The Arabidopsis-accelerated cell death gene ACD2 encodes red chlorophyll catabolite reductase and suppresses the spread of disease symptoms

Abstract

accelerated cell death 2 (acd2) mutants of Arabidopsis have spontaneous spreading cell death lesions and constitutive activation of defenses in the absence of pathogen infection. Lesion formation in acd2 plants can be triggered by the bacterial toxin coronatine through a light-dependent process. Coronatine-triggered and spontaneous lesion spreading in acd2 plants also requires protein translation, indicating that cell death occurs by an active process. We have cloned the ACD2 gene; its predicted product shows significant and extensive similarity to red chlorophyll catabolite reductase, which catalyzes one step in the breakdown of the porphyrin component of chlorophyll [Wüthrich, K. L., Bovet, L., Hunziger, P. E., Donnison, I. S. & Hörtensteiner, S. (2000) Plant J. 21, 189–198]. Consistent with this, ACD2 protein contains a predicted chloroplast transit peptide, is processed in vivo, and purifies with the chloroplast fraction in subcellular fractionation experiments. At some stages of development, ACD2 protein also purifies with the mitochondrial fraction. We hypothesize that cell death in acd2 plants is caused by the accumulation of chlorophyll breakdown products. Such catabolites might be specific triggers for cell death or they might induce cellular damage through their ability to absorb light and emit electrons that generate free radicals. In response to infection by Pseudomonas syringae, transgenic plants expressing excess ACD2 protein show reduced disease symptoms but not reduced growth of bacteria. Thus, breakdown products of chlorophyll may act to amplify the symptoms of disease, including cell death and yellowing. We suggest that economically important plants overexpressing ACD2 might also show increased tolerance to pathogens and might be useful for increasing crop yields.

Plant senescence and disease are linked by common features including cell death and chlorophyll breakdown. Although chlorophyll degradation may a priori seem to be a peripheral consequence of cell death in disease resistance, several observations link both light and chlorophyll degradation to disease resistance. As part of cell death in disease, or programmed cell death in disease resistance [hypersensitive response (HR)], chlorophyll degradation happens as the lesion turns yellow or brown. Chlorophyll breakdown causes both a decrease in photosynthetic capacity and production of light-absorbing chlorophyll breakdown products. Intriguingly, light is required for the HR in the inoculation of rice with Xanthomonas (1). In other reports, the lack of light can also cause a stronger HR (2). Light may be required in the HR for the production of adequate ATP to allow programmed cell death, or it might activate phototoxic porphyrin chlorophyll breakdown products, which accelerate cell death.

Porphyrin compounds, the precursors and breakdown products of both chlorophyll and heme, are extremely phototoxic; thus, their synthesis and degradation are highly compartmentalized and regulated (3, 4). As chlorophyll can absorb light and donate active electrons, so can its porphyrin relatives. In the absence of productive outlets for these active electrons, they can be donated to other compounds, including molecular oxygen, forming free radicals that can be toxic and can act as cellular signals (5–7). Accumulation of porphyrin compounds, caused by inactivation of enzymes in the porphyrin synthesis pathway, can cause cell death in animals and plants (8). For example, Les22 mutant maize accumulate uroporphyrin III and form light-dependent cell death lesions on their leaves (9). Interestingly, in mammalian cells, protoporphyrin IX specifically triggers the mitochondrial permeability transition (10, 11) and thus apoptosis (12). The phenotype of porphyrin precursor accumulation in plants appears similar to the induction of defenses in pathogen resistance, including cell death and defense gene transcription. This is termed a lesion mimic phenotype (13, 14). Lesion mimic mutants have been isolated in many species, including the accelerated cell death (acd) and lesions simulating disease (lsd) series of Arabidopsis mutants (13–16). Many have been cloned, including the LSD1 gene, which encodes a putative zinc finger protein (17).

