The N-terminal region of Pseudomonas type III effector AvrPtoB elicits Pto-dependent immunity and has two distinct virulence determinants

Authors


(fax +1 607 255 6695; e-mail gbm7@cornell.edu).

Summary

Resistance to bacterial speck disease in tomato is activated by the physical interaction of the host Pto kinase with either of the sequence-dissimilar type III effector proteins AvrPto or AvrPtoB (HopAB2) from Pseudomonas syringae pv. tomato. Pto-mediated immunity requires Prf, a protein with a nucleotide-binding site and leucine-rich repeats. The N-terminal 307 amino acids of AvrPtoB were previously reported to interact with the Pto kinase, and we show here that this region (AvrPtoB1-307) is sufficient for eliciting Pto/Prf-dependent immunity against P. s. pv. tomato. AvrPtoB1-307 was also found to be sufficient for a virulence activity that enhances ethylene production and increases growth of P. s. pv. tomato and severity of speck disease on susceptible tomato lines lacking either Pto or Prf. Moreover, we found that residues 308–387 of AvrPtoB are required for the previously reported ability of AvrPtoB to suppress pathogen-associated molecular patterns-induced basal defenses in Arabidopsis. Thus, the N-terminal region of AvrPtoB has two structurally distinct domains involved in different virulence-promoting mechanisms. Random and targeted mutagenesis identified five tightly clustered residues in AvrPtoB1-307 that are required for interaction with Pto and for elicitation of immunity to P. s. pv. tomato. Mutation of one of the five clustered residues abolished the ethylene-associated virulence activity of AvrPtoB1-307. However, individual mutations of the other four residues, despite abolishing interaction with Pto and avirulence activity, had no effect on AvrPtoB1-307 virulence activity. None of these mutations affected the basal defense-suppressing activity of AvrPtoB1-387. Based on sequence alignments, estimates of helical propensity, and the previously reported structure of AvrPto, we hypothesize that the Pto-interacting domains of AvrPto and AvrPtoB1-307 have structural similarity. Together, these data support a model in which AvrPtoB1-307 promotes ethylene-associated virulence by interaction not with Pto but with another unknown host protein.

Introduction

Pseudomonas syringae pv. tomato, the causative agent of bacterial speck disease in tomato (Solanum lycopersicum), enters the intercellular spaces of plant tissues via wounds or stomata (Preston, 2000). Recent evidence suggests that a key plant defense at this early stage of bacterial infection is the detection of pathogen-associated molecular patterns (PAMPs) by host pattern recognition receptors, which then activate basal defense responses (Abramovitch et al., 2006a; Chisholm et al., 2006). A counter-strategy of the pathogen appears to be the production and delivery via a type III secretion system of a battery of ‘effector’ proteins, many of which act to suppress basal defenses (Alfano and Collmer, 2004; Buttner and Bonas, 2006; Chang et al., 2004; He et al., 2006; Kim et al., 2005; Li et al., 2005). P. s. pv. tomato DC3000, for example, expresses approximately 30 type III effector proteins that share little or no sequence similarity, although, based on mutagenesis studies, many of them appear to have redundant functions in promoting bacterial virulence (Badel et al., 2003, 2006; Chang et al., 2005; Lin and Martin, 2005; Mudgett, 2005; Petnicki-Ocwieja et al., 2002; Schechter et al., 2004). Plants have evolved resistance (R) proteins that either directly or indirectly detect the presence of certain effector proteins, and thereby activate a strong immune response typically associated with rapid localized cell death (the hypersensitive response, HR) (Greenberg and Yao, 2004; Lam et al., 2001). As an apparent counter-attack to this defense response, some effectors have evolved to act either at or downstream of R protein recognition to impede signaling events leading to the HR (Abramovitch and Martin, 2004; Abramovitch et al., 2003; Jamir et al., 2004; Rosebrock et al., 2007). This emerging view of plant–pathogen interactions highlights the importance of understanding the structural and functional aspects of virulence activities of pathogen effector proteins, the host targets of these proteins, and the mechanisms by which R proteins recognize avirulence effectors and activate defense responses.

The ‘guard’ hypothesis of plant disease resistance postulates that R proteins detect virulence activities of effectors via their structural features or enzymatic functions (Innes, 2004; Van der Biezen and Jones, 1998). If this is the case, then the avirulence determinants of effectors are expected to be identical to their virulence determinants. Otherwise, an effector could evolve to retain virulence activity while evading R protein detection and thereby defeat host defenses. This prediction is supported by studies that have relied upon either random or targeted mutagenesis to identify amino acid residues required for avirulence and to assess their contribution to virulence (Lim and Kunkel, 2004; Ong and Innes, 2006; Shan et al., 2000a; Yang et al., 2005). For example, mutant forms of the effector AvrB that are no longer recognized by the RPM1 or Rpg1-b proteins of Arabidopsis and soybean, respectively, generally lose virulence activity in ecotypes of those species that lack the cognate R gene (Ong and Innes, 2006). However, there are several reports of instances where avirulence and virulence determinants were found to be separable (Gassmann et al., 2000; Lim and Kunkel, 2004; Yang et al., 2005). For example, a naturally occurring mutant allele of avrBs2 no longer elicits Bs2-mediated disease resistance in pepper and yet is unaffected in virulence activity (Gassmann et al., 2000). Because the functionally important plant virulence targets of bacterial effectors are mostly unknown, and the mechanisms by which effectors are recognized by R proteins are unclear, it has been difficult to directly examine whether R proteins are detecting effector virulence determinants.

The interaction of tomato leaves with P. s. pv. tomato provides an excellent model for investigating these questions because much is known in this pathosystem about both the effector proteins and the host components that mediate recognition of these effectors (Pedley and Martin, 2003). Resistance in tomato to P. s. pv. tomato is conferred by the Pto serine/threonine protein kinase which is encoded by a member of a small, clustered family of five genes (Martin et al., 1993; Pedley and Martin, 2003). Embedded within this Pto gene family is the Prf gene, which encodes a nucleotide-binding site/leucine rich-repeat (NBS-LRR) protein characteristic of the largest class of plant R proteins (Salmeron et al., 1996). The Pto kinase physically interacts with the P. s. pv. tomato effector proteins AvrPto or AvrPtoB (also named HopAB2), and mutagenesis of Pto has revealed a critical requirement for residues in the kinase P loop for both of these interactions (Frederick et al., 1998; Kim et al., 2002; Scofield et al., 1996; Tang et al., 1996). The mechanism by which the Pto/AvrPto or Pto/AvrPtoB interaction triggers host defenses is unknown, but the region of Pto that binds these proteins also appears to serve as a negative regulator of Prf-mediated Pto defense activation and the effector proteins might interfere with this regulation in some way (Pedley and Martin, 2003; Wu et al., 2004). Interestingly, there are a few differences in the Pto amino acid residues required for recognition of AvrPto or AvrPtoB, suggesting that some structural specificity may exist for these R protein–effector protein interactions (Bernal et al., 2005; Wu et al., 2004).

AvrPto is an intensively studied 18 kDa type III effector that, after delivery by P. s. pv. tomato into the plant cell cytoplasm, appears to be targeted to the plasma membrane by host-mediated myristylation of its N-terminus (Shan et al., 2000b). In susceptible plants, AvrPto promotes bacterial virulence, as indicated by a 10- to 100-fold increase in growth compared with isogenic P. s. pv. tomato strains lacking the effector (Chang et al., 2000; Shan et al., 2000a). Over-expression of AvrPto in stable Arabidopsis transformants interferes with cell-wall-based basal defenses (Hauck et al., 2003). Host-mediated phosphorylation of serine 149 in the AvrPto C-terminus contributes to this virulence activity and also affects AvrPto avirulence activity (Anderson et al., 2006). Extensive random mutagenesis of AvrPto has revealed that a group of clustered amino acids play a key role in its interaction with the Pto kinase (Chang et al., 2001; Shan et al., 2000a). Determination of the NMR solution structure of AvrPto showed that most of these residues lie in a 19-residue Ω loop (referred to hereafter as the CD loop) lying between helices C and D of the four-helix structure that comprises the core of the protein (Wulf et al., 2004). Mutation of some of the residues in this loop was previously reported to disrupt avirulence but not the virulence activity of the effector (Shan et al., 2000a).

