The Type II Secreted Lipase/Esterase LesA is a Key Virulence Factor Required for Xylella fastidiosa Pathogenesis in Grapevines

Pierce’s disease (PD) of grapevines is caused by Xylella fastidiosa (Xf), a xylem-limited gamma-proteobacterium that is responsible for several economically important crop diseases. The occlusion of xylem elements and interference with water transport by Xf and its associated biofilm have been posited as the main cause of PD symptom development; however, Xf virulence mechanisms have not been described. Analysis of the Xf secretome revealed a putative lipase/esterase (LesA) that was abundantly secreted in bacterial culture supernatant and was characterized as a protein ortholog of the cell wall-degrading enzyme LipA of Xanthomonas strains. LesA was secreted by Xf and associated with a biofilm filamentous network. Additional proteomic analysis revealed its abundant presence in outer membrane vesicles (OMVs). Accumulation of LesA in leaf regions associated positively with PD symptoms and inversely with bacterial titer. The lipase/esterase also elicited a hypersensitive response in grapevine. Xf lesA mutants were significantly deficient for virulence when mechanically inoculated into grapevines. We propose that Xf pathogenesis is caused by LesA secretion mediated by OMV cargos and that its release and accumulation in leaf margins leads to early stages of observed PD symptoms.

similar and are additive in effect, Xf accumulation in grapevine leaves does not correlate with local PD symptom severity, and in some infections of grapevine and other Xf hosts, severe necrotic symptoms are associated with only minimal vessel occlusion 5 .
The secretion of virulence factors by pathogens is an important trigger mechanism for many plant diseases. Unlike closely related pathogens from the genus Xanthomonas, Xff lacks the type III secretion system (T3SS) 14 . However, Xanthomonas and Xf have in common a similar type II secretion system (T2SS) for a battery of important extracellular enzymes involved in nutrient acquisition and virulence 15 . In Xff, genes have been identified that code for plant cell wall degrading enzymes (CWDEs) such as polygalacturonase, cellulase, and proteases 14,16,17 . These enzymes may aid Xff migration inside xylem vessels by degrading the pit membrane and releasing the carbohydrates necessary for bacterial growth and survival 17 . Cell wall degradation by CWDEs releases oligosaccharides as a bacterial nutrient source, which can also induce potent plant innate immune responses leading to cell death. Such plant defense responses include production of phytoalexins, fortification of cell walls through callose deposition, oxidative burst, and induction of programmed cell death [18][19][20] .
We report here our analysis of the Xff Temecula1 secretome, including comparison with the bacterial surfaceome and outer membrane and outer membrane vesicle proteomes. The uncharacterized protein PD1703 was identified as the most abundant secreted protein and its characteristics and likely participation in Xff-initiated symptoms of PD are discussed.

Results
Lipases are a highly abundant component of the Xff secretome. The bacterial proteome was investigated to characterize the subcellular localization of Xff proteins as described (Fig. 1A). The identities of soluble supernatant proteins (SSPs) in culture medium were determined after removing bacterial cells and cell fractions through a series of ultracentrifugation and concentration steps. Twenty-four SSPs were identified by LC-MS/MS in the culture supernatant (Table 1). These include many potential plant cell wall-degrading enzymes necessary for nutrient acquisition such as lipases/esterases (PD1703, PD1702 and PD1211), proteases (PD0313, PD0657, PD0950, PD0956 and PD1850) and cellulase (PD0529). SSPs potentially related to cell adhesion (PD1792 and PD0731), bacterial toxicity (PD1427), and pathogenesis (PD1506 and PD0855) were also identified. Separation of SSPs proteins by one-dimensional SDS-PAGE (Fig. 1B) revealed a prominent 40 to 50 KDa band, indicating that a few proteins were highly abundant in Xff secretome. After in-gel digestion, mass spectrometry analysis of this particular band revealed two 42 KDa (PD1703, PD1702) proteins and one 46 KDa (PD1211) putative uncharacterized protein. Outer membrane proteins were extracted using 0.1 M sodium carbonate buffer (pH 11.0) followed by mass spectrometry. To isolate surface peptides, cells (4 × 10 8 cells/mL) were harvested from a four-to six-day-old culture and subjected to tryptic digestion (cell shaving). Peptides were concentrated (5× ) and subjected directly to LC/MSMS. OMVs were purified from the culture supernatant using two ultracentrifugation steps (38,000 × g for 1 h to pellet cell debris followed by 150,000 × g for 3 h for OMVs precipitation). The remaining supernatant was concentrated (∼ 75 to 100× ) using Amicon Ultra-15 3 K filter units to identify SSPs by mass spectrometry. (B) SDS-PAGE 12% resolution of 10 μ g Xff total protein (TP), OMP (Table S2) and SSPs (Table 1). Three putative lipase/esterases (PD1703, PD1702, and PD1211) were identified in the highlighted band, which was composed mostly of PD1703 (LesA) and PD1702. (C) Venn diagram quantifying the number of proteins identified in each subcellular proteome. The major outer membrane protein MopB was found in all samples, although it was very poorly represented in the secretome. The lipase/esterase LesA was found in the secretome, surfaceome and OMV proteome.
