Activation of a COI1-dependent pathway in Arabidopsis by Pseudomonas syringae type III effectors and coronatine


For correspondence (fax +1 785 532 5692; e-mail


Gram-negative bacteria use a variety of virulence factors including phytotoxins, exopolysaccharides, effectors secreted by the type III secretion system, and cell-wall-degrading enzymes to promote parasitism in plants. However, little is known about how these virulence factors alter plant cellular responses to promote disease. In this study, we show that virulent Pseudomonas syringae strains activate the transcription of an Arabidopsis ethylene response factor (ERF) gene, RAP2.6, in a coronatine insensitive 1 (COI1)-dependent manner. A highly sensitive RAP2.6 promoter-firefly luciferase (RAP2.6-LUC) reporter line was developed to monitor activities of various bacterial virulence genes. Analyses of P. syringae pv. tomato DC3000 mutants indicated that both type III secretion system and the phytotoxin coronatine are required for RAP2.6 induction. We show that at least five individual type III effectors, avirulence B (AvrB), AvrRpt2, AvrPphB, HopPtoK, and AvrPphEPto, contributed to RAP2.6 induction. Gene-for-gene recognition was not involved in RAP2.6 induction because plants lacking RPM1 and RPS2 responded normally to AvrB and AvrRpt2 in RAP2.6 expression. Interestingly, the role of coronatine in RAP2.6 induction can be partially substituted by the addition of avrB in DC3000, suggesting that AvrB may mimic coronatine. These results suggest that P. syringae type III effectors and coronatine act by augmenting a COI1-dependent pathway to promote parasitism.


The recent completion of genome sequences for several phytopathogenic bacteria, including Xanthomonas campestris, Ralstonia solanacerum, and Pseudomonas syringae, uncovered a large number of genes with a putative role in bacterial pathogenesis. For example, the P. syringae pv. tomato DC3000 contains 276 genes encoding potential virulence determinants such as type III secretion system (TTSS), effector proteins secreted by TTSS, cell-wall-degrading enzymes, enzymes required for the biosynthesis of the phytotoxin coronatine, exopolysaccharide alginate, and receptors responsible for sensing the host environment (Buell et al., 2003).

A major challenge that follows the exciting breakthrough brought about by genome sequencing is to identify the function of these virulence factors in planta. In particular, type III effector genes that are thought to play a crucial role in bacterial virulence and host range determination often encode proteins of unknown function. Many plant bacterial effectors were first identified as avirulence (Avr) proteins that specify resistance (R) gene-mediated disease resistance (White et al., 2000). The biochemical basis of type III effector virulence function is only beginning to emerge for a handful of effectors. For example, P. syringae effectors AvrB, AvrRpm1, AvrPto, and AvrRpt2 confer virulence function on plants lacking cognate resistance genes (Ashfield et al., 1995; Chen et al., 2000; Ritter and Dangl, 1995; Shan et al., 2000b). AvrRpt2 also suppresses defense responses specified by AvrRpm1–RPM1 interaction (Chen et al., 2000; Reuber and Ausubel, 1996; Ritter and Dangl, 1996). AvrRpt2 appears to be a cysteine protease whose presence triggers the degradation of the RIN4 protein (Axtell and Staskawicz, 2003; Axtell et al., 2003; Mackey et al., 2003). Pseudomonas effectors that suppress gene-for-gene resistance also include VirPphA and AvrPphF from P. syringae pv. phaseolicola and AvrPtoB from DC3000 (Abramovitch et al., 2003; Jackson et al., 1999; Tsiamis et al., 2000). AvrPto appears to function as a virulence factor by suppressing host genes involved in callose deposition (Hauck et al., 2003). AvrPto, AvrB, and AvrRpm1 target plant plasma membrane through myristoylation (Nimchuck et al., 2000; Shan et al., 2000a). At least AvrPto requires myristoylation for the virulence function (L. Shan and X.T., unpublished results). However, the majority of type III effectors do not produce measurable virulence effects. Consequently, their role in bacterial parasitism has been difficult to study.

Many P. syringae strains, such as P. syringae pv. glycinea, P. syringae pv. maculicola, and P. syringae pv. tomato, also produce the phytotoxin coronatine. In DC3000, type III effectors and coronatine biosynthetic genes are regulated in a coordinated fashion (Boch et al., 2002), but TTSS per se is not required for coronatine secretion (Penaloza-Vazquez et al., 2000). Coronatine contributes to virulence in P. syringae pv. tomato by promoting bacterial growth and chlorosis in the plant (Mittal and Davis, 1995). Coronatine consists of two moieties, coronafacic acid (CFA) and coronamic acid (CMA). They are structurally similar to jasmonic acid (JA) and aminocyclopropane carboxylic acid (ACC), the immediate precursor of ethylene, respectively (reviewed by Bender et al., 1999). Coronatine is known to mimic JA (Feys et al., 1994), and it may also mimic ethylene because of the similarity of CMA to ACC (Ferguson and Mitchell, 1985). JA and ethylene often act synergistically in response to wounding or pathogen infection (reviewed by Kunkel and Brooks, 2002). Coronatine-producing Pseudomonas bacteria may manipulate the host signaling pathways through these hormones to promote parasitism. However, early molecular events involved in this process remain poorly understood.