Reactive oxygen species and light can affect the phenotype of some lesion mimic mutants. In lsd1 mutants, superoxide can trigger the lesion mimic phenotype (18); also, copper–zinc superoxide dismutase is not properly induced in response to analogs of the defense signal salicylic acid (SA; ref. 19), indicating that a major perturbation of antioxidant defenses may cause the lsd1 phenotype of spreading lesions. A lesion mimic phenotype associated with Cladosporum fulvum resistance in tomato showed reduced symptoms in partial shade (20). Also, lsd1 and lsd3 mutants are suppressed under short day conditions, but lsd2 and, to a lesser extent, lsd5, are suppressed under long day conditions (14). Some light-sensitive lesion phenotypes, such as that of antisense catalase (21), may be caused by excess reactive oxygen produced by chloroplast metabolism. In contrast, the Arabidopsis psi2 mutant shows red light-dependent cell death that requires the photoreceptors phyA and phyB (22). The effects of reactive oxygen and light indicate that some lesion mimic phenotypes may require light energy, light signaling, or active chloroplast metabolism.

We describe here the cloning and characterization of the Accelerated Cell Death 2 (ACD2) gene, which encodes the Arabidopsis homologue of red chlorophyll catabolite (RCC) reductase (RCCR). We hypothesize that accumulation of chlorophyll breakdown products in the mutant causes the acd2 phenotype and show that light induces the cell death phenotype. Moreover, we show that ACD2 protein localizes to chloroplasts and mitochondria and that the removal of SA by the transgene nahG exacerbates the acd2 mutant phenotype. Plants that overexpress ACD2 protein show reduced yellowing and cell death in disease, indicating that chlorophyll breakdown products may amplify cell death, affecting disease symptoms.

Materials and Methods

Arabidopsis Strains and Plant Growth Conditions.

acd2 alleles were described previously (15). Additional alleles were derived from further screening of ethyl methane sulfonate-mutagenized Columbia seeds, except acd2-12E13, which was a gift from S. Y. He (Michigan State University, East Lansing). nahG line B15 was a gift from Novartis (Research Triangle Park, NC). Plants were grown in 16-h day conditions, as described (23). Seedlings for organelle purification were grown from sterilized seed spread to confluence on sterile cheesecloth over sterile soil for 7 days.

Recombination Mapping.

Recombinants between ACD2 and the flanking markers AP2 or CP3 (15) were obtained by selecting acd2, ap2, or cp3 plants from an F₂ population and scoring segregation of the other mutant phenotype in the F₃ progeny. Additional recombinants were obtained by crossing an ap2 acd2 recombinant with cp3 to find plants in which acd2 did not segregate with ap2. We obtained two plants recombinant between ACD2 and AP2 and 82 plants recombinant between ACD2 and CP3. For fine mapping, we used the Cleaved Amplified Polymorphic Sequences (CAPS) method (24). The marker order is: AP2, JM142, JM127, JM110, JM411, and CP3. CAPS information was deposited at www.arabidopsis.org. One of the plants that was recombinant between AP2 and ACD2 was also recombinant at JM142 but not at JM127. One of the plants that was recombinant between ACD2 and CP3 was also recombinant at JM411 but not at JM110. Thus, the ACD2 locus mapped between JM142 and JM411, a 50-kb interval with 11 predicted ORFs.

Allele Sequence and Complementation Constructs.

PCR analysis of the six acd2 mutant alleles showed a small deletion in the AT4 g37000 transcript (Munich Information Center for Protein Sequences designation) in acd2–7, an allele generated during T-DNA transformation (15). Sequencing showed an in-frame deletion of amino acids 181–192 in acd2–7. Sequencing of pooled PCR products from mutant genomic DNA also showed alterations in this transcript in the other acd2 mutant alleles: both acd2–2 and acd2–5 have a G to A change in the 3′ splice acceptor site of the intron. Conceptual translation of the unspliced mRNA shows a stop codon 10 amino acids after the 5′ splice site. acd2–12E13 has a glycine-to-valine change at amino acid 140. Both acd2–6 and acd2–8 have an arginine-to-lysine change at amino acid 279. Sequence changes were confirmed by generation and scoring of a directed CAPS (dCAPS) marker (25) for each allele.

