Chp8, a Diguanylate Cyclase from Pseudomonas syringae pv. Tomato DC3000, Suppresses the Pathogen-Associated Molecular Pattern Flagellin, Increases Extracellular Polysaccharides, and Promotes Plant Immune Evasion

ABSTRACT The bacterial plant pathogen Pseudomonas syringae causes disease in a wide range of plants. The associated decrease in crop yields results in economic losses and threatens global food security. Competition exists between the plant immune system and the pathogen, the basic principles of which can be applied to animal infection pathways. P. syringae uses a type III secretion system (T3SS) to deliver virulence factors into the plant that promote survival of the bacterium. The P. syringae T3SS is a product of the hypersensitive response and pathogenicity (hrp) and hypersensitive response and conserved (hrc) gene cluster, which is strictly controlled by the codependent enhancer-binding proteins HrpR and HrpS. Through a combination of bacterial gene regulation and phenotypic studies, plant infection assays, and plant hormone quantifications, we now report that Chp8 (i) is embedded in the Hrp regulon and expressed in response to plant signals and HrpRS, (ii) is a functional diguanylate cyclase, (iii) decreases the expression of the major pathogen-associated molecular pattern (PAMP) flagellin and increases extracellular polysaccharides (EPS), and (iv) impacts the salicylic acid/jasmonic acid hormonal immune response and disease progression. We propose that Chp8 expression dampens PAMP-triggered immunity during early plant infection.

that is controlled by the extracytoplasmic function sigma factor HrpL (8). The expression of HrpL is strictly controlled by sigma-54 and cooperatively activated through the enhancer binding proteins HrpRS (8,9). Transcriptional control through HrpL and HrpRS is not limited to the hrp-hrc T3SS cluster but extends to other genes, including some which have unknown roles in P. syringae pathogenicity (10). One of these genes is PSPTO_2907, otherwise known as chp8 (co-regulated with hrp 8) (10), whose role in pathogenicity we have investigated in this study.

Chp8 is embedded in the Hrp regulon, and its expression is activated by plant signals.
A functional genomics analysis of P. syringae pv. tomato strain DC3000 identified chp8 as a novel Hrpregulated gene whose expression was upregulated under Hrpinducing conditions, apparently in a hrpRS-dependent but hrpLindependent manner (10,11). To confirm these findings, we measured the activity of the chp8 promoter in strain DC3000 in the presence and absence of hrpS ( Fig. 1) or hrpL (see Fig. S1 in the supplemental material), respectively. Initially, we measured Chp8 induction in HIM (hrp-inducing medium), since it has been shown to induce hrp-hrc gene expression ( Fig. 1, P hrpL ), presumably by mimicking the nutritionally depleted environment encountered by DC3000 in the apoplast (12,13). However, we could not detect upregulation of chp8 induction in DC3000 in HIM alone ( Fig. 1, P chp8 HIM). We reasoned that chp8 induction may, in addition, require plant-derived signals. Indeed, the activity of the chp8 promoter was markedly increased when DC3000 was grown in a plant cell culture ( Fig. 1, P chp8 plant cells). Recent studies have identified that plants produce flavonoids upon infection with P. syringae pv. tomato DC3000 and that this pathogen is susceptible to the plant flavonoid phloretin (14). To determine whether phloretin affects chp8 induction, we measured the activity of the chp8 promoter in HIM supplemented with phloretin ( Fig. 1, P chp8 phloretin). As shown by the results in Fig. 1, the activity of the chp8 promoter was markedly increased in the presence of phloretin. In line with the requirement of HrpRS for chp8 induction, the positive effect of plant cells and phloretin is diminished in the absence of hrpS (Fig. 1, DC3000⌬hrpS). Extending earlier observations (10,11), these results demonstrate that Chp8 is indeed embedded in the Hrp regulon, suggesting that coregulation of Chp8 and T3SS occurs and that induction is responsive to plant signals, implying a role in the infection process.
