Changes in hydrogen peroxide homeostasis trigger an active cell death process in tobacco


  • James F. Dat,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
    • Present address: Laboratoire de Biologie et Ecophysiologie EA 3184, Université de Franche-Comté-INRA, F-25030 Besançon Cedex, France.

  • Riikka Pellinen,

    1. Institute for Biotechnology, University of Helsinki, POB 56, Viikinkaari 5D, FIN-00014 Helsinki, Finland, and
    Search for more papers by this author
    • Present address: A.I. Virtanen Institute, University of Kuopio, POB 1627, FIN-70211 Kuopio, Finland.

  • Tom Beeckman,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
  • Brigitte Van De Cotte,

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author
  • Christian Langebartels,

    1. Institute of Biochemical Plant Pathology, GSF National Research Center for Environment and Health, Ingolstädter Landstraße 1, D-85764 Neuherberg, Germany
    Search for more papers by this author
  • Jaakko Kangasjärvi,

    1. Institute for Biotechnology, University of Helsinki, POB 56, Viikinkaari 5D, FIN-00014 Helsinki, Finland, and
    Search for more papers by this author
  • Dirk Inzé,

    Corresponding author
    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
      For correspondence (fax +32 9 264 5349; e-mail
    Search for more papers by this author
  • Frank Van Breusegem

    1. Department of Plant Systems Biology, Flanders Interuniversity Institute for Biotechnology, Ghent University, K.L. Ledeganckstraat 35, B-9000 Gent, Belgium,
    Search for more papers by this author

For correspondence (fax +32 9 264 5349; e-mail


In transgenic tobacco plants with reduced catalase activity, high levels of hydrogen peroxide (H2O2) can accumulate under photorespiratory conditions. Such a perturbation in H2O2 homeostasis induced cell death in clusters of palisade parenchyma cells, primarily along the veins. Ultrastructural alterations, such as chromatin condensation and disruption of mitochondrial integrity, took place before cell death. Furthermore, enhanced transcript levels of mitochondrial defense genes accompanied these mitochondrial changes. Pharmacological data indicated that the initiation and execution of cell death require de novo protein synthesis and that the signal transduction pathway leading to cell death involved changes in ion homeostasis, (de)phosphorylation events and an oxidative burst, as observed during hypersensitive responses. This oxidase-dependent oxidative burst is essential for cell death, but it is not required for the accumulation of defense proteins, suggesting a more prominent role for the oxidative burst in abiotic stress-induced cell death.


Cell death is an essential event in animal and plant development. In animal cells, two distinct modes of cell death are generally described: apoptosis and necrosis (Kroemer et al., 1998). Although it is becoming gradually more evident that apoptosis and necrosis are only two extreme ends to cell death, the differentiation between upstream events is usually more complex (Fiers et al., 1999; Jones, 2000). ‘Programmed cell death’ (PCD) or ‘active cell death’ are terms used in plant science to qualify cell death events that may include some characteristic features of animal apoptotic cell death, such as chromatin aggregation, cell shrinkage, cytoplasmic and nuclear condensation, and DNA fragmentation (Buckner et al., 2000; Jabs, 1999; O'Brien et al., 1998), but foremost, it is an event that requires intracellular processes for death to occur. PCD events in plants have been best characterized during incompatible plant–pathogen interactions (Pennell and Lamb, 1997), but several abiotic stresses, such as ozone, UV irradiation, chilling, and salt stress also induce DNA laddering and other ultrastructural symptoms typical of PCD (Danon and Gallois, 1998; Katsuhara, 1997; Kratsch and Wise, 2000; Pellinen et al., 1999). In contrast, ‘plant necrosis’ or ‘passive cell death’ are terms used to describe cell death that results from severe trauma during extreme stress situations and occurs immediately and independently of any cellular activity (O'Brien et al., 1998).

Whereas reactive oxygen species (ROS) and associated redox imbalance are key players in the signaling cascades leading to and executing cell death in mammalian systems, the precise role of ROS in plant cell death programmes is still a matter for debate (Jabs, 1999; Lam et al., 1999; Loake, 2001; Noodén et al., 1996). During the hypersensitive response (HR), generation and accumulation of ROS generally coincide with the induction of cell death. In contrast, ROS and, more particularly, hydrogen peroxide (H2O2) have also been implicated in the induction of many plant defense responses (Buckner et al., 2000; Dat et al., 2000; Pennell and Lamb, 1997). It is still unclear whether different cell death signaling pathways overlap during normal development, environmental stress, and pathogen attack (Greenberg, 1996; Wang et al., 1996).

