Abstract
Salicylic acid (SA) is required for resistance to many diseases in higher plants. SA-dependent cell death and defense-related responses have been correlated with disease resistance. The accelerated cell death 5 mutant of Arabidopsis provides additional genetic evidence that SA regulates cell death and defense-related responses. However, in acd5, these events are uncoupled from disease resistance. acd5 plants are more susceptible to Pseudomonas syringae early in development and show spontaneous SA accumulation, cell death, and defense-related markers later in development. In acd5 plants, cell death and defense-related responses are SA dependent but they do not confer disease resistance. Double mutants with acd5 and nonexpressor of PR1, in which SA signaling is partially blocked, show greatly attenuated cell death, indicating a role for NPR1 in controlling cell death. The hormone ethylene potentiates the effects of SA and is important for disease symptom development in Arabidopsis. Double mutants of acd5 and ethylene insensitive 2, in which ethylene signaling is blocked, show decreased cell death, supporting a role for ethylene in cell death control. We propose that acd5 plants mimic P. syringae-infected wild-type plants and that both SA and ethylene are normally involved in regulating cell death during some susceptible pathogen infections.
ROBUST disease resistance in plants is often mediated by host recognition of pathogen-derived proteins, called Avirulence (Avr) proteins, and/or metabolites. In many cases, recognition leads to the coordinated activation of host responses such as hypersensitive cell death (HR), cell wall crosslinking, and defense-related gene induction (Greenberg 1997). Collectively, these responses are referred to as resistance responses. Recently it has become clear that resistance responses are not mediated by a single genetic program. Rather, different genes and signaling molecules are used by plants in response to different pathogens (Centuryet al. 1997; Aartset al. 1998). For example, in Arabidopsis infected with the bacterial pathogen Pseudomonas syringae, the HR requires the defense signal molecule salicylic acid (SA) only in response to some strains carrying avirulence genes (Rateet al. 1999), indicating that the HR can occur by multiple mechanisms. Resistance responses that occur locally can induce long-term resistance at the whole plant level in a process that requires SA and is called systemic acquired resistance (SAR; Ryalset al. 1996). SAR, but not the HR, requires the Nonexpressor of PR1/No Immunity 1 (NPR1/NIM1) gene (Caoet al. 1994; Delaneyet al. 1995). npr1/nim1 mutants of Arabidopsis have extra susceptibility to multiple pathogens (Caoet al. 1994; Delaneyet al. 1995). Attempts to tease apart the role of cell death in the resistance response have been difficult due to the lack of mutations or pharmacological agents that specifically block cell death. It has been suggested that cell death can be uncoupled from the resistance response because in defense no death 1 (dnd1) Arabidopsis mutants, resistance to P. syringae occurs without the HR (Yuet al. 1998). However, since the dnd1 mutant shows constitutively active defenses, it is difficult to infer the normal role of cell death in plants in which defenses are not already active. In barley, one functional allele of the Mlg resistance gene is sufficient to confer resistance to 80% of invading Blumeria graminis without extensive cell death (Gorget al. 1993). However, two copies of the functional Mlg allele provided quantitative resistance to the fungus possibly due to the more extensive HR that occurred (Gorget al. 1993).
Cell death is also associated with host-pathogen interactions that do not involve a resistance response. For example, P. syringae induces water-soaked cell death patches and loss of chlorophyll in a susceptible interaction on Arabidopsis. Concomitant with the onset of these symptoms is the activation of some defense-related responses that require SA for their full induction. These include the low-molecular-weight antimicrobial compound camalexin and transcripts of the Pathogenesis Related-1 (PR-1) gene (Glazebrook and Ausubel 1994; Greenberget al. 1994). Camalexin induction by P. syringae is compromised in nahG plants, in which SA accumulation is blocked by the salicylate hydroxylase activity encoded by nahG (Zhouet al. 1998). Its induction by P. syringae is also compromised in phytoalexin deficient 4 (pad4) mutant plants unless the plants are treated with SA (Zhouet al. 1998). An indication that cell death associated with susceptible interactions in Arabidopsis is genetically conditioned comes from the observation that ethylene insensitive 2 (ein2) plants, in which ethylene signaling is blocked, show attenuated cell death but do not show disease resistance (Bentet al. 1992). ein2 plants are also compromised for the induction of camalexin by Alternaria brassicicola (Thommaet al. 1999) and P. syringae (J. T. Greenberg, unpublished observations), indicating a possible link between the regulation of cell death and other susceptible responses.
