An atypical Phytophthora sojae RxLR effector manipulates host vesicle trafficking to promote infection

In plants, the apoplast is a critical battlefield for plant-microbe interactions. Plants secrete defense-related proteins into the apoplast to ward off the invasion of pathogens. How microbial pathogens overcome plant apoplastic immunity remains largely unknown. In this study, we reported that an atypical RxLR effector PsAvh181 secreted by Phytophthora sojae, inhibits the secretion of plant defense-related apoplastic proteins. PsAvh181 localizes to plant plasma membrane and essential for P. sojae infection. By co-immunoprecipitation assay followed by liquid chromatography-tandem mass spectrometry analyses, we identified the soybean GmSNAP-1 as a candidate host target of PsAvh181. GmSNAP-1 encodes a soluble N-ethylmaleimide-sensitive factor (NSF) attachment protein, which associates with GmNSF of the SNARE complex functioning in vesicle trafficking. PsAvh181 binds to GmSNAP-1 in vivo and in vitro. PsAvh181 interferes with the interaction between GmSNAP-1 and GmNSF, and blocks the secretion of apoplastic defense-related proteins, such as pathogenesis-related protein PR-1 and apoplastic proteases. Taken together, these data show that an atypical P. sojae RxLR effector suppresses host apoplastic immunity by manipulating the host SNARE complex to interfere with host vesicle trafficking pathway.

AvrPto targets RabE to influence plant secretory pathway and suppress plant immunity [26]. The Phytophthora brassicae RxLR effector RxLR24 binds to host RABA-type GTPase to inhibit vesicle-mediated antimicrobial protein secretion [27]. The P. infestans PexRD12/31 effectors associate with Nicotiana benthamiana R-SNARE protein of the VAMP72 family and PexRD31 increases the number of endosomes in N. benthamiana cells [28].
P. sojae is a causal agent of soybean root rot, which poses a great threat to global soybean production [29]. During infection, P. sojae secretes hundreds of RxLR effectors to modulate plant immunity [30]. In the present study, we analyzed whether P. sojae effectors target protein secretion systems to modulate host immunity. PsAvh181 was induced at the early stage of infection and causes yellowing in the leaves of N. benthamiana [30]. In this study, we found PsAvh181 contains an atypical RxLR-dEER motif (RSLAAASEDITVKSSLRYGDALAADEN-DEER) and functions as a virulence factor. Further study found that PsAvh181 localizes to the plasma membrane and binds to GmSNAP-1 to interfere with the interaction between GmSNAP-1 and GmNSF in the SNARE complex. As a result, PsAvh181 suppresses the secretion of apoplastic defense-related proteins. Thus, our study discovered a novel mechanism that exploited by microbial pathogens to modulate plant immunity by disrupting host vesicle trafficking.

Secretion of GmGIP1, P69B and PR1 can be inhibited by PsAvh181
Since GmGIP1 acts as an important resistance component in soybean by inhibiting P. sojae glycoside hydrolase PsXEG1 [7], we determined whether P. sojae counters this defense mechanism by interfering with the secretion of GmGIP1 into the apoplast. In our previous work, we found that the plasma membrane-localized RxLR effector PsAvh240 could inhibit the secretion of GmAP1 [8]. In P. sojae, the other two RxLR effectors PsAvh181 and PsAvh241 also localize to the plant plasma membrane [31]. We developed an assay to monitor the secretion of GmGIP1 using confocal microscopy. We fused green fluorescent protein (GFP) to the C-terminal of GmGIP1. GmGIP1-GFP was co-expressed with PsAvh181-HA, PsAvh240-HA, PsAvh241-HA or empty vector in the leaves of N. benthamiana. Since green fluorescent protein is sensitive to the pH in the apoplast [32], the green fluorescence signal could not be detected in the intercellular space. The GFP signal was present in smaller vesicular structures with their corresponding proteins when GmGIP1-GFP was co-expressed with PsAvh240-HA, PsAvh241-HA or empty vector (EV). When GmGIP1-GFP was co-expressed with PsAvh181-HA, GmGIP1-GFP accumulated in the intracellular space of plant cells, microscopic observation showed the GFP signal was present in the endomembrane compartments like endoplasmic reticulum (ER) network (Figs 1A and S1). The RxLR effectors PsAvh240 and PsAvh241, two plasma membrane-localized effectors, failed to suppress the secretion of GmGIP1 and were used as negative controls in this assay. To determine whether PsAvh181 specifically suppresses the secretion of GmGIP1, we co-expressed GmGIP1-HA with GFP-PsAvh181, GFP-PsAvh240, GFP-PsAvh241 or GFP control in N. benthamiana, and then extracted the apoplast fluid and detected the levels of GmGIP1-HA protein. The results showed that when GmGIP1-HA was co-expressed with GFP-PsAvh181, the levels of GmGI-P1-HA protein in the apoplast were lower than when it was co-expressed with GFP-PsAvh240, GFP-PsAvh241 or GFP control (Fig 1B). The above results showed that PsAvh181 inhibits the secretion of GmGIP1.
To confirm that PsAvh181 inhibits the secretion of plant apoplastic proteins, we coexpressed GFP-PsAvh181 with GmGIP1-HA, P69B-HA, PR1-HA or GmAP1-HA in N. benthamiana and detected the accumulation of these proteins in the apoplast fluid. The accumulation of GmGIP1-HA, P69B-HA and PR1-HA in the apoplast was significantly lower when co-expressed with GFP-PsAvh181 than with the GFP control (S2 Fig). The accumulation of GmAP1 did not differ when it was co-expressed with GFP-PsAvh181 versus the GFP control, but was significantly lower when co-expressed with GFP-PsAvh240 (S2 Fig). These results showed that PsAvh181 can inhibit the secretion of the apoplastic proteins GmGIP1, P69B and PR1.