Here, we describe the use of a transgenic reporter line carrying the RAP2.6 promoter fused with the firefly luciferase gene LUC to investigate host cellular activities modulated by DC3000 type III effectors and coronatine. RAP2.6 (At1g43160) is an Arabidopsis ethylene response factor (ERF) family transcription factor (Okamuro et al., 1997) that is strongly induced by virulent P. syringae strains. The induction of RAP2.6 largely depended on a coronatine insensitive 1 (COI1)-mediated pathway. An intact TTSS, a number of type III effectors, and coronatine are all required for the RAP2.6 induction by the wild-type DC3000 bacterium. Bacterial mutants unable to produce coronatine or defective in TTSS poorly induce the RAP2.6 promoter. The coronatine and TTSS mutants fully complemented each other to activate the RAP2.6-LUC reporter when co-inoculated into plants, suggesting that type III effectors and coronatine acted coordinately to induce RAP2.6 expression. However, the addition of an exogenous effector gene, avrB, allowed DC3000 to induce RAP2.6 in the absence of coronatine, suggesting that AvrB acted, at least in part, to mimic coronatine. The results support a view that bacterial type III effectors and coronatine target the COI1 pathway to promote parasitism.


Association of RAP2.6 expression with disease susceptibility

A number of Arabidopsis ERF genes are induced upon pathogen infection (Chen et al., 2002; Onate-Sanchez and Singh, 2002). One of these ERF genes, RAP2.6, was induced strongly by two virulent strains, DC3000 and P. syringae pv. maculicola ES4326 (R. Warren, P.H., and J.Z., unpublished results). Interestingly, microarray analysis indicated that RAP2.6 was more responsive to virulent bacteria in the npr1 and pad4 mutants and NahG transgenic plants (Chen et al., 2002). These plants are enhanced in disease susceptibility because of a deficiency in salicylic acid (SA)-mediated defense. These experiments examined gene expression at a relatively late time point (30 h) after infection. To further understand the potential role of RAP2.6 expression in plant–bacterial interactions, we examined the RAP2.6 transcripts in the wild type Columbia (Col-0), the resistant mutant cpr5-1 (Bowling et al., 1997) and the susceptible mutant npr1-1 (Cao et al., 1994) at earlier time points after inoculation with DC3000. As shown in Figure 1, RAP2.6 transcripts were induced 12 h after inoculation in all plants tested. The transcript level was the lowest in cpr5-1 plants, and it further declined at 24 h. In contrast, npr1 plants showed the greatest RAP2.6 transcript level at 24 h. Thus, the expression of RAP2.6 in response to DC3000 appeared to be positively correlated with disease susceptibility and negatively with disease resistance of plants. cpr5 and npr1 are known to be altered in SA accumulation in plants during bacterial infection. The latter may suppress RAP2.6 expression through SA/JA antagonism, a possibility that remains to be tested.

Figure 1.

Association of RAP2.6 expression with disease susceptibility.

Wild-type Col-0 (WT), cpr5-1, and npr1-1 plants were infiltrated with 10 mm MgCl2 (mock), or P. syringae pv. tomato DC3000 at 2 × 106 cfu ml−1, and RNA was isolated from tissues collected before (0 h) or after (6, 12, and 24 h) infiltration. The ethidium bromide gel pictures indicate equal loading of RNA samples. The Northern experiments were independently repeated two times with similar results.

The RAP2.6 promoter is bacterial inducible

To better monitor the RAP2.6 gene activation in response to Pseudomonas infection, we developed a reporter transgenic line by fusing a 2-kb fragment upstream of the RAP2.6 open-reading frame to the reporter gene LUC. The resulting construct, RAP2.6-LUC, was transformed into Arabidopsis Col-0, and 20 independent transgenic lines were obtained. To determine if the transgenic plants confer bacterial-induced LUC expression, we inoculated the plants with P. syringae pv. maculicola ES4326 (avrB). Previous experiments had indicated that this strain gave a strong and early induction of RAP2.6 transcripts (Figure 2a). The RAP2.6 transcript level reached a maximum at 6 h after bacterial inoculation, and it was not significantly affected by the bacterial concentration (from 2 × 108 to 2 × 106 cfu ml−1). Mock-inoculated plants did not show an induction at 6 h, but a transient induction was observed 1 h after infiltration (data not shown). Figure 2(b) shows the induction of LUC transcripts in three independent transgenic lines after mock and bacterial infiltration. The LUC transcripts increased dramatically 6 h after pathogen infiltration in all transgenic lines compared with mock infiltration, which is consistent with the induction of the endogenous RAP2.6 gene. Lines 13 and 14 that contained multiple copies of transgene (data not shown) displayed a stronger induction of the reporter gene than did line 12 that contained a single-copy transgene. All three lines also displayed a strong bacterial-inducible luciferase activity. Figure 2(c,d) shows bacterial-induced luciferase activity in line 12, as determined by in vivo luminescence imaging and luminometer assay. These results demonstrated that the RAP2.6 promoter is pathogen inducible. The homozygous line 12 was thus chosen for all luciferase experiments described below.

Figure 2.

The RAP2.6 promoter is induced by bacterial infection.

(a) Wild-type Col-0 plants were vacuum-infiltrated with P. syringae pv. maculicola ES4326 (avrB) at the indicated concentrations, and tissues were collected at different times for RNA isolation. Control plants were infiltrated with 10 mm MgCl2 (mock).

(b) RAP2.6-LUC transgenic plants (Lines 12, 13, and 14) were inoculated with water (M) or ES4326 (avrB) (B) at 2 × 107 cfu ml−1. RNA was isolated from tissues collected before (U) or 6 h after inoculation, and the RNA blot was hybridized with a radiolabeled LUC DNA fragment.

(c) A pseudo-color luminescence image (left) and bright-field image (right) of untreated transgenic leaves (U), or leaves 6 h after inoculation with water (M) or ES4326 (avrB) (B).

(d) Relative luciferase activity after inoculation with water (mock) or 2 × 107 cfu ml−1 ES4326 (avrB) (bacteria). Luciferase assays were performed 6 h after treatment. Each data point consisted of four leaves. Error bars indicate SEs. The ethidium bromide gel pictures in panels (a) and (b) indicate equal loading of RNA samples. The experiments were repeated four times with similar results.