Xff secretes LesA as cargo in outer membrane vesicles. Gram-negative bacteria produce outer membrane vesicles (OMVs) that contain biologically active proteins and perform diverse biological functions 23,24 . Production of OMVs by Xff was recently reported elsewhere 25 . The complete OMV protein cargo has been described for many Gram-negative bacterial species 23 , including Xff's closely related pathogen Xanthomonas campestris pv. campestris (Xcc) 26 , but not for Xff. The OMV fraction was negatively stained and examined using transmission electron microscopy (TEM) ( Fig. 2A), revealing the expected vesicles. Our investigation of OMV protein cargo by LC-MS/MS identified 11 proteins (Table 2). Interestingly, the most abundant SSP, LesA (PD1703), was also part of the OMV cargo. Immunoblot analysis confirmed the localization of Xff LesA in the secretome and OMV proteome (Fig. 2C). In addition to LesA, the secreted proteins PD1702, PD0731 and PD1427 were identified in the OMV cargo, indicating that delivery of these proteins in the host could also be mediated by OMVs. Four outer membrane proteins (PD1709, PD1807, PD0313 and PD1283) were identified in the Xff OMV proteome. We previously identified MopB (PD1709) as the major Xff outer membrane protein (OMP) 27 . PD1807 and PD1283 were also found in our OM preparation (Table S2). The presence of MopB in both the Xff OM and OMV was confirmed by immunoblot (Fig. 2D). The elongation factor Tu (EF-Tu), which is a PAMP in many gram-negative bacteria 28,29 , was abundant in the Xff outer membrane fraction, but was not identified in Xff secretome or as part of the OMV cargo (Table S2; Fig. 2E). LesA is localized in the secreted filamentous network. To visualize the distribution pattern of Xff-secreted material, we examined bacterial cells in culture using scanning and transmission electron microscopy (SEM and TEM). Negatively stained cell clusters show that Xff produces cell aggregates embedded in a dense secreted material possibly composed of extracellular polysaccharide (EPS) and proteins that are weakly attached to the bacterial surface (Fig. 3A). The same dense material surrounds planktonic cells (Fig. 3B). SEM analysis of the bacterial culture revealed that Xff secretes a filamentous network of unknown composition similar to the matrix seen surrounding cells aggregates in the xylem vessels of infected grapevines 30 (Fig. 3C,D). TEM analysis of the secreted matrix revealed its close localization to Xff cells in vitro (Fig. 3E). Immunogold labeling and TEM revealed that LesA is embedded in the Xff secreted network (Fig. 3F).  (Table S1). (C-E) Immunoblot detection of LesA, MopB, and EF-Tu in Xff total proteins (TP), outer membrane proteins (OMP), soluble supernatant proteins (SSP), and outer membrane vesicle proteins (OMV). For TP, OMP and SSP (defined in Fig. 1A), 10 μ g protein was loaded in the gel; however, only 2 μ g protein was loaded in OMV due to the high non-protein content of this sample. This may explain the weak detection of LesA in C.