ACD2 cDNAs were derived from a Columbia library (26). The two longest cDNAs stopped short of the first two amino acids. For in vitro translation, we used PCR to restore the first two amino acids to the cDNA. Reverse transcription–PCR with primers 130 nucleotides upstream of the first AUG gave a specific product, indicating that the transcript extends upstream of the first methionine. The genomic complementation construct included the complete AT4 g37000 ORF plus approximately 1 kb upstream and 0.5 kb downstream, but none of the adjacent predicted genes. We used Pfu polymerase (Stratagene) to amplify (5′-aacaaaaccgatgaacaaagtagtcg-3′; 5′-ccgggatgaaaagaattatgtgg-3′), cloned the product into the HindIII site of pBI121 (CLONTECH), and introduced it into plants by Agrobacterium-mediated transformation (27). Four independent transformants of acd2–2 with this construct showed a wild-type phenotype.

Sense and Antisense Constructs.

For sense and antisense expression constructs of ACD2, the longest cDNA insert was cloned into the SmaI site of pBI121, downstream of the CaMV ³⁵S promoter. Of 17 transformants with the antisense construct, 10 showed a cell death phenotype resembling acd2 mutants. Two independent transformants of the sense construct into wild type showed increased levels of ACD2 protein on Western blots and decreased disease symptoms. Six transformants of the sense construct into acd2–2 complemented the cell death phenotype.

Western Blotting.

Polyclonal anti-ACD2 antibodies were generated in two rabbits by using inclusion body-purified bacterially expressed ACD2 protein (Covance, Denver, PA). The bacterial expression construct fused the second exon of ACD2, beginning with the amino acids IDFV, in-frame into the BamHI site of pET24b (Novagen). Anti-ACD2 antiserum was used for Western blotting at a concentration of 1:5,000 in BLOTTO (20 mM Tris, pH 7.5/154 mM NaCl/0.1% Tween 20/5% Carnation nonfat dry milk) and detected by donkey anti-rabbit horseradish peroxidase (Pierce) followed by chemiluminescent detection, per the manufacturer's instructions (Pierce).

Mitochondrial and Chloroplast Purification.

Chloroplasts were purified as in ref. 28; mitochondria were purified as in ref. 29. Organelle purity was assessed by chlorophyll content, by staining with the mitochondria-specific dye Mitotracker CMX-Ros (Molecular Probes), by Western blotting with antibody to the mitochondrial protease AtLON (ref. 30; gracious gift of S. Mackenzie, University of Nebraska, Lincoln), and by activity gels (31) for cytoplasmic, chloroplast, and mitochondrial isoenzymes of aspartate aminotransferase (32). This experiment was repeated three times with similar results.

Bacterial Strains, Pathogenicity Tests, and Growth Curves.

Derivatives of Pseudomonas syringae pv. maculicola strain PsmES4326 with vector (PsmDG3), avrRpm1 (PsmDG34), or avrRpt2 (PsmDG6) integrated at the recA locus are described in ref. 33. For Western blot analysis, avr genes cloned into the plasmid pLAFR were used; these constructs were a gift of F. M. Ausubel (Harvard University and Massachusetts General Hospital, Boston, MA). Hand inoculations, ion leakage, and growth curves were performed as described (23, 33). Growth curves were repeated twice with similar results.

Coronatine and Cycloheximide Treatments.

Fifteen microliters of 10 μg/ml coronatine (kind gift from R. Mitchell, Horticultural and Food Research Institute, Aukland, New Zealand), 0.1% methanol, or 20 μg/ml cycloheximide (34) was pipetted on top of the leaf. Dark-incubated leaves were covered in aluminum foil. This experiment was repeated three times with similar results.

Results

Cloning ACD2.