Association of RAP2.6 promoter activity with P. syringae pathogenicity

There are over 40 pathovars of P. syringae classified according to their host range among different plant species (Hirano and Upper, 1990). Only certain strains of P. syringae pv. tomato and P. syringae pv. maculicola are known to infect Arabidopsis (Davis et al., 1989). To test whether the RAP2.6 promoter activity was correlated with bacterial pathogenicity, we inoculated the RAP2.6-LUC transgenic plant with different P. syringae strains that are either highly virulent or non-pathogenic on Arabidopsis. DC3000 is a highly virulent pathogen on Arabidopsis (Davis et al., 1989; Dong et al., 1991); P. syringae pv. tomato strain T1, P. syringae pv. tabaci R11528 race 0, and P. syringae pv. phaseolicola NPS3121 are non-pathogenic on Arabidopsis (Davis et al., 1989; Lu et al., 2001; Yu et al., 1993; P.H. and J.Z., unpublished results). DC3000 bacteria dramatically induced LUC at 12 h after inoculation (Figure 3a). The RAP2.6 promoter activity increased further at 24 h and was about 20-fold higher than that in mock-inoculated leaves. A strong induction of luciferase was also observed when plants were inoculated with another virulent strain, ES4326 (see Figure 5). In contrast, the three non-pathogenic strains showed no significant LUC induction throughout the entire experiment. No bacterial growth in the plant was detected for any of the strains in the first 12 h (Figure 3b), indicating that the induction of LUC by DC3000 was not a result of bacterial growth. Overall, the results indicate that the RAP2.6 promoter activity was correlated with the pathogenicity of the P. syringae strains used.

Figure 3.

Association of RAP2.6 promoter activity with P. syringae pathogenicity.

(a) Leaves were collected 12 and 24 h after inoculation, and the luciferase activity was analyzed by using a luminometer.

(b) Bacterial growth in the plant. Each data point consisted of four replicates. Error bars indicate SEs. The experiments were repeated four times with similar results.

RAP2.6-LUC transgenic plants were hand-injected with water (mock), P. syringae pv. tomato DC3000 (DC3000), P. syringae pv. tomato T1 (P. s. t. T1), P. syringae pv. tabaci R1152 race 0 (P. s. tabaci), and P. syringae pv. phaseolicola NPS3121 (P. s. ph.) at the concentration of 2 × 106 cfu ml−1.

Figure 5.

avr genes accelerate RAP2.6 activation.

(a) avr genes in DC3000 and ES4326 accelerate RAP2.6 expression. RAP2.6-LUC transgenic plants were hand-injected with water (mock), DC3000, DC3000 (avrB), DC3000 (avrRpt2), DC3000 (avrPphB), DC3000 (avrRps4), P. syringae pv. maculicola ES4326 (P. s. m), or P. s. m. (avrB) at 2 × 106 cfu ml−1. Luciferase activity was determined by using a luminometer at 6 and 9 h after inoculation. Each data point consisted of at least three leaves. Error bars indicate SEs. The experiments were repeated three times with similar results.

(b) avr genes in P. syringae pv. phaseolicola do not activate RAP2.6 promoter. RAP2.6-LUC transgenic plants were hand-injected with water (mock), or 2 × 106 cfu ml−1 of DC3000, DC3000 (avrB), P. syringae pv. phaseolicola NPS3121 (P. s. ph.), P. s. ph. (avrB), P. s. ph. (avrRpt2), P. s. ph. (avrPphB), or P. s. ph. (avrRps4). Luciferase activity was determined as in (a). The experiments were repeated three times with similar results.

(c) avr genes in P. syringae pv. phaseolicola induce HR. Plants were inoculated with 1 × 108 cfu ml−1 bacteria (same strains as in (b) and photographed 20 h after inoculation.

DC3000 virulence mutants are severely reduced in RAP2.6 promoter activation

To further test if the RAP2.6 promoter activity could be used to detect bacterial virulence, we inoculated RAP2.6-LUC transgenic plants with DC3000 mutants hrcC, ΔCEL, and COR. hrcC encodes an outer membrane protein of the TTSS, and its mutation abolishes pathogenicity (Yuan and He, 1996). CEL is a conserved effector locus in the hrp pathogenicity island and is required for pathogenicity on tomato (Alfano et al., 2000). The COR mutant is unable to produce coronatine (Ma et al., 1991) and shows reduced bacterial growth in the plant when inoculated by dipping or spraying (Penaloza-Vazquez et al., 2000). The loss of pathogenicity in hrcC and ΔCEL was confirmed by measuring the bacterial growth on Arabidopsis plants (Figure S1). ΔCEL bacterial growth was reduced by approximately 30-fold, whereas the growth of the hrcC mutant was reduced by at least 100-fold compared with the wild-type DC3000 strain. The COR mutant and the wild-type DC3000 did not differ in bacterial growth when inoculated by syringe infiltration (Figure S1). This is consistent with the previous report that the COR mutant display reduced bacterial growth only when plants are inoculated by dipping or spraying, but not by infiltration (Mittal and Davis, 1995; Penaloza-Vazquez et al., 2000). Nonetheless, the COR mutant failed to cause chlorosis in the plant in our assay, indicating reduced virulence. Figure 4(a) shows that all three mutants were poor inducers of the RAP2.6 promoter, with the COR mutant showing the least induction. Interestingly, the hrcC mutant showed slightly higher induction of the RAP2.6 promoter than the ΔCEL and COR mutants. The hrcC mutant appears to overproduce coronatine (Penaloza-Vazquez et al., 2000), whereas the CEL deletion is not known to affect coronatine production. The requirement of both hrcC and coronatine for RAP2.6 induction suggests that the phytotoxin and type III effectors acted together to modulate host cellular activities. We further tested this possibility by co-inoculating plants with the hrcC and COR mutant strains. Figure 4(b) clearly shows that the two mutant strains complemented each other in bacterial mixing experiments, and the RAP2.6 promoter activity was completely restored, suggesting that coronatine and TTSS effectors modulate the same pathways in plants.