Chp8 exhibits a functional c-di-GMP synthase activity in vivo and promotes a sessile lifestyle of P. syringae pv. tomato DC3000. In silico analyses of Chp8 (see Fig. S2 in the supplemental material) infer that it belongs to the diguanylate cyclase (DGC) and/or the phosphodiesterase (PDE) family of proteins. The presence of a GGDEF (characteristic of a DGC) and an EAL (characteristic of a PDE) domain indicates that Chp8 has active cyclic di-GMP (c-di-GMP)-synthesizing (DGC) and/or -degrading (PDE) activities (15)(16)(17). As a second messenger, c-di-GMP often controls the switch between planktonic and sessile lifestyles (15)(16)(17). DGCs, as c-di-GMP producers, promote biofilm formation and decrease motility, while PDEs, as c-di-GMP degraders, promote motility and decrease biofilm formation (15)(16)(17). To determine which of the two opposing activities of Chp8 predominates in vivo, we measured the (i) cellular c-di-GMP levels, (ii) biofilm formation, and (iii) motility of P. syringae pv. tomato DC3000 in the presence and absence (Fig. S3) and upon ectopic expression of wild-type Chp8 and two Chp8 variants with either the DGC or the PDE domain inactivated (Fig. 2). We used ectopic expression instead of phloretin-induced Chp8 expression since phloretin had such a strong effect on the phenotypes tested that it masked any Chp8-specific changes. Consistent with our earlier observations that chp8 induction required plant-derived signals, heterologous ectopic expression was needed to study Chp8 function ex planta ( Fig. 2 and Fig. S3). Strikingly, cells expressing wild-type chp8 (P. syringae pv. tomato DC3000⌬chp8/pSEVAchp8 DGC ϩ PDE ϩ ) showed a marked increase in cellular c-di-GMP ( Fig. 2A), a slightly more extensive biofilm (Fig. 2B), and decreased motility (Fig. 2C), in line with net c-di-GMP production by Chp8 in vivo.
To test the activities of the DGC and PDE domains of Chp8 independently, we replaced the critical signature amino acids GGDEF and EAL (see Fig. S2 in the supplemental material) with alanine to create Chp8 DGC Ϫ PDE ϩ (disrupting the GGDEF but maintaining the integrity of the EAL motif) and, conversely, Ϫ (disrupting the EAL but maintaining the integrity of the GGDEF motif). Mutating the Chp8 GGDEF motif (P. syringae pv. tomato DC3000⌬chp8/pSEVAchp8 DGC Ϫ PDE ϩ ) impairs c-di-GMP production and biofilm formation ( Fig. 2A), demonstrating that Chp8 indeed encodes a functional DGC domain. Interestingly, the Chp8 PDE domain appears to be functional (P. syringae pv. tomato DC3000⌬chp8/pSEVAchp8 DGC Ϫ PDE ϩ ), causing a marked increase in the motility of the cells compared to that of the vector control (Fig. 2C). Remarkably, inactivation of the Chp8 PDE domain (P. syringae pv. tomato DC3000⌬chp8/ pSEVAchp8 DGC ϩ PDE Ϫ ) also interferes with c-di-GMP production ( Fig. 2A) despite an unmodified DGC domain, indicating that both domains are required for maximal c-di-GMP synthase activity of Chp8. However, the DGC domain of Chp8 alone, in the absence of the intact PDE domain, retains its characteristic phe- Promoter activity was reported via production of GFP and expressed as the ratio of fluorescence intensity at 520 nm and OD 600 . P hrpL , cells contain a reporter fusion of the hrpL promoter to gfp; P chp8 , cells contain a reporter fusion of the chp8 promoter to gfp. Error bars show standard errors of the means. Statistical analysis of P chp8 activity using unpaired t test gave results as follows (significant if P value is Ͻ0.05): DC3000 (HIM) versus DC3000⌬hrpS (HIM) was not significant, P ϭ 0.0544; DC3000 (HIM) versus DC3000 (plant cells) was significant, P Ͻ 0.0001; DC3000 (plant cells) versus DC3000⌬hrpS (plant cells) was significant, P ϭ 0.0005; DC3000 (HIM) versus DC3000 (phloretin) was significant, P ϭ 0.0078; DC3000 (phloretin) versus DC3000⌬hrpS (phloretin) was significant, P ϭ 0.0478.