Most studies on the role of ROS in plant cell death have used indirect methods to generate ROS, but the catalase-deficient plants (CAT1AS) used in this study provide an ideal tool to investigate the effect of in planta changes in H2O2 homeostasis in a non-invasive way. The CAT1AS plants, which have a reduced catalase activity in the peroxisomes, were originally produced to study the role of catalases in plants (Willekens et al., 1995). Under photorespiratory conditions created by high light (HL) exposure, H2O2 homeostasis is altered. These HL-dependent H2O2 changes have been used previously to study the role of H2O2 signaling during the induction of defense responses (Chamnongpol et al., 1996; Dat et al., 2001; Takahashi et al., 1997; Willekens et al., 1997). However, the role of in planta changes in H2O2 homeostasis on plant cell death has never been thoroughly investigated. Here, we describe a detailed investigation of cellular and signaling characteristics of H2O2-induced cell death in CAT1AS plants.


Changes in H2O2 homeostasis precede cell death in periveinal palisade cells

When transgenic tobacco plants (CAT1AS) that retain only 10% of wild-type catalase activity (Chamnongpol et al., 1996) are exposed for at least 8 h to HL (1000 µmol m−2 sec−1), patches of gray, chlorotic areas consisting of dying cells develop initially in leaf periveinal regions. These patches then extend to the entire interveinal area. This characteristic development of the cell death pattern was monitored by time-lapse video imaging. A single image was obtained every 10 min after the transfer to HL conditions for 24 h. Figure 1 shows the development of cell death at specific time points. The complete time series was used to generate a stack, which was then transformed into a movie (available as supplementary material on the web at Development of cell death was also assessed in leaf cross-sections. The determination of the number of dead cells in each cell layer showed that after 8 h of HL, dead cells were found only in the palisade layer. Even after 24 h of HL, when over 97% (SE ± 1.2%) of the palisade cells were dead, only 3.5% (SE ± 0.6%) of the spongy mesophyll cells had collapsed. Hence, cell death was almost solely localized in palisade parenchyma cells.

Figure 1.

Selected pictures from a time-lapse video analysis showing the changes in phenotype occurring in the leaf during a 24-h exposure to high light (HL) in CAT1AS leaves.

Images were obtained every 10 min and then selected. Supplementary material is available at

In an attempt to assess the potential role of HL-dependent changes in H2O2 homeostasis in this cell death pattern, H2O2 production was visualized by staining leaf segments with 3′3-diaminobenzidine (DAB), a histochemical reagent that polymerizes and turns brown in the presence of H2O2 (Orozco-Cardenas and Ryan, 1999; Thordal-Christensen et al., 1997). Leaves of 6-week-old CAT1AS and wild-type plants were vacuum-infiltrated with DAB and subsequently placed under HL. H2O2 production in both CAT1AS and wild-type plants exposed to HL was analyzed in function of time (0, 15, 30, 45 min and 1, 2, 4, 8, 24 h). No or very weak staining was visible in CAT1AS or wild-type leaves harvested prior to the HL treatment. Within 45 min, clusters of cells along the veins became more darkly stained than interveinal regions in CAT1AS leaves (Figure 2a). The DAB staining was primarily located in the palisade parenchyma layer and not in the epidermal layer, as viewed under a light microscope (data not shown). The experiment was repeated six times with similar results: patches of dark-stained cells occurred systematically and predominantly along the veins. The DAB staining of CAT1AS leaves increased with time under HL. After 2 h, staining was clearly visible in most of the leaf sections, although patches of more darkly stained periveinal cells could still be distinguished. HL treatment of wild-type plants resulted only in a weak background DAB staining. DAB staining was not affected by a rate-limiting peroxidase reaction with H2O2 (Thordal-Christensen et al., 1997) because staining of either wild-type or CAT1AS leaf sections infiltrated with 10 mm H2O2 was similar under low light (LL) and HL. Furthermore, DAB staining depended on H2O2 because infiltration of catalase (2000 U) completely inhibited the staining (data not shown; Willekens et al., 1997).

Figure 2.

Hydrogen peroxide (H2O2) accumulation and cell death pattern in CAT1AS plants exposed to high light (HL) for various time periods.

(a) Diaminobenzidine (DAB) staining of CAT1AS leaf segments (exposed to HL for 0, 15, 30, 45 min and 1, 2, 8 and 24 h) and a representative SR1 leaf segment after 24 h of exposure to HL. The dark-brown areas indicate regions of high H2O2 production. Staining occurs primarily along secondary veins and in clusters of cells.

(b) Trypan blue-stained leaf segments of CAT1AS plants exposed to HL for 0, 8 and 24 h. Dead cells (stained blue) are predominant along the veins after 8 h of HL. Almost all cells of the palisade parenchyma are dead after 24 h of HL.

(c) Toluidine blue-stained sections of CAT1AS after 0, 8 and 24 h of HL. The sections show cell death and migration of cellular constituents after 8 h of HL, mainly in the palisade parenchyma after 24 h of HL. The cytosol is absent in the upper parts of the palisade cells bordering the epidermal cells. Bars = 200 µm (a,b), 25 µm (c).