PsAvh181 is required for full virulence of P. sojae
PsAvh181 is induced during the early stage of P. sojae infection [30], and is conserved among the four sequenced P. sojae isolates (S3 Fig). PsAvh181 contains an atypical RxLR domain (RSLAAASEDITVKSSLRYGDALAADENDEER), to determine whether the N-terminal of PsAvh181 can translocate effector into plant cells, we replaced the N-terminal of Avr1b with the N-terminal of PsAvh181 (before DEER) and generated the mutant PsAvh181Nt+Avr1bCt-GFP (S4A Fig). We overexpressed Avr1b-GFP, Avr1bCt-GFP (the C-terminal of Avr1b) and PsAvh181Nt+Avr1bCt-GFP in P. sojae and examined whether the N-terminal of PsAvh181 can deliver Avr1b into plant cells. Inoculated soybean hypocotyl with PsAvh181Nt+Avr1bCt-GFP, and PsAvh181Nt+Avr1bCt-GFP showed haustoria localization (S4B Fig). The inoculation assays on soybean hypocotyls showed that the transformants overexpressing Avr1b-GFP and PsAvh181Nt+Avr1bCt-GFP were unable to infect the soybean cultivar HARO13 containing the Rps1b resistant gene, but can still infect the susceptible cultivar, Hefeng47 (S4C Fig). In contrast, the P. sojae WT strain (P6497) and the transformants that overexpress Avr1bCt-GFP could infect the soybean cultivar HARO13 containing the Rps1b resistant gene (S4C Fig). Accumulation of each protein was detected by western blotting (S4D Fig). Taken together, these data suggest that the N-terminal of PsAvh181 can translocate effector into plant cells during infection.