We cloned the ACD2 locus by standard Arabidopsis positional cloning methods, recombination mapping the locus to a 50-kb interval between two CAPS markers (see Materials and Methods). To identify the locus, we examined the putative transcripts in the region by PCR and found a small deletion in acd2–7. We isolated and sequenced cDNA clones corresponding to the ACD2 mRNA and used reverse transcription–PCR to determine the extent of the ORF. We proved that the affected ORF encodes the ACD2 gene. First, all of the previously isolated acd2 alleles had lesions in this ORF (see Materials and Methods). Second, either the genomic region covering the ACD2 locus or a full-length cDNA under the control of CaMV 35S promoter, introduced into mutant plants, fully complemented the acd2 phenotype (Fig. 1 and data not shown). Third, Western blot analysis showed that ACD2 protein is reduced or absent in acd2 mutants (Fig. 2A). Additionally, the mutant phenotype was induced by antisense expression of the ACD2 cDNA (data not shown). The predicted ACD2 protein shows extensive homology to RCCR (35). Indeed, the gene we isolated as ACD2 was designated AtRCCR, and its gene product catalyzes the reduction of RCC in vitro (35). We will refer to this gene as ACD2 or ACD2/RCCR.

Figure 1.

Complementation of the acd2 mutant phenotype. acd2–2 plants with a control transgene containing the gene adjacent to ACD2 (AT4 g36990) (A) and the ACD2 (AT4 g37000) genomic transgene (B).

Figure 2.

(A) Western blot of acd2 mutant plants with anti-ACD2 antiserum. Lanes contain protein from approximately equal weights of leaf tissue. Lanes: 1, Columbia; 2, acd2–2; 3, acd2–5; 4, acd2–6; 5, acd2–8; 6, acd2–12E13; 7, acd2–7. The doublet above the ACD2 band is not detected by antiserum from an independently inoculated rabbit or by affinity-purified antiserum (data not shown); thus, we conclude that it is not specific to the ACD2 protein. (B) ACD2 localizes to chloroplasts and mitochondria. Lanes: 1, crude extract; 2, purified mitochondria; 3, purified chloroplasts. M, mitochondrial form of ACD2; X, chloroplast form of ACD2; CHL, chlorophyll in mg/ml; MITO, mitochondria/microliter.

ACD2 Protein Localizes to the Chloroplast and Mitochondrion.

We had several lines of evidence that ACD2 localized to chloroplasts or mitochondria. First, ACD2 protein from leaves migrates at a smaller size than ACD2 protein in vitro translated from a full-length cDNA on a Western blot, consistent with cleavage of a transit peptide (data not shown). Second, analysis with the psort algorithm (36) predicts that ACD2 protein localizes to chloroplasts and mitochondria. Third, in vitro translated AtRCCR can be imported into isolated chloroplasts (35). To address the subcellular localization of ACD2, we purified chloroplasts and mitochondria from 7-day-old seedlings and analyzed the purified organelles by Western blot analysis. The mitochondrial fraction was substantially free of chloroplast contamination, as indicated by chlorophyll content (Fig. 2B). Also, the chloroplast fraction was substantially free of mitochondrial contamination, as indicated by staining with the mitochondrial-specific dye Mitotracker Red CMXRos (Fig. 2B) and by abundance of the mitochondrial isoenzyme of aspartate aminotransferase (32). ACD2 protein was substantially enriched in purified chloroplasts and mitochondria in Arabidopsis seedlings. The two organellar forms migrated at slightly different sizes, both smaller than in vitro translated product (data not shown), indicating that transit peptide cleavage may occur at different residues or that there is another differential modification (Fig. 2B). The two forms migrated as a doublet on Western blots, allowing us to monitor their presence and ratio. The mitochondrial form was less abundant than the chloroplast form (Fig. 2B, lane 1). Furthermore, the mitochondrial form was found only in 7-day-old seedlings, not in plants older than 2 weeks. Thus, ACD2 localizes mainly to the chloroplast.

