This mechanism is not limited to plant-microbe/parasite interactions and has also been observed in animal-pathogen/parasite systems [4][5][6] . In many of these interactions, the transferred microbial sRNAs can utilize host AGO proteins to silence host immune response genes 3,8,[11][12][13] .
Given the importance of these sRNA effectors in pathogen virulence, anti-fungal strategies that target and silence genes in the RNAi machinery, such as the Dicer-like (DCL) proteins that generate sRNAs, have been shown to be effective for plant protection 15,[17][18][19][20][21] . Further protection could likely be achieved by inhibiting the transport processes that enable cross-kingdom trafficking of these sRNAs. However, how these transferred sRNAs are delivered and internalized by plant cells is still a mystery. Extracellular vesicles (EVs) are used across the kingdoms of life to package and deliver functional molecules into recipient cells or organisms 2,6,22-24 . These EVs can provide protection to their cargos from degradation by extracellular enzymes, which is especially important for RNA trafficking. Indeed, EVs have been utilized by plants to deliver plant small RNAs to fungal pathogens during cross-kingdom RNAi 2 . Furthermore, many fungal species have been shown to secrete EVs, including yeast and filamentous fungi [25][26][27][28] . Although, only recently have EVs from plant pathogenic fungi such as the maize smut pathogen Ustilago maydis 29 , the wheat pathogen Zymoseptoria tritici 30 , and the cotton pathogen Fusarium oxysporum f. sp. vasinfectum 31 been characterized. Still, it remains unclear whether plant fungal pathogens secrete EVs to deliver sRNAs into interacting plant cells.
Both BcPLS1-YFP and Arabidopsis exosome marker TET8 were concentrated in the same fractions (Fig. 1f). To further confirm that B. cinerea secretes EVs during infection, BcPLS1-YFP or BcTSP3-YFP strains were used to inoculate Arabidopsis TET8-CFP transgenic plants. EVs were isolated from infected plants using sucrose gradients, and fraction four was examined under confocal microscopy. The EVs from BcPLS1-YFP-infected TET8-CFP plants exhibited two distinct fluorescent protein-labeled signals (Fig. 1g). These results demonstrate that during infection, B. cinerea can secrete BcPLS1-positive EVs.

