The type III effector repertoire of Pseudomonas syringae pv. syringae B728a and its role in survival and disease on host and non-host plants


  • Boris A. Vinatzer,

    1. Department of Plant Pathology, Physiology, and Weed Science, Virginia Polytechnic Institute and State University, Fralin Biotechnology Center, West Campus Drive, Blacksburg, VA 24061-0346, USA.
    Search for more papers by this author
  • Gail M. Teitzel,

    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
    Search for more papers by this author
  • Min-Woo Lee,

    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
    Search for more papers by this author
  • Joanna Jelenska,

    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
    Search for more papers by this author
  • Sara Hotton,

    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
    Search for more papers by this author
  • Keke Fairfax,

    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
    Search for more papers by this author
  • Jenny Jenrette,

    1. Department of Plant Pathology, Physiology, and Weed Science, Virginia Polytechnic Institute and State University, Fralin Biotechnology Center, West Campus Drive, Blacksburg, VA 24061-0346, USA.
    Search for more papers by this author
  • Jean T. Greenberg

    Corresponding author
    1. Department of Molecular Genetics and Cell Biology, The University of Chicago, 1103 East 57th Street, EBC410, Chicago, IL 60637, USA.
      *E-mail; Tel. (+1) 773 834 1908; Fax (+1) 773 702 9270.
    Search for more papers by this author

*E-mail; Tel. (+1) 773 834 1908; Fax (+1) 773 702 9270.


The bacterial plant pathogen Pseudomonas syringae injects a large repertoire of effector proteins into plant cells using a type III secretion apparatus. Effectors can trigger or suppress defences in a host-dependent fashion. Host defences are often accompanied by programmed cell death, while interference with defences is sometimes associated with cell death suppression. We previously predicted the effector repertoire of the sequenced bean pathogen P. syringae pv. syringae (Psy) B728a using bioinformatics. Here we show that PsyB728a is also pathogenic on the model plant species Nicotiana benthamiana (tobacco). We confirm our effector predictions and clone the nearly complete PsyB728a effector repertoire. We find effectors to have different cell death-modulating activities and distinct roles during the infection of the susceptible bean and tobacco hosts. Unexpectedly, we do not find a strict correlation between cell death-eliciting and defence-eliciting activity and between cell death-suppressing activity and defence-interfering activity. Furthermore, we find several effectors with quantitative avirulence activities on their susceptible hosts, but with growth-promoting effects on Arabidopsis thaliana, a species on which PsyB728a does not cause disease. We conclude that P. syringae strains may have evolved large effector repertoires to extend their host ranges or increase their survival on various unrelated plant species.


Strains of the Gram-negative bacterial species Pseudomonas syringae have been isolated as causal agents of leaf spot, leaf blight, leaf speck or bacterial canker disease from the majority of cultivated crop and ornamental plant species all over the world (Agrios, 1997). Thus, P. syringae ranks among the most adaptive and successful plant pathogens. The most important virulence mechanism this species uses to manipulate their plant hosts is a type III secretion system (T3SS), which is used to secrete proteins, called effectors, into the extracellular plant spaces (the apoplast) and directly into the host cell interior (Jin et al., 2003). Genomic sequence data of three P. syringae strains (Buell et al., 2003; Feil et al., 2005; Joardar et al., 2005) in combination with in vitro and in vivo screens and bioinformatics (Guttman et al., 2002; Petnicki-Ocwieja et al., 2002; Greenberg and Vinatzer, 2003; Chang et al., 2005) have revealed that strains of P. syringae contain dozens of effectors. Genetic analysis has shown that most individual effectors play only modest, quantitative roles in promoting bacterial growth and disease symptoms when studied in controlled conditions (White et al., 2000). Redundancy of function among effectors present in individual strains is often claimed as main cause for this finding (Alfano and Collmer, 2004).

Interestingly, effector repertoires vary in size and composition between strains (Greenberg and Vinatzer, 2003; Chang et al., 2005). Only 13 effectors are shared between the three sequenced P. syringae strains P. s. pv. tomato (Pto) DC3000, P. s. pv. syringae (Psy) B728a and P. s. pv. phaseolicola (Pph) 1448a. The remaining approximately 40 effectors are either unique to one of these strains or only shared between two of them (based on data available at These observed differences in effector repertoires between strains are believed to be the main determinants of host range in P. syringae (Alfano and Collmer, 2004).

Particularly important for host range is a subset of P. syringae effectors known as ‘avirulence’ (avr) genes.These genes confer avirulence (the reduced ability to cause disease) to virulent (disease-causing) pathogens on plants that recognize the Avr protein through resistance (R) proteins. R proteins activate plant defences upon recognition of the Avr protein. Such an interaction between the pathogen and the plant is called incompatible (as opposed to a compatible interaction that results in disease). Several avr genes have been shown to increase disease severity in compatible interactions on plants missing the corresponding R gene (Alfano and Collmer, 2004). Some R proteins indirectly recognize Avr proteins; they detect a change in a plant protein caused by the activity of the Avr protein. These R proteins are hypothesized to guard the plant protein targeted by the effector (Dangl and Jones, 2001). In some incompatible interactions, the plant defence response includes programmed cell death, which becomes visible as total leaf collapse when a high dose of pathogen is infiltrated into the host leaf. This response is generally referred to as the hypersensitive response (HR) [recently reviewed by Greenberg and Yao (2004)].

Most characterized bacterial avr genes confer dominant qualitative phenotypes; their presence is sufficient to convert a compatible strain into one that causes little or no disease. In some cases, mutation of an individual avr gene in a strain can also render that strain compatible [see examples in the study by Tsiamis et al. (2000)]. In other cases, deletion of an avr gene in a P. syringae strain does not cause that strain to become compatible. For example, deletion of the effector gene avrRpt2 from PtoJL1065 abolishes the ability of this strain to induce an HR on Arabidopsis thaliana ecotype ‘Columbia’ but the strain is still unable to cause disease on this host (our unpublished results). In this case, another effector in PtoJL1065 may induce defences without inducing an HR, as was found for the effector HopPsyA (Gassmann, 2005). Allelic differences between strains in genes coding for pathogen associated molecular patterns (PAMPs) like flagellin can also determine the outcome of a P. syringae–plant interaction (Takeuchi et al., 2003). Additionally, several effectors either interfere specifically with the defences elicited by avr effectors or by PAMPs [reviewed by Nomura et al. (2005)] also contributing to host range. In contrast, toxin production by P. syringae strains has been found to contribute to disease severity, but not to change host range (Bender et al., 1999).

Although cell death has been studied more thoroughly in incompatible interactions when avr effectors induce the HR, cell death is also important in symptom formation during disease. Recently, an effector was reported to contribute also to this latter cell death (Badel et al., 2003). Interestingly, two pathways involved in modulating cell death during an HR also modulate symptom formation during disease (del Pozo et al., 2004; Yao and Greenberg, 2006). Additionally, the effector HopN1PtoDC3000 suppresses cell death during the HR and during disease (Lopez-Solanilla et al., 2004). Finally, the morphology of many dying cells in the HR or disease caused by low-dose infection by P. syringae is very similar (Yao and Greenberg, 2006), supporting the idea that the main difference between cell death during the HR and during disease is its timing.

We used the P. syringae strain PsyB728a isolated from Phaseolus vulgaris (snap bean) to study the role that its individual effectors play in host range and disease formation. We previously predicted the effector repertoire of PsyB728a (Greenberg and Vinatzer, 2003). It only contains about half the number of predicted effectors compared with PtoDC3000 and is therefore more amenable to genetic studies than PtoDC3000. We confirm here 22 predicted effectors. Surprisingly, we find PsyB728a to be highly virulent on Nicotiana benthamiana, although three effectors induce cell death that depends on plant defence signalling components. We also show that two of the PsyB728a effectors interfere with cell death on N. benthamiana. We further show that the contributions of PsyB728a effectors to growth and/or disease are different on the two hosts snap bean and N. benthamiana and the non-host A. thaliana. Surprisingly, three effectors had quantitative avirulence functions in one or both susceptible host plants. Our data suggest that P. syringae harbours a large effector repertoire to enhance its ability to cause disease and/or grow and survive in diverse host environments.


Confirmation and annotation of the predicted PsyB728a type III secretome

We previously predicted 29 PsyB728a effectors based on the presence of the conserved hrp-promoter element upstream of effector-encoding genes and biased amino acid content in the 50 most N-terminal amino acids, two characteristics typical of P. syringae T3SS effectors (Guttman et al., 2002; Greenberg and Vinatzer, 2003). To experimentally determine if these effectors are secreted, we used the T3SS reporter AvrRpt2101−255 (Vinatzer et al., 2005). We considered effectors to be secreted when their fusion to AvrRpt2101−255 expressed from the consititutive nptII promoter in PtoDC3000 gave an HR when bacteria were injected into the ‘Columbia’ accession of A. thaliana. An additional criterion was that the strain expressing the effector-AvrRpt2101−255 chimera did not cause an HR when infiltrated into the A. thaliana rps2 mutant that does not recognize AvrRpt2. We did not determine the levels to which these proteins were expressed during infection from their native promoters. Table 1 shows that 22 predicted PsyB728a effectors were secreted in our assay [including the effectors HopAG1, HopAH1, HopI1 and HopAE1, whose secretion we previously confirmed (Vinatzer et al., 2005)].