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection 98 and Avh181D-106) for gene functional analysis (Figs 2A and S5). These four PsAvh181 deletion mutants showed no significant growth difference compared to the wild type (WT) or control (CK) when cultured on V8 medium (S5C and S5D Fig). Infection assays revealed that PsAvh181 deletion mutants produced smaller lesions on soybean hypocotyl compared to the WT and CK strains (Fig 2A). Phytophthora biomass analysis also showed that knocking out PsAvh181 reduced P. sojae infection in soybean hypocotyl (Fig 2B). To confirm the virulence function of PsAvh181, we expressed PsAvh181 (without signal peptide) fused N-terminal GFP in soybean hairy roots, and inoculated the transformed roots with red fluorescent protein The bar graph shows the quantified biomass of P. sojae based on the results of genomic DNA qPCR. Data are the mean ± SEM of three replicates. Different letters at the top of bars indicate significant differences (P < 0.01; one-way ANOVA). (C) Expression of PsAvh181 in the soybean hairy roots promotes P. sojae infection. EV (GFP) and GFP-PsAvh181 were co-expressed in the soybean hairy roots, and inoculated with the RFP-labeled P. sojae strain P6497 (WT-RFP). Scale bars, 0.2 mm. (D) Quantification of oospores 48 hours after infection inoculation of P6497 WT-RFP. Data are the mean ± SEM of three replicates. Asterisks at the top of the bars indicate significant differences (P < 0.01; one-way ANOVA). (E) Relative biomass of P. sojae in the transformed soybean hairy roots was determined by qPCR at 48 h post inoculation. Data are the mean ± SEM of five independent biological replicates. Asterisks at the top of the bars indicate significant differences (P < 0.01; one-way ANOVA). (F) Subcellular localization of PsAvh181. C-terminal RFP tagged PsAvh181 was co-expressed with GFP in N. benthamiana. GFP-PsAvh181 and was co-expressed with PsAvh240-RFP in N. benthamiana. Epidermal cells in the infiltrated tissues were investigated using confocal microscopy at 48 h post agroinfiltration. Scale bars, 20 μm. (G) Fluorescence analysis of GFP/PsAvh181-RFP and GFP-PsAvh181/PsAvh240-RFP in membrane transects (white arrowheads). y axis, relative fluorescence intensity of GFP or RFP; x axis, transect length (μm). (H) Protein expression detection of the samples showed in (F) by western blotting using anti-GFP and anti-RFP antibodies. https://doi.org/10.1371/journal.ppat.1010104.g002

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection (RFP)-labeled P. sojae. Compared to the EV (GFP) control, P. sojae infection produced more oospores in the hairy roots that expressing GFP-PsAvh181 (Fig 2C and 2D). Consistently, the biomass of P. sojae in the PsAvh181-expressed hairy roots was much higher than in the EV control ( Fig 2E). Together these data demonstrated that Avh181 is essential for full virulence of P. sojae.

Plasma membrane localization of PsAvh181 is required for virulence
To further study the virulence of PsAvh181, we determined the subcellular localization of PsAvh181 in N. benthamiana using confocal microscopy. PsAvh181 (without signal peptide) fused with a C-terminal RFP was co-expressed with GFP in N. benthamiana as shown in Fig  2F. PsAvh181-RFP and GFP-PsAvh181 localized preferentially to the plasma membrane under normal conditions or plasmolysis (S6D and S6E Fig). We co-expressed GFP-PsAvh181 and PsAvh240-RFP, a previously reported plasma membrane-localized effector in P. sojae [8], in N. benthamiana. Proteins were detected by western blotting (Fig 2H). The merged image and the fluorescence intensity of cross-sections of a cell showed that GFP-PsAvh181 co-localized with PsAvh240-RFP (Fig 2F and 2G). We also used remorin as a marker of plasma membrane localization, as it was reported to localize to plasma membrane [33,34]. Again, we observed colocalization of remorin and PsAvh181 in the plasma membrane (S6A and S6B Fig). These results showed that PsAvh181 localized to the plasma membrane in planta.
PsAvh181 encodes an RxLR effector, and we predicted the protein tertiary structure of PsAvh181 by the structural homology modeling server Swiss-model (https://swissmodel. expasy.org/) (S7 Fig). PsAvh181 is consisted of a N-terminal α-helix (70-95 amino acids) followed by three WY domains at the C-terminal. Based on the predicted tertiary structure, we constructed two mutants, PsAvh181-M1 (deletion of amino acids 70-95 in PsAvh181) and PsAvh181-M2 (amino acids 70-95 in PsAvh181) ( Fig 3A). We co-expressed PsAvh181-RFP and the two mutants with GFP-PsAvh240, a plasma membrane-localized effector, and examined the localization of PsAvh181-RFP and the mutants, proteins were detected by western blotting (S8A Fig). We found that the mutant PsAvh181-M1, which deleted the N-terminal αhelix domain of PsAvh181 (without signal peptide), localized to the nucleus and cytoplasm but not to the plasma membrane (Fig 3B and 3C). However, the mutant PsAvh181-M2, which contained an N-terminal α-helix domain, still localized to the plasma membrane (Figs 3B and S6A). We also extracted the membrane fractions and only PsAvh181-RFP and PsAvh181-M2-RFP, but not PsAvh181-M1-RFP were detected in the membrane fractions by western blotting (S8B Fig).
To determine whether these domains are essential for the virulence function of PsAvh181, we individually overexpressed GFP-PsAvh181, GFP-PsAvh181-M1 and GFP-PsAvh181-M2 in soybean hairy roots. Protein accumulation was detected by western blotting (S8C Fig). Infection assays showed that expression of GFP-PsAvh181 significantly increased P. sojae infection, but the GFP-PsAvh181-M1 and PsAvh181-M2 mutants failed to do so (Fig 3D-3F). These results showed that plasma membrane localization of PsAvh181 is essential for its virulence function.