Engl et al.
notypic impact, evident through an extensive biofilm (Fig. 2B) and decreased motility (Fig. 2C). In summary, we conclude that Chp8 is a composite diguanylate cyclase in which both the DGC and PDE domains are active and required for maximal c-di-GMP synthase activity in vivo and that Chp8 is involved in the switch toward a sessile lifestyle of P. syringae pv. tomato DC3000 by promoting biofilm formation and decreasing motility.
Chp8 downregulates flagellin and upregulates EPS production of P. syringae pv. tomato DC3000. The Chp8-dependent changes in motility prompted us to investigate the impact of Chp8 on flagellin production. Flagellin is the principal constituent of bacterial flagella, which confer bacterial motility (18). Flagellin is also one key pathogen-associated molecular pattern (PAMP) used by plants to detect the presence of a pathogen (19)(20)(21)(22)(23). Central to pathogenicity, therefore, is the link between pathogen detection and plant disease resistance via changes in the phytohormone homeostasis (24)(25)(26). Once detected by the PAMP system, flagellin results in the accumulation of the phytohormone salicylic acid   (21)(22)(23). Consequently, Arabidopsis plants that are unable to detect flagellin exhibit more severe disease symptoms and are less resilient to infection (22). Interestingly, we found that the flagellin levels decreased significantly upon the expression of Chp8 in P. syringae pv. tomato DC3000⌬chp8 (Fig. 3A, pSEVAchp8 DGC ϩ PDE ϩ ). The data are fully in line with our phenotypic observations of a Chp8-dependent decrease in the motility of strain DC3000 (Fig. 2) and point toward a role for Chp8 in undermining the SAdependent plant immune system.
The detection of PAMPs, such as flagellin, generates a cytosolic influx of Ca 2ϩ into the plant cell (27). Here, Ca 2ϩ acts as a second messenger modulating SA biosynthesis and SA-dependent immune responses (28). Bacteria, in turn, chelate Ca 2ϩ , suppressing PAMP-triggered plant immunity through the production of polyanionic extracellular polysaccharides (EPS) (29). Notably, EPS production is c-di-GMP dependent and is thus interlinked with DGC action (30). Chp8's DGC activity (Fig. 2) prompted us to assess the impact of Chp8 on EPS production. We utilized the observation that EPS increases the cell's ability to retain Congo red and to form a "wrinkly" colony (31). As shown by the results in ϩ ) retained more Congo red and were markedly more wrinkly in colony morphology than cells lacking Chp8 (Fig. 3B, pSEVA). Taken together, the data show that Chp8 downregulates flagellin and increases EPS production. Chp8 may therefore hinder the detection of P. syringae pv. tomato DC3000 by the plant and so help to circumvent PAMP-triggered immunity and promote DC3000's pathogenicity.
Chp8 promotes P. syringae pv. tomato DC3000's pathogenicity. Our data show that Chp8 is embedded in the same regulon as the T3SS and that its expression is induced by plant signals and causes a decrease in flagellin and an increase in EPS production. Together, these results strongly indicate a role for Chp8 in the pathogenesis of a P. syringae pv. tomato DC3000 infection. To test this proposal, we infected Arabidopsis thaliana plants with strain DC3000 or the DC3000⌬chp8 mutant using a plate-flooding technique (32) and, in each case, monitored plant health postinfection. P. syringae pv. tomato DC3000 relies on motility to enter the apoplast of the host plant through openings on the surface of the leaves (e.g., stomata), and thus, the infectivity of immotile cells is markedly reduced (33)(34)(35)(36). Since Chp8 decreases the motility of P. syringae pv. tomato DC3000, we chose to flood the plants with low-titer bacterial suspensions as an alternative to leaf wounding or infiltration methods that permit passive entry, in order to encourage an infection route that requires an active movement of the bacterial cells into the apoplast through the stomata.