To understand more about what regulates pathogen susceptibility and disease symptom development, we isolated and characterized a mutant called accelerated cell death 5 (acd5). We report here that acd5 plants are modestly more susceptible to P. syringae early in development and show SA-dependent cell death and defense-related responses spontaneously late in development. However, although acd5 accumulates SA, it does not induce SAR, suggesting that SA can control cell death without inducing disease resistance. We also show that the NPR1 and EIN2 genes play key roles in controlling cell death in acd5, and we propose that these genes may play a role in controlling disease symptoms in pathogen-infected wild-type plants.
MATERIALS AND METHODS
Plant growth, pathogenicity assays, chemical treatments and metabolite measurements: Arabidopsis thaliana plants were grown in a 12-hr light/12-hr dark cycle for all experiments except ethylene measurements, which were made with plants grown in a 16-hr light/8-hr dark cycle. All experiments done in the 12-hr day showed similar results in the 16-hr day (data not shown). Plants were grown on Pro Mix BX supplemented twice weekly with Peter's 15-16-17. Infections with P. syringae were done using 1-ml blunt syringes to hand-inoculate leaves as described (Rateet al. 1999). Unless otherwise indicated, plants were grown in 50% relative humidity conditions. P. syringae pv. maculicola strain PsmES4326 and P. syringae pv. tomato strain PstDC3000 were obtained from F. M. Ausubel (Harvard University and Massachusetts General Hospital, Boston). A hrcU mutant of PstDC3000 was obtained from B. Staskawicz (University of California, Berkeley, CA). Xanthomonas campestris pv. campestris strain BP109 was obtained from Spencer Benson (The University of Maryland, College Park, MD). Leaf discs for pathogen growth assays were 6 mm in diameter. SA in the form of sodium salicylate was used at concentrations of 0.5 and 5 mm. Treatments with SA or benzo(1,2,3) thiadiazole-7-carbothioic acid (BTH) were performed as described (Rateet al. 1999). BTH was a gift from Novartis, Inc. (Research Triangle Park, NC). Camalexin determinations were performed according to Glazebrook and Ausubel (1994). SA levels were determined from 0.5 g of tissue as described (Seskaret al. 1998). Yields, determined from samples spiked with SA, were 52%. Values were adjusted for yield losses. Ethylene measurements were made on whole plants by placing plants in their pots in a sealed quart mason jar. Plants were kept at 22° in the light for 18 hr and the ethylene composition of the head space was determined by sampling using a syringe and injecting into an HP 5890 GC with a packed Hayesep T column coupled to a flame ionization detector. The ethylene content of samples was determined by comparison with known standards.
Seeds of npr1-1, ein2-1, Landsberg erecta, Nossen (Nos), Wassilewskija (Ws), Columbia (Col-0), Col-0 with a BGL2-uidA transgene, and M2 seeds of Col-0 mutagenized with ethyl methanesulfonate were from F. M. Ausubel. Seeds of Cape Verdi Island (Cvi) were obtained from Daphne Preuss (The University of Chicago, Chicago). Seeds of nahG line “B15” in the Col-0 background was from Novartis, Inc. acd5 was backcrossed to its parent (Col-0) four times and used for the studies described here. acd5npr1 plants were constructed by using npr1 as a female recipient for acd5 pollen. acd5 homozygous plants from the F2 population were self-fertilized and screened in the F3 generation for the npr1 mutation by polymerase chain reaction (PCR) as described (Caoet al. 1997). acd5-nahG plants were constructed by pollinating nahG plants with acd5 pollen. F2 plants from the cross were sprayed with 100 μm BTH to identify potential nahG-acd5 plants. F2 plants showing cell death after 4 days were tested for the presence of nahG by PCR of a linked kanamycin resistance marker (nptII). F3 progeny of individuals showing BTH-induced cell death were retested for spontaneous and BTH-induced cell death separately. Plants in the F3 that showed no spontaneous cell death but showed BTH-induced cell death and scored positive for the nahG transgene were test-crossed with wild-type and acd5 plants, respectively, and followed for two generations to confirm that the original plant was homozygous for acd5. To construct acd5ein2 plants, ein2-1 plants were pollinated with acd5 pollen. F2 progeny that were homozygous acd5 were examined for large curled leaves typically seen in ein2 homozygous plants. acd5ein2 double mutants were confirmed for the presence of ein2 by outcrossing to ein2 plants and testing the progeny for lack of a triple response after growth in the dark for 3 days on MS medium (Mirashige and Skoog basal salts with B5 vitamins and 0.7% bactoagar plates, pH 5.8) with 50 μm 1-aminocyclopropane-1-carboxylic acid.