B. cinerea secretes small RNA effectors through EVs
Our previous work demonstrated that B. cinerea could transfer sRNA effectors into host plant cells to suppress host immunity and achieve infection 3 . To study whether B. cinerea uses EVs to transport these sRNA effectors, we examined fungal sRNAs using sRNA stem-loop reverse transcription-polymerase chain reaction (RT-PCR) analysis in EVs isolated from B. cinerea liquid culture 50 . Three previously discovered sRNA effectors, Bc-siR3.1, Bc-siR3.2, and Bc-siR5, were detected in B. cinerea EVs isolated by sucrose gradient fractionation. These sRNAs were concentrated primarily in fractions four and five, which also contained high levels of BcPLS1 (Fig.   2a). Additionally, EVs from fraction four were used for a Micrococcal Nuclease (MNase) protection assay. As shown in Fig. 2b, the sRNA effectors can still be detected after MNase treatment unless vesicles were ruptured with Triton X-100 ( Fig. 2b and Supplementary Fig. 3).
This result indicates that B. cinerea sRNAs are localized within and protected by EVs.
To assess the role of tetraspanin proteins in B. cinerea small RNA delivery, deletion mutants of BcPLS1 and BcTSP3 were generated by targeted gene replacement in wild-type B. cinerea strain B05 ( Supplementary Fig.4a). Two independent mutant strains of each gene Δbcpls1 and Δbctsp3 were complemented by expressing BcPLS1 and BcTSP3, respectively ( Supplementary   Fig.4a). On minimal medium (MM), but not potato dextrose agar, the growth rate of the Δbcpls1 strain was reduced compared to the wild-type and Δbctsp3 strains. (Supplementary Fig.5). The spores of both mutants were collected and used to inoculate wild-type Arabidopsis plants to assess their virulence. The Δbcpls1 mutant strain showed a significant reduction in virulence with the lesion size being reduced by 84% compared with the wild-type strain. (Fig. 2c). However, Δbctsp3 resulted in lesions similar to the wild-type strain (Fig. 2c). To further understand the functions of BcPLS1 and BcTSP3 in EV-mediated sRNA secretion from B. cinerea, EVs were isolated from . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint the culture supernatants of both the Δbcpls1 and Δbctsp3 strains and Bc-sRNA amounts were examined by real-time PCR and sRNA stem-loop RT-PCR. Compared with the wild-type strain, significantly lower levels of Bc-sRNAs were detected in Δbcpls1 EVs, while similar amounts of Bc-sRNAs were detected in Δbctsp3 EVs ( Fig. 2d and Supplementary Fig. 6a). Further, we found the EV amount released by Δbcpls1 was significantly reduced compared with wild-type and Δ bctsp3 strains (Supplementary Fig. 6b To determine whether sRNAs in the EVs contribute to the fungal virulence, we premixed spores from a B. cinerea mutant dcl1dcl2 strain with EVs isolated from wild-type B. cinerea culture supernatants and then inoculated Arabidopsis plants. The dcl1dcl2 mutant cannot generate sRNA effectors, such as Bc-siR3.1, Bc-siR3.2, and Bc-siR5, and exhibits a very weak virulence phenotype 3 . However, when premixed with EVs from wild-type B. cinerea, which contain sRNA effectors, B. cinerea dcl1dcl2 mutant pathogenicity was dramatically recovered. The lesion size increased from 24% of the wild-type strain to 61% of the wild-type strain (Fig. 2f). We then further used EVs isolated from the dcl1dcl2 mutant to complement the phenotype of Δbcpls1. Results showed that B. cinerea dcl1dcl2 mutant EVs could only restore the virulence of Δbcpls1 from 18% of the wild-type strain to 32% of the wild-type strain ( Supplementary Fig. 7). This modest increase may be attributed to the other cargoes in the dcl1dcl2 EVs, because fungal EVs have been found to serve as a mode of transport for various virulence-related proteins and toxic metabolites 51 .
These results indicate that fungal DCL1/DCL2-generated sRNA effectors encapsulated in B.
cinerea EVs play an important role in fungal virulence.
Clathrin-mediated endocytosis is involved in plant immunity against B. cinerea . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made In the mammalian system, multiple endocytic pathways are used to internalize EVs and their cargoes from various origins 52 . To determine whether endocytosis pathways are also involved in the uptake of B. cinerea EVs and their sRNA cargoes, we first treated wild-type Arabidopsis leaves with plant CME-specific inhibitor ES9-17 or CIE-specific inhibitor MβCD 53,54 . Vesicle internalization was strongly suppressed by ES9-17 but not by MβCD ( Supplementary Fig. 8a,c), suggesting that CME is likely the major pathway for EV uptake. However, we found that ES9-17 can also suppress the growth of B. cinerea ( Supplementary Fig. 8b). So, the chemical treatment may cause a broad impact on both plants and fungi. Therefore, we chose genetic methods to ascertain the involvement of the CME pathway in the plant immune response to B. cinerea infection. We performed B. cinerea infection on two Arabidopsis CME pathway mutants, chc2 and ap2σ, which carry mutations in CLATHRIN HEAVY CHAIN 2 and ADAPTOR 2 SIGMA SUBUNIT, respectively. Compared with wild-type plants, both chc2-1 and ap2σ mutants displayed reduced disease susceptibility to B. cinerea (Fig. 3a). The relative lesion size on chc2-1 and ap2σ mutants was significantly reduced by 60% and 73% compared with wild-type plants, respectively ( Fig. 3a). Furthermore, we took advantage of inducible lines that expressed the C-terminal part of clathrin heavy chain (INTAM>>RFP-HUB1) or the clathrin coat disassembly factor Auxilin2 (AX2) (XVE>>AX2). Overexpression of both lines blocked CME in Arabidopsis 39,55 . After induction, both INTAM>>RFP-HUB1 and XVE>>AX2 lines showed significantly higher resistance to B. cinerea infection (Fig. 3b,c). The relative lesion size on INTAM>>RFP-HUB1 and XVE>>AX2 was reduced by 45% and 59% compared with control plants. These results confirm that CME contributes to plant susceptibility to B. cinerea infection. Next, we tested the possible function of the CIE pathway in plant immunity against B. cinerea infection. We examined the response of CIE-pathway mutants flot1/2, remorin 1.2, and remorin 1.3 after infection of wild-type B. cinerea. Because there is a close homolog of FLOT1 in Arabidopsis named FLOT2, we generated flot1/2 double mutants by knocking out FLOT2 in the flot1 mutant background using the Crispr-Cas9 system ( Supplementary Fig.9). Two independent lines were inoculated by B. cinerea, both of them showed a slightly more susceptible phenotype compared with wild-type (Fig. 3d). The homozygous knockout mutants of rem1.2 and rem1.3 T-DNA insertion lines were available and used for pathogen assays. No noticeable difference was observed between rem1.2 or rem1.3 and wild-type plants when challenged by B.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint cinerea spores (Fig.3e). These results indicate that CME plays a more important role than CIE in plant susceptibility and the internalization of B. cinerea sRNAs into plant cells.

