Cyclic-di-GMP controls Type III effector export and symptom development in Pseudomonas syringae infections via the export ATPase HrcN

TL;DR

The study shows that cyclic-di-GMP regulates effector secretion in Pseudomonas syringae, linking it to disease symptoms and bacterial virulence.

Contribution

The discovery that cyclic-di-GMP binding to HrcN regulates effector secretion and disease symptoms in Pseudomonas syringae.

Findings

Mutating the cyclic-di-GMP binding site in HrcN leads to reduced disease symptoms despite normal bacterial proliferation.

The effector HopAA1–2 is particularly important for symptom development in Pseudomonas syringae infections.

Cyclic-di-GMP binding to HrcN may represent a novel regulatory mechanism for effector secretion.

Abstract

Pseudomonas syringae is a destructive bacterial pathogen that infects a wide variety of plants. Following apoplastic entry, P. syringae uses its type 3 injectisome (T3I) to secrete host-specific effectors into the cytoplasm, enabling tissue wetting and immune suppression and leading to bacterial proliferation, chlorosis and necrosis. P. syringae strains encode dozens of highly specialised effectors, whose composition defines strain specificity and host range. Effective plant infection depends on the tight temporal and hierarchical control of effector delivery through the T3I. Effector secretion is driven by HrcN, an ATPase complex that interacts with the base of the T3I and is essential for plant infection. HrcN binds specifically to the bacterial signalling molecule cyclic-di-GMP, although the impact of binding on T3I function and P. syringae virulence is currently unknown. To address this, we examined the influence of mutating the predicted cyclic-di-GMP-hrcN binding site on plant infection and effector secretion. Despite maintaining effective bacterial proliferation in Arabidopsis thaliana leaves, two hrcN mutants showed severely compromised disease symptoms, a phenotype linked to reduced translocation of a specific subset of T3I-effectors, with HopAA1–2 particularly important for symptom development. We propose that cyclic-di-GMP binding may represent a novel regulatory mechanism for effector secretion during bacterial infections.

Pseudomonas syringae is a bacterial plant pathogen that uses a molecular needle; the type 3 injectisome (T3I), to secrete specialised effector proteins into plant cells. Effector secretion through the T3I is vital for successful infection, enabling bacterial proliferation, immune system suppression and plant tissue degradation. HrcN is an ATPase protein that sits at the base of the T3I, and drives effector delivery through it. We previously showed that HrcN binds specifically to the bacterial signalling molecule cyclic-di-GMP, although the purpose of this is currently unknown.

In this study, we use a combination of genetics, plant infection assays and biochemistry to investigate the role of cyclic-di-GMP in HrcN function. We show that mutations in the HrcN cyclic-di-GMP binding site can effectively decouple bacterial proliferation from disease symptoms during infection, producing rare asymptomatic plant infections. This phenomenon is due to reduced secretion of a subset of T3I-effectors in bacteria with mutated HrcN proteins. Of these compromised effectors, one in particular; HopAA1–2, is shown to be highly important for symptom development. Cyclic-di-GMP binding to HrcN may therefore represent a newly discovered mechanism for controlling effector secretion during P. syringae plant infections.

Figures

(A) AlphaFold3 structural predictions for hexameric HrcN complexes.Left: C-terminus facing; Right: N-terminus facing. (B) Upper panel: Residues mutated in this study, highlighted on the HrcN AlphaFold3 structure. Lower panel: Location of mutated residues (black font) surrounding predicted cdG binding residues in red (orange subunit) and purple (blue subunit). Note that each binding site is predicted to incorporate residues from two adjacent subunits. (C) A. thaliana Col-0 disease phenotypes at 6 days post-infiltration infection with Pto DC3000 strains containing deletions in hrcC/hrcN, alongside mutations leading to residue substitutions surrounding the predicted hrcN cdG binding site and an uninfected control. Leaves taken at random from four independent plants are shown in each case. (D) Anti-HrcN (bottom) Western blot of A. thaliana Col-0 leaf tissue infected with Pto DC3000 containing hrcC/hrcN mutant alleles. Infected leaf tissue was collected three days post-infection. Colony forming units (CFU) recovered from infected leaves are quantified in S1A Fig. The location of HrcN (present in all samples apart from ∆hrcC and ∆hrcN) are indicated. FliI appears as a contaminating band with a slightly higher mass than HrcN. Both protein identities were confirmed by mass spectrometry.

(A) ATPase activity demonstrated for purified Δ1-18 HrcN variants using a PK/LDH ATPase assay. This reaction has been plotted as a Michaelis-Menten saturation curve where [HrcN] is 1 µM. Error bars show standard deviation. (B) Upper panel: Fluorescent-CdG DRaCALA assays with increasing concentrations of purified Δ1-18 WT HrcN (top) and Δ1-18 HrcNG176A (bottom). Fluorescent-CdG is included at a fixed 0.6 µM final concentration. Representative assays are shown in each case. Lower panel: ImageJ quantification of HrcN DRaCALA datasets (n = 3). No significant differences were observed in binding affinity between WT and G176A HrcN variants. Error bars show standard deviation. (C) Competition DRaCALA assays with excess unlabelled nucleotides. Δ1-18 WT HrcN/ HrcNG176A and fluorescent-CdG are present at fixed final concentrations of 1.25 µM and 0.6 µM, respectively. Non-labelled competitor nucleotides, as indicated, were added to a final concentration of 1mM. (D) Fluorescent-CdA DRaCALA assays with increasing concentrations of purified Δ1-18 WT HrcN and Δ1-18 HrcNG176A. Fluorescent-CdA is included at a fixed 0.6 µM final concentration. B-D: Control samples contain either H2O or 10 µM BSA in place of HrcN, as indicated.

(A-B) Visual disease phenotypes of Pto DC3000 infiltrated A. thaliana Col-0 leaves, 6 days post-infection.Pto strains harbour mutant hrcC/hrcN alleles as indicated, alongside A: pBBR4-wspR19 (inducing high cdG levels) and B: pBBR4-bifA (inducing low cdG levels). (C) Disease phenotypes of Pto DC3000 infiltrated A. thaliana bak1-5,bkk-1,cerk1 (bbc) leaves, 6 days post-infection. Pto strains harbour mutant hrcC/hrcN alleles as indicated. Imaged leaves were taken at random from three independent plants in each case.