Figure 4.

DC3000 virulence mutants are severely reduced in RAP2.6 promoter activation.

(a) hrcC, CEL, and coronatine synthesis are essential for RAP2.6 activation. Luciferase activity of RAP2.6 plants was measured after inoculation with water (mock), DC3000, and mutant bacteria hrcC, ΔCEL, and COR at 2 × 106 cfu ml−1. Leaves were collected at the indicated time points and analyzed by using a luminometer. Each data point consisted of four leaves.

(b) hrcC and COR co-inoculation restores RAP2.6 inducibility. RAP2.6-LUC plants were co-inoculated with the two mutants at 106 cfu ml−1 per strain. Error bars indicate SEs. The experiments were repeated four times with similar results.

(c) The RAP2.6 promoter activity reports the activity of individual DC3000 effector genes. RAP2.6-LUC transgenic plants were hand-injected with water (mock), DC3000, and the hopPtoK and avrPphEPto mutants at 2 × 106 cfu ml−1. Luciferase activity was measured at 12 and 24 h after inoculation. Each data point consisted of at least three leaves. Error bars indicate SEs. The experiments were repeated three times with similar results.

The RAP2.6 promoter activity is modulated by individual effector genes in DC3000

The requirement of TTSS and CEL in the induction of RAP2.6 prompted us to examine the role of individual effector genes in RAP2.6 induction. There are at least 36 experimentally confirmed Hrp outer proteins (Hops) in DC3000, most of which are type III effectors (Collmer et al., 2002). However, the virulence function for most effectors examined to date remains unclear. To determine if the RAP2.6 promoter could be used to monitor effector activities in plants, we tested Campbell insertion (single cross-over) mutants for 10 individual DC3000 effector genes for RAP2.6 inducibility (J.R.A., manuscript submitted). These include avrPphEPto, avrPpiB1Pto, avrPpiB2Pto, hopPsyAPto, hopPtoD1, hopPtoD2, hopPtoC, hopPtoK, hopPtoJ, and hopPtoF. The generation of these mutants will be published separately. None of these mutants exhibited observable changes in bacterial growth or disease symptoms on Arabidopsis plants (Figure S2; L. Shan and X.T., unpublished results). However, the hopPtoK and avrPphEPto mutants were weakened in RAP2.6 activation, and the luciferase inducibility was reduced by 60–70% compared with that of the wild-type DC3000 strain at 12 and 24 h after inoculation (Figure 4c). Mutations in hopPsyAPto, hopPtoD1, and hopPtoD2 had a smaller, but reproducible, reduction in RAP2.6 promoter activity (data not shown). The remaining effector mutants were identical to the wild-type DC3000 in RAP2.6 activation.

avr genes accelerate the activation of the RAP2.6 promoter

As mentioned above (Figure 2a), P. syringae pv. maculicola bacteria carrying the avrB gene, a type III effector gene isolated from P. syringae pv. glycinea, induced the RAP2.6 transcripts in plants earlier than the strain without avrB. We further tested DC3000 strains carrying heterologous avr genes for RAP2.6 promoter activation. At 6 h after inoculation, DC3000 strains carrying avrB, avrRpt2, or avrPphB strongly induced RAP2.6-LUC expression, while DC3000 without an added avr gene did not induce LUC until 9 h (Figure 5a). DC3000 (avrRps4) showed a marginal, but reproducible, induction of LUC. ES4326 carrying avrB also strongly induced LUC at 6 h. The small effect of avrRps4 may be a result of gene redundancy caused by the endogenous hopPtoK that is highly homologous to avrRps4 (Collmer et al., 2002). Among the four avirulence genes tested, avrB was the most potent in RAP2.6 promoter activation. At the later time points (9–24 h), the LUC activity in DC3000 (avrB)-inoculated plants declined, while it steadily increased in DC3000-inoculated plants (Figure 5b). The maximum induction upon DC3000 (avrB) infection at 6 h was nearly two times of that induced by DC3000 at 24 h (Figure 5b).

R gene-specified resistance is not required for avrB-mediated RAP2.6 induction

RAP2.6-LUC transgenic plants were constructed in the Col-0 ecotype that carries the cognate R genes for avrB, avrRpt2, avrPphB, and avrRps4 (Gassmann et al., 1999; Innes et al., 1993; Kunkel et al., 1993; Warren et al., 1998). Thus, the enhanced RAP2.6 expression could be a result of gene-for-gene interaction. However, the primary function of avr genes is thought to promote parasitism. In fact, many of the cloned avr genes are known to possess a measurable virulence function in plants (White et al., 2000). Therefore, the enhanced induction of the RAP2.6 promoter equally could be caused by the virulence effect of the avr genes. We tested whether the early induction of the RAP2.6 gene by avr genes is because of the gene-for-gene-based resistance reaction. P. syringae pv. phaseolicola NPS3121 is a non-host strain on Arabidopsis and does not induce the hypersensitive response (HR; Lu et al., 2001; Yu et al., 1993). However, heterologous avr genes introduced into NPS3121 can be recognized by corresponding R genes in Arabidopsis plants to initiate typical HR resistance reaction. We transferred avrB, avrRpt2, avrPphB, and avrRps4 into NPS3121 and inoculated them in the RAP2.6-LUC transgenic plants. Similar to DC3000 (avrB), NPS3121 (avrB) induced an HR 5–8 h after inoculation (Figure 5c). NPS3121 (avrRpt2) and NPS3121 (avrPphB) induced an HR 16–20 h after inoculation. The RPS4 gene in Col-0 is weak and does not show a detectable HR in response to the avrRps4 gene (Gassmann et al., 1999). None of the avr genes in NPS3121 had a significant effect on RAP2.6 promoter activation over the course of 24 h (Figure 5b). These results indicated that the recognition of the avr genes by corresponding R genes was not responsible for the induction of the RAP2.6 promoter.