Scientific RepoRts | 6:18598 | DOI: 10.1038/srep18598 LesA accumulates abundantly in leaf regions with minimal Xff titer and is associated with early stages of PD symptom development. Although LesA accumulated in both OMVs and the secreted matrix of Xff cells, its localization in Xff-infected host tissue was unknown. To search for LesA and other potential secreted virulence factors in the infected host, we compared the total leaf proteome of Xff-infected and non-infected grapevines. Of the 524 proteins found, six were of Xff origin (Table 3). LesA (the most abundant secreted protein) and MopB (the most abundant outer membrane protein) were identified in infected host tissue. The surface protein Hsf (PD0744) was found in infected grapevine leaves, although it was not previously identified in the secretome of cultured Xff. This suggests that both expression and secretion of Hsf may be triggered during infection of grapevine leaves. As expected, no Xff protein was found in non-infected grapevine leaves. The accumulation of LesA in Xff-infected host tissue confirmed our hypothesis that LesA is highly expressed and secreted not only in vitro but also during the Xff infection/colonization process. We postulated that LesA has an important role in Xff pathogenesis in grapevines similar to that seen in Xoo, where it elicits callose deposition and programmed cell death in rice 21,31 . We evaluated whether the presence and abundance of LesA correlates with PD symptom development in grapevine leaves 12 weeks post inoculation (wpi). We divided infected leaves into three major areas (inside, middle and outside) corresponding to the pattern of symptom development (Fig. 4A). Viable Xff cells were abundant in the inside portion of infected leaves, but were less abundant in the middle and outside areas, especially close to the leaf margins (Fig. 4B). The most characteristic early PD symptom in grapevines, the marginal leaf necrosis called leaf scorch, was associated with low Xff titer as previously reported 32 . We postulate that Xff could be releasing either LesA-containing OMVs or SSP, or both, from its colonized region near the petiole. These could deliver LesA to the leaf margins, causing PD symptoms. To test this hypothesis, we localized LesA using a polyclonal antibody in the three areas. LesA was slightly more abundant in the leaf edges than the middle and inside portions (Fig. 4C). When the amount of LesA per viable Xff cell was calculated, a difference of ∼ 100-fold (p < 0.05) was observed (Fig. 4D).
LesA elicits a hypersensitive response in grapevine. LipA from Xoo elicits an innate immune response in rice mediated by cell wall degradation, induces callose deposition, and triggers programmed cell death 21,31 . In addition, LipA is required for wild-type levels of Xcv virulence on tomato 22 . Analysis of the high-resolution crystal structure of Xoo LipA revealed the canonical catalytic triad residues Ser-176, Asp-336 and His-377. Ser-176 is embedded in the hydrolase conserved motif Gly-X-Ser-X-Gly. Moreover, mutation in the residue Ser-176 reduced Xoo virulence on rice, confirming that the enzymatic activity of LipA, rather than LipA as a protein, is essential for optimal virulence 21 . In silico analyses mapped LesA from Temecula1 (PD1703) and 9a5c (XF0357) Xf strains on the Xoo LipA structure, assigning the catalytic triad of both Xf LesA molecules to residues Ser-200, Asp-360 and His-402 (Fig. S1). To test whether LesA contributes to Xff virulence in grapevine, we expressed both Xff LesA and its alanine substitution mutant S200A-LesA in E. coli and pressure-infiltrated leaves with protein extract from E. coli. Heterologous expression of both LesA and S200A-LesA constructs was confirmed by immunoblot detection. Extracts from LesA-producing E. coli, but not from the empty-vector control, induced necrosis in the infiltrated area, indicating that LesA may contribute to Xff pathogenesis in grapevines (Fig. 5A,B). Interestingly, infiltrated E. coli extracts producing the Xff S200A-LesA mutant elicited only limited HR-like symptoms compared to wild-type LesA (Fig. 5C), consistent with enzymatic activity of Xff LesA being essential for maximal virulence, as for Xoo LipA.
LesA is required for virulence of X. fastidiosa in grapevines. To test whether LesA has a role in Xff virulence, we generated a lesA mutant using homologous recombination and compared the performance of both proteins after inoculation in grapevines. Our Xff lesA mutant lacked lipase and esterase activities (Fig. 6A,B) and expresses no LesA protein (Fig. 6C). Interestingly, the Xff lesA mutant was predominantly found in the biofilm mode of growth when cultivated in liquid media, unlike the wild-type strain (Fig. 6D), and formed large cell  aggregates (Fig. 6E). The inability of the lesA mutant to elicit typical PD symptoms was confirmed by significantly reduced symptomatic leaves (50%; p < 0.05) observed in infected grapevines compared with plants inoculated with the parental strain (Fig. 6F).