To map acd5 initially, Landsberg erecta was crossed with pollen from acd5. F2 progeny showing the acd5-conferred cell death phenotype were used for recombination analysis. To score recombinant plants, we used the technique of cleaved amplified polymorphic sequences using 34 published markers on all five chromosomes (Konieczny and Ausubel 1993). We detected linkage to LFY on chromosome 5 (see text). Additional mapping was performed on the same F2 population that was treated with 100 μm BTH to induce the acd5-conferred cell death phenotype and with untreated plants from the F2 progeny of crosses of acd5 to Ws. These additional experiments indicated linkage of acd5 to DFR (11%), LTI78 (2.5%), and LFY (15.5%) on chromosome 5.
RNA gel blot analysis: RNA was isolated and gels were run as described (Rateet al. 1999). The probe used for detecting PR-1 was described previously (Greenberget al. 1994). A PhosphorImager (Molecular Dynamics, Sunnyvale, CA) was used to record the PR-1 expression data.
Visualization of dead cells and histochemical analysis: Fresh tissue was boiled in lactophenol (10 ml of lactic acid, 10 ml of glycerol, 10 ml of liquid phenol, and 10 ml of distilled H2O) containing 10 mg trypan blue and cleared as described previously (Rateet al. 1999). Stained tissue was visualized on a petri dish with a Wild M3Z binocular microscope (Leica, Inc., Rockleigh, NJ) with a magnification of ×40. Photographs were taken with Kodak 100 Elite Chrome slide film, scanned with a Microtek ScanMaker 4 using Adobe PhotoShop 4.0LE software (Adobe Systems Inc.), assembled into a composite, and annotated using Canvas 5 (Deneba Software).
Transgenic plants harboring a BGL2-uidA fusion were stained with x-gluconase (Rose Scientific, Toronto, Canada) for 48 hr as described (Bowlinget al. 1994), cleared in an ethanol series, and photographed.
Statistical analysis: All analyses presented were done with a statistical software package from StatView (SAS Institute, Inc., Cary, NC).
RESULTS
Identification of novel mutant with increased pathogen susceptibility and spontaneous cell death: To identify new genes important for disease resistance and/or disease symptom development, we screened 5000 M2 ethyl methanesulfonate-mutagenized Arabidopsis ecotype Columbia for mutants with altered disease symptoms after P. syringae pv. maculicola strain PsmES4326 infection as described previously (Rateet al. 1999). One mutant (named acd5, see below), infected at week 3, showed enhanced disease symptoms after PsmES4326 attack relative to that seen in wild type (Figure 1, A and B). Similar results were obtained with P. syringae pv. tomato strain PstDC3000 (data not shown). Mock inoculation or infection with a nonpathogenic version of PstDC3000 (due to a mutation in hrcU) did not elicit any symptoms (data not shown), indicating that the mutant was not generally more sensitive to inoculation stress or the presence of bacteria. The enhanced symptom development after pathogen attack was accompanied by a moderate increase in pathogen growth of strain PsmES4326 or PstDC3000 (Figure 2, A and B). This increase was manifest as a small but reproducible increase in the highest bacterial titers achieved in acd5 vs. wild-type plants. Infection of acd5 and wild-type plants with X. campestris pv. campestris strain BP109 caused a mild increase in symptom development in acd5 (data not shown); however, the growth of BP109 in acd5 was not different from that seen in wild type in most experiments (Figure 2C). Inoculation with PsmES4326 carrying either of two avr genes (avrRpt2 and avrRpm1) elicited the normal hypersensitive response, indicating that this type of cell death response was not altered in acd5 plants (data not shown).