(A) Pto hrcNG176A effector-CyaA translocation rates shown relative to WT hrcN. (B) Pto hrcNG311A effector-CyaA translocation rates shown relative to WT hrcN. In each case, values above 1 indicate an increase in translocation for a given effector in the hrcNG176A/hrcNG311A mutant compared to WT, values below 1 show a decrease. Values compared were cAMP/µg protein (normalised against non-specific binding controls) in Pto infected A. thaliana Col-0 leaves 6 hours post-infection, where cAMP/µg protein directly correlates with effector translocation. Asterisks (*) denote statistical significance compared to the mean WT value, where ‘*’ = p ≤ 0.05 (unpaired two-tailed t-tests, n = 3). (C) Effector-CyaA fusion reporter assays for key Pto effectors showing cAMP/µg protein (normalised against non-specific binding controls) in Pto DC3000 infected A. thaliana Col-0 leaves 6 hours post-infection, where cAMP/µg protein directly correlates with effector translocation. In each case, asterisks (*) denote statistical significance (one-way ANOVA with Tukey’s multiple comparisons displayed, n = 6) compared to the mean WT value, where ‘*’ = p ≤ 0.05 and ‘*’ = p ≤ 0.01. (D) Effector-CyaA fusion reporter assays showing cAMP/µg protein in Pto DC3000 infected A. thaliana Col-0 leaves over 24 hours post-infection. Top graph: HopH1 effector translocation, bottom graph: HopAA1-2. Pto strains harbour mutant hrcN alleles as indicated.

(A-B) Visual disease phenotypes of Pto DC3000 infiltrated A. thaliana Col-0 leaves, 6 days post-infection.A: Pto strains harbour the hrcNG176A allele/∆hrcN and effector expression plasmids as indicated. pBBR2: empty vector. B: Pto strains harbour the hrcNG176A allele or effector/hrcN deletions as indicated. In each case, control indicates uninfected leaves, and imaged leaves were taken at random from three independent plants.

Full text

Introduction

Pseudomonas syringae is a diverse species complex of Gram-negative foliar plant pathogens, comprising over 100 distinct pathovars [1] with the ability to infect a wide variety of plants, including almost all economically important crops [2]. P. syringae is a widely used model organism for the molecular analysis of plant–microbe interactions [2,3], with the tomato pathogen P. syringae pv. tomato DC3000 (Pto) particularly well-studied due to its ability to infect the model plants Arabidopsis thaliana and Nicotiana benthamiana [4,5]. Upon entry into the apoplast through surface wounds and leaf stomata, P. syringae infections proceed via tissue wetting accompanied by rapid bacterial proliferation, then chlorosis and eventual tissue necrosis [6].

A key organelle in the P. syringae infection process is the type 3 injectisome (T3I), with a functional T3I necessary for plant infection and apoplastic proliferation [7]. This large, multiprotein complex spans the inner and outer membranes, connecting the bacterial cytoplasm to that of its host and enabling the translocation of host-specific protein effectors into the host cytoplasm [8]. Much of our mechanistic understanding of T3I assembly and function is derived from studies of human pathogens such as Escherichia coli, Salmonella and Yersinia, and the secretory system of the bacterial flagellum [9–15]. However, the core T3I components share a high level of structural similarity and a close evolutionary lineage with each other [13], supporting working hypotheses for the functions of many P. syringae T3I components. T3I assembly is tightly regulated, with extracellular subunits exported through the central pore of the lengthening organelle [16].

Recruitment and secretion of type 3 effectors (T3Es) proceeds via the recognition of highly variable secretion signals at the effector N-terminus [17–19] in a chaperone-dependent or independent manner [20]. Effector proteins are then recruited to a sorting platform at the base of the T3I [19,20]. This sorting platform consists of a C-ring (HrcQ and HrpD), export apparatus (HrcRST, HrcU and HrcV) and an ATPase complex (HrpE and HrcN) [10,12]. The HrcN ATPase is a central component of T3I function, whose deletion almost entirely abolishes bacterial cytotoxicity in the plant pathogen Xanthomonas campestris [21].

HrcN forms a dodecamer at the base of the sorting platform [22] and controls substrate entry into the export gate by ensuring that ATP hydrolysis occurs in concert with secretion. It also ensures efficient substrate unfolding in the export channel [11,23,24]. In E. coli and Salmonella species, T3I ATPase monomers comprise a central ATPase domain flanked by an N-terminal domain containing a 5 stranded β-sheet and a predominantly α-helical C-terminal domain [9,14,25]. The cryo-EM structure of the closely related E. coli ATPase EscN reveals an asymmetric multimer, suggesting a rotational catalysis mechanism similar to that of F1 ATPases [14].

T3Es encoded by P. syringae pathovars fulfil diverse roles in plant infection, facilitating bacterial colonisation while repressing the host immune response [1,26]. T3Es can trigger apoplastic wetting [27], manipulate host NAD(+) metabolism [28], suppress plant immune responses [29–33] or interfere with other traits including organelle function, cell development, nutrient distribution and metabolism [34]. P. syringae strains typically contain 15–35 well-expressed T3Es (~36 in Pto [35]), drawing from at least 58 distinct effector families [36,37]. The effector repertoire of an individual strain is highly specialised towards colonisation of its host plant, with these differences one of the main factors defining strains and pathovars [38].

T3E delivery in Salmonella is under tight temporal and hierarchical control [15], and there is some evidence for similar mechanisms in P. syringae. For example, effectors enabling apoplastic wetting show a response in the initial stages of infection [27], while early- and late-stage immune responses are targeted by different T3Es. AvrPtoB targets the PAMP immune receptor FLS2 for degradation [39], while HopM1 disrupts late-stage immunity by inducing degradation of the vesicle trafficking-related protein AtMIN7 [40]. Leaf chlorosis and necrosis are apparently enabled by targeted translocation of a few key T3I effector proteins, rather than with a larger fraction of the effector protein repertoire [41–43]. Coordinated effector delivery to the T3I is enabled by specific hrp-associated (hpa) chaperones [44]. Furthermore, T3I deployment and activity are also subject to extensive regulatory control [16,45].

Our previous research showed that the ubiquitous dinucleotide signalling molecule cyclic-di-GMP (cdG) binds specifically to the ATPase proteins of several bacterial secretory systems, including the flagellar protein FliI and HrcN from Pto [46]. CdG is associated with virulence and infectious persistence in many human and plant pathogens [47] and exerts its effects by binding to diverse protein and RNA targets, which in turn control an array of bacterial traits [48]. In P. syringae, cdG tightly controls the transition between motile, unicellular and communal, biofilm-forming lifestyles [49], with T3I-mediated virulence among the regulated traits [45]. CdG binding to HrcN-like ATPases appears to be widespread, with examples seen with type II, III, IV and VI secretory systems from various species [46,50–52]. At this stage, however, the impact of cdG-HrcN binding on T3I activity and Pto virulence is unknown.