One characteristic symptom of P. syringae pv. tomato DC3000 infection is yellowing (chlorosis) of leaves (2,6). Both strains elicited chlorosis of Arabidopsis thaliana and ultimately caused plant death. However, the prevalence of disease symptoms was markedly delayed upon infection with the DC3000⌬chp8 mutant compared to the results with DC3000 (Fig. 4A), indicating that Chp8 negatively affects the resilience of the plants.
To elucidate the molecular basis of the Chp8-dependent differences in disease progression, we quantified the levels of three key hormones, abscisic acid (ABA), salicylic acid (SA), and jasmonic acid (JA), employed by the plant to modulate its immune response against infection (24,37). ABA regulates plant development in response to abiotic stresses (38,39); it also increases the plant's susceptibility to pathogens, and thus, P. syringae pv. tomato DC3000 employs T3SS effectors during infection to increase ABA (40,41). Accordingly, we observed a marked increase in ABA levels upon infection with DC3000 (Fig. 4B). However, given that similar ABA levels were observed upon infection with DC3000⌬chp8 (Fig. 4B), it seems that the effect Chp8 has on pathogenicity is not associated with the ABA system.
As described above, plants respond to an attack by (hemi-)biotrophic pathogens like P. syringae pv. tomato DC3000 by accumulating SA (42,43) and to herbivores and necrotrophic pathogens by accumulating JA (44). High levels of either SA or JA, which are regulated antagonistically (42), trigger a range of plant immune responses to combat the infection (45)(46)(47)(48). Unsurprisingly, SA accumulated upon infection with both strains (Fig. 4B). Strikingly, however, since larger quantities of SA were recovered from plants infected with DC3000⌬chp8, it would appear that Chp8 restricts SA accumulation ( Fig. 4B and see Fig. S5 in the supplemental material). This is fully in line with our observations that Chp8 decreases flagellin and increases EPS production, which are known to affect SA levels ( Fig. 3) (21)(22)(23)29). Cross-talk between SA and JA has been demonstrated (42,44). During systemic acquired resistance, an initial wave of JA signaling precedes a wave of SA signaling (49), and JA levels then decrease (44,49,50). In agreement with this antagonism between SA and JA (42), we found that JA levels were higher in plants infected with DC3000 ( Fig. 4C and Fig. S5) than in those infected with the DC3000⌬chp8 mutant. Moreover, we conclude that Chp8 decreased JA indirectly through elevated SA levels. Elevated SA would result in negative regulation of JA and explain the observations reported here.
To investigate at what stage during the infection Chp8 is particularly important, we compared apoplast colonization after single and coinfections with P. syringae pv. tomato DC3000 and DC3000⌬chp8 (Fig. 4C). Notably, after stimulating infections with single strains, we recovered significantly more DC3000 cells than DC3000⌬chp8 cells from plants 1 day postinfection (d.p.i.), despite similar initial inoculum densities (Fig. 4C). However, between day 1 and day 2 postinfection, the bacterial load increased similarly for both strains (Fig. 4C). Consistent with this outcome, after a 1:1 coinfection with both strains, the competitive index for DC3000⌬chp8 on day 1 postinfection was only~0.4 (standard error of the mean [SEM], 0.06), but it increased to~0.8 (SEM, 0.15) on day 2 postinfection (Fig. 4C). Apparently, apoplast colonization during early infection events is impaired in DC3000⌬chp8 cells. Chp8, however, appears to have no effect on the survival of DC3000 within the apoplast at later stages of infection (Fig. 4C).