Wild-type Xff and its quorum-sensing mutants have distinct patterns of expression for
lesA. Cell-cell signaling plays an important role in the virulence of many plant pathogenic bacteria. Xff produces the quorum-sensing signaling molecule DSF (diffusible signaling factor) that regulates bacterial pathogenesis. In Xanthomonas campestris pv. campestris (Xcc) and Xanthomonas oryzae pv. oryzae (Xoo), DSF-deficient mutants have reduced virulence 33,34 . Xff rpfF mutants deficient in DSF production possess a hypervirulent phenotype when inoculated into grapevines 30 . The rpfF gene is required for DSF production in both Xff and Xanthomonas species.  Interestingly, Xff rpfC mutants overproduce DSF and have a hyperattachment phenotype, which makes them deficient in virulence and movement in grapevine xylem vessels 35 . For specific virulence-related genes, messenger RNA accumulation for rpfC and rpfF mutants and the wild type was assessed by RT-PCR. The lesA gene was down regulated in the rpfC mutant (Fig. 7).
Agrobacterium harboring X. fastidiosa LesA is hypervirulent. To assess whether LesA increases A.
tumefaciens virulence, we expressed the Xff lesA gene under the control of its own promoter for an in-planta virulence assay (Fig. 8). When inoculated into walnut plants, which are highly susceptible to this pathogen, the infection is usually localized to a region of tumor formation at the site of infection; however, Agrobacterium harboring Xff LesA displayed a systemic necrotic response. The first signs of necrosis appeared in walnut plants inoculated with Agro-A281-LesA at eight weeks post inoculation, but not in plants that were mock-inoculated (PBS) or those that contained an empty vector (Fig. 8A,B). Interestingly, walnut plants inoculated with Agrobacterium expressing Xff LesA were dead at 12 weeks post inoculation (Fig. 8C). When inoculated into grapevines, the same phenotype was not observed (data not shown).

Discussion
PD symptom development has long been thought to result from blockage of xylem vessels by Xf biofilm and associated gels and tyloses, which leads to water stress in the distal parts of the infected plant and to PD 2 . Our results support an alternative mechanism independent of water stress: a phytotoxic effect resulting from the action of a Xff enzyme, the LesA lipase/esterase encoded by the PD1703 locus. Xff LesA is similar to the type II secreted protein LipA, present in all sequenced xanthomonad genomes, and in many other Gram-negative bacteria. LipA from the rice pathogen Xanthomonas oryzae pv. oryzae (Xoo) is a 42 kDa α /β hydrolase fold protein with lipase/esterase function and short-chain specificity. Recently, LipA from Xoo was characterized as a cell wall-degrading enzyme with a carbohydrate-binding domain essential to the protein's virulence function 21 . In Xanthomonas campestris pv. vesicatoria (Xcv), LipA is expressed from an early stage of tomato leaf infection and is required for wild-type virulence 22 . Additionally, treatment with Xoo LipA elicited callose deposition and programmed cell death in rice leaves and roots and an active site serine-to-alanine substitution reduced but did not completely suppress virulence 21 . Is the Xff lipase/esterase secretion associated with the PD pathogenic process? The preponderance of evidence seems to support this hypothesis. The SSP profile identified LesA as the most abundantly secreted protein. The Xff SSP preparations cleaved p-nitrophenol butyrate and tributyrin, but no long chain triacylglycerides (Fig. S2A,B),   Comparison of RNA accumulation in mutants and wild-type by RT-PCR revealed that lesA is regulated by rpfC. RNA was extracted from Xff cells grown in PD3 medium (three flasks/condition; 10 7 -10 8 cells/mL) at 28 °C and 120 rpm. The 16S rRNA gene was used as an endogenous control. Unpaired t test with Welch's correction was used for statistical analysis (onetailed; *p < 0.01). An association is suggested not only by the presence of homologs of PD1703 in Xoo-and Xcv-infected rice 21,31,36 and tomato 22 , respectively, but also by its absence from Xf strain EB92-1, found in elderberry. EB92-1 infects and survives in grapevines for many years but does not cause symptoms and provides effective biocontrol against Xff. A genome draft of the EB92-1 strain revealed that 10 potential pathogenicity effectors were missing, including LesA (PD1703) and another predicted type II secreted enzyme present in our SSP profile, the serine protease PD0956 37,38 . On the other hand, lipase/esterase genes of Xff are present in the genome of an Xf virulent strain responsible for citrus variegated chlorosis (CVC). The CVC 9a5c strain genes XF0357, XF0358 and XF2151 are apparent homologs of the Xff Temecula1 genes PD1703 (LesA), PD1702 and PD1211, respectively. In a previous study, the corresponding CVC proteins were not reported in the Xff 9a5c secretome, possibly because the secreted protein fraction was underrepresented due to depletion by the cell washing procedure used 16,39 . We cannot eliminate the possibility of different secreted virulence factors among various Xf subspecies; however, LesA could also play a role in Xff 9a5c pathogenesis.