To examine this relationship in more detail, we produced mutants in several Pto hrcN residues predicted to play a role in cdG-binding, and tested their impact on plant infection, protein function and T3I effector secretion. Two hrcN mutants, G176A and G311A, showed severely compromised visual disease symptoms, despite maintaining effective bacterial colonisation of A. thaliana Col0 leaves. This asymptomatic phenotype could be reversed by disrupting Pto intracellular cdG levels, and depends on a functional plant immune response. Using cyaA-fusion effector translocation assays [53], we observed significantly reduced translocation of a subset of T3I-effectors in the G176A mutant compared to WT Pto, with the asymptomatic phenotype linked to impaired HopAA1–2 secretion in particular.

Results

Specific HrcN mutations ameliorate disease symptoms upon P. syringae leaf infection

In our previous work we showed that cdG binds specifically to the HrcN ATPase, with dinucleotide binding sites predicted to lie at the interface between two monomers within the larger dodecameric complex at the base of the P. syringae T3I [46]. At this stage, however, the biological function of this interaction is unknown. To address this, we first highlighted several residues predicted in our original study to contribute to the HrcN cdG binding site on an AlphaFold3 model of P. syringae pv. tomato DC3000 (Pto) HrcN (Fig 1A). The predicted positions of these residues were broadly in agreement with our earlier work (Fig 1B, upper panel) [46], with each predicted either to contribute directly (E208), or to coordinate the position of residues contributing to the dinucleotide binding pocket (F174, G176, G311, L338). The predicted cdG binding site is comprised of several arginine, aspartate and glutamate residues positioned at the interface between the two HrcN subunits (Fig 1B, lower panel), consistent with cdG binding motifs in other proteins [54].

(A) AlphaFold3 structural predictions for hexameric HrcN complexes.Left: C-terminus facing; Right: N-terminus facing. (B) Upper panel: Residues mutated in this study, highlighted on the HrcN AlphaFold3 structure. Lower panel: Location of mutated residues (black font) surrounding predicted cdG binding residues in red (orange subunit) and purple (blue subunit). Note that each binding site is predicted to incorporate residues from two adjacent subunits. (C) A. thaliana Col-0 disease phenotypes at 6 days post-infiltration infection with Pto DC3000 strains containing deletions in hrcC/hrcN, alongside mutations leading to residue substitutions surrounding the predicted hrcN cdG binding site and an uninfected control. Leaves taken at random from four independent plants are shown in each case. (D) Anti-HrcN (bottom) Western blot of A. thaliana Col-0 leaf tissue infected with Pto DC3000 containing hrcC/hrcN mutant alleles. Infected leaf tissue was collected three days post-infection. Colony forming units (CFU) recovered from infected leaves are quantified in S1A Fig. The location of HrcN (present in all samples apart from ∆hrcC and ∆hrcN) are indicated. FliI appears as a contaminating band with a slightly higher mass than HrcN. Both protein identities were confirmed by mass spectrometry.

Next, we constructed a series of mutants, introducing conservative substitutions of each selected residue into Pto hrcN, with the intention to either disrupt cdG binding, or to alter its downstream effects on protein function. We then tested the capacity or each mutant to infect Col0 Arabidopsis plants, relative to WT Pto and ΔhrcC & ΔhrcN negative control strains. Bacterial load was measured for each strain at 2 and 3 days post-leaf infiltration. The ΔhrcC/ΔhrcN negative controls showed little increase in bacterial load from the initial inoculum, consistent with the loss of T3I function (S1A Fig). Meanwhile, a 3–4 log increase in load was observed for WT Pto and all five tested mutants, indicating that the hrcN point mutants do not affect bacterial proliferation in planta. Immunoblotting of infected leaf tissue 3 days post-infiltration confirmed the presence of HrcN in all samples except for ΔhrcC, ΔhrcN and the uninfected plant tissue controls (Figs 1D, S1B). Assembly of the HrcC baseplate is thought to be necessary for HrcN production and structural stability [13], explaining the absence of HrcN in the ΔhrcC control sample.

Six days post-bacterial infiltration with WT Pto, infected leaves displayed systemic chlorosis and tissue necrosis, in contrast to the negative controls where no symptoms of infection were observed. Strikingly, however, while the disease symptoms for hrcN^E208D^, hrcN^L338V^ and hrcN^F174Y^ infections did not differ noticeably from WT Pto, minimal disease symptoms were observed for the hrcN^G176A^ and hrcN^G311A^ mutants, with little visible chlorosis or necrosis in either case (Fig 1C). Observed differences in leaf discolouration were quantified using an average pixel intensity calculation applied to each leaf image using ImageJ (version 1.52a), with highly significant differences in the proportion of discoloured leaf tissue observed between WT Pto, hrcN^G176A^ and hrcN^G311A^ (S1C Fig).

We then complemented the hrcN/hrcC Pto mutations by integrating a WT copy of hrcN (or hrcC) under the control of its native promoter into the neutral chromosomal att::Tn7 site of each strain. Full recovery of tissue chlorosis and necrosis was observed upon leaf infection with the complemented hrcN^G176A^ and hrcN^G311A^ strains, suggesting that the asymptomatic infection phenotype resulted from the hrcN mutation in each case. ΔhrcC and ΔhrcN also displayed WT infection phenotypes when complemented, with no differences in infection phenotypes observed for the other complemented mutants (S1D-S1E Fig).

The G176A mutation does not compromise HrcN ATPase activity or cdG binding

To attempt to probe the relationship between cdG-binding and HrcN function in more detail, we next expressed and purified N-terminal truncated, His6-tagged HrcN alongside G176A, G311A and a P142Q variant whose non-conservative residue substitution is predicted to disrupt protein folding. Removing the N-terminal 18 residues of S. enterica FliI improves in vitro protein stability [55], but abolishes higher-order multimerization [56]. P. fluorescens FliI ∆1–18 retains both ATPase activity and cdG binding ability [46], influencing our decision to purify truncated HrcN variants. Despite this, expression levels of all HrcN variants were low, likely due to heterologous host incompatibility. The G311A variant also expressed poorly and appeared to be slightly insoluble.

WT HrcN, HrcN^G311A^ and HrcN^G176A^ retained ATPase activity, unlike HrcN^P142Q^ and in line with our previous findings [46] (Fig 2A). Unfortunately, given the difficulty of working with HrcN^G311A^, we were forced to restrict our subsequent analysis to the G176A variant only. Next, we confirmed cdG binding to WT HrcN using DRaCALA assays [57] with fluorescent cdG (Fig 2B). Curiously, the HrcN^G176A^ variant also bound cdG (Fig 2B-2C), without apparent loss of binding affinity, in contrast to earlier results seen for P. fluorescens FliI [46]. CdG binding to both HrcN variants was specific; fluorescent cdG binding could be disrupted by the addition of unlabelled cdG, but not by a range of other nucleotides (Fig 2C). Furthermore, neither HrcN variant could bind the fluorescent dinucleotide cyclic-di-AMP (cdA, Fig 2D).