Taken together, our plant infection studies showed that Chp8 specifically affected the SA/JA hormone levels, ultimately affecting P. syringae pv. tomato DC3000 pathogenesis in a manner that was particularly apparent in the early stages of infection.
Extending the results of previous reports (10, 11), we show that Chp8 is embedded in the Hrp regulon in a way that suggests that signal transduction downstream from HrpRS bifurcates into HrpL-dependent (T3SS) and HrpL-independent (e.g., Chp8) pathways. Bifurcation thereby appears to occur in response to nutritional (HrpL-dependent pathway) (12) or plant-derived signals (HrpL-independent pathway).
Chp8 is a composite GGDEF-EAL protein. Despite the overriding activity of the Chp8 DGC domain, its EAL domain retained PDE activity (illustrated by the increased motility of the Chp8 DGC Ϫ PDE ϩ variant) but, more importantly, appeared to functionally interact with the DGC for maximal c-di-GMP production. Similar functional requirements for full DGC and/or PDE activities have also been reported for the composite GGDEF-EAL proteins FimX from Pseudomonas aeruginosa (57) and MSDGC-1 from Mycobacterium smegmatis (58). In addition to the GGDEF motif in the active (A) site, many DGCs also contain a secondary inhibitory (I) site to regulate c-di-GMP production through feedback inhibition (15, 16) (see Fig. S2 in the supplemental material). Recent in silico analyses point to a correlation between I site conservation and the presence of an EAL domain, where 66% of GGDEF-only and 48% of composite GGDEF-EAL proteins contain intact I and A sites (59). The EAL domain may therefore compensate for loss of the I site by directly regulating the output of the GGDEF domain. Since Chp8 also lacks an intact I site (due to replacement of the RXXD motif with an SXXV motif) (Fig. S2), this may explain the cooperative effect of the EAL domain. Our observations that the biofilm and motility phenotypes of wild-type Chp8 resemble an intermediate state between the phenotypes of the Chp8 DGC ϩ PDE Ϫ and Chp8 DGC Ϫ PDE ϩ variants further supports this notion. Interestingly, the role of c-di-GMP in virulence is consistent with the proposed subdivision of the DGC and PDE pool of a bacterial cell to specific physiological processes (60)(61)(62)(63). Hence, this may be why c-di-GMP can negatively affect the virulence of the plant pathogens Xanthomonas campestris (64,65) and Erwinia amylovora (66) and yet positively (by means of Chp8) affect P. syringae pv. tomato DC3000 virulence and disease progression.
The consequences of Chp8 expression are increased c-di-GMP production, decreased flagellin production (and by association, motility), and increased EPS production-ultimately escalating the susceptibility of Arabidopsis thaliana to infection. The mechanism underpinning Chp8-dependent P. syringae pv. tomato DC3000 pathogenicity, i.e., c-di-GMP production resulting in evasion of flagellin-triggered plant immune responses, seems generalizable. Strikingly, the accumulation of interleukin-8, a proinflammatory cytokine involved in innate immunity, is impaired by high c-di-GMP levels upon Salmonella enterica serovar Typhimurium infection of HAT-29 cells (67) and is flagellin triggered in intestinal epithelial cells upon enteroaggregative Escherichia coli infection (68).
Our findings that chp8 expression is responsive to plant signals are supported through recent transcriptional studies which showed that the expression of Psyr_2711, the chp8 homologue in P. syringae pv. syringae B728a, increased during epiphytic growth (69). These data further indicate that Chp8 acts prior to the passage of P. syringae pv. tomato DC3000 through the stomata (in line with reports that phloretin, which we show induces Chp8 expression, is a cuticular flavonoid [70]) during the early stages of infection. Recall that, compared to the results for DC3000, the onset of disease symptoms and apoplast colonization by DC3000⌬chp8 are reduced on day 1 but similar in later stages of infection.