Table 1.  Predicted and confirmed effectors in the genome sequence of P. syringae pv. syringae B728a (Accession No. NC_007005).
EffectoraPrevious namebLocus tagSecretion confirmedcCharacterized in this paperID no.d
  • a. 

    The indicated name is the name of the protein family as described in the study by Lindeberg et al. (2005). To indicate the PsyB728a member of a protein family add PsyB728a as subscript, for example AvrB3PsyB728a.

  • b. 

    Name used in the study by Greenberg and Vinatzer (2003).

  • c. 

    ‘YES’ and ‘NO’ in capital letters indicate that the PsyB728a member of this effector family was tested for translocation in planta using full-length fusions of the effector to the AvrRpt2101−255 reporter expressed from the nptII promoter on the high-copy number plasmid pBAV178 in PtoDC3000 (Vinatzer et al., 2005). YES indicates that secretion was reported in the study by Vinatzer et al. (2005). ‘Yes’ means that a family member in another P. syringae strain was confirmed as reviewed in the study by Lindeberg et al. (2005).

  • d. 

    For the effectors characterized in this paper an ID number is indicated. The ID number refers to the pDONR207 plasmids containing this gene.

  • e. 

    PsyB728a family members of this gene family were also previously confirmed to be secreted by Deng et al. (2003).

  • f. 

    Only the N-terminal region of these genes have been cloned to conduct translocation assays as fusions to the AvrRpt2101−255 reporter. Cloning of full-length genes failed because of their length.

  • g. 

    These genes have been cloned separately for expression in P. syringae and for Agrobacterium-mediated expression in planta.

  • h. 

    Data on secretion in culture and translocation in plants of HopAJ2 family members are conflicting. HopAJ2 family members may be either secreted in a T3SS-dependent manner, by the sec system, or both (Vinatzer et al., 2005).

  • i. 

    The HrpA1 family codes for the main structural component of the T3SS pilus. Full-length fusion of HrpAPsyB728a to the AvrRpt2101−255 reporter did not give an HR in our hands. We hypothesize that this fusion interfered with the assembly of the hrp pilus.

  • j. 

    HrpWPsyB728a was also previously shown to be secreted in culture by Charkowski et al. (1998)18.

  • k. 

    HopAI1′PsyB728a has an early STOP codon after amino acid 21. The fusion of HopAI1′PsyB728a to the AvrRpt2101−255 reporter did not give an HR in our hands, but the PtoDC3000 family member HopAI1PtoDC3000 did (Vinatzer et al., 2005).

  • l. 

    These proteins fused to AvrRpt2101−255 could not be detected by Western blot and may therefore be secreted but were not expressed high enough in our experiments to allow the elicitation of a visible HR.


All confirmed effectors were analysed by blastp (Altschul et al., 1997) and by Phyre (, an improved version of 3D-PSSM (Kelley et al., 2000), to find homologous effectors in other pathogens and predict possible enzymatic functions. Phyre is a protein structure prediction program that compares the predicted protein structure of the query protein with a database of known structures of proteins with known function. The results of this analysis are shown in Table S1. Ten PsyB728a effectors have an annotated effector homologue in another plant pathogenic bacterial species, three have putative effector homologues in animal pathogens, while one (HopH1) has predicted hypothetical protein homologues in the plant pathogen Ralstonia solanacearum and in a prophage of Escherichia coli. Based on published data, blastp and Phyre results, four effectors have predicted functions in the intercellular spaces (HrpW, HopAH1, HopAK1 and HopAJ2), three effectors aid in the delivery of effectors into the cell (HrpA, HrpK and HrpZ) and 15 effectors have similarities with known intracellular proteins or effectors with avirulence activity indicating intracellular localization.

The bean pathogen PsyB728a is highly virulent on N. benthamiana

PsyB728a was isolated from a diseased snap bean plant in Wisconsin, USA (Loper and Lindow, 1987). To determine which PsyB728a effectors are important for causing disease or restricting host range, we first sought to better define the host range of PsyB728a. Therefore, we tested the ability of PsyB728a to cause disease on the model plants A. thaliana, Nicotiana tabacum and N. benthamiana. Performing low-dose infections, PsyB728a caused no disease symptoms on nine of 10 accessions of A. thaliana (Van-0, Zu-0, Col-0, Ws, Po, Uk3, Fb1953, Nok3, Nd-0). An exception was the ‘Landsberg (Ler)’ accession, on which we occasionally observed tiny cell death spots less than 2 mm in size. On Ler, bacterial populations of PsyB728a after 3 days multiplied at least 100-fold less than P. syringae strains that caused good disease, such as PtoDC3000 (data not shown). On N. tabacum‘Burley’, PsyB728a caused only mild disease symptoms and grew significantly less than P. syringae pv tabaci strains (data not shown). We conclude that A. thaliana and N. tabacum are not good hosts for PsyB728a.

Surprisingly, disease symptoms caused by PsyB728a on N. benthamiana were comparable to those caused by the tobacco pathogens P. syringae pv. tabaci (Pta) 11528 and PtaPTBR2004. These results were obtained using dipping and syringe-infiltration as infection methods (Fig. 1A and B). Three days after infection, bacterial population size of PsyB728a on N. benthamiana was even slightly higher than that of PtaPTBR2004 (Fig. 1C). Infiltrating bacteria into leaves at a concentration of only 1000 cfu ml−1, the endophytic population size of PsyB728a grew from less than 10 cfu per 0.5 cm2 leaf disc at an hour after infiltration to nearly 107 cfu within 3 days. This maximum bacterial titre was 10-fold higher than what we typically observed for PsyB728a on bean using the same concentration of inoculum (data not shown). Furthermore, growth of bacteria to high levels preceded overt disease symptoms, as is typical for bona fide pathogenic P. syringae strains. We conclude that N. benthamiana is a host of PsyB728a.

Figure 1.

PsyB728a originally isolated from snap bean is a strong, T3SS-dependent pathogen on N. benthamiana.
A. Leaves of 4-week-old N. benthamiana plants were dipped into 10 mM MgSO4, 0.02% silwet (surfactant) containing a suspension of PsyB728a or Pta11528 at an OD600 of 0.0003. Pictures of upper and lower sides of leaves were taken on day 7.
B. Nicotiana benthamiana leaves were syringe-inoculated with a suspension of PsyB728a or PtaPTBR2004 at an OD600 of 0.00001. Pictures were taken on day 7.
C. Leaves were infiltrated as in B and each day endophytic bacterial growth was determined from six 0.5 cm2 leaf discs. The asterisk indicates that bacterial populations of PsyB728a were statistically significantly higher than bacterial populations of PtaPTBR2004 on day 3 (P = 0.0005).
D. Leaves of 4-week-old N. benthamiana plants were dipped into 10 mM MgSO4, 0.02% silwet containing a suspension of PsyB728a or the T3SS-deficient strain PsyB728a hrcC::nptII at an OD600 of 0.001. The asterisk indicates that PsyB728a grew significantly more than PsyB728a hrcC::nptII (P = 0.04) 2 days post infection (only endophytically grown bacteria were counted from six 0.5 cm2 leaf discs), showing that growth of PsyB728a on N. benthamiana is T3SS-dependent.
E. PsyB728a and PsyB728a hrcC::nptII sprayed on leaves of 3-week-old N. benthamiana at an OD600 of 0.01 without surfactant. After 3 days epiphytic bacterial growth was determined taking eight 0.5 cm2 leaf discs per strain. All experiments shown in this figure were repeated at least twice with similar results.

Importantly, growth of PsyB728a on N. benthamiana was dependent on an intact T3SS (Fig. 1D). PsyB728a is thought to have an important T3SS-dependent epiphytic growth phase on its bean host in the field (Hirano et al., 1999). On N. benthamiana leaf surfaces, robust survival of PsyB728a was also T3SS-dependent (Fig. 1E). These results indicated that the model plant species N. benthamiana could be used as a host to study the function of PsyB728a effectors. Interestingly, the tobacco pathogen PtaPTBR2004 was also pathogenic on bean, reaching similar population densities as PsyB728a, but causing less severe disease symptoms (data not shown).