In a parallel experiment, the time course of cell death in CAT1AS exposed to HL was visualized with Trypan blue (a dye that specifically stains dead cells). No Trypan blue staining was observed in wild-type plants exposed for up to 24 h to HL (time points: 0, 15, 30, 45 min and 1, 2, 4, 8, 24 h HL). In contrast, exposing CAT1AS plants to HL for 8 h or more, resulted in staining of palisade parenchyma cells (time points: 0, 15, 30, 45 min and 1, 2, 4, 8, 24 h HL). The pattern of Trypan blue staining (dead cells) in CAT1AS leaves was very similar to that observed in DAB-stained segments at 45 min. Patches of palisade parenchyma cells died predominantly in periveinal regions, in accordance with the development of patches of gray tissue in periveinal regions as visualized previously by the time-lapse video imaging (Figures 1 and 2b). These results indicate that in CAT1AS leaves exposed to HL, H2O2 accumulation and cell death execution are spatially correlated.

Characteristic features of active cell death in CAT1AS leaves

In order to gain more insight into cell death events, cross-sections of CAT1AS leaves exposed for different time periods to HL (0, 15, 30, 45 min and 1, 2, 4, 8, 24 h) were analyzed. Differential interference contrast optics and transmission electron micrographs (TEMs) were used to detect early cytological alterations that take place prior to the execution of cell death. The movements of chloroplasts towards the bottom of the palisade parenchyma cells was observed after 8 h HL (Figure 2c). Leaf cells of wild-type plants exposed to HL did not show such changes. TEMs of leaf sections revealed changes in chromatin distribution in the nucleus (Figure 3a,b). After 8 h of HL, the chromatin of CAT1AS nuclei aggregated. In addition, when 4′,6-diamidino-2-phenylindole (DAPI)-stained nuclei, isolated from leaves of CAT1AS exposed to HL for various periods, were scanned by flow cytometry, additional peaks with lower fluorescence appeared in front of the G1 and G2 peaks (Figure 4). In contrast, no laddering was noted after gel blot analysis of DNA extracted from HL-exposed CAT1AS leaves (data not shown). The observed changes (in nuclei and chromatin) suggest the activation and execution of an active cell death process in HL-exposed CAT1AS plants.

Figure 3.

Transmission electron micrographs (TEM) of nuclei and mitochondria of palisade parenchyma cells of CAT1AS plants exposed to high light (HL).

(a) Nucleus after 0 h of HL.

(b) Nucleus after 8 h of HL showing chromatin aggregates (arrows).

(c) Mitochondria after 0 h of HL.

(d) Bursting of mitochondria (arrow) after 4 h of HL. a, apoplast; c, chloroplast; cw, cell wall; cyt, cytoplasm; m, mitochondrion; n, nucleus; s, starch grain; v, vacuole. Bars = 200 nm.

(e) Relative increase of MnSOD and cytochrome c (cyt. c) in cytosolic fractions of CAT1AS and SR1 leaves exposed to HL for 3, 8 and 24 h, determined by protein blot analysis and densitometry.

Figure 4.

Flow cytometry of DAPI-stained nuclei of CAT1AS plants exposed to high light (HL) for various time periods.

DNA fragmentation is detectable after 8 h of HL in CAT1AS plants. FL1, fluorescence; LS1, forward scatter.

Transmission electron micrographs of leaf sections exposed to HL for the same time points mentioned above revealed that after 4 h of HL, the mitochondrial cristae became disorganized, the electron density of the matrix decreased, and the membrane sometimes became disrupted (Figure 3c,d). Because mitochondria are important integrators during the induction of apoptotic cell death in animal systems (Kroemer et al., 1998), we monitored molecular changes related to mitochondrial oxidative stress during H2O2-induced cell death. Accordingly, transcript levels of both alternative oxidase (aox) and manganese superoxide dismutase (mnsod) genes accumulated in CAT1AS plants within a few hours after the start of the HL treatment (Figure 5). However, we cannot exclude that aox levels are independently upregulated by elevated salicylic acid levels in HL-treated CAT1AS plants (Chivasa et al., 1997; Rhoads and McIntosh, 1992; Willekens et al., 1997). Because mitochondrial cytochrome c release is often considered a key event of PCD in both plants and animals (Hancock et al., 2001; Tiwari et al., 2002), we assessed cytochrome c release during H2O2-induced cell death. Leaf cytosolic protein fractions of both CAT1AS and SR1 plants exposed to HL for 0, 3, 8 and 24 h were separated by SDS–PAGE, and the relative amount of cytochrome c levels was determined by Western blotting. The level of MnSOD, a mitochondrial matrix protein, was detected after stripping and re-hybridizing the same filter. The values obtained were used to get an indication of mitochondrial matrix protein content in the cytosolic fraction. Cytosolic cytochrome c levels increased significantly in CAT1AS plants after 3 h of HL exposure (Figure 3e). This timing coincided with mitochondria bursting and was well before the first cells died (as evidenced by Trypan blue staining; Figure 2b). In contrast, cytosolic cytochrome c levels did not change significantly in SR1 control plants during the 24-h HL exposure. Three independent experiments produced similar results. The compiled cytological and transcriptional data clearly show that variations in H2O2 homeostasis in the CAT1AS plants affect mitochondrial metabolism and function, and that these events precede the cell death event.