PsAvh181 interacts with the vesicle trafficking-related GmSNAP proteins
To further explore how PsAvh181 achieves virulence, we expressed GFP-PsAvh181 in N. benthamiana and assayed the target proteins using co-immunoprecipitation (Co-IP) followed by liquid chromatography-tandem mass spectrometry (LC-MS). According to the LC/MS data of PsAvh181 co-immunoprecipitation, multiple secretory pathway related proteins were detected (S1 Table). There are three copies of SNAP in soybean, as well as in N. benthamiana,

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection we cloned three soybean GmSNAPs (S9A Fig) and three N. benthamiana NbSNAPs (S10A Fig). We then tested whether GmSNAPs interact with PsAvh181 in planta. The N-terminal GFP-tagged GmSNAP-1 was co-expressed with C-terminal HA-tagged PsAvh181 or PsAvh240 in N. benthamiana. We found that GFP-GmSNAP-1 interacted with PsAvh181-HA, but not with the PsAvh240-HA (S9B Fig). In addition, all the three GmSNAP paralogs interacted with PsAvh181 (S9B and S9C Fig). These results showed that PsAvh181 can interact with GmSNAPs in vivo. By expressing these three soybean SNAPs in N. benthamiana leaves, we found that all three soybean SNAPs contributed to plant defense against P. sojae (S9D and S9E  Fig).
Since all three tested SNAPs contributed to plant defense, we focused on GmSNAP-1 for further analyses. We further investigated the interactions between GmSNAP-1 and the PsAvh181mutants. GmSNAP-1 interacted with PsAvh181 and PsAvh181-M2 in vivo and in vitro, but not with PsAvh181-M1 (Fig 4A and 4B). This indicated that the N-terminal 70-95 amino acids domain is the key region for PsAvh181 interaction with GmSNAP-1. We then coexpressed GFP-GmSNAP-1 with PsAvh181-RFP, PsAvh181-M1-RFP or PsAvh181-M2-RFP

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection and investigated their subcellular localization using confocal microscopy. We found that GFP-GmSNAP-1 co-localized to the plasma membrane with PsAvh181-RFP and PsAvh181-M2-RFP, but not with PsAvh181-M1-RFP (Fig 4C and 4D). Proteins were detected by western blotting (Fig 4E). Together, these data demonstrated that PsAvh181 interacts with the vesicle trafficking-related protein GmSNAP-1, and that the 70-95 amino acids domain of PsAvh181 is essential for this interaction.

PsAvh181 interferes with the GmSNAP-GmNSF complex
SNAP proteins are components of the SNARE complex and play important roles in vesicle trafficking [11]. After membrane fusion, SNAP recruits N-ethylmaleimide-sensitive factor (NSF) to the SNARE complex to provide energy for SNARE complex dissociating and recycling [11,13,35]. To determine whether PsAvh181 interacts with SNAPs to interfere with the

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection SNARE complex, we evaluated the interaction between GFP-GmSNAP-1 and GmNSF-HA in planta with/without the presence of PsAvh181-RFP. As shown in Fig 5, PsAvh181-RFP competed with GmNSF-HA for binding with GFP-GmSNAP-1 in a dose-dependent manner ( Fig  5A). In addition, we performed competitive binding assays in vitro using His-GmSNAP-1, GST-GmNSF and MBP-PsAvh181 purified from Escherichia coli. After co-incubation, adding MBP-PsAvh181 significantly reduced the interaction between His-GmSNAP-1 and GST-GmNSF (Fig 5B).
Since the PsAvh181-M2 mutant interacted with GmSNAP, we tested whether PsAvh181-M2 influenced the interaction between GmSNAP-1 and GmNSF. Co-IP assays showed that neither PsAvh181-M1 nor PsAvh181-M2 disrupted the interaction between GmSNAP and GmNSF in vivo and in vitro (S11A and S11B Fig). These results showed that PsAvh181 disrupts the interaction between GmSNAP-1 and GmNSF in vivo and in vitro.