The effector repertoires of PsyB728a and PtaPTBR2004 are divergent

As PsyB728a was pathogenic on N. benthamiana, we compared its effector repertoire by DNA–DNA hybridization (Dot Blot) with that of the tobacco pathogen PtaPTBR2004. A similar repertoire would support the idea that these two pathogens are closely related and cause disease on N. benthamiana using similar effectors. Based on the signal strength of spots corresponding to different effector orthologues when probed with PsyB728a genomic DNA, the minimum identity at the DNA level between an effector on the filter and an effector in the probe had to be at least 85% to give a signal above background. Only five PsyB728a effectors gave a signal above background when probed with PtaPTBR2004 DNA: hopAE1, hopAG1, hopAH1, hopI1 and hrpK1 (Table S2). We also found two PtoDC3000 effectors present in PtaPTBR2004, but absent from PsyB728a: hopO1 and hopT1. In a parallel Dot Blot experiment, we found that P. syringae pv. tomato and P. syringae pv. maculicola strains closely related to PtoDC3000 hybridized to at least 27 of 40 effectors (data not shown). Based on these results, PsyB728a is not closely related to PtaPTBR2004. It likely causes disease on N. benthamiana by using largely different effectors from PtaPTBR2004, by using effectors whose sequences have diverged from those of PtaPTBR2004, and/or by relying on the very few effectors that it has in common with PtaPTBR2004.

Several PsyB728a effectors induce cell death when transiently expressed in planta on the non-host N. tabacum as well as on its hosts snap bean and N. benthamiana

The ability of PsyB728a to cause disease on some plant species might be due to the different defence-eliciting activities of effectors on specific hosts. To address this, we tested the ability of each effector to induce cell death on bean, N. tabacum and N. benthamiana. To avoid epistatic interactions with other effectors or virulence factors in P. syringae, we recombined all cloned PsyB728a effectors into plant expression vectors (Table 2) and expressed them as C-terminally HA (haemagglutinin) epitope-tagged proteins in Agrobacterium tumefaciens-mediated transient assays on the three plants. C-terminal HA tags have been widely used to tag P. syringae effectors and have not been found to interfere with effector function [for example Nimchuk et al. (2000)]. We scored the induction of cell death as an indication of a possible defence-eliciting (avirulence) activity. To obtain reliable results, we expressed most effectors in two different A. tumefaciens strains (Table 2) from two different promoters (35S and DEX, see Table 3 for vectors). Figure 2 shows examples of typical plant responses on the three plant species and Table 4 shows the results. Seven effectors caused rapid cell death on the non-host N. tabacum, four elicited rapid cell death on N. benthamiana and four on snap bean. One effector, HopAB1, caused slow and weak cell death on N. benthamiana. Interestingly, HopM1 caused cell death on all three plants. An orthologous effector from PtoDC3000 was previously found to contribute to symptom formation on tomato (Badel et al., 2003).

Table 2.  Bacterial strains.
StrainParent strainDescriptionAntibioticSource/reference
PsyB728a Isolated from P. vulgaris (snap bean)RifampicinAlan Collmer/Loper and Lindow (1987)
PsyB728a hrcC::nptIIPsyB728aT3SS-deficientRifampicinSusan Hirano/Hirano et al. (1999)
PtaPTBR2004 Isolated from N. tabacum Carol Bender
Pta11528 Isolated from N. tabacumRifampicinBrian Staskawicz/Obukowicz and Shaw (1985)
Escherichia coli DH5alpha  Nalidixic acidInvitrogen (Carlsbad, CA)
Escherichia coli DB3.1  Invitrogen (Carlsbad, CA)
Agrobacterium tumefaciens C58C1/pCH32  Kanamycin/ tetracyclineRichard Michelmore/Hamilton (1997)
Agrobacterium tumefaciens GV3101/pMP90  Rifampicin/ gentamycinKoncz and Schell (1986)
PsyB728a ΔhopAB1PsyB728aUnmarked deletion of hopAB1RifampicinThis study
PsyB728a ΔhopM1PsyB728aUnmarked deletion of hopM1RifampicinThis study
PsyB728a ΔhopAA1PsyB728aUnmarked deletion of hopAA1RifampicinThis study
PsyB728a ΔhopZ3PsyB728aUnmarked deletion of hopZ3RifampicinThis study
Table 3.  Plasmids.
VectorParent vectorDescriptionAntibioticSource/reference
  • a. 

    See Table 1 for effector IDs.

  • b. 

    Vector with no origin of replication for P. syringae to express effectors from their native promoters after integration of the construct at the native effector locus.

pDONR207GatewayTM cloning vectorGentamycinInvitrogen (Carlsbad, CA)
pDONR207-effector ID no.apDONR207pDONR207 containing PsyB728a effectorsGentamycinThis study
pCB302-3Binary plant expression vector (35S promoter)Kanamycin, BASTAXiang et al. (1999)
pGREEN Binary plant expression vectorKanamycin, BASTAHellens et al. (2000)
pBAV123pCKTRGatewayTMP. syringae expression vector with C-terminal HA-tagbKanamycinThis study
pBAV139pCB302-3GatewayTM binary plant expression vector (35S promoter and C-terminal HA-tag)Kanamycin, BASTAThis study
pTA7001NABinary plant expression vector (DEX-inducible promoter)Kanamycin, HygromycinAoyama and Chua (1997)
pBAV150pTA7001GatewayTM binary plant expression vector (DEX-inducible promoter and C-terminal GFP-tag)Kanamycin, BASTAThis study
pBAV154pTA7001GatewayTM binary plant expression vector (DEX-inducible promoter and C-terminal HA-tag)Kanamycin, BASTAThis study
pBAV169pCKTRGatewayTMP. syringae expression vector with C-terminal GFP-tagbKanamycinThis study
pBAV179pME6012GatewayTMP. syringae expression vector (nptII promoter and C-terminal HA-tag)KanamycinThis study
pBAV226pME6010GatewayTMP. syringae expression vector (nptII promoter and C-terminal HA-tag)TetracyclineThis study
pCKTR P. syringae integration vector Guttman and Greenberg (2001)
pBSL118mini-Tn5 transposon vectorAmpicillinAlexeyev et al. (1995)
pBAV209pCKTRVector to make unmarked deletionsKanamycinThis study
pBAV211pBAV209HopAB1PsyB72a deletion constructKanamycinThis study
pBAV212pBAV209HopM1PsyB728a deletion constructKanamycinThis study
pBAV214pBAV209HopA1PsyB728a deletion constructKanamycinThis study
pBAV216pBAV209HopZ3PsyB728a deletion constructKanamycinThis study
pBAV179- effector ID no.apBAV179Constructs for ectopic expressionTetracyclineThis study
pBAV226-effector ID no.apBAV226Complementation constructs for deletionsTetracyclineThis study
Figure 2.

Several transiently expressed effector proteins from PsyB728a give differential cell death (CD) responses on the hosts N. benthamiana and snap bean, and the non-host N. tabacum. N. benthamiana, N. tabacum and P. vulgaris (snap bean) leaves were infiltrated with A. tumefaciens strains C58C1 or GV3101 containing plasmids expressing HA- or GFP-tagged effectors from the 35S- or DEX-inducible promoter. For strains expressing effectors from the DEX-inducible promoter, leaves were sprayed with DEX 48 h after infiltration with A. tumefaciens. All leaves were scored for cell death 72 h after infiltration. For each plant species one example of a non-CD-eliciting and a CD-eliciting effector are shown: N. benthamiana (avrRpm1 and avrPto1 respectively), N. tabacum (hopAB1 and avrPto1 respectively), snap bean (avrPto1 and hopM1 respectively). While the cell death in the tobacco species was usually very strong, cell death on snap bean leaves was quite mild. For all effectors experiments were repeated multiple times in at least two different vectors and/or bacterial strains. Only effectors that consistently induced cell death were considered as cell death elicitors.

Table 4. Agrobacterium-mediated transient assays of PsyB728a effectors for cell death elicitation.
NameaN. benthamianaP. vulgaris (snap bean)N. tabacum
  • a. 

    Names based on new unified nomenclature (Lindeberg et al., 2005).

  • b. 

    Cell death caused by HopAB1 was delayed by 1 day and weaker compared with the other cell death elicitors.

  • × indicates that the effectors elicited cell death consistently in at least three assays using A. tumefaciens strains GV3101 and C58C1 with 35S and DEX promoters. Effectors that are not listed did not elicit cell death on any of the three plant species.

AvrPto1× ×
AvrRpm1 ××
HopAE1× ×
AvrB3  ×
HopZ3 ××
HopAA1× ×

To make sure that the effectors that caused no cell death on any of the three plants were expressed, we fused all effectors to green fluorescent protein (GFP) and repeated the assay on N. benthamiana. The same cell death results were obtained. Figure 3 shows that 19 of 21 effectors were detectable by Western blot using anti-GFP antibody. The size of the bands corresponds to the predicted molecular weight of the effector::GFP fusions. All effector::GFP fusions visible on the Western were also fluorescent when A. tumefaciens infiltrated leaf areas were viewed by fluorescence microscopy (data not shown). HrpK1 and HopAK1 were not detected. Therefore for these two proteins, we cannot exclude that they would have caused cell death if they had been expressed.