AvrPtoB is a 59 kDa modular protein with a C-terminal region that functions to suppress R protein-associated HR (Abramovitch et al., 2003). The three-dimensional structure of this C-terminal region was recently determined and found to contain a domain that structurally mimics host E3 ubiquitin ligases (Abramovitch et al., 2006b; Janjusevic et al., 2006). In contrast, two distinct avirulence domains are encoded by the N-terminus. One is a Pto-dependent recognition domain consisting of amino acids 1–307 [previously termed Δ7 (Abramovitch et al., 2003), and now referred to as AvrPtoB1-307]. The other is the Rsb-dependent recognition domain consisting of amino acids 308–387 (Rosebrock et al., 2007) (the region comprising amino acids 1–387 was previously termed Δ6, and is now referred to as AvrPtoB1-387). The AvrPtoB1-387 region lacks sequence motifs to suggest function and its structure is unknown. The possible avirulence or virulence activity of AvrPtoB1-307 or AvrPtoB1-387 when expressed in P. s. pv. tomato has not been previously reported. An initial alignment found little overall similarity in amino acid sequences between AvrPtoB and AvrPto; however, four residues (GINP) that are present in the AvrPto CD loop are conserved in AvrPtoB (Kim et al., 2002). Mutation of two of these residues (G325, I326) in full-length AvrPtoB decreased, but did not abolish, interaction with Pto (Kim et al., 2002). Later work showed that AvrPtoB1-307, which lacks the GINP motif, can still interact with Pto (Abramovitch et al., 2003). Thus we have hypothesized that the GINP residues may contribute to Pto binding, along with other residues in AvrPtoB1-307 (Abramovitch et al., 2003).

Recently, AvrPtoB has been shown to have different virulence activities in different plant hosts. In susceptible tomato plants, AvrPtoB promotes disease-associated host cell death by enhancing ethylene production via up-regulation of ethylene biosynthesis genes (Cohn and Martin, 2005). In Arabidopsis, AvrPtoB suppresses basal defense by blocking PAMP-mediated early defense gene induction and MAP kinase signaling (He et al., 2006; de Torres et al., 2006). Finally, AvrPtoB has been found to suppress several other PAMP-elicited defense responses in Nicotiana benthamiana (Hann and Rathjen, 2007). Although the contribution of the various domains of AvrPtoB to these diverse activities is mostly unknown, it is clear that at least the basal defense suppression activity in Arabidopsis does not require the E3 ligase activity of AvrPtoB, suggesting this activity is encoded by the N-terminal region of AvrPtoB (He et al., 2006).

Here we investigate recognition of the N-terminal region of AvrPtoB by the Pto kinase and the role of this region in promoting P. s. pv. tomato virulence. We present evidence that the AvrPtoB N-terminal region promotes virulence by two different mechanisms. One of them, involving the first 307 amino acids of AvrPtoB, promotes the production of ethylene, which has been previously reported to be required for enhancement of bacterial speck disease (Cohn and Martin, 2005). The other virulence activity, which requires amino acids 308–387, acts by suppressing PAMP-induced basal defenses. We show that Pto does not appear to be a virulence target of AvrPtoB as neither of these activities is dependent on Pto.

Results

An N-terminal region of AvrPtoB has both avirulence and virulence activity

We previously reported that the N-terminal 307 amino acids of AvrPtoB (previously called Δ7; hereafter AvrPtoB1-307) is sufficient for the yeast two-hybrid interaction of AvrPtoB with Pto, as well as for elicitation of Pto-dependent cell death in an Agrobacterium-mediated transient assay (Abramovitch et al., 2003). To determine whether AvrPtoB1-307 is able to elicit Pto-mediated resistance when expressed by P. s. pv. tomato, we introduced avrPtoB1-307 under the control of the avrPtoB native hrp promoter into the strain DC3000ΔavrPtoΔavrPtoB (carrying deletions of both avrPto and avrPtoB; Lin and Martin, 2005). The resulting strain was infiltrated into RG-PtoR tomato leaves (Pto/Pto, Prf/Prf), which express a functional Pto pathway. DC3000ΔavrPtoΔavrPtoB strains carrying full-length avrPtoB or an empty vector served as positive and negative controls, respectively. A longer N-terminal fragment, AvrPtoB1-387 (previously called Δ6; hereafter AvrPtoB1-387), that elicits Rsb immunity was also tested. DC3000ΔavrPtoΔavrPtoB strains harboring avrPtoB1-307, avrPtoB1-387 or full-length avrPtoB elicited resistance on RG-PtoR plants as evidenced by the lack of bacterial speck disease symptoms and limited bacterial populations 4 days after inoculation (Figure 1a,b). In contrast, DC3000ΔavrPtoΔavrPtoB carrying an empty vector caused typical symptoms of speck disease, and the bacterial population reached a high level (108 cfu cm−2) after 4 days. AvrPtoB1-307 is therefore a minimal N-terminal region that has full avirulence activity on P. s. pv. tomato in Pto-expressing tomato leaves.

Figure 1.

 The N-terminal region of AvrPtoB has both avirulence and virulence activities in Pseudomonas syringae pv. tomato.
(a) Disease symptoms or immunity on leaves from RG-PtoR tomato plants after inoculation with DC3000ΔavrPtoΔavrPtoB carrying either empty vector, or plasmids expressing AvrPtoB, AvrPtoB1-307 or AvrPtoB1-387. Leaves from a similar position on each plant were detached, de-pigmented, and photographed 4 days after inoculation.
(b) Bacterial populations of leaves as in (a). P. s. pv. tomato strains were vacuum-infiltrated into RG-PtoR tomato plants with an inoculum of 104 cfu ml−1. Error bars represent the SE of three replicates. Experiments were repeated three times with similar results.
(c) Disease symptoms on RG-prf3 tomato plants 4 days after inoculation with the P. s. pv. tomato strains as in (a). The arrows indicate severe necrosis observed on lower leaves.
(d) P. s. pv. tomato populations in RG-prf3 leaves at 0 and 2 days after inoculation using a starting inoculum of 104 cfu ml−1. Note that older leaves were sampled and that severe necrosis of these leaves prevented sampling later than day 2. Data from three independent experiments were used for statistical analysis using the SAS general linear model procedure. Values marked by the same letters were not significantly different from each other based on a least significant difference test (= 0.05). Error bars indicate SE.
(e) Disease symptoms on RG-pto11 tomato plants 4 days after inoculation with the P. s. pv. tomato strains as in (a). AvrPtoB1-387 was not used in this experiment because it activates Rsb immunity in RG-pto11 (Abramovitch et al., 2003). The arrows indicate severe necrosis observed on lower leaves.
(f) P. s. pv. tomato populations and statistical analysis in RG-pto11 leaves as in (d).

To test whether the N-terminal region of AvrPtoB contributes to P. s. pv. tomato virulence, DC3000ΔavrPtoΔavrPtoB strains carrying avrPtoB1-307, avrPtoB1-387, full-length avrPtoB or an empty vector were infiltrated into RG-prf3 tomato leaves (Pto/Pto prf/prf). RG-prf3 plants have a 1 kb deletion in the Prf gene, rendering the plants susceptible to avirulent P. s. pv. tomato strains despite their having a functional Pto gene (Salmeron et al., 1996). P. s. pv. tomato disease symptoms were monitored 4 days after infiltration. At the low inoculum level of 104 cfu ml−1, which allows us to distinguish relatively subtle differences in disease severity, DC3000ΔavrPtoΔavrPtoB strains expressing AvrPtoB1-307, AvrPtoB1-387 or AvrPtoB caused more severe disease symptoms on RG-prf3 plants compared with the empty vector control strain (Figure 1c). Specifically, on plants infiltrated with these strains, the lower four leaves developed extensive necrosis, and severe disease later occurred on the upper leaves (red arrows in Figure 1c). To assess whether this AvrPtoB N-terminus virulence activity is dependent on Pto, we inoculated a tomato line, RG-pto11 (pto/pto Prf/Prf), that lacks a functional Pto kinase, and visually assessed disease symptoms. As on the RG-prf3 plants, DC3000ΔavrPtoΔavrPtoB strains expressing either AvrPtoB or AvrPtoB1-307 caused substantially worse disease symptoms than the empty vector strain (Figure 1e) [note AvrPtoB1-387 was not used in this experiment because it activates Rsb immunity in RG-pto11 (Abramovitch et al., 2003)].

We assessed bacterial growth of the strains in leaves of RG-prf3 and RG-pto11 plants and found that, at 2 days after infiltration, the DC3000ΔavrPtoΔavrPtoB strains expressing AvrPtoB, AvrPtoB1-387 or AvrPtoB1-307 grew to statistically significantly higher levels than the empty vector control strain, demonstrating that AvrPtoB1-307, like intact AvrPtoB, accelerates bacterial growth (Figure 1d,f). Taken together, these data show that AvrPtoB1-307 is a minimal N-terminal region that has both full avirulence activity in Pto/Prf-expressing resistant tomato lines, and, significantly, also full virulence activity in susceptible lines independent of Pto or Prf. Based on this observation, we focused our initial attention on this region for further study of its avirulence and virulence activities.