Our proteomic analysis (Table 3) revealed that LesA is one of six Xff proteins identified in infected grapevine leaves, clearly indicating that LesA is abundantly secreted by Xff during plant infection/colonization in addition to in liquid culture (Tables 1,2). If LesA is partially responsible for PD symptoms, there should be a spatial association of LesA with the leaf symptom pattern of radial transition from green tissue around the point of petiole attachment to scorching that begins at the leaf margins (Fig. 4A). Previous findings 32 showed a gradual decrease in Xff titer proceeding from the leaf blade center to the leaf margin, counter to the pattern of symptom development. We have demonstrated that on a per-Xff-cell basis, LesA accumulates most abundantly near the leaf margins in a distribution positively correlated with the development of scorch symptoms (Fig. 4D).
As a direct demonstration of the ability of LesA to induce symptoms, LesA was produced in E. coli and a crude extract was pressure-infiltrated into grapevine leaf blade, inducing local necrosis. The empty-vector control did not induce necrosis and the catalytic triad control, S200A-LesA, induced only minute regions of necrosis (Fig. 5), showing further similarity to Xoo LipA 21 . The weak reaction of grapevine leaf to the extract from E. coli expressing Xff S200A-LesA may result from residual lipase/esterase activity. Another serine esterase retained a few percent of the wild-type activity in an active site serine-to-alanine replacement mutant 40 .
The co-localization of LesA and leaf marginal necrosis at a distance from the central leaf region of greatest Xff cell accumulation raised questions about LesA secretion and translocation. Two broad hypotheses were considered: i) secretion and transport of LesA as a soluble protein, or ii) entrapment of LesA in OMVs for transport toward the leaf margins. OMVs are small spherical structures that allow interaction of Gram-negative bacteria with their environment by releasing the vesicular contents adjacent to prokaryotic and eukaryotic cells. OMVs are a vehicle by which Gram-negative pathogens communicate with and intoxicate host cells 24 . The first hypothesis is fragile because if LesA is a phytotoxin, necrotic lesions should occur everywhere, including in proximity to bacterial aggregates in the central part of the leaf. But necrosis appears first at the leaf margins, favoring the second hypothesis of OMV-facilitated transport of LesA toward the leaf margin with subsequent release of the OMV cargo.
The presence of LesA was highly concentrated in the leaf margins and its concentration gradually decreased toward the petiole, where the bacteria are typically found as a biofilm network. SEM and TEM imaging (Fig. 3) identified a major filamentous network formed in the presence of LesA, but at low concentration and well distributed. Biofilm behavior has been well studied in Xff 2 . Biofilms represent mainly dead cells and are down regulated for genes that are associated with disease. In contrast, the planktonic bioform is motile, spreads long distances, and expresses genes associated with disease 2 . Biofilm behavior and its relation to PD have been studied using two specific mutations, rpfF and rpfC. The rpfF gene is responsible for synthesis of diffusible signal factor (DSF). DSF is an unsaturated fatty acid that modulates Xff virulence and biofilm formation by cell-cell signaling. Xff rpfF mutants are deficient in DSF production, possess a hypervirulent phenotype, and are found mainly in a planktonic bioform 30 . Xff rpfC mutants exhibit a hyperattachment phenotype (increased biofilm formation) associated with their inability to migrate in xylem vessels and to cause PD 35 . Interestingly, gene expression analysis of lesA in both rpfF and rpfC quorum-sensing mutants found that lesA is highly down regulated in the non-pathogenic rpfC mutant (Fig. 7). We hypothesize that lesA provides specific nutrients needed to sustain planktonic growth through its cell wall degrading lipase/esterase activity, much like that observed for the pectin degrading activity of pglA 17 . As a consequence, lesA mutants that do not make LesA display a biofilm mode of growth similar to rpfC mutants. We believe that the inability of Xff rpfC mutants to cause PD symptoms in grapevine may be related in part to decreased expression of LesA and other potential virulence factors regulated by rpfC, as observed in Xcc rpfC mutants. Xcc rpfC mutants are deficient in the production of virulence factors such as extracellular polysaccharide (EPS) and in the secretion of extracellular enzymes 52 . Formation of OMVs by Xff is significantly higher in the rpfF mutant, which is disrupted for DSF 53 . This hypervirulent mutant also displayed XadA on the OMV's surface, correlating with our proteomic data showing XadA abundantly present in Xff OMVs (Table 2). If LesA is delivered by OMVs, then increased OMV production may be partially responsible for increased rpfF mutant virulence. We propose that the release of LesA through OMVs to distant sites, without biofilm formation, may be a major cause of tissue destruction and result in relocation of nutrients from the leaf margins toward bacterial colonization sites.