Figure 3.

Most effector proteins transiently expressed as fusions to GFP in N. benthamiana leaves were detected by Western blots using polyclonal GFP antibody. N. benthamiana leaf discs were collected 3 days after infection with A. tumefaciens C58C1 expressing effectors as fusions to GFP from the DEX-inducible promoter. Protein was extracted and run on an SDS-PAGE gel. Western blots were hybridized using polyclonal GFP antibody. Eighteen of 21 effectors could be detected (indicated by the white asterisks). AvrPto1 (indicated by double asterisks) was not detected because the cell death it induces probably caused its own degradation. In fact, we could detect AvrPto1 when cell death was inhibited by coexpressing HopZ3 (see Fig. 6). HrpK1 and HopAK1 were not detected (indicated by an asterisk). There are three strong bands in the HopAK1 lane, but these are background bands. Unlike the other lanes, there are no bands corresponding to the expected size of HopAK1. Therefore, we conclude that HopAK1 did not accumulate well.

One effector with cell death-inducing activity on N. benthamiana has a quantitative avirulence effect

The cell death observed in the transient expression assay could result from an effector that contributes to disease symptom formation, phytotoxicity, or an HR due to recognition of an avirulence effector. One way to test for a possible avirulence or virulence effect is to create deletion mutants of the different cell death effector genes and test the phenotypes of the resulting strains. Loss of an avirulence effector should result in increased virulence, measured by increased bacterial growth and/or disease symptom development. We tested the phenotype of two individual deletion mutants of hopM1 and hopAA1 respectively. We could not detect any change in growth or symptom formation due to these deletions when infected by inoculating bacteria directly into intercellular spaces of N. benthamiana (data not shown). Spray inoculation of the hopM1 deletion strain also showed no consistent changes in disease or growth relative to wild type (data not shown). However, spray inoculation of the hopAA1 mutant, onto N. benthamiana leaf surfaces repeatedly allowed increased disease lesions accompanied by a trend of higher bacteria counts from leaf washes (Fig. 4). These phenotypes were complemented by a plasmid-borne copy of HopAA1 (Fig. 4). It is impossible to say if the higher number of bacteria caused the lesions or if the lesions allowed bacteria from the apoplast to reach the leaf surface. In either case, this result suggests that HopAA1 has quantitative avirulence activity.

Figure 4.

Deletion of hopAA1 from PsyB728a increases growth on N. benthamiana by spray infection, confirming that HopAA1 has an avirulence activity. PsyB728a, a hopAA1 deletion strain, and a hopAA1 deletion strain complemented with HopAA1 on plasmid pBAV226 were sprayed onto 3-week-old N. benthamiana plants at an OD600 of 0.01. After 3 days epiphytic bacterial counts were determined from leaf washes of eight 0.5 cm2 leaf discs per strain. Pictures were taken on day 4. An asterisk indicates differences in the trends of bacterial growth (P = 0.066). The hopAA1 deletion strain caused increased disease relative to wild type in eight additional repetitions of this experiment. Arrows indicate disease lesions.

Other strong cell death effectors were tested using a different assay due to difficulties in creating stable deletions of avrPto1 and hopAE1 respectively. Specifically, we expressed avrPto1 and hopAE1 in virulent strain PtaPTBR2004 and tested whether the resulting strains caused bacterial growth or symptom changes on N. benthamiana. We found no evidence for avirulence activities of these two effectors in this assay. In contrast, the effector AvrRpt2, used as positive control, had strong avirulence activity, evidenced by the strong reduction in disease of strain PtaPTBR2004 (and PsyB728a) carrying this effector (data not shown).

The cell death caused by the PsyB728a effectors on N. benthamiana is partially dependent on defence signalling components

We used a second test to determine if the rapid cell death elicited by AvrPto1, HopAA1, HopAE1 and HopM1 on N. benthamiana in the transient assays reflected avirulence activities. Specifically, we tested the requirement of defence-related plant genes for effector-induced cell death. We chose four different genes (EDS1, NPR1, RAR1, SGT1) known to be required for the HR induced by some Avr-R recognition events (Azevedo et al., 2002; Liu et al., 2002). We silenced these genes individually in different N. benthamiana plants using virus-induced gene silencing (VIGS) as described by Dinesh-Kumar et al. (2003) using constructs kindly provided by S.P. Dinesh-Kumar (Yale University). A week after introducing the constructs, silencing was confirmed by reverse transcription polymerase chain reaction (RT-PCR) and the transient assays with the four effectors that elicited cell death were repeated. Cell death induced by AvrPto1 was not dependent on any tested defence gene. Figure 5 shows that cell death induced by HopAA1 and HopAE1 was EDS1-dependent and cell death induced by HopM1 was SGT1-dependent. These data support the hypothesis that the cell death induced by HopAA1, HopAE1 and HopM1 was due to their avirulence activities.

Figure 5.

The cell death induced by the PsyB728a effectors HopAA1, HopAE1 and HopM1 depends on known plant defence signalling components. Three-week-old N. benthamiana plants were infected with A. tumefaciens carrying virus-induced gene silencing (VIGS) constructs to silence the defence signalling genes NbEDS1, NbNPR1, NbRAR1 and NbSGT1. Ten days later A. tumefaciens strains expressing AvrPto1, HopAA1, HopAE1 and HopM1 were infiltrated in leaves above the first set of leaves infected with the VIGS strains. Silencing was confirmed by semi-quantitative RT-PCR (M = marker, 1, 2, 3 = DNA was loaded on the gel after 30, 35 and 40 PCR cycles respectively). Cell death was scored 3 days later. Silencing of EDS1 reduced cell death induced by HopAA1 and HopAE1, while silencing of SGT1 reduced cell death induced by HopM1. This experiment was repeated twice with similar results.

HopAB1 and HopZ3 interfere with effector-elicited cell death on N. benthamiana in transient expression assays

How can we reconcile the results from the VIGS experiments suggesting an avirulence activity for HopAE1 and HopM1 with the fact that no apparent avirulence activity during disease was detected? Recently, many P. syringae effectors with cell death-interfering and defence-interfering activities have been reported (Abramovitch et al., 2003; Bretz et al., 2003; Espinosa et al., 2003; Jamir et al., 2004; Lopez-Solanilla et al., 2004). Epistatic effector interactions are thought to influence the interaction of P. syringae with its host. PsyB728a contains a homologue of the related genes hopAB2PtoDC3000 and hopAB1Pph1449B[formerly known as avrPtoB (Abramovitch et al., 2003) and virPphA (Jackson et al., 1999) respectively]. hopAB2PtoDC3000 interferes with cell death induced by various effectors on N. benthamiana (Abramovitch et al., 2003). Therefore, we sought to determine if cell death elicited by AvrPto1, HopAA1, HopAE1, and/or HopM1 on N. benthamiana in the transient assays could be abrogated by other P. syringae effectors. Such effectors with cell death-interfering activity could explain why avirulence activity of some effectors during infection of P. syringae on N. benthamiana was not detected.

We infiltrated A. tumefaciens strains carrying an empty binary vector or binary vectors expressing all cloned PsyB728a effectors with no cell death-eliciting activity, including hopAB1 that showed a delayed and weak cell death in some assays (Table 4), in leaves of N. benthamiana. Two hours later, we infiltrated A. tumefaciens strains expressing the cell death-eliciting effectors AvrPto1, HopAA1, HopM1 and HopAE1. HopAB1 interfered with the rapid cell death elicited by AvrPto1, HopAA1 and HopAE1, but not with the cell death elicited by HopM1. HopZ3 interfered with cell death elicited by all four rapid cell death-eliciting effectors (Fig. 6A and B shows interference of HopAB1 and HopZ3 with cell death caused by AvrPto1).

Figure 6.

The PsyB728a effectors HopAB1 and HopZ3 interfere with cell death caused by the cell death-eliciting effectors in Agrobacterium-mediated transient assays on N. benthamiana leaves. Leaves were infiltrated with A. tumefaciens strains carrying either empty vectors or plasmids expressing PsyB728a effectors that did not give a cell death response in previous assays. Two hours later A. tumefaciens strains expressing PsyB728a effectors that gave a cell death response in previous assays were infiltrated in the same leaf area. Cell death was assayed 3 days later.
A. In the picture on the left A. tumefaciens carrying an empty vector was infiltrated prior to infiltration of A. tumefaciens expressing AvrPto1. AvrPto1 induced cell death. In the picture on the right A. tumefaciens expressing HopAB1 was infiltrated prior to infiltration of A. tumefaciens carrying an empty vector. HopAB1 did not elicit cell death. In the picture in the middle infiltration of A. tumefaciens expressing HopAB1 was followed by infiltration of A. tumefaciens expressing AvrPto1. No cell death was observed in repeated experiments.
B. Same as A, but HopAB1 was replaced with HopZ3. AvrPto1 was expressed as a fusion to GFP. Expression of AvrPto1 in the co-infection assay with HopZ3 (picture in the middle) was confirmed by fluorescence microscopy and Western blot using GFP antibody. AvrPto1 was not detected when expressed by itself (picture on the left). It was probably degraded because of the cell death it induced before it could be detected. All interference assays were repeated at least three times with similar results.