Amino acids 121–200 of AvrPtoB are sufficient for interaction with Pto in yeast

To investigate the specificity underlying AvrPtoB/Pto-mediated resistance, we generated a series of avrPtoB1-307 truncations to determine the minimum region of AvrPtoB1-307 sufficient for interaction with Pto in the yeast two-hybrid assay. These truncations express the N-terminal 280, 260, 244, 220, 200 or 190 amino acids, and are referred to as AvrPtoB1-280, AvrPtoB1-260, AvrPtoB1-244, AvrPtoB1-220, AvrPtoB1-200 and AvrPtoB1-190, respectively. Each of the truncations was expressed well as a prey fusion in yeast, and the yeast two-hybrid assays revealed that the region of 1–200 amino acids of AvrPtoB was sufficient for interaction with Pto (Figure 2a). Based on previous data showing that the N-terminal 120 amino acids of AvrPtoB are dispensable for interaction with Pto (Kim et al., 2002), a clone expressing amino acids 121–200 (AvrPtoB121-200) was generated for analysis. A yeast two-hybrid assay indicated that this 80 amino acid region interacts with Pto (Figure 2a). Deletion of ten additional amino acids from either end of this region abolished the interaction, even though these truncated proteins were expressed well in yeast (Figure 2a). Thus, this 80 amino acid region encompasses the minimal contact surface in AvrPtoB1-307 for interaction with Pto in the yeast two-hybrid system.

Figure 2.

  A small N-terminal region of AvrPtoB is sufficient for interaction with Pto kinase.
(a) Yeast two-hybrid interaction assays of AvrPtoB truncation mutants (N-terminally tagged with an hemagglutinin (HA) epitope) and Pto (upper panel). Pto was expressed as the bait and AvrPtoB truncation mutants as prey. Protein accumulation of the AvrPtoB proteins in yeast was detected by Western blotting using anti-HA antibody (lower panel). The origin of the higher molecular mass band cross-reacting with HA in the 1-190 lane is unknown. Positions of protein molecular mass markers are shown.
(b) Cell death assays of AvrPtoB truncation mutants using Agrobacterium-mediated transient expression. Agrobacterium containing individual truncation mutants were syringe-infiltrated into RG-PtoR tomato leaves at a concentration (OD600) of 0.03. Photographs were taken 3 days after infiltration.
(c) Bacterial populations in RG-PtoR leaves 4 days after inoculation with DC3000ΔavrPtoΔavrPtoB strains harboring the empty vector or plasmids expressing AvrPtoB1-307 or AvrPtoB1-200 at inoculum of 104 cfu ml−1. Error bars represent the SE of three replicates. Experiments were repeated three times with similar results.
(d) De-pigmented leaves from a similar position on each plant as used for the bacterial population measurements in (c).
(e) Expression and secretion of AvrPtoB1-307 and AvrPtoB1-200 from DC3000ΔavrPtoΔavrPtoB grown in minimal medium under hrp induction conditions (Lin et al., 2006). Western blotting was performed using anti-AvrPtoB or anti-NptII antibody. Cytoplasmically localized NptII was used as a control for cell lysis. No NptII protein was detected in the culture medium using an anti-NptII antibody.

Although AvrPtoB121-200 interacts with Pto in yeast, Agrobacterium-mediated transient expression of this region in RG-PtoR tomato leaves did not elicit cell death (Figure 2b). Because we were not able to detect AvrPtoB121-200 in Agrobacterium-transformed tomato leaves (data not shown), it is possible that this truncated protein is simply unstable. Further experiments established that AvrPtoB1-200 elicits cell death on RG-PtoR leaves upon Agrobacterium-mediated transient expression (Figure 2b). We therefore examined whether AvrPtoB1-200 elicits disease resistance on RG-PtoR when expressed in DC3000ΔavrPtoΔavrPtoB (note that type III-mediated secretion of AvrPtoB from P. s. pv. tomato requires some N-terminal residues of the effector). We found that AvrPtoB1-200 did not elicit disease resistance or cause an HR (data not shown) on RG-PtoR, even though AvrPtoB1-200 was expressed and secreted from P. s. pv. tomato at a similar level as AvrPtoB1-307 (Figure 2c–e). The cell death caused by Agrobacterium delivery of AvrPtoB1-200 might therefore be due to the higher protein expression levels associated with this system. Taken together, these data indicate that amino acids 121–200 encode specificity for interaction of AvrPtoB1-307 with Pto, and suggest that amino acids that lie outside this region also contribute to avirulence activity upon delivery by P. s. pv. tomato.

F173 is a key amino acid mediating interaction of AvrPtoB1-307 with Pto

We recently identified and characterized several AvrPtoB homologs from P. s. pv. tomato, P. s. pv. syringae and P. s. pv. maculicola (Lin et al., 2006). Like AvrPtoB from P. s. pv. tomato DC3000, each of these AvrPtoB homologs interacts with Pto in yeast and elicits Pto-mediated host immunity in RG-PtoR tomato plants (Lin et al., 2006). The homologs have 21 identical amino acids in the region between AvrPtoB amino acids 121–200, and we reasoned that they might therefore be useful in further defining the basis of Pto interaction specificity (Figure 3a). Fragments corresponding to AvrPtoB121-200 (as aligned by Lin et al., 2006) were PCR-amplified from two representative alleles, avrPtoBT1 and hopPmaL (also named HopAB3), and examined in the yeast two-hybrid assay for interaction with Pto. The 121–200 amino acid regions from both AvrPtoBT1 and HopPmaL interacted with Pto, suggesting that the Pto interaction capability of this small region is functionally conserved (Figure 3b).

Figure 3.

  Phenylalanine 173 is required for interaction of AvrPtoB1-307 with Pto.
(a) Alignment of the amino acids in the AvrPtoB121-200 regions from AvrPtoB and three homologs. Identical amino acids are shaded in black. Positions of the AvrPtoB1-307 amino acids that were later found to be required for interaction with Pto (see Figure 5) are indicated by asterisks.
(b) Yeast two-hybrid interaction between Pto and the 121–200 amino acid region from AvrPtoBT1 and HopPmaL. Pto interacts with the wild-type 121–200 region from AvrPtoBT1 and HopPmaL, but not F173A mutants of this region. Pto was expressed as the bait protein and the AvrPtoB121-200 fragments were expressed as prey. Expression of the HA-tagged AvrPtoB proteins was confirmed by Western blotting using anti-HA antibody (lower panel).
(c) Yeast two-hybrid interaction between Pto and AvrPtoB1-307 mutants F173A, H172A or P174A. Pto was expressed as the bait. The three mutants were expressed as prey. Western blotting using anti-HA antibody indicates that the HA-tagged AvrPtoB1-307 and three mutants were expressed in yeast (lower panel).
(d) Disease symptoms or plant immunity of RG-PtoR tomato plants 4 days after inoculation with DC3000ΔavrPtoΔavrPtoB harboring the empty vector or plasmids expressing wild-type AvrPtoB1-307 or the F173A mutant.
(e) Bacterial populations in leaves of plants as in (d). Each P. s. pv. tomato strain was vacuum-infiltrated into RG-PtoR tomato plants with an inoculum of 104 cfu ml−1. Error bars represent the SE of three replicates. Experiments were repeated three times with similar results.
(f) Expression and secretion of AvrPtoB1-307 and the AvrPtoB1-307(F173A) mutant from DC3000ΔavrPtoΔavrPtoB grown in minimal medium under hrp induction conditions. The P. s. pv. tomato secretion assay was carried out as described in Figure 2(e).

To identify AvrPtoB residues contributing to the interaction with Pto, alanine substitution mutagenesis of the 18 non-alanine residues among the 21 conserved amino acids was carried out on the AvrPtoB1-307 template, and the mutants were examined for interaction with Pto in the yeast two-hybrid assay. Of these mutants, only AvrPtoB1-307(F173A) (hereafter F173A) lost the ability to interact with Pto in yeast, whereas the others retained Pto interaction (Figure 3c and data not shown) (note we were unable to generate a AvrPtoB1-307(L195A) mutant; however, L195 was later found to be important for avirulence activity by random mutagenesis; see below). Western blot analysis indicated that F173A was expressed well in yeast (Figure 3c). Significantly, mutants with alanine substitutions at the two adjacent amino acids, H172A and P174A, retained the Pto interaction (Figure 3c), suggesting the loss of Pto interaction is unlikely to be due to a gross conformational change of AvrPtoB1-307 by the F173A mutation.

We next tested whether the importance of F173 for AvrPtoB1-307 interaction with Pto is functionally conserved among AvrPtoB homologs. The F173A mutation was introduced into the 121–200 amino acid region of AvrPtoBT1 and HopPmaL, and interactions with Pto were examined. The F173A mutants were expressed in yeast at a similar level as the wild-type proteins but their interaction with Pto was abolished (Figure 3b), indicating that the role of F173 is conserved among these AvrPtoB homologs. Finally, we examined the specificity of F173 for interaction with Pto by analyzing the yeast two-hybrid interaction of the F173A mutant with an arbitrary AvrPtoB tomato-interacting (Bti) protein isolated previously (Bti6 is a putative lactate dehydrogenase; R.B.A. and G.B.M., unpubl. results). The F173A mutant interacted with Bti6 as strongly as wild-type AvrPtoB1-307, further supporting the hypothesis that the F173A protein retains its overall conformation and that F173 mediates a specific interaction of AvrPtoB1-307 with Pto (Supplementary Figure S1).