Bc-sRNA loading into host AGO1 and subsequent host target gene suppression is attenuated in CME mutants
B. cinerea sRNAs can hijack the host plant RNAi machinery and silence plant immunityrelated genes by binding to Arabidopsis Argonaute protein AGO1 3 . If CME is a critical factor in the internalization of EVs and Bc-sRNAs into plant cells, the deficiency of CME is likely to impact the level of sRNAs incorporated into the AGO1 protein. To determine whether CME is crucial for the uptake of B. cinerea sRNAs, we immunoprecipitated AGO1 from B. cinerea-infected wildtype Arabidopsis and CME mutants, including chc2-1 and ap2σ. As shown in Fig. 4a, the fungal sRNA effectors, Bc-siR3.1, Bc-siR3.2, and Bc-siR5, were all detected in the AGO1-associated fraction precipitated from infected wild-type plants. However, the amount of B. cinerea sRNAs loaded into AGO1 was significantly reduced by around 60% and 80% in chc2-1 and ap2σ, respectively, as compared with wild-type plants. This observation was further supported by similar results obtained from INTAM>>RFP-HUB1 and XVE>>AX2 lines (Fig. 4b,c).
We also assessed whether CIE contributes to Bc-sRNA uptake by challenging flot1/2, rem1.2, and rem1.3 mutants with wild-type B.cinerea. Unlike CME mutants, the amount of AGO1-associated Bc-sRNAs in flot1/2, rem1.2, and rem1.3 mutants was similar to wild-type plants (Fig. 4d,e). As the total amount of AGO1 protein was not decreased in any of the CME and CIE mutants ( Supplementary Fig.10), the decrease of Bc-sRNAs associated with AGO1 in CME mutants can be attributed to a reduction in the Bc-sRNA internalization of plant cells, rather than a decrease in the amount of AGO1 protein. Thus, CME is a key pathway for Bc-sRNA trafficking into host plant cells.
B. cinerea sRNA effectors have a suppressive effect on predicted host target genes during infection 3,56 . If CME is truly essential for cross-kingdom trafficking of Bc-sRNAs, we expect to observe attenuated suppression of host target genes in CME mutants. Thus, we quantified the (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint XVE>>AX2 mutants shown by no reduction in target gene transcript levels 24 hours after infection ( Fig. 4f and Supplementary Fig. 11a,b). We performed a similar experiment to explore the role of the CIE pathway in target gene suppression using the flot1/2, rem1.2, and rem1.3 mutants. Unlike CME mutants, all target genes were still significantly suppressed in the CIE mutants ( Supplementary Fig. 12a,b). These results again demonstrate that CME is the major pathway for Bc-sRNA uptake.  (Fig. 5a,b). Meanwhile, we detected colocalization between CLC1-GFP or CHC2-YFP with the FM4-64 signal on vesicles at the infection sites, with co-localization rates of 52% and 60%, respectively (Fig. 5a,b). These results suggest that B. cinerea infection induces the accumulation of CCVs around infection sites. In addition to clathrin, AP2 adaptor complexes are important components for CME in plant cells 57 . AP2A1-GFP plants were also treated with wild-type B. cinerea spores. Similar to CHC2-YFP and CLC1-GFP plants, clear colocalization (46%) was observed on the endocytic vesicles at the infection sites between AP2A1-GFP with FM4-64 ( Fig. 5c). We also generated the YFP-tagged FLOT1 transgenic line and monitored the endocytosis through FM4-64 staining following inoculation with B. cinerea. As shown in Fig. 5d, endocytotic FM4-64 signals can also be observed, but the percentage of colocalization (10%) between YFP-FLOT and FM4-64 was limited. These results further confirm that CME is more involved than CIE in B. cinerea infection of Arabidopsis.