ACD2 Protein Abundance in Senescence and Pathogen Attack.

ACD2/RCCR should function during the breakdown of chlorophyll in senescence and pathogen attack. To determine whether ACD2 protein levels change during senescence, we collected the ninth true leaf from wild-type plants, from a day 3 leaf that was expanding to a day 28 leaf that was yellowing and collapsed. Western blot analysis of equal amounts of protein from leaf extracts showed constant levels of ACD2 protein (Fig. 3A).

Figure 3.

Western blot analysis of ACD2 protein abundance in senescence and pathogenesis. Each lane contains equal amounts of protein by Bradford assay (Bio-Rad). (A) ACD2 in senescence. Duplicate samples of the ninth true leaf of wild-type plants through its development. At day 3, the leaf was approximately 3 mm long. The chlorophyll measurement is the average of the two leaves. (B) ACD2 in the HR. Panels show wild-type leaves mock inoculated (MgSO₄) or inoculated with 0.01 A₆₀₀ of P. syringae containing vector only (pLAFR) or containing pLAFR with avrRpm1 or avrRpt2. (C) ACD2 in disease. As in B, infection with P. syringae containing pLAFR only compared with mock inoculation. Strong disease symptoms were seen at 48 h.

To determine whether ACD2 protein levels change during pathogen attack, we inoculated wild-type plants with derivatives of P. syringae strain PsmES4326 containing either the pLAFR vector only (virulent) or pLAFR containing either avrRpm1 or avrRpt2 (avirulent). Infection with strains bearing avrRpm1 and avrRpt2 provoked an HR 6 and 16 h after inoculation, respectively. Virulent bacteria showed disease symptoms after 48 h. Western blot analysis showed no significant differences in ACD2 protein levels in any of the infections (Fig. 3 B and C). Thus, although ACD2 likely plays a key role in chlorophyll catabolism, its levels do not change during pathogen attack or senescence.

Light Activates the acd2 Mutant Phenotype.

A predicted phenotype of a mutation in ACD2/RCCR is an accumulation of porphyrin compounds such as RCC and, possibly, its precursor pheophorbide a. The phototoxic properties of these compounds should depend on light. To determine whether the acd2 phenotype requires light, we induced the phenotype in mutant leaves by treating with the P. syringae toxin coronatine. In Arabidopsis, this toxin partially mimics the hormone methyl jasmonate, alters chloroplast functions, and induces the expression of chlorophyllase, the first enzyme in the chlorophyll breakdown pathway (37, 38). In wild-type Arabidopsis, 1.5–150 ng of coronatine induces anthocyanin accumulation but no significant cell death (ref. 39 and data not shown). In acd2 mutants, as little as 1.5 ng induced cell death within 24 h (data not shown). acd2 mutant leaves were treated with 15 ng of coronatine and either covered or exposed to light for 24 h. The absence of light completely suppressed cell death in acd2 (Fig. 4A).

Figure 4.

Effects of light, translation, and SA on acd2 cell death. (A) Light activates the acd2 mutant phenotype. The center of each indicated acd2–2 leaf was spotted with either 0.01% methanol as a control (M) or coronatine (D and L). The leaf marked D (dark) was covered. The other leaves (including L, light) remained in the light for 24 h. Similar treatment of the wild-type control did not induce any cell death or chlorosis. (B) The acd2 phenotype requires active translation. Wild-type (COL) or acd2–2 mutant (acd2) plants were spotted with either coronatine (C) or coronatine and cycloheximide (C + CHX) and were left in the light for 24 h. Depletion of SA exacerbates the acd2 mutant phenotype. acd2–2 (C) or acd2–2 nahG (D).

The acd2 Mutant Phenotype Requires Active Translation.