(A) ATPase activity demonstrated for purified Δ1-18 HrcN variants using a PK/LDH ATPase assay. This reaction has been plotted as a Michaelis-Menten saturation curve where [HrcN] is 1 µM. Error bars show standard deviation. (B) Upper panel: Fluorescent-CdG DRaCALA assays with increasing concentrations of purified Δ1-18 WT HrcN (top) and Δ1-18 HrcNG176A (bottom). Fluorescent-CdG is included at a fixed 0.6 µM final concentration. Representative assays are shown in each case. Lower panel: ImageJ quantification of HrcN DRaCALA datasets (n = 3). No significant differences were observed in binding affinity between WT and G176A HrcN variants. Error bars show standard deviation. (C) Competition DRaCALA assays with excess unlabelled nucleotides. Δ1-18 WT HrcN/ HrcNG176A and fluorescent-CdG are present at fixed final concentrations of 1.25 µM and 0.6 µM, respectively. Non-labelled competitor nucleotides, as indicated, were added to a final concentration of 1mM. (D) Fluorescent-CdA DRaCALA assays with increasing concentrations of purified Δ1-18 WT HrcN and Δ1-18 HrcNG176A. Fluorescent-CdA is included at a fixed 0.6 µM final concentration. B-D: Control samples contain either H2O or 10 µM BSA in place of HrcN, as indicated.

The hrcNG176A and hrcNG311A infection phenotypes are linked to bacterial cdG levels and plant immunity

To investigate the hrcN^G176A^ and hrcN^G311A^ asymptomatic infection phenotypes in more detail, we conducted additional rounds of leaf infection with strains over-expressing the well-characterised cdG synthase and phosphodiesterase genes wspR^19^ and bifA [49,58], producing strains with elevated and suppressed cdG levels, respectively. CdG overproduction by the autoactive diguanylate cyclase variant WspR^19^ fully recovered tissue chlorosis and necrosis in both hrcN^G176A^ and hrcN^G311A^ infections (Figs 3A, S2A-S2B), suggesting that excess cdG is able to override the effects of the hrcN mutation in each case. Curiously, symptom recovery was also observed for low-cdG, *bifA-*expressing strains, albeit only partially in the case of hrcN^G176A^ (Figs 3B, S2C-S2D). Disruption of cdG levels did not produce noticeable effects on infections with any of the other tested mutants or control strains.

(A-B) Visual disease phenotypes of Pto DC3000 infiltrated A. thaliana Col-0 leaves, 6 days post-infection.Pto strains harbour mutant hrcC/hrcN alleles as indicated, alongside A: pBBR4-wspR19 (inducing high cdG levels) and B: pBBR4-bifA (inducing low cdG levels). (C) Disease phenotypes of Pto DC3000 infiltrated A. thaliana bak1-5,bkk-1,cerk1 (bbc) leaves, 6 days post-infection. Pto strains harbour mutant hrcC/hrcN alleles as indicated. Imaged leaves were taken at random from three independent plants in each case.

We next examined the impact of hrcN mutation on infections of immune compromised plants. Infection of Col0 bbc plants; missing the bak1–5, bkk-1, and cerk1 immune receptor genes [27], with WT Pto and all five hrcN mutants led to strong infection phenotypes, in contrast to the ΔhrcC, ΔhrcN and water controls (Figs 3C, S3A-S3B). Similar results were seen with Col0 fec plants (with fls2, efr and cerk1 immunity genes deleted [59]). Here, strong infection phenotypes were observed for WT Pto, hrcN^G176A^, hrcN^E208D^ and hrcN^G311A^ in contrast to the ΔhrcC negative control (S3C-S3E Fig). These data suggest that the asymptomatic phenotype of hrcN^G176A^ & hrcN^G311A^ mutants may stem from reduced secretion of effector proteins that target the Arabidopsis immune system.

Pto hrcNG176A and hrcNG311A mutants display compromised T3I effector secretion

To probe the relationship between the hrcN^G176A/G311A^ infection phenotypes and effector secretion, we constructed CyaA-fusion vectors for 23 randomly selected P. syringae effectors [53]. These effector fusions are unfolded until they translocate into plant tissue, where translocation can be detected using ELISA assays for cAMP synthesis [53]. The vectors were transformed into WT Pto, ΔhrcN, hrcN^G176A^ and hrcN^G311A^, plants were infected, and relative effector translocation measured in each case (Fig 4A, 4B). Strikingly, both hrcN^G176A^ and hrcN^G311A^ showed substantial differences in the extent of translocation relative to WT Pto for multiple effectors. No effector translocation was observed for any ΔhrcN strain, as expected.

(A) Pto hrcNG176A effector-CyaA translocation rates shown relative to WT hrcN. (B) Pto hrcNG311A effector-CyaA translocation rates shown relative to WT hrcN. In each case, values above 1 indicate an increase in translocation for a given effector in the hrcNG176A/hrcNG311A mutant compared to WT, values below 1 show a decrease. Values compared were cAMP/µg protein (normalised against non-specific binding controls) in Pto infected A. thaliana Col-0 leaves 6 hours post-infection, where cAMP/µg protein directly correlates with effector translocation. Asterisks () denote statistical significance compared to the mean WT value, where ‘’ = p ≤ 0.05 (unpaired two-tailed t-tests, n = 3). (C) Effector-CyaA fusion reporter assays for key Pto effectors showing cAMP/µg protein (normalised against non-specific binding controls) in Pto DC3000 infected A. thaliana Col-0 leaves 6 hours post-infection, where cAMP/µg protein directly correlates with effector translocation. In each case, asterisks () denote statistical significance (one-way ANOVA with Tukey’s multiple comparisons displayed, n = 6) compared to the mean WT value, where ‘’ = p ≤ 0.05 and ‘**’ = p ≤ 0.01. (D) Effector-CyaA fusion reporter assays showing cAMP/µg protein in Pto DC3000 infected A. thaliana Col-0 leaves over 24 hours post-infection. Top graph: HopH1 effector translocation, bottom graph: HopAA1-2. Pto strains harbour mutant hrcN alleles as indicated.*