Figure 5.

Transcript levels of aox and mnsod in CAT1AS plants exposed to high light for 0, 1, 2, 4 and 8 h.

HL-induced cell death in CAT1AS plants is H2O2 specific

There is an on-going debate on the role of H2O2 versus equation image during PCD in plants (Alvarez et al., 1998; Jabs et al., 1997). Therefore, we undertook a series of experiments to verify that cell death in CAT1AS plants was indeed triggered by H2O2. Infiltration of antioxidants (catalase, ascorbate, and glutathione) in CAT1AS leaves inhibited H2O2 accumulation and protected them from cell death (Chamnongpol, 1997; Table 1). However, when catalase was infiltrated 6 h after the beginning of the HL treatment, it could not prevent cell death anymore. Infiltration with a glucose/glucose oxidase solution in both CAT1AS and wild-type plants provoked cell death under LL with a similar phenology as that observed in HL-treated CAT1AS plants. In contrast, injection with superoxide generators (xanthine/xanthine oxidase, rose bengal, diethyldithiocarbamate; for details, see Table 1) under LL generated another type of cell death (Figure 6a). The superoxide-dependent cell death was typified by complete collapse of tissues because all cell layers were destroyed (palisade, spongy, and epidermal), whereas the HL-induced cell death in CAT1AS mainly took place in the palisade layer, as described above. These data indicate that H2O2 is the initial and prime signal responsible for the cell death that develops in CAT1AS plants exposed to HL.

Table 1.  Compounds used for the pharmacological experiments to test the presence of known signaling cascade components after injection prior to a 24-h HL treatment
AgentaPresumed mode of actionInfiltrated concentrationEffect on HL-induced cell death in CAT1AS
Ascorbic acidAntioxidant10–50 mmNone
100 mmProtection
GlutathioneAntioxidant1–5 mmNone
10–20 mmProtection
CatalasebSpecific H2O2 scavenger2000 U ml−1Protection
Heat-inactivated catalase 2000 U ml−1None
Dithiothreitol 0.1–1 mmNone
zVAD-fmkCaspase inhibitor100 mmNone
Peroxidase 25–2500 U ml−1None
H2O2 (X) 1–5 mmNone
10–100 mmIncreased cell death
Xanthine/xanthine oxidase (X)equation image generator1 mm per 0.5 UNone
10 mm per 0.5 UIncreased cell death (different phenotype)
Diethyldithiocarbamate (X)SOD inhibitor1–5 mmNone
10–100 mmIncreased cell death (different phenotype)
Rose bengal (X)ROS generator0.1–1 µmIncreased cell death (different phenotype)
Diphenylene iodiniumcNADPH-oxidase inhibitor0.1–10 µmNone
25–50 µmModerate protection
100 µmProtection
QuinacrinedOxidase inhibitor0.1–10 µmNone
25 µmModerate protection
50 µmProtection
Oxalate 5–20 mmProtection
CaffeineRelease Ca2+25–50 µmNone
100 µmIncreased cell death
CaCl2Release Ca2+1–100 mmNone
LaCl3eCa2+ channel blocker1–5 mmModerate protection
10 mmProtection
BAPTACa2+ chelator1–5 mmModerate protection
EGTACa2+ chelator10 mmProtection
Gd3+fCa2+ chelator10 mmProtection
Okadaic acidPhosphatase inhibitor0.03–0.1 µmNone
0.3 µmProtection
1–3 µmIncreased cell death
H7Protein kinase inhibitor1-µMNone
10–100 µmProtection
250–500 µmNone
CycloheximideProtein synthesis inhibitor0.1–20 µmNone
50 µmProtection
100 µmIncreased cell death (different phenotype)
Figure 6.

Effect of diethyldithiocarbamate, diphenylene iodinium (DPI), quinacrine (Q), and LaCl3 on cell death in leaf segments of CAT1AS plants.

(a) Leaf segments of CAT1AS injected with 50 µm diethyldithiocarbamate under low light (LL), 100 µm DPI and 50 µm Q, exposed to a 24-h high light (HL) treatment. Arrowheads indicate injected segments, adjacent segments were injected with water as a control.

(b) Microscopic evaluation of the effect of DPI on cell morphology. Toluidine blue-stained cross-sections of CAT1AS leaf segments injected with either water (H2O), 100 µm DPI, and exposed to 24 h HL. DPI blocked cell death in the palisade parenchyma layer and no obvious cellular changes were observed. Bar = 50 µm.

(c) Transcript levels of rbohD in CAT1AS plants exposed to HL for 0, 1, 2, 4 and 8 h. Methylene blue staining of Northern blot is shown as loading control.