F173 is required for the avirulence activity of AvrPtoB1-307

To determine whether abolishing AvrPtoB1-307 interaction with Pto results in a loss of avirulence activity when the protein is expressed in P. s. pv. tomato, the avrPtoB1-307(F173A) gene was introduced into DC3000ΔavrPtoΔavrPtoB and the resulting strain infiltrated into RG-PtoR plants. This strain caused similar disease symptoms and grew to the same population level in RG-PtoR tomato plants as the DC3000ΔavrPtoΔavrPtoB strain with an empty vector (Figure 3d,e). A P. s. pv. tomato secretion assay showed that the F173A protein was expressed and secreted at a similar level in the DC3000ΔavrPtoΔavrPtoB strain as wild-type AvrPtoB1-307 (Figure 3f). Thus, these results indicate that F173 is required for AvrPtoB1-307 avirulence activity.

Four additional amino acids in the 121–200 region are critical for the avirulence activity of AvrPtoB1-307

In order to identify other amino acids essential for the avirulence activity of AvrPtoB1-307, we performed a PCR-based random mutagenesis screen to identify AvrPtoB1-307 mutants deficient in eliciting the HR in RG-PtoR tomato leaves (see Experimental procedures). Among 386 mutagenized clones examined in an Agrobacterium-mediated transient expression assay, 22 were found to be defective in HR elicitation. Of these clones, 18 were not studied further because they either had mutations within the first 200 N-terminal amino acids causing a stop codon or a frameshift or they had multiple point mutations (Supplementary Table S1). Four avrPtoB1-307 clones that had single point mutations in the 121–200 amino acid region (causing substitutions E165 K, F169S, G180 V or L195H) were found to be unable to elicit an HR in RG-PtoR leaves using the Agrobacterium infiltration assay and were therefore further characterized (Figure 4a). Interestingly, of these four residues, only L195 is conserved at the corresponding position of each homolog (Figure 3a). Analysis of the sequences, however, suggests some amino acid residue conservation at two of the other three positions. All of the AvrPtoB homologs have hydrophobic residues phenylalanine or leucine at position 169 (Lin et al., 2006). The G180 residue is conserved in all the homologs except AvrPtoBT1, where the glycine is substituted for a glutamate residue. However, in AvrPtoBT1, there is a glycine in the position prior to the glutamate, and this might structurally accommodate the glutamate substitution (Lin et al., 2006).

Figure 4.

  Four AvrPtoB1-307 proteins with single amino acid substitutions are compromised in avirulence activity.
(a) HR assay of wild-type AvrPtoB1-307 or E165 K, F169S, G180 V, L195H mutants using Agrobacterium-mediated transient expression in RG-PtoR tomato leaves. Photographs were taken 3 days after Agrobacterium infiltration.
(b) Transient expression of wild-type AvrPtoB1-307, E165 K, F169S, G180 V or L195H mutants in RG-PtoR protoplasts. Plasmids expressing wild-type AvrPtoB1-307 or the mutants were transformed into RG-PtoR protoplasts and cell death was monitored by Evans blue staining 24 h after transformation (upper panel). Approximately 200 randomly selected protoplast cells were counted for determination of the dead cell fraction. Protein expression was analyzed by Western blotting using an anti-AvrPtoB antibody (lower panel).
(c) P. s. pv. tomato assay of AvrPtoB1-307, E165 K, F169S, G180 V or L195H mutants. Disease symptoms or plant immunity of RG-PtoR tomato leaves 4 days after inoculation with DC3000ΔavrPtoΔavrPtoB harboring the empty vector or plasmids expressing wild-type AvrPtoB1-307, E165 K, F169S, G180 V or L195H (upper panel). Bacterial populations in leaves of plants inoculated with the indicated strains (lower panel). Each P. s. pv. tomato strain was vacuum-infiltrated into three RG-PtoR tomato plants with an inoculum of 104 cfu ml−1. Error bars represent the SE of three replicates. Experiments were repeated three times with similar results.
(d) Yeast two-hybrid interaction between Pto and AvrPtoB1-307 or the AvrPtoB1-307 mutants. Wild-type AvrPtoB1-307 or the four mutants were expressed as prey and Pto was expressed as the bait. Protein expression of the HA-tagged AvrPtoB proteins was confirmed by Western blot using an anti-HA antibody.
(e) Expression and secretion of AvrPtoB1-307 mutants from DC3000ΔavrPtoΔavrPtoB grown in minimal medium under hrp induction conditions. The P. s. pv. tomato secretion assay was carried out as described in Figure 2(e).

To further examine the loss of AvrPtoB avirulence activity due to these four mutations and to verify normal expression of these proteins, we tested them in a RG-PtoR tomato protoplast transformation assay. In this assay, transient expression of wild-type AvrPtoB1-307 is recognized by the endogenous Pto protein, and immunity-associated cell death is monitored by Evans blue staining 24 h after transformation. We found that RG-PtoR tomato protoplasts transformed with each of the four mutants or with an empty vector control maintained reasonably high viability (<40% dead cells) at 24 h (Figure 4b). In contrast, 90% of the protoplasts transformed with wild-type AvrPtoB1-307 died within 24 h due to immunity-associated cell death by Pto recognition (Figure 4b). Western blot analysis showed that all four mutants were expressed well at the 24 h time point, whereas wild-type protein was not detected, probably due to general protein degradation caused by cell death (Figure 4b).

We next tested whether the four mutations affected avirulence activity when expressed by P. s. pv. tomato. The mutant genes were introduced on plasmids into the DC3000ΔavrPtoΔavrPtoB strain, and the expression and secretion of their proteins was confirmed by a P. s. pv. tomato secretion assay (Figure 4e). Infiltration of these strains into RG-PtoR tomato plants revealed that none of the four mutants elicited Pto-mediated immunity, as evidenced by formation of bacterial speck symptoms and final population levels that were not different to those with the control DC3000ΔavrPtoΔavrPtoB empty vector strain (Figure 4c).

These four mutations lie in the 121–200 amino acid region that is sufficient for AvrPtoB1-307 interaction with Pto. As anticipated, based on their location in this region and their loss of avirulence activity, the yeast two-hybrid interaction between Pto and these four point mutants was abolished even though each of the proteins was expressed well in yeast (Figure 4d). Collectively, our mutagenesis data show that five clustered residues (E165, F169, F173, G180 and L195) residing in a small (approximately 30 amino acid) domain play an important role for AvrPtoB1-307 interaction with Pto and correspondingly for avirulence activity of AvrPtoB1-307.

Most AvrPtoB1-307 avirulence-deficient mutants retain virulence activity

It has been hypothesized that R proteins might have evolved to recognize structural or enzymatic features of type III effectors that are required for their virulence activity (Dangl and Jones, 2001; Van der Biezen and Jones, 1998). We therefore tested whether the five avirulence-defective AvrPtoB1-307 mutants were also defective in virulence activity. DC3000ΔavrPtoΔavrPtoB strains harboring the five AvrPtoB1-307 mutants, wild-type AvrPtoB1-307 or an empty vector were infiltrated into susceptible tomato RG-prf3 plants. Virulence activity was evaluated after infiltration, as measured by severity of disease symptoms and level of bacterial growth (Figure 5). DC3000ΔavrPtoΔavrPtoB strains carrying wild-type AvrPtoB1-307 or the mutations E165 K, F169S, G180 V or L195H all caused severe disease symptoms and resulted in similar population levels (Figure 5a; note the four lowest leaves, which are necrotic and completely desiccated 4 days after inoculation). Interestingly, the DC3000ΔavrPtoΔavrPtoB strain expressing F173A (F173A was earlier shown to be expressed and secreted normally from P. s. pv. tomato; Figure 3f) caused much less severe disease symptoms and resulted in a bacterial population level at 2 days post-inoculation that was not statistically different from that of the empty vector control strain (Figure 5b). These results therefore indicate that, with one exception (F173), the AvrPtoB1-307 residues involved in the interaction with Pto do not significantly contribute to AvrPtoB1-307-mediated virulence.

Figure 5.

 Virulence activities of AvrPtoB1-307 mutants.
(a) Disease symptoms of tomato RG-prf3 plants inoculated with 104 cfu ml−1 of DC3000ΔavrPtoΔavrPtoB carrying the empty vector or plasmids expressing AvrPtoB1-307, F173A, E165 K, F169S, G180 V or L195H. Representative de-pigmented leaves from each plant are shown in the bottom left corners. Plants were photographed 4 days after inoculation.
(b) Measurement of bacterial populations in leaves of the plants in (a) at 0 and 2 days after inoculation using a P. s. pv. tomato starting inoculum of 104 cfu ml−1. Data pooled from three independent experiments (n = 9) were used for statistical analysis using the SAS general linear model procedure (see Experimental procedures). Values marked with the same letters were not significantly different from each other based on the least squares difference test (= 0.05). Error bars indicate SE.