We further tested the requirement of gene-for-gene resistance in the avrB-mediated induction of RAP2.6 transcripts in the rps3/rpm1 and ndr1 mutants by using Northern analysis (Figure 6a). The rps3/rpm1 mutant carries a mutation in the RPM1 gene and is unable to initiate the avrB-specified disease resistance (Bisgrove et al., 1994; Innes et al., 1993). NDR1 is an important signaling component essential for resistance mediated by several R genes, including RPM1 (Century et al., 1995). Similar to wild-type Col-0 plants, a strong avrB-dependent RAP2.6 induction was observed in both rps3/rpm1 and ndr1 plants 6 h after bacterial inoculation (Figure 6a). In contrast, RAP2.6 transcripts in DC3000-inoculated plants were indistinguishable from the mock-inoculated plants at 6 h. The results demonstrate that avrB induced RAP2.6 expression independent of the RPM1 resistance function.

Figure 6.

Gene-for-gene resistance is not required for RAP2.6 induction.

(a) The RPM1-specified resistance is not required for RAP2.6 induction by avrB. Wild-type Col-0 (WT), rps3, and ndr1 plants were infiltrated with 10 mm MgCl2 (mock), DC3000, or DC3000 (avrB) at 2 × 106 cfu ml−1, and RNA was isolated at the indicated times.

(b) RPS2 is not required for RAP2.6 induction by avrRpt2. Four- to 5-week-old Col-0 rps2-101C plants and Col-0 rps2-101C plants carrying the avrRpt2 transgene under the control of the RPS2 promoter were dip-inoculated with the indicated bacteria at 5 × 108 cfu ml−1 in 10 mm MgCl2 and 0.02% Silwet L-77; leaves were harvested at the indicated time for Northern blot analysis. The Northern blots were hybridized to the RAP2.6 cDNA probe. The ethidium bromide gel pictures indicate equal loading of RNA samples. The experiments were repeated three times with similar results.

Similarly, we tested if RPS2 is required for the enhanced RAP2.6 expression by avrRpt2. rps2 mutant plants (Yu et al., 1993) were dip-inoculated with DC3000 bacteria with or without avrRpt2. Figure 6(b) shows that the presence of avrRpt2 in DC3000 enhanced the RAP2.6 expression 2–3 days after inoculation. rps2 plants expressing the avrRpt2 transgene display enhanced susceptibility to DC3000 (Chen et al., 2000). When inoculated with DC3000 bacteria, these plants showed much greater RAP2.6 induction than non-transgenic rps2 plants (Figure 6b). These results demonstrated that avrRpt2 also enhanced RAP2.6 expression independent of RPS2 function.

Interestingly, the strong induction of RAP2.6 by DC3000 (avrB) at 6 h was followed by a dramatic decline, and the transcripts reduced to the basal level after 24 h (Figure 6a). This was not a result of general transcriptional cessation upon HR induction because the expression of PR1 was dramatically induced at 12–48 h (Figure S3). A decline of LUC activity was also seen in RAP2.6-LUC plants inoculated with DC3000 (avrB) following the strong induction at the 6-h time point (Figure 5). In contrast, a slower but steady induction over the course of 24 h was observed following the DC3000 inoculation in both the Northern analysis and the LUC reporter assay (Figures 5 and 6). The decline of RAP2.6 expression between 6 and 24 h was dependent on the gene-for-gene resistance because the decline was blocked in both rps3/rpm1 and ndr1 plants (Figure 6a). These results were repeatedly observed and suggest that the avrB-RPM1-mediated resistance actively suppressed the RAP2.6 expression, reinforcing the notion that the expression of RAP2.6 gene is negatively modulated by disease resistance.

Requirement of the JA pathway in RAP2.6 induction

The requirement of coronatine production for RAP2.6 induction by DC3000 suggested the involvement of JA and ethylene signaling because the phytotoxin is thought to mimic both hormones (Bender et al., 1999). An earlier study using microarrays indicated that the JA-signaling mutant coi1 and ethylene-signaling mutant ein2 showed partial RAP2.6 transcript induction 30 h after P. syringae pv. maculicola inoculation (Chen et al., 2002). It was not clear whether JA and ethylene signaling are required for the bacterial-induced RAP2.6 expression at earlier time points that are more relevant to bacterial virulence. We examined RAP2.6 transcript levels in DC3000-inoculated wild-type, coi1 (Feys et al., 1994), and ein2-1 (Guzmann and Ecker, 1990) plants. The wild-type plants showed a strong induction of RAP2.6 12 h after inoculation. In contrast, the coi1 plants did not accumulate any detectable transcripts, while the ein2 plants showed reduced transcripts (Figure 7a). P. syringae pv. maculicola also strongly induced RAP2.6 in wild-type plants 9 and 18 h after inoculation (Figure 7b). In coi1 plants, RAP2.6 transcripts were not detectable at 9 h and were only partially induced at 18 h. The ein2 mutant showed reduced RAP2.6 transcripts at both time points. The absolute requirement of COI1 for the bacterial-induced RAP2.6 expression at early hours prior to bacterial multiplication suggests that the JA signaling is critical for early activities of bacterial virulence factors in host cells.

Figure 7.

Requirement of JA and ethylene pathways in RAP2.6 induction.

(a) RAP2.6 induction in response to DC3000.

(b) RAP2.6 induction in response to P. syringae pv. maculicola ES4326.