(d) Protein accumulation of PR-1a, GPX, and heat shock protein HSP17.6 in CAT1AS after injection with H2O (control), 10 mm LaCl3, or 100 µm DPI, 24 h after either an LL or HL treatment. The signs + and – indicate occurrence and inhibition of cell death, respectively, as assessed in Table 1.

Cell death in CAT1AS plants is an active process involving an oxidative burst

To assess the requirement of de novo protein synthesis for in planta H2O2-induced cell death, interveinal leaf regions were injected with a range of cycloheximide concentrations (Jonak et al., 2000; Solomon et al., 1999). Injection of 50 µm cycloheximide, prior to a 24-h HL exposure, effectively prevented cell death in the injected region from a subsequent 24-h HL treatment (Table 1). These results suggest that de novo protein synthesis is required for HL-induced cell death to develop in CAT1AS plants.

The specific location of the H2O2 staining pattern (palisade parenchyma cells in the proximity of the veins) and subsequent cell death bear similarity to the cell death phenotype caused by ozone stress observed in tobacco and birch seedlings (Pellinen et al., 1999; Schraudner et al., 1998). The fact that ozone-induced cell death has been shown to be amplified by the activation of a diphenylene iodinium (DPI)-sensitive oxidative burst (Pellinen et al., 1999; Rao and Davis, 1999) prompted us to investigate whether a similar mechanism was induced during H2O2-induced cell death in CAT1AS plants. Leaf segments were injected with a range of concentrations of two inhibitors, which are commonly used to study the potential function of the oxidative burst in regulating plant responses: DPI and quinacrine (Q) (Auh and Murphy, 1995; Doke et al., 1994; Levine et al., 1994; Rao and Davis, 1999; Van Gestelen et al., 1997). Figure 6(a,b) shows that protection from HL-induced cell death was best with concentrations of either 100 µm DPI or 50 µm Q. The DPI concentrations may seem high when compared to those commonly used in other studies; however, concentrations of 0.1–20 µm DPI were utilized in cell suspension cultures and not in intact plants (Cazaléet al., 1999; Jabs et al., 1997; Long and Jenkins, 1998; Romeis et al., 2000; Table 1). We also found that oxalate injections protected leaves from cell death (Table 1). Oxalate inhibits a signaling step positioned upstream of the oxidase assembly/activation during Sclerotinia infection and, thus, the oxidative burst (Cessna et al., 2000). Finally, transcript levels encoding a tobacco homolog (NtrbohD) of the human superoxide-producing NADPH oxidase (Lambeth, 2002) were induced in the CAT1AS plants exposed to HL (Figure 6c). These data support the requirement for an oxidase-dependent burst to initiate H2O2-induced cell death in our model system. Analysis of the effect of the inhibitors revealed that H2O2-induced accumulation of the pathogenesis-related 1a (PR-1a) and glutathione peroxidase (GPX) proteins was only slightly reduced by DPI injections, whereas levels of the heat shock protein HSP17.6 were not (Figure 6d). This observation suggests that the oxidase-dependent burst is required for cell death but is not essential for defense protein accumulation.

Further evidence for the involvement of an oxidative burst was obtained with a selection of commonly used inhibitors and activators that block the signaling pathway leading to the oxidative burst (Table 1). The molecules were injected in CAT1AS and wild-type plants just before HL treatment to test their ability to block HL-induced cell death. Pilot experiments were undertaken to determine the minimum concentration that gave the maximum response. The leaves were scored for cell death development after a 24-h HL treatment.

Reactive oxygen species production during plant–pathogen interaction is compromised in the presence of Ca2+-chelating or Ca2+ channel-inhibiting compounds (Scheel, 1998). Therefore, we tested the effect of injecting Ca2+ channel blockers or Ca2+ chelators on the induction of cell death by HL treatment. The Ca2+ channel blockers and chelators tested (LaCl3, Gd3+, 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA), ethylene glycol-bis(β-aminoethyl ether) N,N,N′,N′-tetraacetic acid (EGTA)) (Bauer et al., 1998; Cazaléet al., 1998; Price et al., 1994; Romeis et al., 2000; Suzuki et al., 1999) effectively inhibited cell death. Furthermore, HL-induced accumulation of GPX, ascorbate peroxidase (data not shown), and HSP17.6 proteins remained unaltered by LaCl3 injections, whereas PR-1a accumulation was only slightly reduced, suggesting that as with the DPI injections, cell death and accumulation of these defense proteins require different signaling components downstream of the HL-induced changes in H2O2 homeostasis (Figure 6c). Gain-of-function experiments confirmed a role for ion fluxes in the cell death process, as caffeine (≥50 µm), a Ca2+-releasing agent (Bauer et al., 1998) activated cell death (Table 1). A broad-range protein kinase inhibitor (1-(5-isoquinolinesulfonyl)-2-methylpiperazine) (H7; Levine et al., 1996; Moutinho et al., 1998) also very efficiently inhibited cell death in leaves when injected just before the HL treatment. Finally, okadaic acid, an inhibitor of type 1 and type 2A protein phosphatases (Johansson et al., 1998; Kuo et al., 1996), protected leaves from HL-induced cell death, when injected 12 h before the HL treatment.