Cell death and defense phenotypes of acd5 and acd5npr1 plants. (A and B) Young leaves from 24-day-old wild-type (A) and acd5 (B) plants infected with P. syringae strain PsmES4326 at a dose of OD600 = 0.002 photographed 3 days after the infection. Similar symptom enhancement was seen with infections using lower doses of P. syringae. (C and D) Five-week-old ACD5 BGL2-uidA (C) and acd5 BGL2-uidA (D) leaves grown in 95% relative humidity. Leaf in C was infected with PsmES4326 at a dose of OD600 = 0.0002 and photographed 4 days later. Arrows indicate representative disease lesions. Leaf in D shows spontaneous cell death in acd5 that resembles leaf spot disease shown in C. Arrows indicate representative spontaneous lesions. acd5 plants without BGL2-uidA looked indistinguishable from plants with the transgene (data not shown). (E) acd5 (left side) and acd5npr1 (right side) plants photographed at 9 wk. (F and G) Five-week-old ACD5 BGL2-uidA (F) and acd5 BGL2-uidA (G) leaves grown in 95% relative humidity. Leaf in F was infected with PsmES4326 at a dose of OD600 = 0.0002 and photographed 4 days later. Leaves were stained for β-glucuronidase activity. Mock-treated wild-type and acd5 leaves without lesions showed no staining (data not shown). (H–K) Three-week-old acd5 (H and I) or wild-type (J and K) leaves treated with water (H and J) or 100 μm BTH (I and K) and photographed 4 days later. Note lesions induced in leaves in I only. (L) Twenty-nine-day-old npr1 (left) and acd5npr1 (right) plants photographed 8 days after treatment with 100 μm BTH. Water-treated controls were indistinguishable from BTH-treated plants (data not shown). These experiments were repeated twice under short-day and twice under long-day conditions with similar results.
At week 5 after planting, the acd5 mutant showed spontaneous disease-like lesions, which resembled leaf spot symptoms caused by P. syringae (Figure 1, C and D), on the youngest leaves of the rosette (sink leaves). These spontaneous lesions typically showed modest spreading. Similar disease-like lesions could also be precociously induced by treatment of acd5 plants with the SA agonist BTH (see below). Trypan blue staining showed that the spontaneous lesions were composed of dead cells (Figure 3, A and B). Over time, additional leaves showed spontaneous cell death as did stems (Figure 1E) and siliques (data not shown). acd5 is named for its spontaneous accelerated cell death phenotype and falls into the category of “disease lesion mimic” mutants found in numerous plant species (Greenberg 1997). Prior to the onset of cell death, acd5 plants had normal sized rosettes (data not shown), but mature plants were shorter than wild type (Figure 4) due to premature death of acd5 floral inflorescences, not altered internode length (data not shown). acd5 had reduced fitness as evidenced by the lower seed yield (Table 1), probably as a result of the cell death. In acd5 plants, spontaneous cell death cosegregated with enhanced disease symptoms after pathogen attack. The spontaneous cell death phenotype was not due to P. syringae spreading, as P. syringae was not recovered from the spontaneous lesions. In addition, aseptically grown mutant plants developed spontaneous lesions independent of infection (data not shown). Based on these phenotypes, and the observation that acd5 is recessive (see below), the ACD5 gene appears to negatively regulate cell death and pathogen susceptibility.
Analysis of the growth of P. syringae strains PsmES4326 and PstDC3000 and X. campestris stain BP109 in acd5 plants. Lesion-free plants were inoculated with bacteria at 3 wk of age. The mean value of the growth of bacteria in six leaves is indicated in each case. Bars indicate standard error. (A) Plants were inoculated with PsmES4326 (OD600 = 0.002). *Growth of PsmES4324 in acd5 was significantly different from the growth in wild type (P < 0.002, unpaired t-test day 2). (B) Plants were inoculated with PstDC3000 (OD600 = 0.002). *Growth of PstDC3000 in acd5 was significantly different from the growth in wild type (P = 0.03, unpaired t-tests days 2 and 3). (C) Plants were inoculated with X. campestris BP109 (OD600 = 0.002). There was no significant difference in the growth of BP109 in acd5 and wild type. This experiment was repeated once under short-day and three times under long-day conditions with similar results.