GmSNAPs-mediated soybean defense against Phytophthora depends on the interaction between GmSNAP and NSF
To determine whether SNAPs are involved in P. sojae-soybean infection, we overexpressed GmSNAP-1 fused with N-terminal GFP tag in soybean hairy roots (Fig 6A). The transgenic hairy roots expressing GmSNAP-1 were collected and inoculated with the RFP-labeled P. sojae

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection ( Fig 6A). P. sojae infection produced significantly fewer oospores in hairy roots expressing GFP-GmSNAP-1 compared to the EV (GFP) control (Fig 6A and 6B). In addition, P. sojae biomass was also significantly lower in the GmSNAP-1-overexpressing soybean hairy roots compared to the GFP control ( Fig 6C). To further confirm the biological function of GmSNAPs, we silenced the GmSNAPs in soybean hairy roots using a silencing vector that targets GmSNAP-1 and its two homologs, GmSNAP-2 and GmSNAP-3 ( Fig 6E). Quantitative realtime polymerase chain reaction (qRT-PCR) assays confirmed successful silencing of GmSNAPs in the soybean hairy roots (Fig 6H). After inoculation with RFP-labeled P. sojae, the biomass of P. sojae in GmSNAPs-silenced hairy roots was much higher compared to the empty vector control (Fig 6E-6G).
To determine whether GmSNAPs are important targets for PsAvh181, we performed infection assays on GmSNAPs-silenced hairy roots using the PsAvh181-knockout mutant Avh181D-10 ( Fig 6G). Biomass analysis showed that silencing of GmSNAPs in the soybean hairy roots partially restored the virulence of Avh181D-10 ( Fig 6G). These data demonstrated that PsAvh181 achieves virulence by binding to GmSNAPs and interfering with their function. But the P. sojae biomass in GmSNAPs-silenced roots infected with Avh181D-10 was still lower

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection than those inoculated with WT. This may be due to additional functions of PsAvh181 besides its interference with the interaction of GmSNAP and GmNSF.
Furthermore, we determined whether the resistance of GmSNAPs depends on the interaction between GmSNAPs and GmNSF. We predicted the key sites on GmSNAP-1 that interact with GmNSF based on the structure of the SNARE complex [13,[36][37][38]. By mutation of these amino acid sites to produce alanine (A), we obtained the mutant GmSNAP-M3 with 243-264 (DEED to AAAA) and 285-289 (EDDLT to AAAAA). Co-IP assays revealed that GmSNAP-M3 cannot interact with GmNSF, but can still interact with PsAvh181 in vivo (S12A and S12B Fig). To determine whether GmSNAP-M3 retained its biological function, we attempted to overexpress GmSNAP-M3 in soybean hairy roots, but could not obtain transgenic hairy roots overexpressing GmSNAP-M3. Instead, we expressed GmSNAP-1 and GmSNAP-M3 in N. benthamiana and performed infection assays using Phytophthora capsici (S12C and S12D Fig). Compared to N. benthamiana leaves expressing GmSNAP-1, the leaves expressing GmSNAP-M3 had much larger lesions, similar to those expressing the control GFP. In addition, the lesion diameter and relative P. capsici biomass in N. benthamiana leaves expressing GmSNAP-M3 were much greater than those of leaves expressing GmSNAP-1, showing that the GmSNAP-M3 mutant could not defend against Phytophthora (S12C-S12F Fig). These results showed that GmSNAPs contribute to Phytophthora resistance based on their interactions with GmNSF.