In contrast, horseradish peroxidase, CaCl2, DDT, and the broad-range caspase inhibitor, benzyloxycarbonyl-Val-Ala-dl-Asp(OMe)-fluoromethylketone (zVAD-fmk), did not inhibit nor accelerate cell death over a wide range of concentrations (Table 1). Finally, control injections with water, 0.1% dimethylsulfoxide, or heat-inactivated catalase indicated that injection of any substance was not sufficient to inhibit cell death (Table 1). The compiled data obtained show that changes in ion homeostasis and a phosphorylation poise are involved in the signaling pathway leading to cell death. Furthermore, our results demonstrate that cell death events happen downstream of the initial H2O2 signal, and are independently regulated from the accumulation of defense proteins.


The present study shows that an active cell death programme is triggered by perturbating in planta H2O2 homeostasis in transgenic tobacco plants. Cell death was found almost exclusively in the palisade parenchyma periveinal region and, thus, shares spatial location with cell death induced by ozone exposure (Pellinen et al., 1999; Rao et al., 2000; Schraudner et al., 1998), and by some biotic interactions (Alvarez et al., 1998; Tamagnone et al., 1998; Tang et al., 1999). This particular cell death pattern could be attributed directly to changes in H2O2 homeostasis in the tissue because the high H2O2 production and subsequent cell death in CAT1AS plants were both inhibited by exogenously applied catalase and could be mimicked by H2O2-generating compounds. The containment of cell death in the palisade parenchyma layer may be explained by higher photorespiratory activities and/or lower antioxidant capacity in this cell layer, both facilitating the accumulation of the threshold H2O2 levels needed to activate an irreversible cell death programme. Alternatively, additional factors needed to induce or execute cell death are predominantly present in palisade cells (i.e. lipid peroxidation products and other cellular messengers, such as NO or ROS), which is supported by the recently proposed close interplay between H2O2, equation image, and NO during the initiation of cell death (Delledonne et al., 2001; Torres et al., 2002; Wendehenne et al., 2001).

Cell death in CAT1AS leaves is preceded by various early cytological changes: migration of cellular constituents, chromatin aggregation and DNA fragmentation, nuclear hallmarks of PCD (Asai et al., 2000; Mittler et al., 1997; Xu and Roossinck, 2000) and changes in metabolism as evidenced by enhanced transcript levels of several mitochondrial respiratory chain proteins (S. Vandenabeele, unpublished results) and two mitochondrial oxidative stress defense transcripts. Cytochrome c release from mitochondria into the cytosol coincided with some mitochondrial bursting, both events taking place well before cell death execution. Nevertheless, we cannot conclude that cytochrome c release is an active mechanism (formation and opening of permeability transition pores), as cytochrome c release also coincides with some mitochondria bursting and the release of MnSOD into the cytosol. The complete burst of some mitochondria before the occurrence of dead cells suggests that mitochondria in plant cells may behave differently from animal cells during the execution of cell death programmes. Interestingly, the mechanism of cytochrome c release is not completely clarified in the animal field, and whether it takes place with or without rupturing of the outer membrane of mitochondria is still a matter for debate (Bernardi et al., 2001; Halestrap et al., 2002; Vander Heiden et al., 1997; Waterhouse and Green, 1999). There is also a growing body of evidence that ROS production by mitochondria can operate in the initiation phase of PCD (Carmody and Cotter, 2000; Li et al., 1999). At any moment the level of intracellular ROS seem to determine the fate of the cell: low levels can induce PCD whereas high levels promote necrosis or can lead PCD-committed cells towards necrotic-like destruction (Fleury et al., 2002). Therefore, it is noteworthy that several mitochondrial events (increased metabolism, cytochrome c release, and loss of integrity) take place before cell death execution.