The N-terminal region of AvrPtoB uses two distinct mechanisms to promote bacterial virulence

We reported previously that, in susceptible tomato plants, AvrPtoB induces the expression of a set of host genes associated with ethylene biosynthesis and promotes ethylene production during disease development (Cohn and Martin, 2005). To determine whether the N-terminus of AvrPtoB is responsible for this ethylene-dependent virulence activity, we tested whether AvrPtoB1-307 is sufficient for the promotion of ethylene production and the induction of two ethylene biosynthesis genes, S. lycopersicum ACC oxidase 1 and 2 (SlAco1and SlAco2). Using similar conditions as our previous experiments, we found by RT-PCR that expression of AvrPtoB1-307 (or AvrPtoB1-387) in P. s. pv. tomato was sufficient to induce SlAco1 and SlAco2 in RG-prf3 leaves as reported previously for the full-length AvrPtoB (Figure 6a) (Cohn and Martin, 2005)). The AvrPtoB1-307(F173A) mutant, which lacks virulence activity, did not induce expression of either of these two genes. We also examined the level of ethylene produced by susceptible tomato RG-prf3 plants infiltrated with DC3000ΔavrPtoΔavrPtoB strains harboring full-length AvrPtoB, AvrPtoB1-387, AvrPtoB1-307, AvrPtoB1-307(F173A) or an empty vector. The strains expressing AvrPtoB1-307 or AvrPtoB1-387 enhanced ethylene production, whereas the AvrPtoB1-307(F173A) mutant had no effect, like the empty vector control (Figure 6b). Expression of AvrPtoB in DC3000ΔavrPtoΔavrPtoB resulted in slightly higher ethylene production than expression of AvrPtoB1-307 or AvrPtoB1-387 (Figure 6b). Taken together, these data indicate that AvrPtoB1-307, like full-length AvrPtoB, contributes to bacterial virulence by inducing ethylene biosynthesis genes and promoting ethylene production during a compatible disease interaction in tomato.

Figure 6.

 AvrPtoB1-307 is sufficient to induce ethylene biosynthesis genes and promotes ethylene production during disease development in tomato plants.
(a) Tomato ACC oxidase 1/2 (SlAco1/2) gene expression was induced by AvrPtoB, AvrPtoB1-387 and AvrPtoB1-307, but not by the virulence-deficient AvrPtoB1-307(F173A) mutant. RG-prf3 plants were inoculated with 104 cfu ml−1 of DC3000ΔavrPtoΔavrPtoB carrying the empty vector, AvrPtoB, AvrPtoB1-387, AvrPtoB1-307 or AvrPtoB1-307(F173A). Total RNA was isolated at the indicated time points after infiltration, and 1 μg total RNA was used to generate first-strand cDNA. The induction of SlAco1/2 was determined by 30 PCR cycles using SlAco1/2 gene-specific primers, with EF1-α gene-specific primers as a control (Cohn and Martin, 2005; Lin and Martin, 2005). The data shown are representative of two experiments with independent biological replicates that gave similar results. Lane M, 1 kb DNA ladder.
(b) Ethylene production by susceptible tomato RG-prf3 plants infiltrated with 104 cfu ml−1 of DC3000ΔavrPtoΔavrPtoB carrying the empty vector or plasmids expressing full-length AvrPtoB, AvrPtoB1-387, AvrPtoB1-307 or AvrPtoB1-307(F173A). Ethylene was measured by gas chromatography at the time points indicated (Cohn and Martin, 2005; Lin and Martin, 2005). Error bars represent the SE of six replicates. The experiment was repeated twice with similar results.

Recently, He et al. (2006) reported that AvrPtoB promotes P. s. pv. tomato virulence on Arabidopsis by targeting MAP kinase signaling to suppress various PAMP-mediated basal defenses. Importantly, this virulence activity appeared to be encoded by an N-terminal region of AvrPtoB as it was unaffected by disruption of the C-terminal E3 ligase activity. To extend this observation, we tested whether the two N-terminal regions, AvrPtoB1-307 and AvrPtoB1-387, can suppress activation of the basal defense marker gene FRK1-LUC by the PAMP elicitor flg22, a conserved 22 amino acid peptide from bacterial flagellin (Figure 7a). The FRK1-LUC reporter gene was co-transformed with AvrPtoB1-307, AvrPtoB1-387, full-length AvrPtoB or vector only, and Western blotting indicated that all AvrPtoB proteins were expressed in Arabidopsis protoplasts. As reported previously, full-length AvrPtoB strongly suppressed flg22 activation of FRK1-LUC. Interestingly, we found that AvrPtoB1-307 did not suppress flg22 activation of FRK1-LUC whereas AvrPtoB1-387 did (Figure 7a). Thus, basal defense suppression requires an additional domain of AvrPtoB encompassing amino acids 308–387. These results were further supported by real-time RT-PCR analysis on an additional six endogenous genes activated by flg22, including WRKY30, WRKY53, At5 g57220, chitinase, At2 g17740 and At2 g40180. The activation of all six genes by flg22 showed specific suppression by full-length AvrPtoB and AvrPtoB1-387, but not by AvrPtoB1-307 (Supplementary Figure S2).

Figure 7.

 AvrPtoB1-387 suppresses markers of pathogen-associated molecular patterns-induced basal defenses in Arabidopsis.
(a) AvrPtoB1-387, but not AvrPtoB1-307, suppresses flg22-induced FRK1 promoter activity. Arabidopsis Col-0 protoplasts were co-transformed with AvrPtoB or derivatives and a FRK1-LUC reporter gene, incubated for 6 h, and then treated with 100 nm flg22 for 3 h. Luciferase activity was measured and is shown as relative promoter activity compared to a UBQ10-GUS control (He et al., 2006). Protein expression of AvrPtoB and derivatives was analyzed by Western blotting using an anti-HA antibody (right panel).
(b) AvrPtoB and AvrPtoB1-387, but not AvrPtoB1-307, blocked flg22-induced MAP kinase activation. HA-tagged MPK3/6 and AvrPtoB or derivatives were co-expressed in Arabidopsis protoplasts for 6 h, and then 1 μm flg22 was added for 10 min. MAP kinases and effectors were immunoprecipitated with anti-HA antibody, followed by in vitro kinase assay (upper panel). Expression of MAP kinases and effectors were confirmed by Western blotting using an anti-HA antibody (bottom panel) (He et al., 2006).

To test whether suppression of PAMP-mediated basal defense by AvrPtoB1-387 is associated with inhibition of MAP kinase signaling, we examined the flg22-induced activity of the MAP kinases MPK3 and MPK6, in the same assay used in our previous experiments (He et al., 2006). Both full-length AvrPtoB and AvrPtoB1-387, but not AvrPtoB1-307, significantly blocked activation of MPK3 and MPK6 by flg22 in Arabidopsis protoplasts (Figure 7b), again indicating that amino acids 308–387 are required to interfere with MAP kinase signaling, and supporting a role for these residues in basal defense suppression in Arabidopsis.

It is possible that virulence-associated residues within the first 307 amino acids of AvrPtoB might contribute to the basal defense-suppressing activity associated with AvrPtoB1-387. We performed two experiments to test this hypothesis. Firstly, we considered the possibility that removal of 80 amino acids from the C-terminus of AvrPtoB1-387 might impair a basal defense-suppressing activity encoded within the AvrPtoB1-307 region. To examine this, we reduced the amount of flg22 from 1 μm to 100, 10 or 1 nm to induce weaker MPK6 activation, and tested whether AvrPtoB1-307 might interfere with this lower activity (Figure 8a). However, we observed that AvrPtoB1-307 was unable to suppress MPK6 activation induced by these lower flg22 levels, indicating it does not have any detectable activity towards this signaling pathway. Secondly, we assessed whether mutations in the AvrPtoB1-307 Pto interaction domain might affect MPK6 suppression activity of AvrPtoB1-387. Point mutations in AvrPtoB1-387 of the ethylene-associated virulence residue F173 or four avirulence residues E165, F169, G180 or L195 caused no change in the ability of AvrPtoB1-387 to suppress MPK6 activation (Figure 8b). Collectively, these data indicate that the 1-387 region of AvrPtoB has two autonomous virulence activities, one promoting ethylene and one suppressing basal defenses, and, significantly, these two virulence activities are structurally separable from AvrPtoB avirulence activity.

Figure 8.

 Amino acids 308–387 encode a determinant for suppression of pathogen-associated molecular patterns-induced basal defense in Arabidopsis.
(a) AvrPtoB1-307 was unable to suppress even weak activation of MPK6 induced by one-tenth the amount (100 nm) of flg22 used in Figure 7, further supporting a requirement for amino acids 308–387. Note that flg22 at a concentration of 10 or 1 nm did not activate MPK6.
(b) Five mutations (E165 K, F169S, G180 V, L195H and F173A) that compromise the avirulence or ethylene-dependent virulence activities of AvrPtoB do not affect the suppression of flg22-induced MPK6 activation by AvrPtoB1-387.