To determine whether the cell death in acd2 is an active process, we applied cycloheximide, an inhibitor of translation, to acd2 mutant leaves along with coronatine. If acd2 cells die by passive toxic cell death, then cycloheximide should not affect the formation and spread of the lesion. Instead, we found that application of cycloheximide inhibited cell death in response to coronatine (Fig. 4B) and the spread of cell death in spontaneous lesions (data not shown).

Depletion of SA Exacerbates the acd2 Mutant Phenotype.

The defense signal SA is required for cell death induced by some pathogens (40), accumulates in acd2 plants (15), and activates many of the defenses expressed in acd2 mutants. To determine whether the acd2 cell death phenotype requires SA, we crossed acd2–2 plants with nahG plants, in which SA is depleted (41). We isolated acd2 nahG plants by screening for F₂ progeny containing the nptII gene of the nahG transgene and lacking ACD2 protein. acd2 nahG plants showed waves of cell death that initiated at approximately the same time of development as acd2 single mutants but spread to consume young leaves as well as old (Fig. 4 C and D). Thus, depletion of SA amplified the cell death phenotype of acd2, producing a systemic cell death phenotype.

Overexpression of ACD2 Protein Suppresses Cell Death in Bacterial Infection.

If ACD2 acts to suppress cell death, then overexpression of ACD2 might prevent cell death. We engineered plants that produce excess ACD2 protein (Fig. 5A) from a CaMV 35S-driven ACD2 cDNA (35S-ACD2). To test the cell death responses of these plants, we inoculated leaves with avirulent derivatives of PsmES4326 (containing the avrRpm1 or avrRpt2 genes) or the cognate virulent strain PsmDG3. The 35S-ACD2 plants showed no detectable reduction of the HR with the avirulent pathogens but did show reduced yellowing and cell death, as measured by ion leakage, with the virulent pathogen (Fig. 5B). The reduced disease symptoms of 35S-ACD2 plants, if correlated with reduced growth of the pathogen, could indicate activation of the defense response. However, the growth of PsmDG3 was indistinguishable in wild type and 35S-ACD2 (Fig. 5C). Thus, overexpression of ACD2 protein made the plants tolerant but not resistant to bacterial infection.

Figure 5.

Overexpression of ACD2 protein suppresses symptoms of pathogen infection. (A) Western blot of plant extracts, with equal amounts of protein loaded. Lanes 1 and 2, wild type; lane 3, 35S-ACD2. (B) Ion leakage 48 h after inoculation with 0.003 A₆₀₀ of P. syringae strain PsmDG3. White bars are wild type; shaded bars are 35S-ACD2. (C) Growth, in colony-forming units/cm², of virulent PsmDG3 on wild type (○) or 35S-ACD2 (■).

Discussion

Chlorophyll breakdown during senescence detoxifies photoreactive molecules so that other components of the chloroplast can be recycled. Catabolism of chlorophyll (reviewed in ref. 3) begins in the chloroplast, with the removal of the phytol tail and the chlelated magnesium ion from the chlorophyll porphyrin, forming pheophorbide a (Pa). Next, the porphyrin ring is broken by the enzyme Pa oxidase, forming RCC. Plants defective in this step accumulate Pa and its precursors and do not yellow as they age, instead exhibiting a “stay-green” phenotype (42–44). RCC is broken down by ACD2/RCCR, eventually forming fluorescent chlorophyll catabolites, which are moved to the vacuole for storage. RCCR activity is found throughout terrestrial plants (45). acd2 is the first mutant shown to affect the gene for an enzyme of chlorophyll catabolism.

Disruption of chlorophyll catabolism by the acd2 mutant is likely to cause the accumulation of RCC and Pa. This could cause spreading cell death by one of two mechanisms. In the first mechanism, accumulation of phototoxic porphyrin molecules causes cell death by the light-dependent production of free radicals. SA-depleted acd2 plants show enhanced cell death; this may be caused by a lowered antioxidant capacity of the plants, as SA can be important for maintaining the redox capacity of plants in some situations (46). Also, the application of SA analogs produces an increase of copper–zinc superoxide dismutase, an antioxidant enzyme (19). Depletion of SA also exacerbates the phenotype of the lesion mimic mutants lsd2 and lsd4 (47) and the HR in response to some pathogens (ref. 46 and D. N. Rate and J.T.G., unpublished work).