Further corroboration of LesA participation in Xff pathogenesis has been shown through the two-component system of GacS/GacA. This system is involved in environmental signaling and control of secondary metabolites and production of extracellular enzymes. It regulates the virulence of many pathogenic and environmental bacteria, including Xff 54 . GacA is a response regulator that controls various physiological processes and pathogenicity factors through transcriptional activation of genes that regulate pathogenesis, e.g., genes involved in quorum-sensing, toxin production, motility, biofilm formation, and extracellular polysaccharide production [55][56][57][58][59] . Several putative pathogenicity-related genes are regulated by gacA in Xff, including the secreted lipase-esterase LesA. Interestingly, Xff gacA mutants expressed less LesA and developed significantly less severe disease symptoms when inoculated into grapevines 54 . Xff gacA mutants may be deficient in pathogenesis due to low expression of virulence factors such as LesA.
To confirm the correlation between LesA and Xff pathogenicity, we generated a lesA mutant, which induced less severe PD symptoms than the wild-type strain. The lack of virulence of our Xff lesA mutant, similar to those of Xff rpfC, gacA, Xoo, and Xcv lipA mutants 21,22 , supports a significant role for LesA in Xff pathogenicity. In addition, expression of lesA under its own promoter in Agrobacterium tumefaciens propagated a systemic cell death response in inoculated walnut plants. These findings strongly suggest that LesA is part of the Xff repertoire of secreted virulence factors and is directly associated with bacterial pathogenesis. We propose a model for Xf pathogenesis in PD based on the secretion, movement and accumulation of the lipase/esterase LesA in infected grapevine leaves, leading to leaf scorching and chlorosis. In our preferred symptom-inducing mechanism, planktonic forms of Xff secrete LesA by means of OMVs. These vesicles increasingly accumulate in the leaf margins and release their cargo, including LesA, initiating PD symptom development.

Material and Methods
Xff strains and growth conditions. The WT Temecula1 strain of Xylella fastidiosa subspecies fastidiosa (Xff; ATCC 700964) and Xff lesA mutant were grown in PD3 medium 60 with aeration (120 rpm) at 28 °C. Plate cultures were prepared in the same medium with the addition of 1.5% agar. PD3 medium was supplemented with kanamycin (5 μ g/mL) for selective growth of the lesA mutant. The rpfF and rpfC strains used in this study were kindly provided by Prof. Steven E. Lindow.

Isolation of secreted proteins and outer membrane vesicles from culture supernatants. Xff cells
were harvested by centrifugation at 8000 × g for 15 min at 4 °C. The culture supernatant was transferred to 38.5 mL tubes and centrifuged at 38,000 × g for 1 h at 4 °C in a SW28 rotor (Beckman Coulter, USA). The supernatant was collected (the remaining pellet was discarded), transferred to 12 mL tubes and centrifuged at 150,000 × g for 3 h at 4 °C (SW41 Ti rotor, Beckman Coulter). The supernatant, containing SSPs, was concentrated 75 to 100× using Amicon Ultra-15 3 K filters units (Millipore). The pellet, containing OMVs, was resuspended in 300 μ L PBS (pH 7.4) for subsequent SDS-PAGE analysis. For electron microscopy analysis, the vesicle pellet was resuspended in 50 mM HEPES buffer (pH 6.8).
Outer membrane and total protein extraction. Outer membrane protein extraction was performed as described 27 . To prepare total protein extracts, bacteria from a 2 mL culture (four to six days old) were harvested by centrifugation at 8000 × g for 15 min at 4 °C and washed three times with 1 mL washing buffer containing 10 mM Tris-HCl (pH 8.8), 3 mM KCl, 50 mM NaCl, 5 mM EDTA and 1 mM PMSF and were centrifuged for 2 min at 3000 × g. The pelleted cells were then lysed with 200 μ L 10 mM Tris (pH 8.8) with 0.5% w/v SDS, 5 mM EDTA, and 1 mM PMSF. After adding DTT to 100 mM, the sample was boiled 3 min and stored at − 80 °C.