We assessed the potential involvement of an oxidative burst during HL-induced cell death in CAT1AS plants. An oxidative burst is a typical event of incompatible plant–pathogen interactions (Draper, 1997; Lamb and Dixon, 1997; Levine et al., 1994; Low and Merida, 1996), but has also been reported during some abiotic stress conditions (Cazaléet al., 1998; Pellinen et al., 1999; Rao and Davis, 1999). In addition, numerous studies have shown the importance of DPI-dependent oxidases for this oxidative burst (Alvarez et al., 1998; Auh and Murphy, 1995; Doke, 1985; Rao and Davis, 1999; Sagi and Fluhr, 2001; Vera-Estrella et al., 1992). The HL-induced cell death in CAT1AS plants is inhibited by at least two oxidase inhibitors, DPI and Q. In addition, transcript accumulation of a tobacco homolog (NtrbohD) of a mammalian NADPH-oxidase (Lambeth, 2002) was observed in the CAT1AS plants exposed to HL (Figure 6c). The accumulation of these transcripts and the protective effect of the inhibitors suggest a central role for oxidase activity during cell death in our model system. These results are in agreement with the proposed cell death cascades during the HR induced by either ozone or pathogens (Alvarez et al., 1998; Rao et al., 2000; Torres et al., 2002). However, DPI specificity for NADPH oxidase has been questioned recently. To verify that the proximal photorespiratory-dependent H2O2 increase was not also blocked by DPI, we monitored the level of defense protein accumulation following DPI injection. Only a minor reduction in GPX and PR-1a accumulation was observed. These results demonstrate that: (i) the initial change in H2O2 homeostasis is not affected by DPI/Q and not sufficient for the induction and/or execution of cell death; (ii) the DPI/Q-inhibited oxidase-dependent burst is not essential for the accumulation of defense proteins; but (iii) the proximal H2O2 perturbation is enough to trigger an oxidase-dependent burst needed for cell death to develop.

To get an insight into the signaling pathways leading to the observed cell death and to consolidate the requirement/involvement of an oxidase-dependent burst, we used an in planta pharmacological approach. As during an incompatible plant–pathogen interaction, in which the earliest detectable cellular events after pathogen recognition are ion fluxes across the plasma membrane and a burst of oxygen metabolism that produces the oxidative burst (McDowell and Dangl, 2000), changes in ion homeostasis, (de)phosphorylation events, and protein synthesis also affect the development of HL-induced cell death in CAT1AS plants. Therefore, we can conclude that an initial perturbation of H2O2 homeostasis in CAT1AS can stimulate a similar signaling cascade leading to an oxidase-dependent burst necessary for the execution of cell death (Figure 7). This model is also supported by the fact that H2O2 scavengers (such as catalase, ascorbic acid, and glutathione) only inhibit cell death when injected before HL exposure. Thus, once initiated (by a perturbation in H2O2 homeostasis and subsequent oxidative burst), this PCD does not require additional H2O2. Although we did not measure salicylic acid or ethylene levels, both are probably also involved in the induction or execution of oxidative cell death (Dong, 1998; Rao et al., 2000; Van Camp et al., 1998). Levels of both plant growth regulators have indeed been shown to increase under HL in the CAT1AS plants (Chamnongpol et al., 1998). In addition, antagonism and/or synergism with other signaling molecules (such as jasmonic acid, NO, and gibberellin) cannot be excluded.

Figure 7.

Schematic diagram illustrating the proposed role of changes in hydrogen peroxide (H2O2) homeostasis and the two possible signaling pathways leading to cell death and defense activation in CAT1AS plants.


We provide evidence that the oxidative burst plays a more general role than anticipated before in amplifying, relaying, and/or executing oxidative stress-induced cell death during both abiotic and biotic stress conditions. Therefore, it is tempting to speculate that in our system the proximal perturbation in H2O2 homeostasis and the subsequent oxidase-dependent burst and associated cell death are a mirror image of the biphasic burst, classically observed during an incompatible plant–pathogen interaction.

Experimental procedures

Plant growth conditions

Seeds of transgenic CAT1AS and wild-type SR1 Nicotiana tabacum L. plants were germinated and grown as described previously (Chamnongpol et al., 1998). The HL treatments were performed in a Phytotron chamber PLC SCG 970 (Sanyo Gallenkamp, Leicester, UK) with a constant HL of 1000 µmol m−2 sec−1. The LL chamber (Weiss Umwelttechnik GmbH, Reiskirchen-Lindstruth, Germany) had similar relative humidity (70%) and temperature (24/18°C for day/night period) settings with a light intensity of 80 µmol m−2 sec−1. When not otherwise indicated, experiments were performed using the sixth leaf of 6–8-week-old plants.

Time-lapse imaging

Time-lapse analysis was performed using a video camera (4910 CCIR; Cohu, San Diego, CA, USA) fitted with a TV lens (50 mm, 1 : 1.3; Computar, Tokyo, Japan), and mounted in the growth chamber. The camera was connected to a personal computer (Dell, PentiumII) that was fitted with a frame grabber board (LG3 CCIR; Scion Corp., Frederick, MD, USA); the image program was run with ScionImage (WinNT version beta3b; Scion Corp.). Using the built-in macro language, the image program was adapted to obtain one single image every 10 min over a 24-h period. These images were subsequently labeled with date and time and automatically saved to disk. This time series was used to generate a stack, which was then transformed into an avi movie using the publicly available image program imagej (version 1.22d;

H2O2 localization

Hydrogen peroxide was localized in leaves of CAT1AS and wild-type plants using DAB according to an adapted procedure from Orozco-Cardenas and Ryan (1999) and Thordal-Christensen et al. (1997). Briefly, the sixth leaf of CAT1AS and wild-type plants were harvested and immediately vacuum infiltrated for 20 min in 0.5 mg ml−1 of DAB. The leaves were subsequently placed under HL for the periods of the experiment, followed by boiling for 20 min in 80% ethanol. Intensity and pattern of DAB staining were assessed visually. The experiment was repeated six times, each providing similar results.