Due to the low sample sizes used in these initial screening experiments, the results in Fig 4A and 4B showed a high degree of error. We therefore selected four effectors for more detailed analysis. HopAA1–2 and HopAM1 showed the greatest translocation reductions in hrcN^G176A^ and hrcN^G311A^ respectively and were both strongly reduced in hrcN^G176A^, while HopAF1 showed strongly reduced translocation in hrcN^G311A^ and mildly reduced translocation in hrcN^G176A^. HopH1 plays a role in Pto virulence in Arabidopsis [60], but showed little difference in translocation across any of the tested strains and was included as a negative control. Effector translocation assays with these four proteins were then conducted in WT Pto, ΔhrcN and hrcN^G176A^. In line with our previous findings, absolute levels of effector secretion from hrcN^G176A^ were significantly lower than WT Pto for HopAA1–2, HopAM1 and HopAF1, while no significant difference was observed for HopH1 (Fig 4C). When measured over the course of 24 hours post-infection, the onset of HopAA1–2 translocation was delayed by approximately 3 hours in the hrcN^G176A^ mutant compared to WT Pto, and occurred to a significantly reduced extent. No differences were seen for HopH1 translocation over the same time period (Fig 4D).

hrcNG176A-dysregulated effectors contribute to the onset of infection phenotypes

To determine the impact of reduced effector expression on Arabidopsis infection, hopAA1–2, hopAM1, hopAF1, and hopH1 were expressed in trans in WT Pto and hrcN^G176A^ and their impact on infection phenotypes was assessed (Fig 5). Overexpression of hopAA1–2, hopAM1 and hopAF1 had little or no impact on infection phenotypes (Fig 5A) or bacterial load (S4A-S4B Fig) in WT Pto, but led to a partial recovery of chlorosis and necrosis in hrcN^G176A^. Conversely, overexpression of hopH1 had little effect on infection phenotypes with either tested strain.

(A-B) Visual disease phenotypes of Pto DC3000 infiltrated A. thaliana Col-0 leaves, 6 days post-infection.A: Pto strains harbour the hrcNG176A allele/∆hrcN and effector expression plasmids as indicated. pBBR2: empty vector. B: Pto strains harbour the hrcNG176A allele or effector/hrcN deletions as indicated. In each case, control indicates uninfected leaves, and imaged leaves were taken at random from three independent plants.

Finally, we examined the impact of deleting the compromised effectors on Col0 infections by WT Pto. While our efforts to generate ΔhopAF1 were unsuccessful, we were able to produce Pto ΔhopAA1–2, ΔhopAM1 and a ΔhopAA1–2/hopAM1 double mutant. Strikingly, while hopAM1 deletion produced no obvious effects on Pto infection phenotypes, both hopAA1–2 deletion mutants phenocopied the hrcN^G176A^ strain, with a loss of infection phenotypes (Fig 5B) but no accompanying reduction in bacterial load (S4C-S4D Fig). Together, these data strongly suggest a central role for HopAA1–2 in enabling symptom onset during Pto Arabidopsis infections.

Discussion

The coordinated deployment of T3Es is likely to play an important role in effective T3I-mediated infections of both plants and animals [6,15]. In addition to regulating T3I deployment and activity [16,45], and controlling the delivery of specific T3Es to the injectisome [44], our findings suggest that secretion of specific T3Es during P. syringae plant infections is also controlled by allosteric regulation of T3I function, linked to the second messenger cdG. Conservative point mutations in residues linked to hrcN cdG binding [46] led to markedly altered virulence phenotypes upon A. thaliana infection, with Pto hrcN^G176A^ and hrcN^G311A^ mutants showing few visible signs of disease, despite bacterial apoplastic proliferation remaining unaffected.

The asymptomatic infection phenotypes seen for the Pto hrcN^G176A/G311A^ mutants are unlikely to be the result of compromised T3I assembly or a nonspecific reduction in effector secretion. Bacterial proliferation in the apoplast requires a functional T3I [7], and HrcN was readily detected in planta in both hrcN mutants. Furthermore, ATPase activity remains intact in both HrcN^G176A^ and HrcN^G311A^, and secretion of many T3Es is unaffected relative to wild-type Pto. Together, these results strongly argue for the presence of fully assembled, if compromised, T3Is in the hrcN^G176A/G311A^ mutants.

While the majority of effectors are secreted normally in both mutants, Pto hrcN^G176A^ and hrcN^G311A^ each show disrupted export of a specific subset of T3Es, including several responsible for the onset of disease symptoms such as tissue chlorosis and necrosis. For the HopAA1–2 effector, compromised export persists over time, with levels associated with hrcN^G176A^ markedly lower than for WT over a 24 hour time course (Fig 4D). This result suggests that export of at least some T3Es is permanently reduced, rather than delayed in the hrcN^G176A^ and hrcN^G311A^ mutants. Immunocompromised plants showed wild-type infection symptoms upon infiltration with both hrcN mutants, suggesting that the disrupted effectors function by suppressing elements of the plant immune response, which in-turn counter the destruction of plant tissues during infection.

At this stage, the mechanistic relationship between the HrcN mutants, cdG binding and T3I regulatory control remains unclear. It is possible that the slight loss of ATPase activity seen for the HrcN variants in vitro contributes to altered T3I selectivity. However, Minamino and coworkers showed that only residual ATPase activity is needed for flagellar injectisome functionality [23], and proton motive force can bypass the requirement for FliI ATPase activity entirely [24], arguing against this hypothesis. We were unable to confirm cdG binding to HrcN^G311A^, however, binding to HrcN^G176A^ was not significantly different to that seen for WT HrcN. This is not entirely surprising, as this residue does not directly contribute to the predicted cdG binding interface (Fig 1C). Nonetheless, it suggests that the G176A phenotype may not result from reduced cdG binding affinity directly but instead represents a partial decoupling of cdG binding from its downstream impacts on T3I function. How this takes place is currently unknown, but may involve the disruption of HrcN multimerization, downstream allosteric signaling to T3Es or other secretion targets, or altered interactions with other T3I components.

Wild-type infection symptoms were recovered in both HrcN mutants upon diguanylate cyclase expression. This suggests that high levels of cdG may compensate for reduced HrcN binding affinity or inefficient signal transduction, overcoming the reduced responsiveness of the G176A/G311A mutants. The observed symptom recovery upon phosphodiesterase expression is less clear-cut, but suggests that reduced cdG levels may exert indirect, positive impacts on Pto infectivity. An intriguing possibility is raised by the related human pathogen P. aeruginosa, where T3I production is controlled by cdG, with low levels required for production [52,61]. If Pto behaves similarly, then reduced T3I activity in the HrcN mutants might be countered by a general increase in T3I/T3E production upon phosphodiesterase expression. Further research is needed to fully dissect the relationship between cdG and HrcN complex formation in P. syringae.