Bacterial surface digestion. Xff cells were harvested by centrifugation at 8000 × g for 15 min at 4 °C and washed three times with 1 mL PBS. A total of 4 × 10 8 cells were resuspended in 100 μ L filtered and sterilized 50 mM ammonium carbonate buffer (pH 7.5). Digestions were carried out with 20 μ g sequencing grade modified trypsin (Promega) in the presence of 5 mM DTT for 15 min at 37 °C. The digestion mixture was centrifuged at 3500 × g for 10 min at 4 °C and the supernatant (containing the peptides) was collected. The trypsin reaction was stopped by adding formic acid to 0.1%. The reaction was filtered using Amicon Ultra Microcon 3 K filters units (Millipore), and the flow-through peptides were held at − 20 °C for later analysis.
Grapevine leaf protein extraction. Grapevine (Vitis vinifera L. cv. 'Thompson Seedless') leaves from Xff-infected (n = 5) and non-infected (n = 5) plants were collected about one meter above the point of inoculation (12 wpi). Plant tissues were flash frozen in liquid nitrogen, lyophilized and kept at -80°C. Proteins were extracted using a phenol extraction procedure as described 61 . Each leaf segment (Fig. 4A) was ground in liquid nitrogen using a pestle and mortar containing 1% (w/w) PVPP. One hundred mg plant material was resuspended in 600 μ L extraction buffer (0.7 M sucrose, 0.1 M KCl, 0.5 M Tris-HCl pH7.5, 0.5 M EDTA, 1 mM PMSF and 2% β -mercaptoethanol). The suspension was homogenized three times (1 min each) using a MM300 TissueLyser (Qiagen). An equal volume of UltraPure buffer-saturated phenol (Life Technologies, USA) was added and the mixture was rehomogenized as described above. After centrifugation at 12,000 × g for 15 min at 4°C, the upper phenol phase was removed and the remaining pellet used for re-extraction in extraction buffer. Proteins were precipitated from the phenol phase using five volumes of 100 mM ammonium acetate in methanol overnight at -20°C followed by centrifugation at 12,000 × g for 15 min at 4°C. Protein pellets were washed four times with 4 mL 100 mM ammonium acetate in methanol and dried 10 min in the hood. Proteins were solubilized with urea buffer (7 M urea, 2 M thiourea, 40 mM Tris, 2% Chaps and 18 mM DTT). The protein concentration was determined according to Bradford's method using BSA as the standard.
Protein preparation and mass spectrometry analysis. Xff SSPs and OMV proteins were resolved by SDS-PAGE prior to in-gel digestion to reduce the amount of non-protein contaminants in the samples. Peptides from the cell shaving (surfaceome) procedure were desalted using Aspire RP30 desalting tips (Thermo-Fisher Scientific) and subjected directly to LC/MSMS analysis. For the in-gel digestion used for SSPs and OMV proteome analysis, gel pieces were washed twice with 100 to 150 μ L 50 mM ammonium bicarbonate (AmBic; pH 8.0), followed by dehydration with acetonitrile (ACN; three to four times the total volume of gel pieces) for 10 to 15 min. Proteins were reduced for 30 min at 56°C in a solution of 10 mM DTT and 50 mM AmBic. Gel pieces were dehydrated again, followed by replacement of ACN by 55 mM iodoacetamide in 50 mM AmBic. Gel pieces were incubated 20 min in the dark at room temperature, followed by two washes with 150 to 200 μ L of 50 mM AmBic for 15 min each. Gel pieces were dehydrated with ACN, dried by speed vacuum centrifugation and subjected to tryptic digestion overnight. Peptides were extracted by adding 60% ACN and 0.1% trifluoroacetic acid (TFA) in water to the gel pieces, followed by sonication for 10 min. The solution containing the peptides was mixed with the supernatant resulting from the tryptic digestion, followed by speed vacuum centrifugation. Digested peptides were then desalted using Aspire RP30 desalting tips and resuspended in loading buffer.
The digested peptides were analyzed using a QExactive mass spectrometer (Thermo Fisher Scientific) coupled with an Easy-LC (Thermo Fisher Scientific) and a nanospray ionization source. The peptides were loaded onto a trap (100 micron, C18 100 Å 5U) and desalted online before separation using a reverse phased column (75 micron, C18 200 Å 3U). The gradient duration for separation of peptides was 60 min using 0.1% formic acid and 100% ACN for solvents A and B respectively. Data was acquired using a data-dependent ms/ms method with a full scan range of 300 to 1600 Da and a resolution of 70,000. The ms/ms method's resolution was 17,500 with an isolation width of 2 m/z with normalized collision energy of 27. The nanospray source was operated using 2.2 KV spray voltage and a heated transfer capillary temperature of 250°C. Raw data was analyzed using X!Tandem and visualized using Scaffold Proteome Software (Version 3.01). Samples were searched against Uniprot databases appended with the cRAP database, which recognizes common laboratory contaminants. Reverse decoy databases were also applied to the database prior to the X!Tandem searches.