As cell death-interfering activity of a HopAB1 orthologue had previously been reported, we concentrated further studies on HopZ3. First, we made sure that HopZ3 did not interfere with the transfer of DNA from A. tumefaciens to plants or with protein expression in general. For this purpose, we infiltrated an A. tumefaciens strain expressing full-length GFP 2 h after infiltration with A. tumefaciens containing an empty binary vector or a binary vector expressing HopZ3. There was no difference in GFP expression between the two experiments (data not shown). We then specifically confirmed that the cell death elicitors AvrPto1, HopAA1, HopAE1 and HopM1 were still expressed when no cell death was visible. We fused the cell death elicitors to GFP and confirmed the expression of the GFP fusion proteins by fluorescence microscopy. For AvrPto1::GFP, we confirmed expression also by Western blot (Fig. 6B), as we had not been able to determine expression of AvrPto1::GFP in the absence of a cell death interferer (Fig. 3). This may have been due to protein degradation during cell death.

Deletion of hopAB1 or hopZ3 increases epiphytic growth, intracellular growth and/or disease symptoms caused by PsyB728a on N. benthamiana

To determine if the cell death-interfering activities of HopAB1 and HopZ3 were important in PsyB728a growth and symptom formation on N. benthamiana, we separately deleted hopAB1 and hopZ3, respectively, from PsyB728a. Surprisingly, the PsyB728a hopAB1 deletion strain caused more severe symptoms on N. benthamiana than wild-type PsyB728a when inoculated intercellularly. The deletion strains complemented with HopAB1 expressed from a plasmid were indistinguishable from wild-type PsyB728a (Fig. 7A). This suggests that HopAB1 has a quantitative avirulence activity on N. benthamiana, consistent with its delayed and weak cell death-eliciting activity in the transient assay. Although HopZ3 interfered with cell death in the transient assays, we saw no significant change in bacterial growth or disease symptom formation in the PsyB728a hopZ3 deletion strain compared with wild-type PsyB728a in intercellular inoculations (data not shown). However, deletion of hopZ3 resulted in increased disease lesions accompanied by higher bacterial counts in N. benthamiana leaf washes upon spray inoculation (Fig. 7B). The increased disease and bacterial growth was complemented by introduction of the HopZ3 gene on a plasmid. This suggests that HopZ3 also has quantitative avirulence activity, even though it does not induce cell death in the transient assay on N. benthamiana.

Figure 7.

Deletion of the cell death interfering effectors hopAB1 and hopZ3 from PsyB728a has minor but differential effects on the ability to grow and to cause disease on N. benthamiana and snap bean.
A. PsyB728a, PsyB728a ΔhopAB and a complemented strain expressing HopAB1 from the plasmid pBAV226-2C10 were infiltrated using a blunt-end syringe at an OD600 of 0.00001 in N. benthamiana leaves. Endophytic bacterial growth was determined 2 days after infection from eight 0.5 cm2 leaf punches per strain. Pictures were taken on day 5.
B. PsyB728a, PsyB728a ΔhopZ3 and a complemented strain expressing HopZ3 from plasmid pBAV226-1B09 were sprayed onto N. benthamiana leaves at on OD600 of 0.01. After 3 days, epiphytic bacterial counts from the washes of eight 0.5 cm2 leaf discs per strain were determined. Pictures were taken on day 3. For the experiments in A and B, disease symptoms caused by the deletion strains were increased relative to wild type in all of the replicate experiments (at least three). Arrows indicate disease lesions. Additionally, in all trials, the mutant strains showed twofold to 10-fold higher growth relative to the wild-type strains in four or more independent experiments. However, in some trials, these differences in growth were not statistically significant, but nevertheless always showed a consistent trend.
C. PsyB728a, PsyB728a ΔhopAB1 and PsyB728a ΔhopAB1 complemented with HopAB1 expressed from plasmid pBAV226-2C10 were infiltrated into snap bean using a blunt-end syringe at an OD600 of 0.0001. Endophytic bacterial growth was determined 4 days after infection taking eight 0.5 cm2 leaf discs per strain. Pictures were taken on day 5.
D. PsyB728a, PsyB728a ΔhopZ3 and PsyB728a ΔhopZ3 complemented with HopZ3 expressed from plasmid pBAV226-1B09 were infiltrated into snap bean as in C. Endophytic bacterial growth was determined 2 days after infection taking eight 0.5 cm2 leak discs per strain. Pictures were taken on day 5. An asterisk indicates a significant difference compared with the wild-type strain (P < 0.05). For each panel, the leaves shown for each strain are representative of leaves from at least three independent experiments.

Effector genes hopZ3 and hopAB1 have different effects on snap bean compared with N. benthamiana

While HopZ3 and HopAB1 interfered with cell death in the A. tumefaciens-mediated transient expression assays on N. benthamiana, these two effectors both elicited cell death on snap bean (Table 4). Furthermore, HopAA1 induced cell death on N. benthamiana, but not on snap bean. Interestingly, on snap bean, the hopZ3 and hopAB1 deletion strains caused consistently more and less severe disease, respectively, than that caused by wild-type PsyB728a. The growth differences of these strains were repeatedly statistically significant (Fig. 7C and D). This implicates HopZ3 as an avirulence effector and HopAB1 as a virulence effector during snap bean infections. In contrast, the PsyB728a hopAA1 and hopM1 deletion strains showed no bacterial growth or disease symptom differences on snap bean when compared with the wild-type PsyB728a strain (data not shown). Thus, in the context of P. syringae infection, HopZ3 had quantitative avirulence activity in both N. benthamiana and snap bean, even though it only induced cell death on snap bean. However, the role of HopAB1 in the interactions of PsyB728a with snap bean and N. benthamiana was clearly different on these two susceptible hosts.

Effector genes play a role in growth of PsyB728a on the non-host A. thaliana

To determine the role of PsyB728a effectors on a non-host, we tested the effector mutants for their ability to grow on A. thaliana. Surprisingly, we found a significant reduction in the growth of strains deleted for either hopAB1 or hopZ3 (Fig. 8). The reduction in growth was as severe as the hrcC T3SS-deficient mutant that fails to secrete any effectors (Fig. 8). However, we found no difference in the growth of the hopM1 and hopAA1 deletion mutants on A. thaliana (data not shown). This suggests that some effectors can promote P. syringae growth even in a non-host in the absence of significant disease.

Figure 8.

The effector genes hopAB1 and hopZ3 are important for growth of PsyB728a on the non-host A. thaliana. PsyB728a, PsyB728a hrcC::nptII, PsyB728a ΔhopAB1, PsyB728a ΔhopZ3 and the ΔhopAB1 and ΔhopZ3 strains expressing HopAB1 and HopZ3, respectively, from plasmids pBAV226-2C10 and pBAV226-1B09 were infiltrated at an OD600 of 0.001 in leaves of 3-week-old A. thaliana (Ler) plants. Endophytic bacterial growth was determined after 3 days taking eight 0.5 cm2 leaf discs per strain. Growth of the ΔhopAB1 and ΔhopZ3 strains was significantly different from the wild type and respective complemented lines (indicated by an asterisk, P < 0.03). This experiment was repeated twice with similar results.


Bacterial plant pathogens such as P. syringae show strikingly diverse and often large T3SS effector repertoires (Greenberg and Vinatzer, 2003; Lindeberg et al., 2005). We focused here on the strain PsyB728a, whose repertoire is relatively small compared with another commonly studied strain, PtoDC3000 (Greenberg and Vinatzer, 2003; Feil et al., 2005). We found that a large number of effectors previously predicted by bioinformatics are true T3SS substrates. We further established that in addition to being a good pathogen on snap bean, its host of isolation, PsyB728a is also as good a pathogen on the model plant N. benthamiana. Surprisingly, the tobacco pathogen PtaPTBR2004 is also pathogenic on bean, yet is not closely related and has only a few effectors with high similarity to PsyB728a effectors. Thus, either different combinations of effectors (or effector alleles) or few common effectors may promote successful pathogenesis by distinct strains on the same hosts.