SNAPs are required for the secretion of GmGIP1, P69B and PR1
Since SNAPs play important roles in vesicle trafficking [11], we evaluated whether the inhibition of GmGIP1 secretion by PsAvh181 was due to its interaction with SNAPs. We examined whether SNAPs mediate the secretion of GmGIP1, P69B and PR1. Since PsAvh181 interacts with NbSNAPs (S10A and S10B Fig), we silenced NbSNAPs in N. benthamiana (Fig 7A). Silencing NbSNAPs influenced the growth of N. benthamiana (Fig 7B). We overexpressed GmGIP1-HA, P69B-HA, PR1-HA and GmAP1-HA in SNAPs-silenced N. benthamiana and investigated the accumulation of these proteins in the apoplast. Both the apoplastic fluid and intercellular proteins were isolated and detected by western blot analysis, which showed that significantly lower amounts of GmGIP1, P69B and PR1 were collected from the apoplasts of TRV:: SNAPs plants than from EV control plants. In contrast, silencing SNAPs did not influence the secretion of GmAP1 (Fig 7C). We also fused GFP to the apoplastic proteins GmGIP1, P69B, PR1 and GmAP1, and overexpressed these proteins in the TRV:: SNAPs N. benthamiana. Compared to the EV control, GmGIP1-GFP, P69B-GFP and PR1-GFP accumulation in the cytoplasmic space increased significantly in the TRV:: SNAPs-treated N. benthamiana (S13 Fig). These results showed that SNAPs are required for the secretion of GmGIP1, P69B, and PR1.

Discussion
Hosts secrete defense proteins into the apoplast when infected by pathogens [39,40]. P. sojae secretes PsXEG1, a glycoside hydrolase 12 protein and major virulence factor, to help infection [6]. For fighting back, soybean secretes GmGIP1 to apoplast. As an inhibitor of the P. sojae apoplast effector PsXEG1, GmGIP1 was predicted to be an aspartic protease without

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection enzymatic activity and contributes to plant defense by inhibiting the glycoside hydrolase activity and virulence function of PsXEG1 [7]. We chose GmGIP1 to screen the RxLR effectors that can inhibit its secretion. In the previous work of our group, we found PsAvh240, a plasma membrane localized effector, can inhibit the secretion of GmAP1 [8]. In this study, we found that the RxLR effector PsAvh181 inhibits the secretion of GmGIP1.
RxLR effectors are an important group of intracellular effectors secreted by Phytophthora pathogens during infection [24,41]. Multiple studies show that RxLR effectors are secreted into different subcellular compartments to modulate plant immunity [30,42]. In the present study, we showed that PsAvh181, a plasma membrane-localized atypical RxLR effector, acts as a virulence factor by inhibiting the secretion of apoplastic proteases such as GmGIP1, P69B and PR1.
PsAvh181 is induced during the early stages of P. sojae infection [30], and knockout of PsAvh181 reduces infectivity. N-terminal amino acids 70-95 of PsAvh181 are necessary for its plasma membrane localization and virulence function. In addition, a PsAvh181 mutant