B. cinerea EVs and sRNAs are associated with clathrin-coated vesicles
Unlike in animal cells, plant CCVs are not disassembled immediately after scission from the plasma membrane, which allows isolation of CCVs by immunocapture 58 . We isolated plant CCVs from CLC1-GFP transgenic plants after infection with B. cinerea using a two-step procedure comprising differential centrifugation, followed by immunoaffinity purification with antibodies . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint against GFP. GFP signal was clearly detected on the surface of GFP-TRAP agarose beads after incubation with isolated CCVs (Fig. 6a). To examine whether the immunoprecipitated CCVs were intact, the GFP-TRAP beads after pull-down were subjected to scanning electron microscopy (SEM) analysis. Fig. 6b shows that most of the immuno-captured CCVs were intact with typical cage structures on the surface of GFP-TRAP agarose beads. We further inoculated CLC1-GFP plants with the BcPLS1-mCherry B. cinerea strain. After ten hours of inoculation, colocalization of CLC1-GFP and BcPLS1-mCherry signals can be observed around the infection sites (Fig. 6c).
Further, BcPLS1-mCherry can be detected in the CCV fractions isolated from infected plants (Fig.   6d). All three of the Bc-sRNA effectors that we examined previously were also present in the immunopurified CCVs (Fig. 6e). These results provide conclusive evidence that B. cinerea EVs are responsible for transporting Bc-sRNAs and enabling their internalization through plant CME.

Discussion
Cross-kingdom RNAi is emerging as a critical regulatory mechanism across interaction systems. While recent studies show that plant and animal hosts utilize extracellular vesicles to protect and transport sRNAs to their fungal pathogens 2,5 , how fungi transported sRNAs to their hosts was still unknown. Here, we demonstrate that, just like their plant hosts, B. cinerea utilizes EVs to protect and transport sRNAs to plants, providing the first example of fungi utilizing EVs for cross-kingdom RNA transport. Furthermore, we show that plant cells internalize fungal EVs through clathrin-mediated endocytosis. CME is one of the major pathways for plant uptake of environmental molecules, with extensive studies demonstrating its vital role in different biological processes 36,59 , including plant defense responses 37,39,60-62 . While the plant CME pathway has recently been implicated in the uptake of protein effectors from fungal and oomycete pathogens 42,43 , we demonstrate here that it is also critical for the plant uptake of fungal sRNA effectors and fungal EVs. Although in mammals, intraorganismal EV internalization into recipient cells could occur through CME 63 , the mechanisms of internalization of foreign EVs from one organism to another in a different kingdom had never been reported until this study. We found that numerous CCVs accumulated around infection sites, indicating the activation of CME during B. cinerea infection. Additionally, the B. cinerea EV marker protein BcPLS1 is colocalized with plant CME component protein CLC1 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint during fungal infection and can be detected in the immunopurified plant CCVs. These findings strongly support the EV internalization by plant CME pathway. Furthermore, the EV-delivered sRNAs were also detected in the immunopurified CCVs, providing further evidence for the role of plant CME in the uptake of fungal EV-delivered sRNAs. and Magnaporthe grisea showed that Δpls1 mutants exhibit weak virulence 71,72 , and the M. grisea PLS1 protein was found to be mainly expressed during infection and localized to appressorial membranes 72 . Consistently, we found the Δbcpls1mutant showed weak virulence on plants (Fig.   2c) and the BcPLS1 protein can be observed on the appressorial membranes (Fig. 1e). Furthermore, we detected BcPLS1 in the isolated B. cinerea EVs, and found that the delivery of B. cinerea sRNA effectors is compromised in the Δbcpls1 mutant, suggesting that BcPLS1-positive EVs are critical for sRNA effector transport. Thus, we have expanded the repetoire of effective genetic targets, such as PLS1, for anti-fungal strategies such as spray-induced gene silencing. Combining existing effective gene targets with PLS1 and other EV proteins targeting in SIGS will likely . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint provide even more effective crop protection by targeting not only sRNA biogenesis, but also the trafficking mechanism of sRNA effectors from pathogens to plants.
Overall, our study helps to elucidate the underlying molecular mechanisms of EV-mediated sRNA trafficking during "cross-kingdom RNAi" and will further inform efforts in other organisms that also deliver sRNAs into interacting species. Given how widespread the phenomenon of crosskingdom RNAi is in different branches of life, this will have strong rammifications for antipathogen measures in agriculture and human health.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