The availability of the genomic sequence of PsyB728a permitted the use of a genomic approach to gain a provisional understanding of the possible roles of PsyB728a effectors in interactions with different plant species. We first expressed one or two effectors at the same time in plants and determined whether rapid cell death occurred. We reasoned that activation of rapid cell death could reflect a defence-inducing avirulence activity. In contrast, slow cell death or interference with cell death might reflect a virulence function. Of the 15 effectors predicted to act inside host plants, we detected novel quantitative avirulence and/or virulence roles for several. These are summarized and compared with known effector functions in Table 5. Note that in this study, we used HA and/or GFP epitope tags on the effectors for the transient expression experiments. It is possible that for any given effector, the epitope tag interfered with its ability to induce or interfere with cell death during transient expression studies. However, as it was important to be able to monitor effector production on a genome-wide level, we accepted this compromise in order to survey as many effectors as possible. Because we are mainly drawing conclusions on tagged effectors that complemented P. syringae mutants in vivo, we believe the tags did not interfere with most effectors' functions.

Table 5.  Summary of PsyB728a effector functions on different hosts and non-hosts.
PsyB728a effectorPlantFunction of orthologues in other P. syringae strains
P. vulgaris (host)N. benthamiana (host)A. thaliana (Ler, non-host)
  1. An individual effector was rated as an avirulence factor (Avr) if the corresponding PsyB728a mutant showed increased disease symptoms and/or growth. An effector was rated as a virulence factor (Vir) if the corresponding PsyB728a mutant showed decreased disease symptoms or growth. An equal sign indicates no role of the effector in the symptoms or growth of the PsyB728a on the indicated host, based on the behaviour of the corresponding PsyB728a effector mutant upon infection. Brackets ([]) summarize the responses from the transient A. tumefaciens expression assays from Table 4: no response [N], cell death-inducing [CD], cell death-suppressing [SUP] or delayed/weak cell death [DCD] seen in the transient assays with A. tumefaciens infection. ND, not determined.

HopAA1=[N]Avr [CD]= NDPtoDC3000 orthologue promotes colony formation on A. thaliana leaves
HopM1=[CD]=[CD]= NDPtoDC3000 orthologue promotes disease symptoms on tomato
HopZ3Avr [CD]Avr [SUP]Vir NDNo orthologue in P. syringae has been characterized
HopAB1Vir [CD]Avr [DCD; SUP]Vir NDPph1449B and PtoDC3000 orthologues promote growth and disease or have avirulence activity depending on the host

Unexpectedly, we found that even on the susceptible host N. benthamiana, several effectors induced cell death dependent on known defence signalling pathways. This suggested that some effectors have quantitative avirulence activities in the context of a susceptible infection. Indeed, we confirmed this prediction for the HopAA1 effector on N. benthamiana and for HopZ3 on bean. It was also surprising that PsyB728a effectors elicited largely different responses on the two susceptible hosts (Table 5). This suggested that some effectors might play different roles on the two hosts even though the pathogen was successful on both hosts. Indeed, further characterization of the PsyB728a effector mutants supported this interpretation in some cases. For example, unlike on N. benthamiana, HopAA1 did not induce cell death on bean and we found no evidence for an avirulence role on bean in the context of PsyB728a infection. While it is well known that the same effector can have different roles on hosts with qualitative resistance and susceptibility, respectively (Alfano and Collmer, 2004), different roles for the same effector on two susceptible hosts have not been extensively documented.

For some effectors, we could not detect an avirulence role even when they induced cell death dependent on defence signalling components on at least one susceptible host. This could be because of epistatic interactions, effector activities that are redundant, or because our pathogenicity assay conditions were not optimized. The dependence of cell death caused by some effectors on defence signalling could also reflect virulence activities, as at least one signalling component, a mitogen-activated protein kinase kinase kinase alpha, is important for cell death in both disease resistance and susceptibility (del Pozo et al., 2004). Abramovitch and Martin (Abramovitch and Martin, 2004) proposed that programmed cell death caused by effectors late in an infection aids in the release of nutrients from plant cells and in the dissemination of bacteria. This cell death could require known defence components. The cell death elicitor HopM1 could have a role in this late cell death. Although we could not detect a virulence activity in our assay conditions, HopM1PtoDC3000 is known to contribute to disease symptom formation on tomato (Badel et al., 2003). Interestingly, the A. thaliana mutant acd5 that has reduced ceramide kinase activity shows ectopic programmed cell death that also depends on plant defences and confers increased susceptibility to P. syringae (Greenberg et al., 2000; Liang et al., 2003). Whether HopM1 or other effectors target the ceramide pathway is unknown.

Our cell death induction and interference assays with HopAB1 and HopZ3 also revealed some unexpected results when followed up with PsyB728a effector mutant studies. First, slow cell death elicited by HopAB1 on N. benthamiana likely reflects a quantitative avirulence role during PsyB728a infection of this host. However, HopAB1-induced cell death on bean was correlated with a virulence function. Orthologues of HopAB1 have both virulence and avirulence activities (Abramovitch and Martin, 2005; Lin et al., 2006). Our results show, however, that even in the context of two different susceptible interactions, HopAB1 can have either virulence or avirulence functions. Second, interference of cell death by HopZ3 on N. benthamiana does not reflect a virulence activity of this effector. Rather, HopZ3 has avirulence activity. The lack of cell death associated with an avirulence effector has been previously described in a few instances (Parker et al., 1993; Gassmann, 2005).

HopZ3PsyB728a is a member of the YopJ/AvrRxv effector family of cysteine proteases that are hypothesized to functionally mimic small ubiquitin-like modifier proteases (Orth, 2002; Roden et al., 2004). Therefore, HopZ3 may modify a plant protein(s) that is recognized in plants to trigger a defence response. Once modified by HopZ3, the host protein(s) may be unavailable for interactions with other effectors. This could explain why HopZ3 blocks cell death induced by other effectors. Competition between avirulence effectors has been previously described (Reuber and Ausubel, 1996; Ritter and Dangl, 1996). In the best-studied example, AvrRpt2 interferes with AvrRpm1-elicited defences by proteolysing the plant protein RIN4 that is required for activation of AvrRpm1-dependent defences (Kim et al., 2005). It is also possible that HopZ3 is a general cell death inhibitor like HopAB1PtoDC3000, which interferes with programmed cell death in both plants and yeast (Abramovitch et al., 2003). However, the observation that HopZ3 has an avirulence function during PsyB728a infection of both N. benthamiana and bean during disease and induces cell death on bean does not support this simple view.

Orthologues of some effectors that we studied here have previously been tested for their ability to cause and/or suppress cell death on N. benthamiana (Abramovitch et al., 2003). Of note are the phenotypes induced by AvrPto1PsyB728a and HopAB1PsyB728a respectively. The PtoDC3000 orthologues of these two effectors both interact with the tomato disease resistance protein Pto (Kim et al., 2002). However, the PsyB728a allele of AvrPto1 elicits cell death on N. benthamiana, whereas other related proteins do not, unless the AvrPto-interacting tomato resistance protein Pto is introduced (Abramovitch et al., 2003). Similarly, HopAB1PsyB728a induces cell death on N. benthamiana, whereas HopAB2PtoDC3000 does not.

Interestingly, the HopAB2PtoDC3000 effector has both cell death-inducing and cell death-suppressing domains. However, the cell death-suppressing domain overrides its cell death-inducing activity on N. benthamiana (Abramovitch et al., 2003). Cell death induction likely results from the interaction of the HopAB2PtoDC3000 amino terminus with the Rsb resistance protein predicted to be encoded by a Pto gene family member (Abramovitch and Martin, 2005). The cell death suppression activity of HopAB2PtoDC3000 requires an E3 ubiquitin ligase activity that resides at its carboxy terminus (Abramovitch et al., 2006; Janjusevic et al., 2006). HopAB1PsyB728a may function oppositely from HopAB2PtoDC3000, showing a stronger ability to induce avirulence/cell death than to promote virulence/cell death-interference during disease. Because neither AvrPto1PtoDC3000 nor HopAB2PtoDC3000 induce cell death on N. benthamiana, whereas both AvrPto1PsyB728a and HopAB1PsyB728a do, the predicted N. benthamiana Rsb protein may have higher affinity for the PsyB728a effectors than for their PtoDC3000 orthologues. We did not determine whether the cell death-suppressing activity of HopAB1PsyB728a is as general as that found for HopAB2PtoDC3000 (Abramovitch et al., 2003). We cannot exclude that, as hypothesized for HopZ3, HopAB1PsyB728a suppresses faster, stronger and possibly avirulent cell death effectors in the transient assay by modifying a defence component that is also targeted by other effectors.