Genetic analysis of acd5: acd5-conferred cell death segregated as a single recessive trait in a backcross to its wild-type Col parent (Table 2). Using cleaved amplified polymorphic sequences (see materials and methods) for mapping, acd5 was tightly linked to LFY on chromosome 5 (2 out of 66 chromosomes or 3% ± 2.2% recombination). Although acd5 segregated as a single Mendelian locus in backcrosses to Col, in the F2 progeny from crosses to the Landsberg ecotype, plants with a mutant phenotype were underrepresented (segregating at a 16:1 ratio), suggesting reduced penetrance of the mutant phenotype, reduced viability of acd5 seedlings, or the presence of a modifying gene (Table 2). Despite a survey of the entire Arabidopsis genome (see materials and methods), we were unable to identify a modifying locus; only the previously identified LFY region cosegregated with the acd5 phenotype. In addition, to test whether acd5 homozygotes might have reduced viability, we monitored the segregation of the LFY marker in the Ler F2 population. This marker segregated 1:2:1 (45 LFYCol: 91 LFYCol/Ler: 44 LFYLer), suggesting that the under-representation of acd5 homozygous phenotypic plants was due to the lack of penetrance of the cell death phenotype in the mapping cross. Interestingly, the lack of acd5 penetrance was relieved when F2 individuals from the same cross were treated with 100 μm BTH; acd5 phenotypic plants (scored as plants with cell death patches) were found in a quarter of the plants, as expected for a single recessive trait (Table 2). In the BTH-treated F2 population, the acd5-conferred cell death phenotype showed 15.5% recombination (39 out of 252 chromosomes) with LFY, 2.5% recombination (7 out of 252 chromosomes) with LTI178a, and 11% recombination (26 out of 242 chromosomes) with DFR. The larger sample size of the second mapping experiment is likely to be largely responsible for the different recombination frequency of acd5 with LFY seen in the BTH-treated vs. the untreated plants.
Microscopic cell death in acd5, acd5npr1, and acd5ein2 leaves. Representative leaves from 7-wk-old plants stained with trypan blue are shown. Dark spots are condensed, dead cells. Bar, 1 mm. (A) Wild-type; (B) acd5; (C) npr1; (D) acd5npr1; (E) ein2; and (F) acd5ein2. This experiment was repeated once under short-day and once under long-day conditions with similar results.
Height analysis of acd5 and other genotypes of plants affecting SA and ethylene signaling. Three-month-old plants of the indicated genotypes were used for height measurements. At least 12 plants were used for each determination. Box plots show the mean (center line of the box) and the second and third quartiles, which indicate the dispersion of 50% of the data points (shaded boxes) and the range (vertical lines above and below the boxes). Statistical outliers are indicated with x's. The distribution of heights for all possible combinations were compared using Fisher's least-squares test. Each letter represents a height class that is different from other height classes with a different letter designation (P < 0.002). This experiment was repeated once under short-day conditions and once under long-day conditions with similar results.
Similar reductions in acd5 penetrance were seen in crosses of acd5 to the ecotypes Ws, Cvi, and Nos with recovery of acd5 phenotypic plants in 21, 20, and 2.4% of the F2 progeny, respectively. In agreement with the linkage data from the BTH-treated Ler × acd5 F2 progeny, we detected 18% linkage of acd5 to LFY (19 out of 104 chromosomes) in the F2 progeny of the cross of acd5 to Ws. Taken together, these data suggest that acd5 maps 2.5 cM north of the LTI78 marker on chromosome 5. The variable penetrance of the acd5 phenotype in the different crosses may indicate that there are many modifiers that collectively behave as quantitative traits, as no distinct loci have yet been detected that are responsible for modifying the acd5-conferred cell death phenotype. No other mutants affecting spontaneous cell death that we know of have been reported to map to this region. Thus, acd5 represents a new cell death mutant.
Seed yield in acd5 plants
Genetic analysis of the acd5 mutant
Camalexin levels in acd5 plants
Activation of multiple defense-related responses without disease resistance in acd5: Cell death activation often requires the accumulation of and/or sensitivity to the defense signal SA. To test whether other defense-related markers that require SA were activated in acd5, we analyzed camalexin levels and steady-state PR-1 transcript levels. Additionally, we analyzed β-glucuronidase activity in acd5 plants harboring a SA-inducible BGL2-uidA gene fusion (Bowlinget al. 1994). Camalexin levels in acd5 were elevated to levels similar to those seen in PsmES4326-infected wild-type plants (Table 3). The steady-state gene transcript level of PR-1 was strongly elevated in acd5 plants showing lesions relative to the wild-type control (Figure 5). Leaves taken from plants prior to lesion formation showed no PR-1 mRNA accumulation (data not shown). The defense-related gene fusion BGL2-uidA was also induced in acd5 around the spontaneous lesions in a similar fashion to what was found in P. syringae-infected wild-type plants (Figure 1, F and G). Consistent with these defense-related phenotypes, acd5 accumulated high levels of free and total SA relative to wild-type plants (Figure 6).