(c) RAP2.6 induction in response to DC3000 with (+) or without (−) avrB. Wild-type Col-0 (WT), coi1, and ein2 plants were infiltrated with the indicated bacteria at 2 × 106 cfu ml−1, and RNA was isolated at the indicated times after inoculation. The Northern blots were hybridized with the RAP2.6 cDNA probe. The ethidium bromide gel pictures indicate equal loading of RNA samples. The Northern experiments were repeated two times with similar results.

(d) Exogenous JA, ethylene, and SA weakly activate RAP2.6 promoter. RAP2.6-LUC plants were dipped in a 0.015% Silwet 77 solution containing 0.5 mm SA, 0.5 mm ACC, or 50 µm MeJA. The luciferase activity was measured at the indicated times with a luminometer. Each data point consisted of four leaves. Error bars indicate SEs. Similar results were obtained in two independent experiments.

We next tested if the avrB-dependent RAP2.6 induction also required COI1 and EIN2. As shown in Figure 7(c), while the RAP2.6 transcripts were strongly induced at 6 h in an avrB-dependent manner, they were abolished in coi1 and reduced in ein2, indicating that the AvrB effector required both JA and ethylene pathways to fully activate the RAP2.6 expression.

JA, ethylene, and SA weakly activate RAP2.6 promoter

The apparent involvement of the JA and ethylene pathways in bacterial-induced RAP2.6 expression prompted us to test if these phytohormones alone were able to induce the RAP2.6 promoter. Figure 7(d) shows that application of methyl jasmonic acid (MeJA) or ACC clearly activated the promoter within 6 h. The maximal induction occurred at 12 h for MeJA and 24 h for ACC. Interestingly, SA also activated the promoter to a similar level. However, the RAP2.6 promoter activity induced by these hormones was only about 20% of that induced by bacterial inoculation. This is consistent with the previous microarray study showing that JA only marginally induced the RAP2.6 transcripts 2–12 h after treatment (Chen et al., 2002).

AvrB enhances RAP2.6 expression in the absence of coronatine

The results described above suggested a crucial role of coronatine in RAP2.6 induction. We therefore tested if coronatine is also required for the enhanced RAP2.6 expression mediated by avrB. The original avrB plasmid, which is kanamycin resistant, was modified by integrating a tetracycline resistance transposon and introduced into DC3000 and the COR mutant strain DC3682, which is kanamycin resistant. The transposon insertion rendered the plasmid somewhat less effective in RAP2.6 induction. Nevertheless, the DC3000 strain carrying the new avrB plasmid was capable of inducing RAP2.6-LUC at 6 h. Surprisingly, the COR strain carrying avrB repeatedly gave identical LUC induction at 6 h (Figure 8), demonstrating that coronatine is not needed for the AvrB-dependent enhancement of RAP2.6 expression.

Figure 8.

AvrB enhances RAP2.6 expression independent of coronatine.

RAP2.6-LUC transgenic plants were infiltrated with the DC3000 and COR mutant strains with or without the avrB gene at 2 × 106 cfu ml−1. Relative luciferase activity was measured 6 h after infiltration. The experiment was repeated four times with similar results.


In this report, we demonstrate that the RAP2.6-LUC reporter gene was a reliable indicator for activities associated with virulence factors defined by at least five DC3000 mutants, including those defective in the biosynthesis of the phytotoxin coronatine, TTSS, and several effector genes. In addition, RAP2.6-LUC also reported the activity of three avr genes, avrB, avrRpt2, and avrPphB, when added into DC3000. At least the avrB-mediated RAP2.6 activation was independent of gene-for-gene resistance, suggesting that the activation is related to the virulence activity. Furthermore, we suggest that the JA-signaling pathway plays a large role at the early stages of pathogenesis and that at least some type III effectors require the JA-signaling pathway for function.

Traditionally, bacterial virulence or plant susceptibility is measured by visual disease symptoms and bacterial multiplication in the plant, often monitored at a late stage of disease development, and is difficult to quantify. The use of the RAP2.6-LUC reporter provided a highly sensitive and reliable measurement of bacterial activity in the plant that can be easily visualized at the early hours during infection, demonstrating its utility in bacterial pathogenesis studies. The ability to monitor the early cellular activities of single bacterial effectors with ease will significantly facilitate the molecular dissection of their functions in the host cell.

RAP2.6 belongs to the ERF family of plant-specific transcription factors, many of which are pathogen inducible at the transcript level (Gu et al., 2000; Onate-Sanchez and Singh, 2002; Thara et al., 1999; Zhou et al., 1997; R. Warren and J.Z., unpublished results). Several members of the ERF family, including the tobacco Tsi1 and tomato Pti4, Pti5, and Pti6, play a positive role in plant defense against P. syringae (Gu et al., 2002; He et al., 2001; Park et al., 2001). Overexpression of these genes enhances defense gene expression and disease resistance to Pseudomonas bacteria. However, not all ERFs function positively in bacterial resistance. For example, the Arabidopsis ERF1 gene appears to play a negative role in resistance to Pseudomonas bacteria. Overexpression of ERF1 enhances susceptibility to P. syringae, although it increases resistance to fungal pathogens (Berrocal-Lobo et al., 2002). The function of RAP2.6 in disease susceptibility remains to be determined, but its expression appeared to be associated positively with disease susceptibility and negatively with disease resistance. The RAP2.6 expression was stronger in the npr1 and pad4 mutants that are more susceptible to DC3000 than wild-type plants (Figure 1; Chen et al., 2002). Interestingly, P. syringae pv. tabaci that was unable to induce RAP2.6 transcripts in wild-type plants strongly induced RAP2.6 in the nho1 mutant (data not shown). nho1 compromises Arabidopsis resistance to several non-host Pseudomonas bacteria including P. syringae pv. tabaci (Kang et al., 2003; Lu et al., 2001). Conversely, the cpr5 mutant that is more resistant to DC3000 than wild-type plants showed diminished RAP2.6 transcript induction during infection. Furthermore, the RAP2.6 expression was suppressed upon the activation of gene-for-gene resistance. We have shown recently that plants activate NHO1 gene expression for general resistance, but virulent bacteria actively suppress this to promote parasitism (Kang et al., 2003). While the biological function of RAP2.6 is yet to be tested genetically, it is intriguing to speculate that the activation of RAP2.6 may be also modulated by bacteria to promote parasitism.