Discussion

A core domain of AvrPtoB1-307 is required for Pto recognition and its activity is conserved in multiple AvrPtoB homologs

Homologs of avrPtoB are present in many P. syringae pathovars, and one of these, HopPmaL, encodes a truncated effector consisting just of a region corresponding to the N-terminal 354 amino acids of AvrPtoB (Guttman et al., 2002). We reported previously that HopPmaL interacts with Pto and confers avirulence to P. s. pv. tomato on Pto-expressing tomato plants (Lin et al., 2006). This observation suggested that the N-terminal region of AvrPtoB might be sufficient for avirulence if expressed in P. s. pv. tomato. Indeed, in a study directed at understanding the cell death-suppression activity of AvrPtoB, it was found that AvrPtoB1-307 is sufficient for interaction with Pto, and elicits Pto/Prf-dependent cell death in a transient Agrobacterium assay (Abramovitch et al., 2003). Together, these observations suggested that avrPtoB1-307 would be sufficient to elicit Pto/Prf-mediated resistance when introduced into DC3000ΔavrPtoΔavrPtoB. The data presented here confirmed this to be the case.

Random mutagenesis of AvrPtoB1-307 indicated that a localized region is required for Pto interaction. Other random mutagenesis studies of effector proteins have also defined either localized stretches of amino acids or spatially clustered amino acids that are required for avirulence activity (Lim and Kunkel, 2004; Ong and Innes, 2006; Shan et al., 2000b). For example, random mutagenesis of AvrRpt2 identified four residues required for avirulence associated with RPS2, with three lying within a 20 amino acid stretch and another one just 30 residues away (Lim and Kunkel, 2004). A three-dimensional structure is available for AvrB, and random mutagenesis of that effector revealed that residues required for its avirulence activity map to a spatially close upper lobe and pocket region of that protein (Ong and Innes, 2006). It is possible that these clustered avirulence determinants represent contact points for host proteins acting in concert with R proteins (and, in some cases, R proteins themselves), or catalytic residues if enzymatic activities are involved. Indeed, a recent report of the co-crystal structure of AvrB and a portion of RIN4, a host protein required for recognition of AvrB by RPM1, revealed that two AvrB residues, T125 and R266, required for avirulence serve as direct binding sites for RIN4 (Desveaux et al., 2007).

The 80 amino acid region of AvrPtoB that is sufficient for interaction with Pto in yeast was not sufficient to elicit cell death upon Agrobacterium-mediated transient expression. Nor were the first 200 amino acids of AvrPtoB (which includes the N-terminal region needed for secretion) sufficient to elicit disease resistance in Pto-expressing plants when delivered by P. s. pv. tomato DC3000, despite the fact that the AvrPtoB1-200 truncated protein was secreted from the bacterium at normal levels. Together, these observations suggest that additional amino acids of AvrPtoB lying outside the Pto-interacting core region are required for avirulence activity in addition to being required for secretion from P. s. pv. tomato. In the plant cell, regions flanking the Pto-interacting core region could be acting to properly localize the effector, stabilize the interaction with Pto, or facilitate other protein–protein interactions that are required to elicit Pto/Prf-mediated immunity.

Our study focused mostly on amino acids 1–307 of AvrPtoB because we found that this region interacted with Pto, elicited avirulence, and promoted virulence in tomato in a fashion indistinguishable from full-length AvrPtoB. However, our data do not exclude the possibility that, in full-length AvrPtoB, other regions outside this region might also affect avirulence activity. For example, we reported previously that an alanine substitution at G325 weakened the interaction of full-length AvrPtoB with Pto and diminished its avirulence activity (Kim et al., 2002). It is possible therefore that intramolecular interactions between the regions containing F173 and G325 influence interaction with Pto and other activities of full-length AvrPtoB. Interestingly, evidence for a possible intramolecular interaction was also observed with AvrPto. In that effector, a mutation at the C-terminal phosphorylation site (S149) weakened the avirulence activity mediated by I96 in the CD loop (Anderson et al., 2006). Ultimately, it will be necessary to obtain a three-dimensional structure of full-length AvrPtoB to understand whether such intramolecular interactions play a role in AvrPtoB avirulence activity.

Two N-terminal virulence activities of AvrPtoB

We found that AvrPtoB1-307 is sufficient to account for the increased disease symptoms and enhanced bacterial growth that were previously found to be associated with full-length AvrPtoB virulence activity when DC3000ΔavrPto is inoculated onto susceptible tomato plants (Cohn and Martin, 2005; Lin and Martin, 2005). We investigated the molecular basis underlying this virulence activity and found that AvrPtoB1-307 induces expression of ethylene biosynthetic genes and promotes ethylene production associated with enhanced disease symptoms. The virulence-defective AvrPtoB1-307(F173A) mutant abolished induction of SlAco1/2 genes and ethylene production (Figure 6). In addition to this ethylene-associated virulence in tomato, two recent reports show that AvrPtoB also enhances the virulence of P. s. pv. tomato DC3000 on Arabidopsis (He et al., 2006; de Torres et al., 2006). These reports showed that AvrPtoB suppresses basal defense-associated activation of MAP kinases and callose deposition, and this virulence activity did not require the C-terminal E3 ligase activity (He et al., 2006; de Torres et al., 2006). Significantly, our experiments now show that AvrPtoB residues 1–307 promote ethylene production associated with disease symptoms and bacterial growth, whereas residues 308–387 are required to suppress PAMP-mediated basal defenses. Thus these two virulence activities are encoded by distinct structural domains of AvrPtoB. We were unable to detect enhanced ethylene production from Arabidopsis plants infected by AvrPtoB-expressing P. s. pv. tomato (F.X. and G.M., unpubl. results). In addition, we are currently unable to investigate whether amino acids 308–387 suppress basal defenses in tomato as we lack a robust assay for this activity. Future studies will therefore be needed to clarify whether certain AvrPtoB virulence activities are general or specific to different plant species.

It is notable that we did not observe a significant difference in virulence activity conferred by AvrPtoB1-307 or AvrPtoB1-387 with regard to disease severity or bacterial growth rates. This could simply be due to a lack of sensitivity of our bacterial growth and disease assays, or alternatively it could be indicative of some functional redundancy of the two virulence activities. In this regard, AvrPtoB has been shown recently to act as suppressor of multiple plant defense responses in N. benthamiana, including ROS burst, callose deposition, calcium burst and MAP kinase activation (Hann and Rathjen, 2007). It will be interesting to determine which AvrPtoB domains are responsible for these various activities.

AvrPtoB1-307 does not require Pto for virulence activity, suggesting alternative models for the guard hypothesis in Pto-mediated recognition

The guard model, which was originally inspired by the Pto/Prf system, postulated that Pto contributes to host basal defense, that the P. s. pv. tomato effectors AvrPto and AvrPtoB evolved to interact with Pto and thereby perturb its activity, and that Prf evolved to recognize (‘guard’) this perturbation of Pto and to activate host defenses (Dangl and Jones, 2001; Van der Biezen and Jones, 1998). Accordingly, avirulence and virulence activities of effector proteins are expected to be co-dependent. Several lines of evidence support tenets of the guard hypothesis for the Pto system. Firstly, Pto is a plausible virulence target for Pseudomonas effectors because its over-expression leads to enhanced resistance to virulent pathogens (and it therefore might play a role in basal defense) (Tang et al., 1999). Secondly, Pto physically interacts with AvrPtoB and AvrPto (Kim et al., 2002; Scofield et al., 1996; Tang et al., 1996). Thirdly, AvrPtoB and AvrPto have demonstrable virulence activity (Abramovitch et al., 2003; Chang et al., 2000; Lin and Martin, 2005; Shan et al., 2000a). Finally, the activities of Pto and Prf are inter-dependent (Rathjen et al., 1999; Xiao et al., 2003).

However, we have previously discussed some observations involving AvrPto that are inconsistent with tenets of the guard model (Pedley and Martin, 2003). First, AvrPto exhibits similar virulence activity on tomato leaves whether or not they express Pto (Shan et al., 2000a). Thus Pto does not appear to be an important virulence target of AvrPto. Second, certain mutations in AvrPto cause a loss of avirulence and yet the proteins still retain virulence activity (Shan et al., 2000a). This observation indicates that there is a virulence target other than Pto for AvrPto. Similar results have been reported in the case of interaction between Rin4 and AvrRpt2 and AvrRpm1, in which Rin4, which is required for avirulence of AvrRpt2 and AvrRpm1, is not required for the virulence activity of AvrRpt2 or AvrRpm1 in Arabidopsis (Belkhadir et al., 2004). Importantly, separable avirulence and virulence activities have also been observed with other type III effectors, including AvrXa7, AvrRpt2 and AvrBs2 (Gassmann et al., 2000; Lim and Kunkel, 2004; Yang et al., 2005).