EV isolation
EV isolation from fungal culture supernatant was performed as previously described 25 . In order to eliminate the possibility of contamination from the culture medium, the medium was initially ultracentrifuged at 100,000 g for 1 hour to eliminate any particles. Subsequently, the medium was filtered using a 0.22 μm filter prior to its use for B. cinerea culture. 10 6 B. cinerea spores were cultured for 48 hours at room temperature with shaking on the rocker in 100 ml of purified YEPD (0.3% yeast extract, 1% peptone, 2% glucose, pH 6.5) medium. The culture supernatant was collected by centrifugation at 3,000 g for 15 mins at 4°C. The supernatant was then filtered through a 70 μm cell strainer to further remove cells. The cell-free supernatant was centrifuged at 10,000 g for 30 mins to remove smaller debris and a second round of centrifugation at 10,000 g for 30 mins was performed to ensure the removal of aggregates. The resulting supernatant was filtered through a 0.45 μm filter and ultracentrifuged at 100,000 g for 1 h to precipitate vesicles. Vesicles were washed once in phosphate-buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4), and the final pellets were resuspended in PBS.
EV isolation from infected plants was performed as previously described 2,47 . Apoplastic fluid was extracted from B. cinerea-infected Arabidopsis leaves by vacuum infiltration with infiltration buffer (20 mM MES, 2 mM CaCl2, 0.1 M NaCl, pH 6.0) and then centrifuged at 900 g for 10 mins.
Cellular debris was further removed from the supernatant by centrifugation at 2,000 g for 30 .
CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint minutes, filtration through a 0.45 μm filter and finally centrifugation at 10,000 g for 30 mins. EVs were then collected by ultracentrifugation at 100,000 g for 1 hr and washed with infiltration buffer at 100,000 g for 1 hour to collect the EVs.

Sucrose gradient separation of EVs
EVs from B. cinerea liquid culture and B. cinerea-infected Arabidopsis were purified by discontinuous sucrose density gradient centrifugation 23 . 10-90% sucrose stocks (w/v), including 10,16,22,28,34,40,46,52,58,64,70 and 90%, were prepared using infiltration buffer (20 mM MES, 2 mM CaCl2, 0.1 M NaCl, pH 6.0). The discontinuous gradient was prepared by layering 1 ml of each solution in a 15-ml ultracentrifuge tube. 100 μl EVs were premixed with 1 ml of 10% sucrose stock and loaded on top of the sucrose gradient. Then, samples were centrifuged in a swinging-bucket rotor for 16 h at 100,000 g, 4 °C and six fractions (2 ml each) were collected from top to bottom. Collected fractions were transferred to new ultracentrifuge tubes and each sample was diluted to 12 ml using infiltration buffer, followed by a final centrifugation for 1 h at 100,000g, 4 °C to obtain pellets. The pellet was then resuspended in 50 μl infiltration buffer for further analysis.