It is possible that the activation of the RAP2.6 promoter reflects stress or damage caused by virulent bacteria. However, several lines of evidence argue against this. The RAP2.6 promoter activation occurred prior to bacterial multiplication. DC3000 activated the RAP2.6 promoter as early as 9 h after inoculation. The presence of avrB, avrRpt2, and avrPphB accelerated this induction by at least 3 h. The bacterial multiplication, however, did not occur in the first 12 h after inoculation (Figure 3b), and the disease symptoms did not develop until 3–4 days after DC3000 inoculation at the concentration used. While it had the greatest impact on RAP2.6 induction, the coronatine mutation had smaller effects on disease severity than did the hrcC or ΔCEL mutation. Of the 14 individual effector genes tested, only one (avrRpt2) had a visible virulence effect on Arabidopsis (Chen et al., 2000), but at least five were able to stimulate the RAP2.6 promoter when expressed in DC3000. Therefore, the activation of the RAP2.6 promoter did not appear to be a result of bacterial growth or disease symptom development. Instead, it likely reflects the early cellular activity associated with certain virulence factors.

Coronatine plays an important role in RAP2.6 induction. The COR mutant was completely unable to induce RAP2.6, indicating that coronatine production was essential for the wild-type DC3000 to activate RAP2.6 transcription. This may explain why both DC3000 and ES4326 strongly induced RAP2.6, whereas P. syringae pv. tabaci and P. syringae pv. phaseolicola failed to do so. The first two strains produce coronatine, whereas the latter two do not (Bender et al., 1999).

However, coronatine production by itself may not be sufficient to induce RAP2.6 because P. syringae pv. tomato strain T1 that is fully capable of producing coronatine was unable to activate the RAP2.6 promoter. Similarly, the hrcC mutant produced more coronatine than does the wild-type DC3000 (Penaloza-Vazquez et al., 2000), but it induced very poorly the RAP2.6 promoter. These data suggested that, in addition to coronatine, type III effectors were essential for RAP2.6 activation. This was further demonstrated by the complementation of the COR and hrcC mutants for RAP2.6 activation. Alternatively, these strains did not make sufficient amount of coronatine because of the inability to grow in Arabidopsis. Interestingly, the requirement of coronatine for the wild-type DC3000 to induce RAP2.6 can be obviated by the addition of avrB, suggesting that the latter may function, at least in part, to mimic the effects of coronatine in the host cell.

The role of type III effectors in RAP2.6 induction was directly supported by the study of five individual effector genes, including avrPphEPto, hopPtoK, avrB, avrRpt2, and avrPphB. AvrPphB is a cysteine protease that cleaves the Arabidopsis protein kinase PBS1, and that is required for HR on RPS5-resistant plants (Shao et al., 2002, 2003). AvrRpt2 also appears to act as a cysteine protease to trigger resistance in RPS2-resistant plants (Axtell et al., 2003). AvrPphE and AvrB do not have detectable virulence function in Arabidopsis but are known to function as virulence factors in bean and soybean plants, respectively (Ashfield et al., 1995; Stevens et al., 1998). The ΔCEL mutant contains a deletion spanning at least six genes, avrE, ORF2, hopPtoM, ShcM, hrpW, and hopPtoA1, encoding one harpin and three effectors. (Alfano et al., 2000; Badel et al., 2003; Collmer et al., 2002). hopPtoM is responsible for lesion formation on Arabidopsis and partially accounts for the ΔCEL phenotype (Badel et al., 2003). The biochemical functions for the other effectors tested in this study, however, remain unknown. All of these effectors contributed to RAP2.6 induction. At least avrB and avrRpt2 did not require gene-for-gene recognition for RAP2.6 induction, suggesting a connection to virulence activities of the effectors. These findings are reminiscent of the RPS2-independent degradation of the RIN4 protein mediated by AvrRpt2 (Axtell and Staskawicz, 2003; Axtell et al., 2003; Mackey et al., 2003) and the RPS5-independent cleavage of the PBS1 kinase by AvrPphB (Shao et al., 2003).

It should be pointed out that none of the effectors tested in this report was sufficient for RAP2.6 induction on its own. Direct expression of AvrRpt2 in Arabidopsis did not result in constitutive RAP2.6 expression. Similarly, AvrB, AvrRpt2, AvrPphB, and AvrRps4 delivered by P. syringae pv. phaseolicola were insufficient to induce RAP2.6. It is possible that the presence of other, compatible effectors or coronatine was necessary for RAP2.6 induction by these effectors. This is different from AvrPto that, when overexpressed in the plant, induces/suppresses plant gene expression on its own (Hauck et al., 2003). Perhaps, the difference can be explained by different expression levels of the effector genes.

The JA-signaling pathway may be an important avenue for Pseudomonas bacteria to promote parasitism. Consistent with a role of JA in susceptibility, the JA-insensitive mutant coi1 is significantly more resistant to Pseudomonas bacteria (Feys et al., 1994; Kloek et al., 2001). We showed recently that COI1 is also required by DC3000 bacteria to actively suppress the NHO1 gene expression, which is necessary for resistance to non-host Pseudomonas bacteria (Kang et al., 2003). Type III effectors and coronatine are likely to interact with the JA- and ethylene-signaling pathways to alter plant cellular activities. Indeed, the type III and coronatine-dependent RAP2.6 induction required COI1 and, to a lesser extent, EIN2. Taken together, the results support that different effectors and cornatine coordinately augment the JA- and ethylene-mediated pathways to promote parasitism.