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection containing amino acids 70-95 of PsAvh181 could still localize to the plasma membrane, but lacked virulence function. Collectively, these results showed that amino acids 70-95 of PsAvh181 determine plasma membrane localization, and that plasma membrane localization and the C-terminal effector domain are essential for virulence function of PsAvh181.
Here, we demonstrated that PsAvh181 targets GmSNAP-1, a vesicle trafficking protein, to interfere with the secretion of extracellular proteases. Although some other secretory pathway related proteins were detected in the LC/MS data of PsAvh181 co-immunoprecipitation (S1 Table), only GmSNAP interacts with PsAvh181 in planta. SNAP plays a role in plant resistance during pathogen infection [15,43,44]. We also confirmed that GmSNAP-1 contributes to plant resistance against Phytophthora pathogens. GmSNAP-1 is a component of the SNARE complex. After the SNARE complex helps vesicles fuse to the plasma membrane, SNAP combines with NSF to form SNARE, and provides energy for SNARE complex dissociating and recycling [10,11]. We found that PsAvh181 interacts with GmSNAPs and interferes the interaction between GmSNAPs and GmNSF. Although the colonization in GmSNAPs-RNAi of Avh181D-10 didn't restore to the level of WT completely, the colonization of WT is~2.2 times more than Avh81D-10 in EV samples, and the colonization of WT is~1.5 times more than Avh181D-10 in GmSNAPs-RNAi samples (Fig 6G). This may be due to additional functions of PsAvh181 besides its interference with the interaction of GmSNAP-GmNSF. Mutation of key C-terminal sites in GmSNAP-1 abolished the interaction with GmNSF but not with PsAvh181, indicating that the interaction sites of GmSNAPs with GmNSF and PsAvh181 are likely different.
In a previous study, when the soybean Rhg1 (resistance to Heterodera glycines 1) was infected by Heterodera glycines, high levels of resistance-type α-SNAPs interfere with wildtype α-SNAP activities and disrupt vesicle trafficking, contributing to defense by causing cytotoxicity and cell death [45]. The GmSNAP-1 in this study is equivalent to the wild-type nonresistant allele of Rhg1. It also has been reported that SNARE components contribute to host resistance. The SYP132 syntaxin contributes to plant resistance against bacteria and the secretion of pathogenesis-related protein 1 [16]. The interaction between rice OsSYP121 and OsS-NAP32 may contribute to host resistance to rice blast disease [17]. The GmSNAP and GmNSF complex plays important roles in membrane fusion and vesicle trafficking [11,13]. In our previous study, we found PsAvh181 causes yellowing in N. benthamiana, which can be suppressed by another effector PsAvh172 [30], indicating that Phytophthora may secretes additional effectors to suppress the cell death caused by cytotoxicity. We found that PsAvh181 interfered with the interaction between GmSNAPs and GmNSF, and that the mutant GmSNAP-M3 could not interact with GmNSF and did not contribute to plant resistance. This represents a novel mechanism by which an effector suppresses plant immunity by interfering the interaction between two important components of the SNARE complex. Consistent with the findings described above, silencing NbSNAPs in N. benthamiana inhibited the secretion of apoplastic proteases, including GmGIP1, P69B and PR1, but not another apoplastic protease (GmAP1). This was similar to the function of PsAvh181, which inhibits the secretion of GmGIP1, P69B and PR1, but not GmAP1. The secretory pathway for GmAP1 may differ from that of GmGIP1, P69B and PR1, and P. sojae produces other effectors, such as PsAvh240, to inhibit the secretion of GmAP1. What's more, silencing of NbSNAPs didn't influence the localization of PsAvh181 (S15 Fig), which means the plasma membrane localization of PsAvh181 is independent on the interaction with GmSNAPs. Together, these data showed that PsAvh181 achieves its virulence function by interfering with the SNAP and NSF complex to suppress the secretion of apoplastic proteases.
Several effectors have been shown to inhibit the secretion of apoplastic defense-related proteins to suppress plant immunity [5,8,[25][26][27]. PsAvh181 is the first P. sojae RxLR effector shown to suppress the secretion of GmGIP1, P69B and PR1 by binding to GmSNAP-1, an important component of the SNARE complex. These results demonstrate that interference with secretion of the apoplastic defense-related proteins is a conserved strategy employed by different microbial pathogens to counter host defense.
This study provides novel insight into how plant pathogens modulate plant immunity by interfering with host protein secretion. PsAvh181 inhibits the secretion of plant defense proteins such as GmGIP1, P69B and PR1 by influencing the interaction between GmSNAP and GmNSF (Fig 7D). This finding will facilitate genetic engineering to enhance plant defense by modifying the target of PsAvh181.

Plant and pathogen materials
N. benthamiana and soybean (Hefeng 47) were grown in greenhouses at 25˚C. N. benthamiana plants were grown for about 5 weeks. Soybean plants were grown for 5 days. P. sojae and P. capsici were cultured on 10% V8 medium at 25˚C in dark.

Transformation of P. sojae
Gene deletion mutants of P. sojae were got using CRISPR-Cas9 gene replacement strategy [46]. And the polyethylene glycol-mediated protoplast transformation has been described previously [47]. The PsAvh181 gene ligated with two 1.0-kb fragments flanking the target gene was used as donor DNA in homology-directed repair (S4A Fig), the primers used for plasmid construction were listed in the S2 Table. The sequences of sgRNAs used for PsAvh181 gene deletion are: GGAGCAGCGTCGATACATGT and CATGAAGTAGATCTGCGCGT.

Transient Agrobacterium tumefaciens-mediated protein expression in N. benthamiana
Proteins were expressed in N. benthamiana using the Agrobacterium tumefaciens (GV3101) system. A. tumefaciens strains carrying different vectors were cultured in LB medium at 30˚C and 200 rpm for 16 hours. The transformed A. tumefaciens were incubated in LB medium at 30˚C and 200 rpm for about 16 hours. A. tumefaciens was collected and washed with buffer [10 mM MgCl2, 10 mM MES (pH 5.7), and 100 μM acetosyringone] three times. The A. tumefacien cells were resuspended using the buffer described above and infiltrated into leaves of N. benthamiana at appropriate concentrations. The infiltrated N. benthamiana was maintained in a greenhouse for 24-48 hours and collected for protein extraction.