Nanoparticle tracking analysis
A NanoSight NS300 fitted with a blue laser (405 nm) with NanoSight NTA software version 3.1

Transmission electron microscopy analysis
B. cinerea EVs were imaged using negative staining. Briefly, 10 μl of EVs were deposited onto 3.0 mm copper Formvar-carbon-coated grids (EMS) for 1 min and excess sample was removed from the grids using filter paper. The grids were then stained with 1% uranyl acetate for 30 s and blotted before a second round of staining with 1% uranyl acetate 2 min. The samples were airdried before visualization with a Talos L120 transmission electron microscope, operated at 120 kV.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint sRNA expression analyses RNA was extracted using the Trizol extraction method for both total RNA and EV samples. sRNA RT-PCR was performed as previously described 2 . The PCR products were detected on a 12% PAGE gel. Quantitative PCR was performed with the CFX96 real-time PCR detection system (Bio-Rad) using SYBR Green mix (Bio-Rad). All primer sequences are provided in Supplementary Data 1.
To determine if sRNAs were localized within vesicles, EVs from sucrose gradient-purified fraction four were treated with 10 U of micrococcal nuclease (MNase) (Thermo Fisher) with or without Triton-X-100. For Triton-X-100 treatment, vesicles were incubated with 1% Triton-X-100 on ice for 30 minutes before nuclease treatments. Nuclease treatment was carried out at 37°C for 15 minutes followed by RNA isolation.

Transformation
The homologous recombination-based method was used to knock out B. cinerea genes as  The BcPLS1-YFP and BcTSP3-YFP strains were generated by amplifying and inserting the stop codon removed CDS region and 3' UTR region into the modified PBS-HPH, in which the YFP gene sequence has been inserted between BamHI and HindIII. The BcPLS1-mCherry was generated by inserting the stop codon removed CDS region and 3' UTR region into the modified PBS-HPH with the mCherry sequence inserted at the C terminal. The transformation and selection conditions were the same as the homologous recombination-based gene knock-out. All primers used are provided in Supplementary Data 1.

Plant Materials and Growth Conditions
Arabidopsis thaliana wild-type and mutant lines used in this study were all Columbia 0 (Col-0) backgrounds. The chc2-1(SALK_028826) 74

Plasmid construction
The vector for CHC2-YFP was constructed as follows: Arabidopsis genomic DNA was used as a template with primers CHC2-Pentr-F and CHC2-Pentr-R to amplify the whole sequence of CHC2.
The final PCR product was cloned into pENTR/DTOPO (Life Technologies), and then into the

CCV purification
The clathrin-coated vesicle isolation procedure was adapted from previously established protocols 78,79 . 50 grams of Arabidopsis leaves were homogenized for 1 min in 2 volumes of resuspension buffer (100 mM MES, pH 6.4, 3mM EDTA, 0.5 mM MgCl2, 1mM EGTA, protease inhibitors (Roche)) using a Waring Blender. The homogenate was squeezed through eight layers of cheesecloth and three layers of Miracloth (Sigma). Then the homogenate was centrifuged in four increments to obtain a microsomal pellet: 500 g, 5 min; 5,000 g, 10 min; 20,000 g, 30 min; 100,000 g, 60 min. The pellet was then resuspended in 6 ml resuspension buffer. Then the resuspended pellet was loaded on top of the discontinuous sucrose gradients, which were made from three sucrose stocks (10%, 35%, and 50%) in the resuspension buffer. The gradient was composed of 1 ml of 50% sucrose, 5 ml of 35% sucrose, and 3 ml of 10% sucrose in a Beckman 14 × 95 mm Ultra-Clear centrifuge tube. The gradients were centrifuged at 116,000 g 50min.
Samples were collected from 10% and 10/35% interface and resuspended in the resuspension buffer to make the sucrose concentration lower than 5%. The samples were then centrifuged at 200,000 g for 50min, and the pellets were collected as crude CCVs.
Further immunoisolation of CCVs was performed as follows: GFP-Trap® Agarose (ChromoTek) was equilibrated in the resuspension buffer. Plant crude CCVs from CLC1-GFP and YFP were incubated with the GFP-Trap® Agarose for 2 hr at 4 °C. After, beads were washed three times . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made with resuspension buffer. The RNA from purified CCVs was extracted using TRIzol Reagent (Invitrogen) following immunoisolation. Subsequently, the immunoisolated CCVs on GFP-Trap Agarose beads were further analyzed by a scanning electron microscope (SEM).