Experimental procedures

Plants, bacterial strains, and inoculation

Arabidopsis plants used in this study were wild-type ecotype Col-0 and mutants rpm1, ndr1, rps2-101C, rps2-101C with pRPS2-avrRpt2 transgene, and coi1-1 (Century et al., 1995; Chen et al., 2000; Feys et al., 1994; Innes et al., 1993; Yu et al., 1993). Homozygous coi1 plants were identified by using a CAPs marker, as described by Xie et al. (1998). All plants were grown in a growth chamber at 21°C with a 10 h per day photoperiod for 35 days. Wild-type P. syringae strains used in this study included: P. sringae pv. tomato DC3000 (Cuppels, 1986); P. syringae pv. tomato T1 (Ronald et al., 1992); P. syringae pv. maculicola ES4326 (Dong et al., 1991); P. syringae pv. phaseolicola NPS3121 (Lindgren et al., 1986); and P. syringae pv. tabaci R11528 (Willis et al., 1988). DC3000 mutants included: DC3682 (COR; Ma et al., 1991); hrcC(Yuan and He, 1996); ΔCEL (Alfano et al., 2000); avrPphEPto, avrPpiB1Pto,, and avrPpiB2Pto; hopPsyAPto, hopPtoD1, and hopPtoD2; hopPtoC, hopPtoK, hopPtoJ, and hopPtoF (J.R.A., manuscript submitted). Avirulence genes, avrB, avrRpt2, avrPphB, and avrRps4 (Gassmann et al., 1999; Innes et al., 1993; Warren et al., 1998; Whalen et al., 1991), were also incorporated into DC3000, ES4326, and NPS3121 as indicated. To introduce avrB into DC3682, a tetracycline R gene was inserted into the plasmid carrying avrB by using EZ::TN™ < TET-1 > insertion kit (Epicentre, Madison, WI, USA). Bacteria were grown overnight at room temperature in King's B medium containing the appropriate antibiotics, precipitated, washed two times, and diluted to the desired concentration with 10 mm MgCl2 for plant inoculation prior to Northern analysis. For luciferase activity assay and bacterial growth assay, the bacteria were diluted to 2 × 106 cfu ml−1 in H2O supplemented with 0.1 mm luciferin and syringe-infiltrated into leaves. Each data point consisted of at least three replicates. All experiments have been repeated at least two times with similar results.

Construction of RAP2.6-LUC and Arabidopsis transformation

The RAP2.6 promoter fragment was PCR-amplified by using the primers, 5′-AAAAGCTTCCTAGATACATACAGCGAG-3′ and 5′-AAGATATCTTGCGGTGGTAGACAAGTTG-3′, and Col-0 genomic DNA as the template. The promoter fragment was digested with HindIII and EcoRV and placed upstream of the firefly luciferase gene LUC (Promega, Madison, WI, USA). The chimeric DNA was inserted between HindIII and SacI sites of pBI121 to form the pRAP2.6-LUC construct. This construct was introduced into Agrobacterium tumefaciens strain LBA4404 and transformed into Col-0 by floral dip (Clough and Bent, 1998). The transgene copy number was determined by Southern blot and segregation ratio of kanamycin resistance in T2 seedlings.

CCD imaging and luminometer assay for Luciferase activity assays

Arabidopsis leaves were sprayed with 1 mm of the luciferase substrate luciferin supplemented with 0.01% Triton X-100. The leaves were kept in the dark for 6 min before luminescence images were captured in 1-min exposures by using an LN/CCD low-light imaging system (RoperScientific, Trenton, NJ, USA) and the winview imaging software.

Quantitative luciferase assay was performed by using Luciferase Assay System (Promega, Madison, WI, USA) and a microplate luminometer (Turner Biosystems, Sunnyvale, CA, USA) following the manufacturer's instructions.

RNA blot analysis and phytohormone treatment

Leaf tissue was collected either before or after inoculation at the indicated time points. RNA extraction and RNA gel blot analysis were performed as described by Goldsbrough et al. (1990). The SA, JA, and ACC treatments were carried out according to Ton et al. (2002).


The authors thank Drs Frank White and Scot Hulbert for critical review of the manuscript and Douglas Baker for excellent technical assistance. The authors are also grateful to Drs Diane Cuppels and Carol Bender for providing DC3682, and Dr Alan Collmer for sharing results prior to publication. B.N.K. was supported by a USDA grant (2002-35319-12728). J.R.A. and X.T. were supported by a NSF Plant Genome grant (DBI0077622). J.M.Z. was supported by a NSF grant (MCB9808701). This is a contribution of Kansas Agriculture Experimental Station (#03-397-J).

Supplementary Material

The following material is available from

Figure S1.  Bacterial growth assay of DC3000, COR, hrcC, and ΔCEL mutants on Col-0 plants.

Figure S1.

Bacterial growth assay of DC3000, COR, hrcC, and ΔCEL mutants on Col-0 plants.

Figure S2.  Bacterial growth assay of DC3000, hopPtoK, and avrPphRPto mutants on Col-0 plants.

Figure S2.

Bacterial growth assay of DC3000, hopPtoK, and avrPphRPto mutants on Col-0 plants.

Figure S3.PR1 expression in Arabidopsis plants upon bacterial infection. Col-0 plants were infiltrated with DC3000 and DC3000(avrB) at 2 × 106 cfu ml−1. RNA was isolated from leaves at the indicated times, and RNA blots were hybridized with PR1 probe. Ethidium bromide gel picture indicate equal loading of RNA.