Most known avirulence effectors show strong qualitative effects on the interaction between bacteria and host plants. Specifically, the presence of an avirulence effector causes a strong reduction in pathogen fitness on plants with the cognate resistance gene. This is usually the basis for the discovery of avr genes. Our characterization of PsyB728a effectors highlights the fact that at least three effectors (HopAA1, HopZ3 and HopAB1) show quantitative avirulence functions on one or two susceptible hosts. Such quantitative avirulence effectors may be relatively common, albeit not yet well documented. Interestingly, Chang et al. (2002) have argued that some tomato accessions harbour ‘minor recognition determinants’ that recognize AvrPto1. In our system, weak recognition could result from the specific sequence of the effector or the resistance determinant, or both. Which ever is the case, quantitative avirulence–R protein interactions may explain the observation that even during susceptible interactions, P. syringae triggers plant defences, albeit slower and/or weaker than during resistant interactions [for example, see the study by Tao et al. (2003)]. The retention of quantitative avr genes could be a way for pathogens to avoid becoming too aggressive on host plants. Very strong pathogenicity could cause the host tissue to die too fast, leading to reduced host fitness that would deprive the pathogen of its long-term source of nutrition.

In summary, although PsyB728a causes disease on both bean and N. benthamiana, specific individual effectors contribute differently to the outcome of the infection on the two hosts. There is overlap in the roles of some effectors in promoting or suppressing fitness and/or disease symptoms on the different plant species (Table 5). However, in no case does one effector have the same apparent role on all plant species tested. There is also a diversity of cell death-modulating activities on different plant species. Strikingly, two effectors also promote the growth of PsyB728a on the non-host A. thaliana. This suggests that effectors can increase bacterial fitness in the absence of overt disease. Considering that P. syringae probably evolved long before the advent of agriculture, these results may indicate that only pathogens with a broad host range and good ‘survival skills’ on non-hosts could spread through a diverse plant community. We thus suggest that the effector repertoires of P. syringae strains are large to accommodate the different host environments that bacteria may encounter. Only since the beginning of agriculture, some plant pathogens may have reduced their host range by losing effectors and specializing on a single crop plant.

Experimental procedures

Strains and media

Strains and plasmids used in this paper are listed in Table 2. E. coli and A. tumefaciens strains were grown in LB medium at 37°C and 30°C respectively (Sambrook et al., 1989). P. syringae strains were grown at 30°C in KB medium (King et al., 1954). Antibiotics were used at the following concentrations: gentamycin (10 μg ml−1), kanamycin (50 μg ml−1), tetracyclin (12.5 μg ml−1), rifampicin (34 μg ml−1), nitrofurantoin (100 μg ml−1).

Vector construction

We modified the plant expression vectors pTA7001 (Aoyama and Chua, 1997) kindly provided by Nam-Hai Chua (Rockefeller University) and pCB302-3 (Xiang et al., 1999), kindly provided by C. Xiang and D. Oliver (Iowa State University) for use with the GatewayTM phage lambda red-based recombinational cloning system (Invitrogen, Carlsbad, CA, USA). We replaced the hygromycin resistance gene of pTA7001 with the bar gene for BASTA resistance. We restriction-digested pTA7001 with Sbf1 and SpeI, purified the T-DNA fragment and ligated it with pBS SK+ (Stratagene, La Jolla, CA, USA) digested with PstI and XbaI. To replace the hygromycin resistance gene with the bar gene, we amplified the resulting vector using Pfu Turbo (Stratagene, La Jolla, CA) with primers upstream and downstream of the hygromycin resistance gene within the nos promoter and within the nos terminator, respectively, using the primers ‘nos promoter R’ and ‘nos terminator F’, which point away from each other (all primers used for vector construction are listed in Table S3). We used Pfu Turbo to amplify the bar gene from a pGREEN vector (Hellens et al., 2000) with the primers ‘nos promoter F’ and ‘nos terminator R’ located ‘head to tail’ with the primers used to amplify the vector. We ligated the two PCR fragments by blunt-end ligation, selected a clone with the correct insert orientation and sequenced the ligation junctions. We amplified the T-DNA from pBS SK+ with M13 F and Pasta-R, digested the PCR product with SpeI and Sbf1 and ligated it to the vector fragment from the SpeI/Sbf1 digest of pTA7001. We then added the ‘GatewayTM cassette’ including a downstream HA tag or the E-GFP gene into the XhoI and SpeI cloning sites of this vector. To do this we took advantage of similar GatewayTMP. syringae expression vectors with HA tag (vector pBAV123) or a C-terminal E-GFP fusion (vector pBAV169) that we had previously constructed (our unpublished data). We amplified the Gateway cassette including the HA tag or E-GFP, respectively, by PCR using Pfu Turbo from the P. syringae expression vectors with primers to which we had added an XhoI site at the 5′ end of the Gateway cassette (gatex53F) and an SpeI site at the 3′ end of the HA tag (new gatex HA R) or E-GFP (E-GFP R) gene respectively. We digested the PCR products and pTA7001 with XhoI and SpeI and ligated them to each other.

We added the GatewayTM cassette followed by an HA tag to the cloning site of pCB302-3 by digesting the vector with XbaI and ligating it to a PCR product amplified from the P. syringae Gateway expression vector pBAV123 containing the GatewayTM cassette followed by an HA tag using primers with an XbaI and SpeI overhang respectively (gat35S F and gat35S R).

To express vectors in P. syringae we used pBAV179, a vector derived from pBAV178 (Vinatzer et al., 2005) in which the avrRpt2101−255 sequence was replaced by an HA tag. pBAV179 is derived from pME6012, a high-copy plasmid derivative of pVS1 (Heeb et al., 2000). We used a low-copy version of pBAV179, called pBAV226 for complementation of deletion strains. pBAV226 contains a Gateway cassette followed by the HA tag in pME6010 with the wild-type pVS1 origin of replication (Heeb et al., 2000).

To create unmarked deletions, we constructed a vector called pBAV209 based on the nptI-sacB-sacR cartridge (Ried and Collmer, 1987), oriR6K and RP4oriT (Alexeyev et al., 1995), and the origin of replication of pCKTR (Guttman and Greenberg, 2001). We amplified the origin of replication and the nptII gene from pCKTR using the primers kick_npt2 F and kick Cm R and combined it with oriR6K, RP4oriT and Apr from pBSL118 amplified with primers transposase F and Rk6ori R. We then amplified the nptI-sacB-sacR cartridge from the vector pBAV62 (our unpublished data) using the primers R6Kori2F and Trans2F and ligated it to a PCR product obtained by round the vector-PCR (Guttman and Greenberg, 2001) of pBAV208 using the primers kick_npt2 F and kick_npt2R.

Effector cloning

We cloned effectors using the ‘one tube’ PCR protocol described in the Gateway manual (Invitrogen, Carlsbad, CA). We used primer3 at for primer design. We designed the forward primer maintaining 11–14 nucleotides of native 5′ UTR. We changed the default settings in the Primer3 program as follows: Max 3′ Stability 10.0, Primer Size Max 24, Primer Tm Min 50.0, Opt 60.0. Max 70.0. We designed the reverse primer to end with the T of the STOP codon. Whenever Primer3 did not find a forward primer that started exactly at 11 or 14 nucleotides upstream of the START codon, we added nucleotides to the 5′ end of an otherwise acceptable forward primer to reach those positions. Whenever primer3 did not find a reverse primer that ended with the T of the STOP codon we added nucleotides to the 5′ end until we reached that position. To allow N-terminal fusions, we also made sure that there was no in frame STOP codon in the 5′ UTR and no out of frame START codon. If an in frame STOP codon was present in the 5′ UTR, we changed the forward primer in one nucleotide of this STOP codon to eliminate it (only if it was close to the 5′ end of the primer). If the first two nucleotides of the forward primer were either GA, AA or AG we changed the first nucleotide to a T to avoid a STOP codon from the fusion to the Gateway sequence, which we added to the forward and reverse primers as described in the Gateway manual.

For amplification, we used Pfu Turbo and the thermocycler settings proposed in the GatewayTM manual using 25 μl reaction volumes. Bands of the right size from each reaction were cleaned using two units of SAP and two units of ExoI and cloned into pDONR207 following the GatewayTM manual for overnight BP reactions using reaction volumes and reagents of 5 μl total. The next day we added proteinase K to the samples for 10 min and then transformed them into E. coli. At least three gentamycin-resistant colonies were sequenced for each construct.

attL × attR (LR) reactions in 5 μl volumes were performed using a 1:10 or 1:20 fold dilution of the vector containing an effector. E. coli was transformed and resulting colonies were analysed by PCR and restriction digest. Triparental matings from E. coli into P. syringae and electroporation of P. syringae were performed as described (Guttman and Greenberg, 2001; Guttman et al., 2002). Triparental matings from E. coli into A. tumefaciens were performed using E. coli RK600 as helper strain and by simply streaking out donor, helper and recipient strain on the same area of an LB agar plate without selection. After 24 h of growth at 30°C a loop-full of bacteria was taken from the plate and streaked out to single colony on a plate with antibiotics for selection of the recipient strain and the plasmid construct. Single colonies were visible 2 days later and analysed by PCR.