The decoupling we observe between leaf disease symptoms and bacterial proliferation in P. syringae infections appears to be a rare phenomenon, with changes to disease symptom severity typically coupled to a corresponding increase or decrease in apoplastic colonisation [62,63]. Symptomless P. syringae infections have previously been reported where the induction of HR in non-host plants was missing [64]. Decoupling of disease severity and colonisation has also been reported for mutants in phytotoxin pathways. Pto strains lacking syringolin A undergo asymptomatic infections of wheat plants, whereas bacterial colonisation is relatively unaffected [65]. A similar reduction in necrotic lesion formation has been also reported for strains missing the phytotoxin coronatine, albeit in this case accompanied by a slight drop in bacterial load [62].

HopAM1, HopAF1 and particularly HopAA1–2 appear to play key roles in mediating Pto symptom development, and their export is post-transcriptionally regulated by HrcN. Using ELISA analysis of T3E-CyaA reporter fusion strains, we showed that delivery of specific effector proteins into leaves, including HopAA1–2, HopAM1, and HopAF1 was significantly compromised in the hrcN^G176A/G311A^ mutants relative to wild-type. While over-expression of all three effectors partially recovered disease phenotypes in infections with the asymptomatic Pto hrcN^G176^ mutant, only hopAA1–2 appeared to be necessary for full establishment of disease phenotypes, with hopAM1 deletion mutants showing similar visual disease symptoms to wild-type Pto.

HopAA1–2, HopAM1 and HopAF1 have all previously been linked to immune system dysregulation, which aligns with our observation of symptom recovery for hrcN^G176A^ infections in immune compromised plants. HopAA1–2 interacts with EDS1 (enhanced disease susceptibility 1) and PBS3, and is hypothesised to play a role in salicylic acid-mediated defence subversion [66]. EDS1 is involved in transcriptional mobilisation associated with resistance pathways [67], while PBS3 protects EDS1 from proteasome-mediated degradation [68]. Reduced levels of both genes were seen upon co-expression with HopAA1–2 [66]. Deletion of a gene cluster containing hopAA1–2 (alongside hopV1, hopAO1, and hopG1) showed strongly reduced Pto virulence in N. benthamiana and A. thaliana plants [60].

HopAM1 shares high Toll/interleukin-1 receptor (TIR) domain homology to NB-LRR receptors, which can hydrolyse nicotinamide adenine dinucleotide (NAD^+^) [69]. It has been shown to induce meristem chlorosis and local cell death in Arabidopsis [70]. EDS1, alongside HSP90.2 and SGT1b; part of a steady-state NLR chaperone complex were shown by genome-wide association mapping to affect HopAM1-induced cell death [70]. HopAM1 has also been linked to virulence enhancement in water-stressed plants by inducing hypersensitivity to abscisic acid [71].

Finally, HopAF1 can suppress plant immunity by blocking ethylene induction through targeting of methionine recycling [72]. HopAF1 showed an interaction with Arabidopsis methylthioadenosine nucleosidase (MTN1 and MTN2) in yeast 2-hybrid screens [72,73]. MTN enzymes are known to be involved in the Yang cycle, a process essential for high levels of cellular ethylene in A. thaliana [72]. HopAF1 has also been shown to suppress production of defensive reactive oxygen species in the phytopathogen Pseudomonas savastanoi [74].

Bacterial cdG signalling relays diverse environmental inputs into integrated phenotypic responses [48]. This control can take place both at the level of transcription, but also by rapidly influencing the activity of existing proteins in the cell, as seen here. We propose that cdG binding to HrcN enables the allosteric control of T3E export for a subset of effectors involved in disease symptom development. Our model suggests that the coordinated deployment of bacterial virulence factors depends not only on transcriptional hierarchy and bacterial proximity to apoplastic cells, but is controlled actively and rapidly in response to signals from the apoplastic environment.

Materials and methods

Bacterial strains and growth conditions

Unless otherwise stated P. syringae strains were grown at 28°C and E. coli at 37°C with shaking. Bacterial strains and plasmids used in this study are listed in S1 Table. King’s B medium (KB) [75] and L medium [76] were used for bacterial cell culture as specified. Antibiotics were used at final concentrations of rifampicin at 50 μg/ml, nystatin at 25 µg/mL, tetracycline at 12.5 μg/ml, kanamycin at 50 μg/mL and chloramphenicol at 25 μg/mL, unless otherwise stated.

Molecular cloning

Cloning was performed following standard molecular biology techniques [76]. Pto gene deletions and hrcN mutants were generated by allelic exchange, using vectors constructed by amplifying the upstream and downstream flanking regions of each mutagenesis site with primers DC3000-HrcN-FWD and DC3000-HrcN-REV plus appropriate internal point mutant primers (e.g., Mut-FWD/REV-DC3000-HrcN, see S2 Table), combining them by ligation/strand overlap extension PCR as appropriate and cloning into the suicide vector pTS1 between restriction sites XhoI and BamHI [77]. Pto cells were transformed by electroporation with the appropriate deletion/mutagenesis vectors following the method described in [78]. Single crossover integrations of each plasmid into the chromosome were selected on tetracycline plates and re-streaked before single colonies were grown overnight in KB media without selection. Following a tetracycline enrichment step after [79], double crossovers were counter-selected by plating serial dilutions onto L agar containing 10% (w/v) sucrose. Deletion/site-specific hrcN mutants were subsequently confirmed by PCR and sequencing using primers FWDExternalHrcN and REVExternalHrcN (S2 Table).

To complement Pto hrcN/hrcC mutations with WT gene copies, the hrcN/hrcC genes and ~200 bp of upstream DNA incorporating the promoter region were amplified using the HrcNComp FWD and REV primers (S2 Table) and cloned into pUC18-miniTn7-Gm between HindIII and BamHI/PstI. The resulting vectors were transformed into Pto mutant strains by electroporation alongside the helper plasmid pTNS2 [80]. Chromosomal integration at the att::Tn7 site for successful transformants was verified by PCR using primers pTN7R and pGlmS-Down [80]. HrcN purification vectors were constructed by amplifying the hrcN alleles synthesised above using primers OE_DC3000_FWD and OE_DC3000_REV plus relevant internal point mutant primers (S2 Table) and cloning the resulting fragments between the NdeI and XhoI restriction sites of a pETM11 expression vector. The effector-CyaA fusion constructs were produced by gateway cloning of effectors and plasmids, as described previously [53]. pCPP5371-CyaA backbone vectors were Gateway cloned with T3E (lacking stop-codons)-pENTR-SD/D-TOPO vectors via an LR reaction to create pCPP5371::T3E + C-terminal CyaA vectors (pDEST) [53]. Effector protein overexpression vectors were produced by amplifying effector genes using primers, e.g., OvExp HopAF1 FWD/REV (see S2 Table for complete list) before cloning the resulting fragments between the KpnI and XhoI restriction sites of a pBBR1MCS-2 expression plasmid [81].