Steady-state levels of PR-1 gene transcripts in acd5 and wild type. Leaves from 5-wk-old plants were used for RNA gel blot analysis of PR-1 transcript levels. (Left) A PhosphorImager image; (right) ethidium staining of the rRNA from the same blot. Sample in the acd5 lane was from leaves with spontaneous lesions. This experiment was repeated once under short-day and once under long-day conditions with similar results.
Free and total salicylic acid levels in acd5 plants. Leaves from 5-wk-old plants were used for SA extractions. Five replicates were used for each sample. Box plot parameters are the same as described in the legend to Figure 4. SA levels in ACD5 and acd5 were statistically different (P = 0.016 for free SA, P = 0.002 for total SA). This experiment was repeated once under long-day conditions with similar results.
To determine if SA, cell death, and/or defense-related markers were associated with SAR in acd5, we monitored the growth of two strains of P. syringae and one strain of X. campestris pv. campestris on acd5 plants with preformed spontaneous lesions. The growth of both strains of P. syringae grew slightly better in acd5 than wild type (Figure 7, A and B). Additionally the growth of X. campestris was equivalent in acd5 and wild-type plants (Figure 7C). Thus, acd5 was not more resistant than wild type to P. syringae or X. campestris, indicating that SA, cell death, and defense-related responses of acd5 were uncoupled from disease resistance.
A role for SA signaling and the NPR1 gene for the acd5-conferred phenotypes: The cell death, defense-related phenotypes and elevated SA levels of acd5 raised the possibility that SA, an important defense signal, might be required for one or more of the acd5-conferred phenotypes. To determine this, we crossed acd5 with a well-characterized transgenic plant harboring the nahG gene, whose product metabolizes SA to the inactive catechol. The acd5-conferred cell death phenotype segregated 15 to 1 in the F2 progeny (Table 2). The 1 out of 16 plants showing the acd5-conferred cell death phenotype lacked the nahG transgene, while the other homozygous acd5 plants, lacking a cell death phenotype, had 1 or 2 copies of the nahG transgene. This result suggests that nahG dominantly suppressed the cell death of acd5 plants. Treating the F2 progeny with 100 μm BTH, an SA agonist, reversed the suppression and yielded the expected 3 to 1 segregation of plants with the acd5-conferred cell death phenotype (Table 2). Plants homozygous for acd5 and nahG were indistinguishable in size from wild type or nahG (Figure 4) and had wild-type seed yield (Table 1), indicating that removal of SA also suppressed the growth defect and reduced fitness of acd5. Application of 100 μm BTH to acd5-nahG plants induced camalexin synthesis to a similar level as that seen in acd5 plants (Table 3). Interestingly, spraying acd5 with BTH (or SA, data not shown) 1–2 wk prior to the onset of the spontaneous visible phenotype induced cell death (Figure 1, H–K) and camalexin synthesis (Table 3) in the young leaves in the same pattern that occurred spontaneously in older plants.
Analysis of the growth of P. syringae and X. campestris in acd5 plants with lesions. Plants were inoculated with bacteria as described in the legend to Figure 2. Bars indicate standard errors. (A) Plants were inoculated with PsmES4326. *Significant differences between bacterial growth in acd5 and wild type (P = 0.03). (B) Plants were inoculated with PstDC3000. *Significant differences between bacterial growth in acd5 and wild type (P = 0.003). (C) Plants were inoculated with X. campestris strain BP109. There was no significant difference in the bacterial growth in acd5 and wild type. This experiment was repeated once under short-day and twice under long-day conditions with similar results.
The NPR1 gene is required for some aspects of SA signaling and npr1 mutants show highly increased susceptibility to P. syringae (Caoet al. 1994; Delaneyet al. 1995; Shahet al. 1997). Susceptibility of acd5npr1 mutants to P. syringae was indistinguishable from that seen in npr1 alone (data not shown), indicating that the increased susceptibility of acd5 was not additive with npr1. However, in acd5npr1 double mutants, the spontaneous cell death was highly attenuated. In some acd5npr1 plants, no spontaneous cell death was seen associated with leaves, while stems showed a modest amount of cell death. When cell death did occur on acd5npr1 leaves, it was typically very mild (Figure 1E, and Figure 3, compare B and D) and started 2–3 wk after the initiation of cell death of the acd5 single mutant. The attenuated phenotype resulted in plants that were much taller than the acd5 plants, indicating that npr1 partially suppressed the reduced stature of acd5 (Figure 4). npr1 also suppressed the reduced fitness of acd5 and restored the seed yield to that seen in npr1 (Table 1). NPR1 was strictly required for the BTH- or SA-induced cell death of young acd5 plants, as no visible (Figure 1L) or microscopic (not shown) cell death was seen after treatment of the acd5npr1 double mutant. Thus NPR1 was required for mediating BTH- or SA-induced early cell death and was largely required for spontaneous cell death in acd5 plants, while SA was necessary and sufficient to induce cell death.