Plant growth and plant infections

Nicotiana benthamiana and P. vulgaris (bean) cv ‘Blue Lake 274’ were grown at 24°C under 16 h light conditions in walk-in growth chambers or reach-in incubators. Humidity in walk-in growth chambers was set on 80%, whereas humidity in incubators was not regulated, but was kept high by keeping flats of water inside. For syringe inoculations, 1-month-old N. benthamiana plants were infected with a bacterial suspension at an optical density at 600 nm (OD600) of 0.00001 in 10 mM MgSO4 with a blunt-end syringe. Four areas of approximately 1 cm in diameter of each of the two to three largest leaves were infected. N. benthamiana plants were kept at 21°C for 1 day after infiltration to increase relative humidity. Bean plants were infected at 3 weeks of age using syringe inoculations of the two largest leaves. Areas of approximately 1 cm in diameter were infiltrated with a bacterial suspension in 10 mM MgSO4 of an OD600 at 0.00003 or 0.0001. For growth curves, eight punches 8 mm in diameter were taken for each strain, ground up in 200 μl of 10 mM MgSO4 and dilution-plated.

For spray inoculations, P. syringae strains were diluted to an OD600 of 0.01 in 10 mM MgSO4 and sprayed onto 17- to 21-day-old N. benthamiana plants. After 2 or 3 days, eight leaf punches 8 mm in diameter were taken from the second to fourth leaf. Leaf punches were placed into 1 ml of 10 mM MgSO4 and vortexed for 5 s at maximum speed to detach leaf-associated biomass. Colony-forming units were determined after dilution-plating of the samples.

To compare the pathogenicity of PsyB728a with that of the tobacco pathogen Pta11528 and that of the type III secretion deficient PsyB728a hrcC::nptII mutant, whole 4-week-old N. benthamiana plants were dipped in 500 ml of bacterial suspensions containing MgSO4 at a 10 mM concentration and the surfactant silwet at a concentration of 0.02%. The OD600 for the first comparison was 0.0003 and for the second comparison 0.001. For the first comparison, symptoms were compared by taking pictures 7 days after infection. For the second comparison, bacteria were extracted from six leaf punches for each strain, ground up in 200 μl of 10 mM MgSO4 and dilution-plated.

Arabidopsis thaliana accession ‘Landsberg erecta’ plants were grown at 16 h day and leaves of 3-week-old plants were syringe-inoculated at an OD600 of 0.001 as previously described (Greenberg et al., 2000).

For all infections, the optimum day(s) on which differences in growth between the wild type and each mutant were detectable was initially determined by monitoring growth over several days. Replicate experiments then focused on the days on which the differences were optimally different. The starting inocula of mutant and wild-type strains were verified to be equivalent in each infection experiment.

Dot Blots

In this study 2 μl of plasmid DNA (at an approximate concentration of 0.2 μg μl−1) extracted with a Qiagen Plasmid Miniprep kit was diluted in 1 ml 2× SSC (300 mM sodium chloride, 30 mM sodium citrate) in a 96-deep-well plate. A 1 ml aliquot of 0.1 N NaOH was added and mixed by pipetting. This DNA preparation was fixed to a positively charged Nylon membrane (GeneScreen Plus, PerkinElmer, Norwalk, CT, USA) using a Dot Blot apparatus (Minifold I Dot-Blot System, Schleicher and Schuell, Dassel, Germany). A 200 μl aliquot of 2× SSC was vacuumed through the membrane before and after applying 200 μl of the DNA suspension. DNA was cross-linked to the membrane using a GS Gene Linker (Bio-Rad, Hercules, CA, USA). Filters were hybridized following standard procedures (Sambrook et al., 1989). As probe [32P] random prime-labelled AluI-digested whole genomic bacterial DNA was used. Filters were washed once for 20 min in 2× SSC 0.1% SDS and twice for 30 min in 0.5× SSC 0.1% SDS and exposed for up to 4 days.

Transient expression assays

Agrobacterium tumefaciens strains GV3101 and C58C1 containing vector pBAV154 or pBAV150 expressing effectors from the 35S promoter or DEX promoter, respectively, were grown on LB agar plates, then transferred to 5 ml of LB and grown overnight at 30°C. Overnight samples were diluted 1:3 with induction medium [1 litre: K2HPO4 10.5 g; KH2PO4 4.5 g; (NH4)2SO4 1 g; Na Citrate(× 2H2O) 0.5 g; 1 M MgSO4 1 ml; glucose 2 g; glycerol 5 ml; MES 1.95 g; vacuum sterilized] containing 50 μg ml−1 acetosyringone and grown for 5 h at 30°C. Samples were spun down and diluted to OD600 = 0.4 in infiltration medium (1 l: MS medium 2.2 g; MES 1.95 g; pH 5.6 with KOH) containing 150 μg ml−1 acetosyringone. Four-week-old N. benthamiana plants were then infected with A. tumefaciens using a blunt-end syringe. Dexamethasone (30 μM) in a solution of 0.1% Tween 20 in water was sprayed onto plants infected with Dex-inducible vectors 48 h after initial infection and observed 24 h after Dex-induction. All plants were observed 48 h and 72 h after infection. For co-infections (interference assay), the two bacterial samples were diluted to OD600 = 0.4. The effector being tested for cell death interference was infected 2–4 h prior to infection with the cell death elicitor.

Protein extraction and Western blot

Leaf discs were taken at 24 h and 36 h after A. tumefaciens infection. At each time point, two 0.5 cm2 leaf discs were frozen in liquid nitrogen, ground while still frozen, and further homogenized in 75 μl of resuspension buffer [2 ml: 2.22 sampling buffer 1 ml; 50% glycerol 200 μl; beta-mercaptoethanol 0.35 μl; 20% SDS 100 μl; 10× PBS 100 μl; H2O 600 μl]. The samples were then heated at 95°C for 8 min and centrifuged for 1 min at 16 000 g. The supernatants were run on a 12% SDS-PAGE gel and transferred to a polyvinylidene fluoride membrane. GFP polyclonal antibody (Clontech, Mountain View, CA) was used at a 1:500 dilution and secondary anti-rabbit antibody (Pierce, Rockford, IL) at a 1:20 000 dilution in BLOTTO (20 mM Tris, pH 7.5/154 mM NaCl/0.1% Tween 20/5% Carnation non-fat dry milk). Detection was achieved by chemiluminescence (Pierce, Rockford, IL).


Virus-induced gene silencing experiments in N. benthamiana were performed as described by Dinesh-Kumar et al. (2003). We thank Dr Dinesh-Kumar (Yale University) for kindly providing us with all silencing constructs. Cell death assays were performed 7 days after silencing constructs were infiltrated. Total RNA was extracted from silenced and non-silenced N. benthamiana using Trizol solution (Invitrogen, Carlsbad, CA). RT-PCR was performed using a Superscript one-step RT-PCR kit (Invitrogen, Carlsbad, CA). A 100 ng aliquot of RNA was used as template. After DNA amplification in each cycle, the PCR products were examined by electrophoresis in a 0.8% agarose gel.

Construction of individual and multiple effector deletion strains

Approximately 800 bp of upstream and 800 bp of downstream region of each gene to be deleted were cloned into pBAV209 (Table 3). pBAV209 was digested with ClaI and ApaI. The upstream DNA fragment was amplified with a forward primer that had a ClaI restriction site overhang and a reverse primer with an EcoRI overhang. The downstream fragment was amplified with a forward primer with an EcoRI overhang and a reverse primer with an ApaI overhang. PCR products were digested with the respective enzymes. Digested PCR products and digested vector were run on an agarose gel and gel-purified using the Qiagen PCR purification kit (Qiagen, Valencia, CA, USA). Both PCR fragments were ligated to the vector in a single ligation reaction. Plasmids were confirmed by PCR and sequencing.

Plasmids were mated into PsyB728a and colonies were confirmed by PCR. Colonies were re-streaked out to single colonies on KB plates containing 10% sucrose. Colonies were re-streaked on KB plates and KB kanamycin plates to select for second recombination events that lead to vector loss. Kanamycin-sensitive colonies were tested by PCR for gene deletions. Deletion strains were complemented with effectors expressed from the nptII promoter on the low-copy pBAV226 plasmid described earlier.


This work was supported by a subcontract from the National Science Foundation Grant 00RA6325-DBI0211923 from the Plant Genome Program and a University of Chicago Faculty Fund award to J.T.G. B.A.V. was supported by an individual postdoctoral fellowship award from the National Institute of Health (NRSA 1 F32 G066606-02) and Virginia Tech start up funds. We thank The University of Chicago undergraduate student C. Woodard for her help with the cloning of effectors. We thank the DOE Joint Genome Institute for granting access to PsyB728a sequence data ahead of publication. We thank R.W. Michelmore and members of his laboratory for helpful discussions.