In summary, among the host-pathogen interactions that depend upon complex interplays between pathogen-triggered host immunity and pathogen evasion of the host immune response, Chp8, a composite GGDEF-EAL protein with a net c-di-GMP activity, serves to reduce flagellin production and increase EPS production, thus functioning as a contributor to pathogen survival in this finely tuned balancing act.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Unless otherwise indicated, P. syringae pv. tomato DC3000 and its derivatives were grown at 28°C in King's B (KB) medium or Hrp-inducing minimal medium (HIM) supplemented with 50 g/ml rifampin and additional antibiotics as appropriate (50 g/ml kanamycin and 100 g/ml ampicillin).
Construction of gene deletions in P. syringae pv. tomato DC3000. Markerless P. syringae pv. tomato strains DC3000⌬chp8, DC3000⌬hrpS, and DC3000⌬hrpL were constructed via allelic exchange, utilizing a protocol adapted from reference 71. Briefly,~2-to 700-bp sequences, corresponding to the 5= and 3= flanking regions of the target open reading frame (ORF), were PCR amplified and fused by single overlap extension PCR. The fusions were inserted into pGEM-T (Promega) to generate intermediate pFUSE vectors. A BamHI fragment containing an FRT-flanked kanamycin resistance gene (nptII) was obtained from the pGEM-T-nptII-BamHI plasmid and inserted into the pFUSE constructs to yield pKO plasmids. P. syringae pv. tomato DC3000 was electroporated with
Construction of plasmids. To create the green fluorescent protein (GFP) gene-tagged reporter for chp8 promoter activity, a 648-bp fragment upstream from the PSPTO_2907 start site was cloned into pBBR1MCS-4 (73) containing gfp-mut3, including the rbs30 ribosome binding site and transcriptional terminator (74). For ectopic expression, Chp8 was placed under the heterologous control of a lacI q -P trc module by cloning the PSPTO_2907 coding sequence into pSEVA224 (75) to create pSEVAchp8 DGC ϩ PDE ϩ . This vector also contains an RK2 origin of replication, conferring broad host range and low copy number. The GGDEF and EAL motifs of Chp8 were replaced with alanine residues via sitedirected mutagenesis to specifically inactivate the DGC and PDE domains, respectively, leaving the rest of the protein intact. Plasmid pSEVAchp8 DGC Ϫ PDE ϩ encodes Chp8 GGDEF::AAAAA , while pSEVAchp8 DGC ϩ PDE Ϫ encodes Chp8 EAL::AAA . Sequencing confirmed successful construction of the plasmids. All primers and plasmids used are listed in Table S1.
Measurement of chp8 promoter activity ex planta. Overnight cultures of P. syringae pv. tomato DC3000 strains carrying the pBBR1-P chp8gfp reporter were washed twice with 10 mM MgCl 2 and resuspended in HIM medium with 10 mM fructose to an optical density at 600 nm (OD 600 ) of 0.25. To test the effect of phloretin, cell cultures were supplemented with 1 mM of the flavonoid. Fluorescence was measured simultaneously with cell density (OD 600 ) in a black, clear-bottom 96-well tissue culture plate using a BMG FLUOstar fluorometer (485 nm for excitation, 520 Ϯ 10 nm for emission, gain of 1,000). The fluorescence per unit of cell growth was calculated in triplicate at 20-min intervals over 8 h. Growth was at 25°C with orbital shaking at 200 rpm.
Measurement of chp8 promoter activity in plant cell coculture. A suspension of Arabidopsis thaliana (Landsberg erecta) callous cells was kindly provided by Alessandra Dovoto (Royal Holloway University, London). The suspension was maintained in a 16-h light regimen at 20°C and subcultured 10-fold every 7 days into cell suspension medium (3% sucrose, 0.44% MSMO (Murashige & Skoog medium with minimal organics), 2.7 M 1-naphthylacetic acid, 50 g/liter kinetin solution, pH 5.7). Ten milliliters of plant cell suspension was harvested during stationaryphase growth (approximately 2 to 3 days into the 7-day culture cycle) for use in coculture experiments. The cell suspension medium was replaced with HIM, with or without P. syringae pv. tomato DC3000 cells containing the pBBR-P chp8 -gfp reporter (prepared as described above), during a series of three cycles in which the plant cells were allowed to sediment, after which the medium layer was aspirated. The activity of the chp8 promoter was measured using the fluorescent reporter assay as described above, except that the supplementation of HIM with a 10 mM carbon source was omitted.