Our present data extend these points to a second effector, AvrPtoB. First, four of five mutations that disrupt the avirulence activity of AvrPtoB1-307 had no discernible effect on the ability of this region to promote early bacterial growth and exacerbate disease symptoms. We cannot rule out the possibility that some of these residues might contribute to bacterial fitness under field conditions. However, our observations strongly suggest that interaction of AvrPtoB1-307 with Pto is not required for AvrPtoB virulence activity. Second, AvrPtoB1-307 exhibited virulence activity on a tomato line that does not carry the Pto gene, further indicating that Pto is not a functionally important virulence target of this effector. Together, the previous data involving AvrPto and the present data regarding AvrPtoB are inconsistent with some of the central tenets of the guard hypothesis. Participation of Prf in a receptor complex with Pto is possible and even likely. However, our data suggest no reason why Pto should be viewed as a virulence target of AvrPtoB guarded by Prf.

Although recognizing that we still lack key details of this pathosystem (e.g. what the true virulence targets of AvrPtoB1-307 are), we propose an alternative to the guard model to explain our results. In this alternative model, we postulate that the region encompassed by AvrPtoB1-307 evolved to promote virulence by interaction with a host protein other than Pto. This virulence protein may share structural similarity with Pto, as suggested by the fact that one AvrPtoB residue (F173) is required for interaction with Pto and for virulence. However, other, presently unknown, residues of AvrPtoB1-307 probably contribute to targeting of this other host protein. Why then does Pto interact with AvrPtoB1-307? It is possible that Pto evolved to interact with AvrPtoB by mimicking certain structural or enzymatic features of the true virulence target of AvrPtoB (e.g. F173) and by utilizing other AvrPtoB residues to stabilize this interaction (e.g. E165, F169, G180 and L195). Alternatively, because the Pto gene is present in at least two diverse tomato species and appears to have arisen before Solanum speciation (Riely and Martin, 2001), it is possible that the similarities of Pto to the host virulence target were pre-existing and required for other functions of Pto. In this model, Pto (and other Pto family members) might have evolved to function in a signaling complex with Prf that arose to activate defense responses. In this scenario, Prf can be viewed as an important recognition and/or signaling component of the Pto pathway but one that evolved without any specific selection pressure exerted by AvrPtoB (or AvrPto). Future identification of the host virulence target(s) of AvrPtoB should help address many interesting questions that remain unanswered in this system, including what structural motif in the virulence target is required for interaction with AvrPtoB, and why there hasn’t been selection pressure on AvrPtoB for mutation of the residues required for Pto recognition but dispensable for virulence.

Homology modeling suggests that AvrPtoB121-200 may have structural similarity with the Pto-interacting domain of AvrPto

Our initial amino acid alignment of AvrPtoB with AvrPto showed the two effectors had little overall similarity although some regions had short stretches of conserved residues (Kim et al., 2002). Additional mutagenesis of AvrPto and determination of its three-dimensional structure later revealed the importance of the CD loop for interaction with Pto (Chang et al., 2001; Shan et al., 2000a; Wulf et al., 2004). The present mutagenesis of AvrPtoB1-307 indicates a corresponding significance of the region around F173, and caused us to further scrutinize the AvrPtoB1-307 sequence for similarity with the CD loop region of the AvrPto sequence. We focused on the functional core of AvrPto (AvrPto33-128), the minimal Pto-binding 80-residue fragment of AvrPtoB (AvrPtoB121-200), and for comparison the corresponding regions from HopPmaL and AvrPtoBT1. These regions were found to have identical residues conserved at 10 positions, and residues with similar properties conserved at 13 positions (Figure 9a). This sequence similarity suggests that AvrPtoB121-200 may have structural similarity to the three-helix bundle of AvrPto33-128 (Wulf et al., 2004). Indeed, secondary structure predictions are consistent with AvrPtoB121-200 having multiple helices (Figure 9b). The sequence alignment and the AvrPto structure (pdb code 1R5E) were used to construct a homology model of AvrPtoB121-200 (Figure 9c) using Swiss-Model (Peitsch, 1995; Schwede et al., 2003). Overall, the level of sequence identity, helical propensities, and conserved structural features in the αC–CD loop–αD region provide high confidence in this region of the model, whereas the αA–AB loop–αB loop region is more speculative.

Figure 9.

 Sequence alignment and structural model for AvrPtoB121-200.
(a) Sequence alignment of AvrPto36-128, AvrPtoB121-200 and the comparable region of AvrPtoB homologs from P. s. pv. tomato T1 and P. s. pv. maculicola ES4326. Identical residues (red) and similar residues (black) conserved across all four sequences are indicated. I96 in AvrPto and F173 in AvrPtoB are highlighted in yellow.
(b) Secondary structure prediction for AvrPto and AvrPtoB sequences. Categorization into α-helix (H), β-sheet (E), turn (T) or other (x) is based on the relative probabilities of six backbone conformational states as determined by HMMSTR-R (Bystroff et al., 2000). AvrPto residues that actually adopt α-helical backbone conformation in the AvrPto NMR structure (Wulf et al., 2004) are underlined.
(c) AvrPtoB homology model and conserved features of CD loop that may facilitate Pto binding. Ribbon overlay (center) of AvrPto (green) and AvrPtoB (blue), with essential residues (red) and conserved M176 (gray) shown in stick representation. The cluster of hydrophobic residues in the modeled AvrPtoB CD loop includes essential residues F169 and F173 and conserved residue M176 (right; AvrPtoB residues shown in blue). These residues overlie residues M85, I96 and M100, respectively, in the known AvrPto structure (AvrPto residues shown in green). The electrostatic surface potential of the homology model suggests an electrostatic role for essential residue E165 (left). Images were generated using deepview (http://www.expasy.org/spdbv) (Guex and Peitsch, 1997).

The homology model provides a reasonable framework from which to rationalize the AvrPtoB mutations that lead to loss of Pto binding (Figure 9c). A putative CD loop of the AvrPtoB model is shortened by nine residues relative to the corresponding 19-residue loop in the AvrPto structure, and consequently adopts a different shape. However, as in AvrPto, this abbreviated loop would form a pseudo-hydrophobic core that involves the conserved residues M176, F169 and F173. The latter two residues are required for interaction with Pto, and loss of either of them may disrupt the localized conformation. L195 is also required for Pto interaction and may influence fold stability, as it is predicted to be buried in the hydrophobic core. The essential role of E165 may be due to its electrostatic field contribution to the surface potential, as the E165 side chain is adjacent to a large positive surface region of the model (Figure 9c). Another residue identified by mutagenesis, G180, may fulfill additional conformational requirements of the fold. Based on these observations, we hypothesize that the putative CD loop of AvrPtoB and the known CD loop of AvrPto each play a central role as the contact surface for interaction with Pto, and that the kinase employs a common mode of binding to these two effector proteins. Efforts are underway to obtain sufficient amounts of AvrPtoB121-200 for three-dimensional structural analysis to elucidate the actual features in this region that are involved in the interaction with Pto.

AvrPtoB is a multi-functional virulence protein

This study, and earlier work on the cell death-suppression activity of the C-terminus of AvrPtoB, reveals that this effector has at least three independent regions that make significant contributions to bacterial pathogenesis (Abramovitch et al., 2003). Although these virulence activities are separable experimentally, it remains to be determined whether the three domains act independently or in concert during P. s. pv. tomato infection of leaves. Interestingly, processed species of AvrPtoB are detected in vivo, suggesting that the domains might be released and function separately (de Vries et al., 2006; He et al., 2006; R.B.A. and G.B.M., unpubl. results). In addition, the existence in P. s. pv. maculicola of a C-terminus truncated homolog of AvrPtoB (HopPmaL) suggests that at least the N-terminal domains might act autonomously. In this way, AvrPtoB might be similar to viral polyproteins, where one large virulence protein is processed into multiple protein domains that act together in viral pathogenesis. As many type III effectors are large proteins (Desveaux et al., 2006), it is likely that other type III effectors have multiple virulence and enzymatic activities. Careful structural and functional studies of these effectors are required to elucidate the relative contributions of these activities to bacterial virulence.

Experimental procedures

PCR-based random mutagenesis and Agrobacterium-mediated transient assay

Random mutagenesis of AvrPtoB1-307 was carried out using a Gene Morph random mutagenesis kit (Stratagene, http://www.stratagene.com/) according to the manufacturer’s instructions and using primers 5′-ACTCTAGAATGGCGGGTATCAATAGAGCGGGACCATCG-3′ and 5′-TTCTGCAGTCATACATGTCTTTCAAGGGCCGT-3′ (XbaI and PstI sites underlined). The AvrPtoB1-307 mutant alleles were digested with XbaI and PstI, and cloned into the same sites of binary vector pBTEX, followed by electroporation into Agrobacterium tumefaciens strain GV2260. The Agrobacterium-mediated transient expression assay was performed as described by Sessa et al. (2000), with strains carrying AvrPtoB1-307 or mutant alleles being syringe-infiltrated into tomato RG-PtoR leaves at a concentration (OD600) of 0.03. Photographs were taken 3 days after infiltration.