Confocal microscopy
Arabidopsis seedlings were imaged on a Leica TCS SP5 confocal microscope (Leica Microsystems). Briefly, leaves were infiltrated with 10 μM FM4-64 dye for 30 mins before examination. Samples were imaged using a 40x water immersion dip-in lens mounted on a Leica For pathogen assay using a mixture of WT or Bc-dcl1/dcl2 B. cinerea EVs with Δbcpls1 and Bc-dcl1/dcl2, 10 6 B. cinerea spores were cultured for 48 hours at room temperature with shaking on the rocker in 100 ml of YEPD (0.3% yeast extract, 1% peptone, 2% glucose, pH 6.5) medium. B.
cinerea EVs were isolated and purified using sucrose gradients as described above, then EVs from . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

Western blotting
Protein samples were resolved in a 12% SDS-polyacrylamide gel electrophoresis (PAGE) and then transferred to nitrocellulose membranes (GE Healthcare), blocked with 5% skimmed milk in TBST (Tris-buffered saline, 0.1% Tween 20) and probed with anti-GFP (Sigma-Aldrich). Then the signals were detected using the enhanced chemiluminescence method (GE Healthcare).

AGO1 immunoprecipitation
Arabidopsis AGO1 IP was performed with 5 g fresh leaves collected 24 hr after spray inoculation with B. cinerea. Tissues were collected and ground in liquid nitrogen before resuspension in 5 ml of extraction buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 5mM MgCl2, 1 mM DTT, 0.5% NP40, proteinase inhibitor cocktail; Sigma). The extract was then precleared by incubation with 50 ul of Protein A-agarose beads (Roche) at 4°C for 30 min. The precleared extract was incubated with 10 ul of AGO1-specific antibody for 6 hours. 50 ul of Protein A-agarose beads (Roche) was then added to the extract before overnight incubation. Immunoprecipitates were washed three times (5 min each) in wash buffer (20 mM Tris-HCl at pH 7.5, 150 mM NaCl, 5mM MgCl2, 1 mM DTT, 0.5% Triton x-100, proteinase inhibitor cocktail; Sigma). RNA was extracted from the . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.15.545159 doi: bioRxiv preprint bead-bound AGO1 with TRIzol reagent (Invitrogen). Proteins were detected by boiling the beadbound AGO1 in SDS-loading buffer before running in a SDS-PAGE gel.

Scanning electron microscope (SEM)
The structure of CCVs isolated using GFP-Trap Agarose was examined using field emission scanning electron microscopy Nova NanoSEM 450 (FEI) at 15kV. The samples were washed three times with PBS (pH 7.2) and fixed with 2.5% glutaraldehyde overnight at 4 °C. Samples were dehydrated with gradient alcohol solutions (10%, 30%, 50%, 70%, 80%, and 90%) for 10 min, 100% ethanol for 20 min, and 100% acetone for 30 min. Subsequently, samples were dried by CO2 under critical point. Samples were sputter coated with gold and fixed on the stab using carbon tape before observation and imaging.

Statistical analysis
The statistical analyses were performed using analysis of variance (ANOVA), Dunnett's multiple comparisons test, and unpaired two-tailed Student's t-test using GraphPad Prism (9.0.2). The statistical tests and n numbers, including sample sizes or biological replications, are described in the figure legends.

Data availability
The data that support the findings of this study are available within the paper, Supplementary   Information, and Source Data. Source data are provided with this paper.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made        (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made