Measurement of cellular c-di-GMP. For c-di-GMP measurement, bacterial cells were grown on King's B solid agar plates overnight and resuspended in King's B growth medium at an OD 600 of 1. The suspension was supplemented with 200 ng/ml cyclic XMP (cXMP) as an internal standard. Extraction of c-di-GMP and cXMP was done in acetonitrilemethanol-water (40:40:20, vol/vol/vol) as described previously (76). c-di-GMP analysis was by liquid chromatography-tandem mass spectrometry (LC-MS/MS) (76,77). The LC-MS system was comprised of an Agilent 1100 LC system and an ABSciex 6500 Qtrap MS. c-di-GMP was separated on a Phenomenex Luna C 18 (2) column (100 mm by 2 mm by 3 m) at a temperature of 35°C, utilizing a gradient solvent system comprised of solvents A (10-mM ammonium acetate and 0.1% [vol/vol] formic acid) and B (acetonitrile). The compounds were eluted at a flow rate of 400 ml/ min with a gradient from 100% A to 90% A over 5 min. The column was washed with 70% B for 3 min and re-equilibrated with 100% A. Typically, 20-l injections were used for the analysis. The MS was configured with a Turbo Spray IonDrive source; gas 1 and 2 were set to 40 and 60, respectively; the source temperature was 425°C; and the ion spray voltage was 5,500 V. c-di-GMP and cXMP were analyzed by multiple-reaction monitoring (MRM) in positive mode using the following transitions (mass-tocharge ratio [m/z]), with the collision energies (CE) used shown in parentheses after the transitions: 691.1¡152 (60 eV), 691.1¡248 (50 eV), 691.1¡540 (40 eV), 347¡153 (30 eV), and 347¡136 (60 eV). The declustering potential, exit potential, and collision cell exit potential were set at 80 V, 10 V, and 10 V, respectively, for all transitions. c-di-GMP and cXMP eluted with retention times of 4.6 min and 5.0 min, respectively. Data acquisition and analysis were done with Analyst 1.6.1.
Biofilm formation. Biofilm formation was assayed as described previously (78). Briefly, P. syringae pv. tomato DC3000 strains were grown in KB in borosilicate glass tubes for 48 h at 28°C without shaking. Planktonic cells were removed, and the OD 600 was measured. Sessile cells bound to the glass were washed twice with water, dried, and stained through a 15-min incubation with 0.1% crystal violet (Sigma) at room temperature. After washing twice with water, the stained cells were resuspended in 75% ethanol and the OD 590 was measured. Biofilm formation was expressed as the ratio between sessile and planktonic cells (OD 590 /OD 600 ). All assays were performed in triplicate.
Motility assay. For motility, cells from overnight cultures were diluted in fresh medium to an OD 600 of 0.05. Five microliters of the bacterial suspension were spotted onto soft agar plates containing 0.4% agar. The plates were incubated at 28°C for 48 h, and swarming across the plate was measured as the diameter of spread. All assays were performed in triplicate.
Flagellin quantification. Flagellin production was measured via immunoblotting with antibodies against FliC as described previously (79). Briefly, bacteria were pelleted by centrifugation at 5,000 rpm for 10 min. To extract flagellin, pellets were washed and resuspended in 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Tween 20 (vol/vol), 10% glycerol (vol/ vol), 1 mM phenylmethylsulfonyl fluoride. Cells were sonicated, and proteins precipitated with 5% trichloroacetic acid for 1 h on ice, resuspended in the same buffer, and separated by 12% SDS-PAGE. Flagellin was detected via immunoblotting with anti-FliC antibodies (79) and quantified via densitometry. Gel loading was controlled and quantified via SYPRO ruby protein stain (Molecular Probes).