Protein purification

Protein purification was conducted after [79]. Briefly, overnight cultures of E. coli BL21-(DE3) pLysS containing pETM11-hrcN-His_6_ vectors were used to inoculate 2.0 litres LB media (1:50 dilution). These cultures were then grown at 37°C, 220 rpm to an OD_600_ of 0.4, before protein expression was induced overnight at 18°C, 220 rpm with 0.5 mM IPTG. Cells were harvested at 5000 g, 4°C for 10 minutes and re-suspended in 30 mL of ice cold chromatography binding buffer (20 mM HEPES, 250 mM NaCl, 2 mM MgCl_2_, 3% v/v glycerol pH 7.8) supplemented with 1 mg/ml lysozyme, 10 µl DNaseI (Promega) and 1 complete protease inhibitor tablet EDTA-free (Roche), before immediate lysis on ice using an MSE Soniprep 150 ultrasonic disintegrator (15s on, 45s off, 15 cycles). The insoluble fraction was removed by centrifugation (18,000 xg, 35 minutes 4°C). The soluble fraction was loaded onto a HisTrap excel (Cytiva) column equilibrated with binding buffer. Following protein immobilisation, a column wash with 10% elution buffer (20 mM HEPES, 250 mM NaCl, 2 mM MgCl_2_, 3% v/v/ glycerol, 500 mM imidazole pH 7.8) was performed to remove non-specific contaminants. Proteins were eluted over a linear gradient of 10–100% elution buffer. Protein purity was confirmed by SDS-PAGE and concentrations were calculated using Bradford assays [82]. Pooled fractions were desalted using PD-10 (Cytiva) according to the manufacturer’s instructions, with final samples being stored at 4°C in storage buffer (20 mM HEPES, 250 mM NaCl, 2 mM MgCl_2_, 3% v/v glycerol pH 7.8). Proteins remained stable at 4°C for up to 5 days and were taken forward immediately for in vitro analysis.

Protein structural predictions

The hexameric structure of HrcN was modelled by AlphaFold3 [83], using the default settings and specifying 6 monomeric subunits (ipTM = 0.78, pTM = 0.8). The downstream analysis and image generation was performed in Pymol.

Linked pyruvate kinase/ lactate dehydrogenase (PK/LDH) ATPase activity assays

ATPase activity was measured indirectly by monitoring NADH oxidation in a CLARIOstar plate reader (BMG Labtech) at 25 °C. The reaction buffer consisted of 100 mM Tris-Cl (pH 8.0), 2 mM MgCl_2_. Each 100 μL reaction contained 1 μM (final concentration) of purified HrcN, 0.4 mM NADH, 0.8 mM phosphoenolpyruvic acid, 0.7 μL PK/LDH enzyme (Sigma) and was initiated by the addition of 10 μL ATP solutions, 2 replicates of each concentration: 0, 0.125, 0.25, 0.5, 0.75, 1.0, 1.5, 2.0 mM ATP. Enzyme kinetics were determined by measuring A_340_ at 90 second intervals. Kinetic parameters were calculated by plotting the specific activity of the enzyme (nmol ATP hydrolyzed/min/mg protein) versus [ATP] and by fitting the non-linear enzyme kinetics model (Michaelis-Menten) in GraphPad Prism.

DRaCALA nucleotide binding assays

CdG/CdA binding assays were performed as described by Roelofs et al. [57]. Protein samples were incubated for 2 minutes at room temperature with 2’-Fluo-AHC-c-diGMP (BioLog 009) or 2’-Fluo-AHC-c-diAMP (BioLog C 088), in a final concentration of 0.6 μΜ in reaction buffer (25 mM Tris, 250 mM NaCl, 10 mM MgCl_2_, 5 mM β-mercaptoethanol pH 7.5). For the competition experiments, 1 mM of each competing nucleotide was mixed with 1.25 μΜ HrcN and incubated for 30 minutes prior to the addition of the fluorescent di-nucleotides. The tubes were incubated on ice for 10 minutes in the dark. 5 μL of each sample was then spotted on nitrocellulose membrane and left to briefly dry before visualisation using the UV fluorescence mode on a G:Box F3 Imager (Syngene). Average pixel intensity was measured in ImageJ, with the mean pixel intensity subtracted from the background intensity and divided by the area measured. Analysis was performed for three independent replicates.

Plant growth conditions

Col0 A. thaliana plants and immuno-compromised fec (fls2, efr, and cerk1 gene knock-out) and bbc (bak1–5, bkk1–1, and cerk1 gene knock-out) lines [84] were used in this study. Plants were grown in controlled environment horticultural facilities under short day (10-hour) lighting conditions [120–180 μmol m^-2^ s^-2^ light intensity], 70 percent relative humidity, and 22°C growth temperature. All seedlings were grown prior to infection for 4–5 weeks + 1 week 4°C seed vernalisation in mixed Arabidopsis peat (Levington F2 600 ITS Peat, 100 ITS 4mm grit, 196g Exemptor Chloronicotinyl Insecticide). Plant experiments were conducted in containment level 2 controlled environment rooms, under plant health licence 51103/197332/5.

Plant infection assays

Cell cultures were grown overnight in L medium, harvested by centrifugation, then resuspended in 10mM MgCl_2_ and adjusted to a final density (OD_600_) of 0.0002 (equivalent to 10^7^ cells ml^-1^). Cells were gently infiltrated into the leaf apoplast using a 1 ml syringe until leaves appeared wet. Post infiltration, bacterial suspensions were allowed to absorb/dry onto leaf surfaces for approximately 1 h before initial (0 DPI) samples were collected. For calculating bacterial load, two 7mm diameter leaf discs (area 0.384 cm^2^) were collected for each sample at 0, 2 and 3-days post-infection, flash-frozen, and ground using a Qiagen Tissue Lyser II (30 hz, 45 seconds x 2). Lysate was serially diluted 10^-1^ to 10^-5^ and plated onto LB with rifampicin and nystatin. Plates were incubated for 2 days at 28 °C prior to colony counting. Bacterial load was calculated and presented as colony forming units per unit leaf area (CFU/cm^2^). A minimum of three plants were used for each condition and assays were run at least twice independently.