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection centrifugation at 1,000 × g for 2 min and washed with buffer [10 mM Tris-HCl (pH 7.5), 100 mM NaCl, 0.5 mM EDTA] three times. Protein was eluted by boiling for 5 min and analyzed on sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gels followed by western blot analysis.

Protein pulldown assays
N-terminal His-tagged GmSNAP-1 and MBP-tagged PsAvh181 were expressed in E. coli (strain BL21). Purified proteins were dissolved in 1× phosphate-buffered saline (PBS) buffer with 1 mM phenylmethylsulfonyl (PMSF). His-GmSNAP was incubated with Ni-NTA agarose for 2 h at 4˚C. Beads were added to the total protein extracted from the supernatant of E. coli expressing MBP-PsAvh181 and incubated for 3 h at 4˚C. The Ni-NTA agarose was washed four times using 1× PBS buffer. Proteins were eluted by boiling SDS loading buffer for 5 min, and then analyzed by SDS-PAGE and western blot analysis. His-GmSNAP was detected using an anti-His antibody (Abmart, Shanghai, China). MBP-PsAvh181 was detected using an anti-MBP antibody (CMCTAG).

Q-RT-PCR
RNA was isolated using the Total RNA Kit I (Omega Bio-Tek, Norcross, GA, USA). cDNA was synthesized with HiScript II Q RT SuperMix for PCR (Vazyme Biotech Co., Ltd., Nanjing, China) and then used for qRT-PCR with SYBR qPCR Master Mix (Vazyme) and the primers listed in the S2 Table. Apoplastic fluid collection GmGIP1-HA, P69B-HA and GmAP1-HA were transiently expressed in N. benthamiana. N. benthamiana leaves were collected 48 hours after agro-infiltration and soaked in 1× PBS buffer in a vacuum vessel. Then keep vacuum vessel in -10 psi for 1 min, open the intake valve slowly, infiltrate the 1× PBS buffer infiltrate into the tissue of N. benthamiana leaves. The apoplastic fluid was collected by centrifugation at 4˚C and 1,000 × g for 5 min.

Virus-induced gene silencing (VIGS)
The silencing fragments for SNAPs were designed using the SGN VIGS tool (https://vigs. solgenomics.net), and inserted into the TRV2 vectors via homologous recombination. The obtained TRV2 vectors were transformed into A. tumefaciens (GV3101). For infiltration, A. tumefaciens carrying TRV2: SNAPs or TRV2: EV was mixed with A. tumefaciens carrying TRV1 at a 1:1 ratio, and the concentration was adjusted to an OD600 = 1 for each. The mixed A. tumefaciens suspensions were infiltrated into the cotyledons of the 2-week-old N. benthamiana.

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection Confocal microscopy observation N. benthamiana leaves were collected 48 hours after agroinfiltration and examined using a LSM 710 laser scanning microscope (Carl Zeiss, Jena, Germany). The excitation wavelengths of GFP and RFP were 488 nm and 561 nm, respectively. The emission wavelength of GFP and RFP were 495-530 nm and 600-650 nm, respectively.

Relative abundance statistics
The absolute abundances of bands from western blot were counted by the software Odyssey V 3.0. Relative abundances were absolute abundances of rest bands comparing to the absolute abundance of the first band needed to count abundance in each western blot image. benthamiana. Proteins were co-immunoprecipitated with GFP-Trap_A beads, and the coprecipitation of GmSNAP-1-HA was detected by western blot analysis using anti-HA antibodies. (B) PsAvh181-HA was co-expressed with GFP, GmSNAP or GmSNAP M3 in N. benthamiana. Proteins were co-immunoprecipitated with GFP-Trap_A beads, and the coprecipitation of GmSNAP-HA was detected by western blot analysis using anti-GFP and anti-HA antibodies. (C-F) Infection assays of Phytophthora capsici on N. benthamiana leaves expressing GFP-GmSNAP-1, GFP-GmSNAP-M3 or GFP (negative control). P. capsici was inoculated 48 h after agroinfiltration. The lesions were photographed 48 h after inoculation. Lesion diameter (E) and relative Phytophthora biomass (F) were quantified 48 h after inoculation. Data are the mean ± SEM of five replicates. Different letters indicate statistically significant differences

PLOS PATHOGENS
A Phytophthora sojae effector manipulates host vesicle trafficking to promote infection