EPS quantification and colony morphology. Extracellular polysaccharide (EPS) was quantified and colony morphology visualized as described previously (30). Briefly, overnight cultures of P. syringae pv. tomato DC3000⌬chp8/pSEVA and DC3000⌬chp8/pSEVAchp8, respectively, were washed with fresh medium, 40 g/ml Congo red cell stain (Alfa Aesar) was added, and the bacterial suspension was incubated at room temperature for 2 h with shaking. Cells were pelleted, washed, and resuspended in fresh medium. Cells were normalized by protein content, and the absorbance at 490 nm was measured to quantify the amount of Congo red retained by the cells. To visualize colony morphology, cells from overnight cultures were diluted in fresh medium to an OD 600 of 0.05. Five-microliter amounts of the bacterial suspensions were spotted onto solid KB agar plates and incubated at 28°C for 5 days before visualizing colony morphology. All assays were performed in triplicate.
Plant infection assays. Arabidopsis Col-0 seedlings were grown on agar plates composed of 1/2 strength ( 4°C prior to sterilization. Seeds were sterilized as follows: 5 min of 70% ethanol, 5 min of 50% sodium hypochloride, and 4 washes with sterile distilled water (SDW). Seedlings were then grown at 22 Ϯ 1°C and 120 mol photons m Ϫ2 . After 2 weeks, seedlings were flood inoculated as described previously (32). Plates were flooded with 40 ml of 10 mM MgCl 2 containing 0.025% Silwett L-77 and~5 ϫ 10 5 CFU P. syringae pv. tomato DC3000 and/or DC3000⌬chp8 singly or together in a 1:1 ratio for 3 min and then drained. Seedlings were then grown for a further 4 days postinfection (d.p.i.). Seedlings were harvested at 1 and 2 d.p.i. in order to determine in planta bacterial cell counts and seedling chlorosis and quantify JA, SA, and ABA phytohormones. For in planta bacterial cell counts, seedlings were harvested from plates postinfection and surface sterilized in 70% ethanol for 1 min. Leaves were blotted dry and rinsed in SDW for 1 min before being homogenized in 500 l of PBS. Serial dilutions were plated, and CFU per gram of plant determined. All in planta bacterial cell counts were performed in triplicate. To determine chlorosis levels, multiple seedlings were imaged postinfection. Representative images for those taken from each experimental group are shown. ABA, SA, and JA were extracted and quantified as described in reference 37 using seedling material as described in reference 80. Briefly, tissue was harvested and immediately frozen in N 2 . Samples were freeze-dried in a Heto Drywiner DW1.0-60e for 24 h. Samples were extracted in 394 l of extraction solution composed of 25% methanol, 1% acetic acid in water. Internal standards were then added as follows: 2 l jasmonic acid ([ 13 C 2 ]JA, 5 g ml Ϫ1 ), 2 l salicylic acid ([ 2 H 4 ]SA, 100 M), and 2 l abscisic acid ([ 2 H 6 ]ABA, 0.5 g ml Ϫ1 ). A 3 mM tungsten bead was also added. Samples were placed in a Qiagen TissueLyser at 25.5 Hz for 1 min 50 s and incubated on ice for 30 min. Samples were centrifuged at maximum speed, and the supernatant removed. Samples were re-extracted using 400 l of extraction buffer, and both extractions were pooled and transferred to vials for LC-MS/MS analysis. An injection volume of 50 l was used. Analysis was performed on an Agilent 1100LC coupled to an Applied biosystems Q-TRAP LC-MS/MS system. Separation of molecules based on hydrophobicity was achieved using a Phenomenex Luna C 18 (2)  Statistical analysis. Statistical analysis was performed using Graph-Pad Prism software, version 6.