A partial requirement for ethylene signaling for acd5-conferred phenotypes: Ethylene signaling has been shown to be important for some bacterial pathogen-induced symptoms, but not resistance, in susceptible interactions in Arabidopsis (Bentet al. 1992). As acd5 showed lesions that appeared to mimic P. syringae-induced symptoms, we tested the involvement of ethylene signaling in the acd5 phenotypes. First we measured the amount of ethylene evolved from whole acd5 plants with lesions and found that they evolved significantly more ethylene than the wild-type plants (Table 4). acd5 plants without lesions showed a more modest increase (twofold) in ethylene levels (data not shown). To further test the involvement of ethylene in the acd5-conferred phenotypes we crossed acd5 with the ethylene-insensitive ein2 mutant in which ethylene signaling is blocked (Guzman and Ecker 1990). Spontaneous cell death of the acd5ein2 plants lagged 1 wk or more behind those seen in acd5 plants alone. In addition, the severity of the BTH- or SA-induced cell death of young plants (data not shown) and the spontaneous cell death of older plants was reduced relative to the acd5 single mutant (Figure 3, compare B and F). The attenuated phenotype resulted in slightly taller acd5ein2 double mutant plants relative to the acd5 single mutant (Figure 4). In addition, ein2 partially suppressed the fitness defect of acd5, as acd5ein2 seed yield was intermediate between acd5 and ein2 (Table 1). Interestingly, ein2 plants were taller and produced more seed than wild type, suggesting that ein2 conferred improved fitness under these growth conditions (Table 1). However, the magnitude of these effects (comparing ein2 with wild type) was much smaller than that seen when comparing acd5ein2 with acd5 (Table 1 and Figure 4). The ein2 mutation also caused a reduction in camalexin production in acd5 (Table 3). Despite the attenuated spontaneous symptoms of acd5ein2 plants, P. syringae growth in acd5ein2 was not decreased relative to the acd5 single mutant (data not shown). This is consistent with the previous report that ein2 shows attenuated symptoms with P. syringae but does not increase disease resistance (Bentet al. 1992).
Ethylene evolution from acd5 plants
DISCUSSION
We identified a new mutant, acd5, with altered susceptibility to the bacterial pathogen P. syringae. acd5 shows both modestly increased growth of P. syringae and increased symptom development similar to other previously described enhanced disease susceptibility (eds) mutants of Arabidopsis. However, unlike the eds mutants,acd5 also has a spontaneous cell death phenotype that is correlated with decreased fitness of the plants. This cell death phenotype of acd5 is strictly dependent on the defense signal molecule SA, as inferred from the phenotype of the acd5-nahG plants and the reversibility of the suppressed phenotype by application of the synthetic SA analogue BTH. acd5 also accumulated high levels of SA relative to wild-type plants. Several other mutants of Arabidopsis such as lsd6, lsd7, ssi1, acd6, and cpr20cpr21 also show SA-dependent cell death (Weymannet al. 1995; Rateet al. 1999; Shahet al. 1999; Silvaet al. 1999). However, unlike these other disease-resistant cell death mutants, acd5 does not show increased resistance to P. syringae, X. campestris, or P. parasitica (J. T. Greenberg, unpublished data) under any conditions tested despite the activation of several defense-related markers. Thus, acd5 is unique in that it uncouples SA-dependent cell death and defense-related markers from disease resistance. The observation that acd5 mutation is recessive suggests that the ACD5 gene acts to repress some SA-dependent responses.