P. s. pv.tomatoprotein secretion and Western blotting assay

A DNA fragment containing AvrPtoB1-307 and its hrp promoter was amplified by PCR using primers 5′-GGTGTGATGGAACTCTTTCCTGCTC-3′ and 5′-TCATACATGTCTTTCAAGGGCCG-3′, and cloned into pCR2.1 (Invitrogen, http://www.invitrogen.com/). The resulting construct was also used as a template for creating AvrPtoB1-307 mutants by site-directed mutagenesis with the following primer sets: E165 K, 5′-GCTTTTTCTCGAGTAAAACAGAACATATTTCGC-3′ and 5′-GCGAAATATGTTCTGTTTTACTCGAGAAAAAGC-3′; F169S, 5′-GTAGAACAGAACATATCTCGCCAGCATTTCCCG-3′ and 5′-CGGGAAATGCTGGCGAGATATGTTCTGTTCTAC-3′; G180 V, 5′-ACATGCCCATGCATGTAATCAGCCGAGATTCGG-3′ and 5′-CCGAATCTCGGCTGATTACATGCATGGGCATGT-3′; L195H, 5′-GCTCCGTGGGGCGCATCGTCGAGCGGTTCA-3′ and 5-TGAACCGCTCGACGATGCGCCCCACGGAGC-3′. The EcoRI fragment from pCR2.1, containing AvrPtoB1-307 or derived mutants plus the hrp promoter, was cloned into the broad-host-range vector pCPP45 (Lin and Martin, 2005). The resulting constructs were confirmed by sequencing, transformed into DC3000ΔavrPtoΔavrPtoB by electroporation, and analyzed using the P. s. pv. tomato secretion assay as described previously (Lin et al., 2006) Western blotting was carried out using an anti-HA antibody (Roche Applied Science; http://www.roche.com), anti-NptII (US Biological; http://www.usbio.net) or anti-AvrPtoB antibody (Lin et al., 2006). Detection of proteins was carried out using HRP-conjugated secondary antibodies and the ECL Plus detection system (Amersham-Pharmacia, http://www5.amershambiosciences.com/).

Tomato protoplast isolation and transient assay

Tomato protoplasts were isolated and transformed according to protocols described by Xing et al. (2001) and Sheen (http://genetics.mgh.harvard.edu/sheenweb/). Proteins were transiently expressed in protoplasts from the pTEX vector expression cassette. AvrPtoB1-307 and derived mutants were released from the pBTEX vector using EcoRI and HindIII and cloned into the same sites of the pTEX vector. The resulting constructs were used for PEG-mediated transformation into protoplasts, and cell death was monitored by staining with 0.04% Evans blue 24 h after transformation (Asai et al., 2000).

Yeast two-hybrid assays

A LexA yeast two-hybrid system was used to test protein interactions. Pto was expressed from the bait vector pEG202 (Tang et al., 1996), and AvrPtoB1-307, AvrPtoB homologs or the derived mutants were expressed from the prey vector pJG4-5. AvrPtoB truncation mutants and Ala substitution mutant constructs were produced by introduction of a stop codon or Ala codon at the appropriate sites by site-directed mutagenesis with pJG4-5::AvrPtoB1-307 as the template. To create prey constructs containing the 121–200 amino acid region of AvrPtoB and homologs, the 121–200 fragments were PCR-amplified with the following primer pairs: AvrPtoB, 5′-ATGAATTCATGCCGCGTAGAGGGGCGGTTGCACACGCCAA-3′ and 5′-TTGTCGACGTGAACCGCTCGACGAAGCGCCCCA-3′; AvrPtoBT1, 5′-ATGAATTCATGGTACGCACAGAGGGAGGACGCACA-3′ and 5′-TTGTCGACTCAGCGAACCTCCCGACGGAGAGCCGC-3′; AvrPtoBB278a, 5′-ATGAATTCATGGTTGCAGAGCGCATCGTTCAGGAG-3′ and 5′-TTGTCGACTCAGTGAACCACCTGACGAAGCGTTTG-3′; HopPmaL, 5′-ATGAATTCGTGGCACGTGCAGAGAGAGGACGCACG-3′ and 5′-TTGTCGACTCATACATGTCTTTCAAGGGCCGTGCGC-3′ (EcoRI and SalI sites underlined). The PCR products were digested with EcoRI/SalI and cloned into the EcoRI and XhoI sites of pJG4-5. The resulting constructs were also used as the templates for creating the F173A mutants by site-directed mutagenesis using the following primer sets: AvrPtoBT1, 5′-TTTTGATCGAGCATGCCCCTCACCTGCTAGCG-3′ and 5′-CGCTAGCAGGTGAGGGGCATGCTCGATCAAAA-3′; AvrPtoBB278a, 5′-CTTGCTTCAGCATGCTCCTGACATGCATAC-3′ and 5′-GTATGCATGTCAGGAGCATGCTGAAGCAAGAT-3′; HopPmaL, 5′-GTTTTTCTCCAGCATGCTCCTGACATGCTACC-3′ and 5′-GGTAGCATGTCAGGAGCATGCTGGAGAAAAAC-3′. For interaction between Bti6 and AvrPtoB1-307 or the F173A mutant, wild-type AvrPtoB1-307 and the F173A mutant were PCR-amplified with primers 5′-GGGAATTCATGGCGGGTATCAATAGAGCGGGACCATCG-3′ and 5′-AACCATGGTCATACATGTCTTTCAAGGGCCGTGCGCAGGTGC-3′ (EcoRI and NcoI sites underlined) from pJG4-5 constructs and cloned into the same sites of pEG202 as the baits. Bti6 was expressed from pJG4-5 as prey. Yeast cells containing both prey and bait proteins were streaked on X-Gal plates with galactose to analyze the interactions, and photographed after 2 days after incubation at 30°C.

Measurement of bacterial populations on tomato

Three genotypes of Rio Grande tomato lines were used: RG-PtoR (Pto/Pto, Prf/Prf), RG-prf3 (Pto/Pto, prf/prf) and RG-pto11 (pto/pto, Prf/Prf). The method for preparation of P. s. pv. tomato inoculum has been described previously (Anderson et al., 2006). Six-week-old greenhouse grown plants were vacuum-infiltrated with P. s. pv. tomato strains at an inoculum of 104 cfu ml−1. For consistent results, after inoculation with P. s. pv. tomato bacteria, tomato plants were kept in a climate-controlled growth chamber under optimized conditions (see Anderson et al., 2006; for details). Bacterial populations were recovered from plant leaves and quantified at 2 and 4 days after infiltration following methods described by Anderson et al. (2006). To assess virulence activity, analysis of variance of bacterial populations on susceptible plants was based on the pooled data from two or three experiments (as indicated) using a general linear model procedure (SAS Institute Inc.; http://www.sas.com). The least significance difference (LSD) at a 0.05 probability level was used to test the difference between means.

Comparison of AvrPto and AvrPtoB structural similarities

The programs jnet (Cuff and Barton, 2000), a multiple-sequence alignment tool, and AGADIR (Lacroix et al., 1998), which predicts α-helix propensities using the nucleation–condensation model of α-helix formation, were used to predict the secondary structure of AvrPto (AvrPto36-128) and AvrPtoB123-200. The AGADIR propensities were obtained for each protein at pH 6.1, 298 K, and 0.24 M ionic strength. The manual sequence alignment of AvrPtoB123-200 and AvrPto36-128 was used to construct a homology model using Swiss-Model (Peitsch, 1995; Schwede et al., 2003).

Analysis of AvrPtoB virulence activities on tomato and Arabidopsis

Ethylene production by tomato leaves inoculated with various P. s. pv. tomato strains was measured at the indicated time points after vacuum infiltration of an inoculum of 104 cfu ml−1 as described previously (Cohn and Martin, 2005; Lin and Martin, 2005). Basal defense suppression assays and MAP kinase signaling interference by AvrPtoB in Arabidopsis protoplasts were performed as described previously (He et al., 2006). UBQ10-GUS was co-transformed with FRK1-LUC as an internal control, and the promoter activity of FRK1 is presented as LUC normalized to UBQ10-GUS activity. Detailed methods are described in the supplementary experimental procedures of a previous report (He et al., 2006).

Acknowledgements

We thank Tim Devarenne, Tracy Rosebrock, and Kathy Munkvold for critical reading of the manuscript, James Booth (Department of Mathematics, Cornell University, USA) and Pete Pascuzzi for statistical analysis, and Sabih Effendi for technical assistance. R.B.A. was supported by a fellowship from the National Sciences and Engineering Research Council of Canada. This work was supported, in part, by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant 2005-35301-15675 (G.B.M.), National Science Foundation grant DBI-0605059, and National Institutes of Health grant number R01-GM078021 (G.B.M.).

Ancillary