Western blotting of leaf tissue

Two 7mm discs of Pto-infected leaf tissue per sampled plant (3 days post-infection) were incubated with 100 µl SDS-PAGE sample buffer (BioRad) at 100°C for 10 minutes, before the supernatant was collected following centrifugation (16,000 g, 10 minutes, 4°C) to remove the insoluble plant material. Samples were then separated by SDS-PAGE gels and transferred onto a polyvinylidene difluoride (PVDF) membrane (Millipore). Membranes were incubated overnight in blocking solution at 8°C (1x PBS pH 7.4, 0.01% Tween-20, 5% w/v milk powder). Target protein was subsequently detected with 1/10,000 anti-HrcN primary antisera (Minotech Biotechnology) and 1/3000 anti-rabbit secondary antibody (Sigma). Western blots were visualized using ECL chemiluminescent detection reagent on an ImageQuant LAS-500 scanner (GE Healthcare). To verify the identity of non-specific antibody targets, SDS gel slices containing protein samples of interest were washed, treated with DTT and iodoacetamide and digested with trypsin according to standard procedures. Peptides were extracted from the gels and analysed by LC-MS/MS on an Orbitrap Fusion Tribrid Mass Spectrometer (Thermo Fisher, Hemel Hempstead, UK).

ImageJ analysis

The level of chlorosis severity of infected leaves was quantified by average pixel analysis using ImageJ [85]. Images were opened and the leaf surface area was selected using the multi-point tool for each leaf. The measure setting was used to calculate pixel intensity (brightness of a given pixel) over the selected area on a scale of 1–255 (1 showing least pixel intensity and no evidence of chlorosis, 255 showing highest pixel intensity with strong evidence of chlorosis). A minimum of 8 leaves per sample were used to calculate final averages per replicate.

CyaA effector fusion reporter assay

Pto strains containing effector-CyaA fusion constructs (in pCPP5295 stable expression vectors, driven by a hrp-promoter) were grown overnight in KB medium and diluted to an OD_600_ of 0.05 (5 x 10^7^ cfu/mL) in 10 mM MgCl_2_ before infiltration into 4-week-old Col0 leaves as described above. 6 hours post-infection, 4 mm diameter leaf discs from the infected leaves were collected using a leaf corer, flash frozen in liquid nitrogen and ground using a Qiagen Tissue Lyser II (30 hz, 45 seconds x 2). 300 µL of 0.1 M HCl was added to each sample, samples were pelleted by centrifugation (13,000 g, 10 min), vortexed and then pelleted by centrifugation for a further 10 minutes. The supernatant was collected and diluted 1:5 and 1:50 in 0.1 M HCl. A direct cAMP ELISA kit (Enzo) was used to quantify [cAMP] in each sample following the manufacturer’s instructions. Data was analysed by comparing each well against the logarithmic curve generated from the cAMP standards. This value was then divided by the calculated sample protein concentration (via standard Bradford assay) to give pmol cAMP/µg protein.

Supporting information

S1 TableStrains and plasmids used in this study.(DOCX)

S2 TablePrimers used in this study.(DOCX)

S1 FigA. Colony forming units (CFU) recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 hrcC/hrcN mutants at 0, 2 and 3 days post-infection (DPI) as indicated. B. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 hrcC/hrcN mutants after 1 and 3 DPI. 3DPI samples were used for anti-HrcN Western blotting in Fig 1D. C. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 hrcC/hrcN mutants 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. Control indicates uninfected leaf tissue. D. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 hrcC/hrcN complementation strains at 0, 2 and 3 DPI as indicated. E. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 complementation strains 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. Control indicates uninfected leaf tissue. In each case, different hrcC/hrcN alleles are indicated on the X-axis. Error bars show standard error of the mean, and asterisks denote statistically significant differences from the WT/D0 (2 sample t-test) where ‘’ denotes p = ≤ 0.01, ‘’ (‘***’ in B) denotes p = ≤ 0.001. A: n = 4 plants; B, D: n = 3 plants; C,E: n = 8 leaves.(TIF)

S2 FigA. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 *wspR19-*expressing strains at 0, 2 and 3 DPI as indicated. B. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 *wspR19-*expressing strains 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. C. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 *bifA-*expressing strains at 0, 2 and 3 DPI as indicated. D. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 bifA-expressing strains 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. In each case, different hrcC/hrcN alleles are indicated on the X-axis. Error bars show standard error of the mean, and asterisks denote statistically significant differences from the WT/D0 (2 sample t-test) where ‘’ = p ≤ 0.05 and ‘**’ = p ≤ 0.001. A,C: n = 3 plants; B,D: n = 8 leaves.(TIF)

S3 FigA. CFU recovered from A. thaliana bbc (immunocompromised) leaves infiltrated with Pto DC3000 hrcC/hrcN mutants at 0, 2 and 3 DPI as indicated. B. Average pixel intensity analysis for A. thaliana bbc leaves infiltrated with Pto DC3000 hrcC/hrcN mutants 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. C. CFU recovered from A. thaliana fec leaves infiltrated with Pto DC3000 hrcC/hrcN mutants at 0, 2 and 3 DPI as indicated. D. Average pixel intensity analysis for A. thaliana fec (immunocompromised) leaves infiltrated with Pto DC3000 hrcC/hrcN mutants 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. In each case, different hrcC/hrcN alleles are indicated on the X-axis. Error bars show standard error of the mean, and asterisks denote statistically significant differences from the WT (2 sample t-test) where ‘***’ denotes p = ≤ 0.001. A,C: n = 3 plants; B,D: n = 8 leaves. E. Visual disease phenotypes of Pto DC3000 infiltrated A. thaliana fec leaves taken at random from 3 independent plants, 6 days post-infection. Pto strains harbour the hrcC/hrcN alleles as indicated.(TIF)

S4 FigA. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 effector overexpression strains at 0, 2 and 3 DPI as indicated. B. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 effector overexpression strains 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. Control indicates uninfected leaf tissue. C. CFU recovered from A. thaliana Col-0 leaves infiltrated with Pto DC3000 effector deletion mutants at 0, 2 and 3 DPI as indicated. D. Average pixel intensity analysis for A. thaliana Col-0 leaves infiltrated with Pto DC3000 effector deletion mutants 6 days post-infection. Analysis was performed using ImageJ software (version 1.52a) and increased intensity is directly proportional to the extent of leaf yellowing. Control indicates uninfected leaf tissue. In each case, different hrcN/effector mutants are indicated on the X-axis. Error bars show standard error of the mean, and asterisks denote statistically significant differences from the WT/D0 (2 sample t-test) where ‘’ = p ≤ 0.05, ‘’ = p ≤ 0.01, and ‘’ = p ≤ 0.001. A,C: n = 3 plants; B,D: n = 8 leaves.(TIF)

S1 DataRaw data. for all graphs contained in this manuscript.(ZIP)