In addition to requiring SA for developmentally induced cell death and camalexin synthesis, acd5 shows precocious stimulation of these events when the SA pathway is activated. This phenotype of acd5 is similar to what was reported for the Arabidopsis cell death mutant lsd1. Under short-day conditions, the repressed cell death phenotype of lsd1 is induced by SA (Dietrichet al. 1994). However, lsd1, unlike acd5, shows disease resistance and uncontrolled spreading of cell death. One possibility is that SA is required for the induction of distinct cell death pathways, one that is associated with disease resistance and one that is associated with disease susceptibility. Alternatively, acd5 and lsd1 might induce the same SA-dependent cell death, but acd5 might only induce a subset of defense-related responses (not enough to confer disease resistance). Since it is not known which SA-dependent defense(s) is required for resistance to P. syringae, it is difficult to determine whether the appropriate antibacterial defense(s) is activated in acd5. It is also possible that acd5 activates a disease susceptibility factor that overrides the defense response.
NPR1 is required for some aspects of SA signal transduction and for disease resistance (Caoet al. 1994; Delaneyet al. 1995; Shahet al. 1997). We showed previously that NPR1 influences the induction of cell death in the acd6 mutant of Arabidopsis (Rateet al. 1999). acd6npr1 plants show modestly delayed and reduced cell death that is coupled to a cell growth response (Rateet al. 1999). The characterization of acd5 plants further suggests that NPR1 plays a key role controlling cell death. npr1 suppressed much of the spontaneous cell death of acd5 and completely blocked cell death induced by early activation of the SA signaling pathway. This effect of the npr1 mutation raises the possibility that NPR1 functions to control cell death in wild-type plants during pathogen infection. If this is the case, it is unclear why npr1 mutants show earlier and more severe symptom development after attack by pathogens such as P. syringae than wild-type plants. It is possible that when NPR1 is removed, cells under pathogen attack die by a mechanism distinct from that which occurs in cells that contain functional NPR1. An examination of the mechanism of cell death in NPR1 plants and npr1 mutants will be necessary to resolve this question. Interestingly, plants lacking SA due to the presence of the nahG gene die by a different mechanism after ozone treatment than wild-type plants (Rao and Davis 1999). It has been suggested that ozone induces cell death by activating an HR (Sharma and Davis 1997; Sandermannet al. 1998). We found recently that Arabidopsis nahG plants are compromised for inducing the HR after infection with a subset of normally HR-inducing P. syringae strains (Rateet al. 1999). Thus, it is clear that there are both SA-dependent and -independent modes of the HR in plants. Cell death during plant-pathogen interactions that do not involve the HR could similarly show SA-dependent and -independent mechanisms.
Unlike SA, which plays an essential role in acd5-conferred cell death, the hormone ethylene plays a more minor role in the expression of this phenotype. acd5 plants produce more ethylene than wild-type plants. The ein2 mutation, which blocks ethylene signal transduction, partially suppresses the acd5-conferred phenotypes. Thus, the acd5ein2 double mutants were delayed for the developmental induction of cell death, showed reduced camalexin synthesis, showed less intense cell death after BTH or SA treatment, and had higher fitness than acd5 single mutants. ein2 plants were shown previously to have reduced pathogenic symptoms without decreasing the growth of P. syringae (Bentet al. 1992). Recently it was also shown that ein2 plants were compromised for camalexin induction by the fungal pathogen A. brassicicola (Thommaet al. 1999). We have found a similar result with P. syringae (J. T. Greenberg, unpublished observations). Since ein2 partially suppresses both pathogen-induced and acd5-conferred cell death and camalexin synthesis, it is possible that these two incidences of cell death and defense induction share a common mechanism of activation and/or execution. If this were true, it would imply that acd5 plants truly mimic a pathogenic infection and would furthermore suggest that pathogenic symptoms caused by P. syringae are caused largely by host-encoded functions when the ethylene and SA signaling pathways are intact.
Acknowledgments
We thank D. Guttman, J. Mach, D. Rate, H. Vanacker, J. H. Lu, D. Preuss, and J. Malamy for helpful discussions. We thank one of the anonymous reviewers for helpful comments on the mapping experiments. We thank Michael Spaly for help with the mapping experiment and James Cuenca for excellent technical assistance. J.T.G. is a Pew Scholar and an American Cancer Society Research Fellow (no. JFRA623). This research was supported by a grant to J.T.G. from the National Institutes of Health (no. 1R29GM54292-01) and by an award to the University of Chicago's Division of Biological Sciences underthe Research Resources Program for Medical Schools of the Howard Hughes Medical Institute.
Footnotes
-
Communicating editor: V. L. Chandler
- Received February 10, 2000.
- Accepted May 22, 2000.
- Copyright © 2000 by the Genetics Society of America