For the grapevine proteomic analysis, leaf proteins were precipitated using a ProteoExtract protein precipitation kit (Calbiochem) followed by dehydration overnight in a sterile fume hood. The protein pellet was resuspended in 50 mM AmBic (pH 8.0) and subjected to an in-solution tryptic digestion. Digested peptides were then desalted and subjected to LC/MSMS as described above.
Electron microscopy analysis of outer membrane vesicles. Outer membrane vesicles were resuspended in 50 mM HEPES buffer (pH 6.8) and fixed with 4% paraformaldehyde in 1 M Sorenson's phosphate buffer (pH 7.4). Copper grids (400 mesh) supported with formvar coating were used for electron microscopy. Ten μ L fixed previously 65 . Primers outside of PD1703 were used to confirm double crossover events in transformants using gel electrophoresis and sequencing.

Grapevine infection and disease quantification. Grapevines ('Thompson Seedless') were inoculated
with 20 μ L Xff suspension containing ∼ 2 × 10 7 cells. The plants were inoculated with 10 μ L on the first day and reinoculated with 10 μ L on the second day, with an independently grown Xff culture used for each inoculation. The bacteria were introduced into each plant at three to four inches above the soil using a number 0 insect pin. Twenty-five plants were inoculated with Xff wild type strain, 26 plants with lesA mutant, and 26 plants with PBS (mock inoculation). Quantification of symptoms was performed using 18 randomly selected plants/treatment at 14 wpi. We show the average percent scorching of the vine with standard error, with nodes without leaves and petioles being excluded from this analysis. Leaf Point System: 0-24% scorching of leaf = 0 points, 25-49% = 1 pt, 50-74% = 2 pts, 75-100% = 3 pts. Values for individual leaves were then summed and divided by the total possible scorching (3 * number of leaves) to give one value per vine. One-Tailed T-test with Welch's correction was carried out using GraphPad Prism 5.0 (GraphPad Software Inc., San Diego, CA). A value of p < 0.05 was considered statistically significant.
RNA extraction and real-time RT-PCR. Xff 3A2 (WT), rpfC and rpfF cells were grown in 50 mL PD3 liquid medium (three flasks/condition) until the cultures reached 10 7 to 10 8 cells/mL. RNA was extracted using the miRNeasy Mini Kit (Qiagen, USA) following manufacture's instructions. cDNA was synthesized using SuperScript III First-Strand Synthesis SuperMix (Life Technologies, USA). RT-PCR reactions were performed using TaqMan Universal PCR Master Mix (Life Technologies, USA) on the StepOne Real-Time PCR System (PE Applied Biosystems, USA). The level of gene transcription was normalized to 16S rRNA and expressed as a relative difference. The statistical analysis was performed using GraphPad Prism software, version 5 (Graph-Pad Software, San Diego, USA), using the Unpaired t test with Welch's correction. The data was considered significant when p < 0.05.
Agrobacterium virulence assay in walnut plants. BPROM, a bacterial sigma70 promoter recognition program, was used to predict the potential promoter for lesA. This software predicted one promoter region starting at 229 bps 5′ of the lesA start codon. The lesA open reading frame and the 500 bps 5′ of the intergenic space, which was predicted to include the promoter region for lesA, were cloned into the binary vector pDU97.1005. Expression of LesA was tested by detecting protein activity on a tributyrin agar plate. The presence of LesA activity ensured that the predicted 500 bps 5′ of this gene contain the active lesA promoter region, since no other promoter was present in the binary vector. Subsequently, positive colonies with LesA activity were transformed into a disarmed Agrobacterium strain (EHA101-PCH32) by electroporation. The lesA region and empty vector containing Agrobacterium strains were infected into walnut (Chandler) by creating an incision in the bark on the main stem. Ten μ L resuspended bacterial culture (10 8 cells in PBS) was placed on the incision and wrapped with parafilm. The symptoms were observed at eight and 12 weeks post inoculation.