Accumulation of photoactive porphyrin compounds has been shown to be toxic to both plants and animals (8, 48). In plants, accumulation of porphyrin precursors causes a lesion mimic phenotype and induction of defense gene expression (49–51). Interestingly, Les22 mutant maize leaves, which accumulate uroporphyrin III (9), show punctate discrete lesions, in contrast to the spreading lesions of acd2 mutant leaves. This may reflect a difference in porphyrin localization, or porphyrin accumulation alone may not be sufficient to cause cell death. In support of the latter possibility, some strains of grass and legumes accumulate Pa but do not show a cell death phenotype (42–44).

In a second possible mechanism to account for the acd2 cell death phenotype, accumulation of RCC may cause a specific signal that triggers cell death. Interestingly, in mammalian systems, protoporphyrin IX, but not other porphyrins such as porphobilinogen, triggers apoptosis by induction of the mitochondrial permeability transition (10, 11). RCC may specifically trigger cell death in plants, possibly by affecting mitochondria and/or chloroplasts. In mammalian systems, mitochondria integrate cell death signals and trigger cell death (12). Both mitochondrial (52) and chloroplast functions may affect the life–death decision in plants. For example, perturbation of the levels of the chloroplast protease FtsH affects cell death in response to tobacco mosaic virus (53).

What causes the rapid spread of cell death in acd2 plants? We hypothesize that cell death in this mutant activates chlorophyll breakdown, which then causes more cell death. This process requires both active translation and light, which are required for cell death in acd2 mutant protoplasts (54). Translation may be involved in the activation of cell death, in the initiation of chlorophyll breakdown, or in other aspects of the process. Blocking translation inhibited degreening and accumulation of RCC in the alga Chlorella (55). Light may be absorbed by accumulated porphyrins, it may activate photosensitive signaling pathways, or it may be used for photosynthesis, providing energy required for active cell death.

Although we have concentrated on the likely effect of the acd2 mutation on chlorophyll catabolism, the ACD2 gene may have additional, possibly redundant, functions. ACD2 protein is present at all times of development and in all tissues tested, including roots (ref. 35 and data not shown). The mutant phenotype of defense gene transcription and cell death activation begins at the same time as chlorophyll breakdown in senescence. However, acd2 plants show some elevated resistance to P. syringae as early as the four-leaf stage (15). The mitochondrial localization of ACD2 protein, although transient, may also reflect additional functions. Protein localization to either chloroplasts or mitochondria is usually highly specific (56). However, glutathione reductase produces forms that localize to the chloroplast, mitochondrion, and cytoplasm from a single nuclear gene (57). Examination of the temporal and spatial requirement for ACD2 may prove instructive.

Plants with no ACD2 show spreading cell death. Conversely, plants expressing high levels of the ACD2 protein show increased tolerance to infection with virulent P. syringae. An increase in ACD2/RCCR protein levels may alter the flux of chlorophyll catabolites that may normally accumulate during disease and trigger cell death. The ethylene-insensitive mutant ein2 shows similar tolerance to bacterial infection (39). The ethylene-insensitive etr1 and ein2 mutants show delayed senescence (ref. 58 and J.T.G., unpublished observations). Ethylene induces the synthesis of chlorophyllase in Citrus fruit (59). It is unknown whether ein2 alters the flux of chlorophyll catabolites, but it will be interesting to determine whether there are shared mechanisms of tolerance. Manipulation of chlorophyll breakdown may be useful in creating agronomically useful tolerance to pathogens.

Abbreviations

HR

hypersensitive response

Pa

pheophorbide a

RCC

red chlorophyll catabolite

RCCR

RCC reductase

SA

salicylic acid