Light microscopy

Leaf segments (1 cm2) from CAT1AS and wild-type plants were cleared for 2 days in chlorallactophenol (2 : 1 : 1 mixture of chloral hydrate, lactic acid, and phenol; Beeckman and Engler, 1994), and then examined with differential interference contrast optics and a Jenalumar contrast microscope (Jenoptik, Jena, Germany). Large leaf segments were stained with lactophenol–Trypan blue and mounted in chloral hydrate, as described in Koch and Slusarenko (1990), and viewed with a Jenalumar contrast microscope (Jenoptik).

For sectioning, fresh plant material was fixed in formaldehyde (5%), acetic acid (5%), and alcohol (45%) and passed over a graded ethanol series. Infiltration and embedding with Historesin embedding kit (Leica Microsystems, Heidelberg, Germany) were performed according to the manufacturer's instructions. Sections of 5–10 µm were cut on a rotary microtome Minot-Milerotonn 1212 (Leitz, Wetzlar, Germany) with disposable Superlab Knifes (Adamas Instrumenten, Leersum, The Netherlands). Morphological sections of leaves were stained with either 0.05% toluidine blue for 10 min before examination under a light microscope or 10 µg ml−1 4′,6-diamidino-2′-phenylindole (DAPI; Sigma–Aldrich, St. Louis, MO, USA) in water and examined with an Axioskop microscope (Zeiss, Jena, Germany).

Transmission electron microscopy

Leaf samples for electron microscopy were pre-fixed in 2.5% glutaraldehyde in 0.1 sodium phosphate (Na-phosphate) buffer (pH 7.0) overnight at 4°C. Pre-fixed samples were washed three times with 0.1 m Na-phosphate buffer and stored at 4°C. Samples were post-fixed in 1% osmium tetroxide (EMS, Washington, PA, USA), dehydrated in ascending ethanol series, embedded in Epon LX 112 (Ladd Research Industries Inc., Williston, VT, USA), and polymerized. Blocks were sectioned (60 nm) on a Ultracut microtome (Reichert-Jung, Heidelberg, Germany), using a diamond knife (Diatome, Bienne, Switzerland) and mounted on copper slot grids (2 mm × 1 mm). Sections were examined with a transmission electron microscope (Jeol JEM-1200EX; Joel Ltd., Tokyo, Japan) at an accelerating voltage of 60 kV.

Exogenous addition of pharmacological substances

An assay was developed in plants to test whether exogenously added substances could complement CAT1AS deficiency under HL of the selected solution. Using a syringe without a needle, 500 µl of the solution was injected into the intercellular spaces of a leaf segment (4 cm2) (Willekens et al., 1997) and spread through the leaf mesophyll, but was contained by the primary and secondary veins. Plants were subsequently exposed to HL for 24 h. Injection of de-ionized H2O was used as a control. All substances tested were purchased from Sigma–Aldrich or Molecular Probes (Eugene, OR, USA).

Flow cytometric analysis of plant nuclei

Leaf segments (1 cm2) were chopped with a razor blade in 300 µl Galbraith buffer (45 mm MgCl2, 30 mm sodium citrate, 20 mm 3-(N-morpholino)propanesulfonic acid (pH 7), 1% Triton-X100). One microliter of DAPI of a stock of 1 mg ml−1 was added and the mixture was filtered through a 30-µm mesh. The nuclei were analyzed with a BRYTE HS flow cytometer and WynBryte™ software (Bio-Rad, Hercules, CA, USA).

Protein extraction and immunodetection

Total soluble protein extracts were prepared from the sixth leaf of CAT1AS and wild-type plants whose the primary vein had been removed (Chamnongpol et al., 1996). Fifty micrograms of total protein was separated on SDS–PAGE, blotted, and immunodetected as described by Willekens et al. (1997). Primary antibodies used were 1 : 4000 diluted polyclonal anti-acidic PR-1a (Lotan and Fluhr, 1990), 1 : 5000 diluted polyclonal anti-GPX from N. sylvestris (H. Willekens, unpublished data), and 1 : 1000 diluted polyclonal anti-MnSOD from N. plumbaginifolia (Bowler et al., 1991), and 1 : 500 diluted anti-HSP17.6 from Arabidopsis thaliana (Lee et al., 1995).

For determining cytochrome c release, cytosolic fractions were prepared by grinding fresh leaves with a mortar and pestle in ice-cold grinding buffer containing 0.2 m sucrose, 20 mm HEPES, pH 7.5, 20 mm KCl, 1.5 mm MgCl2, 1 mm EDTA, 1 mm EGTA, 1 mm DTT, and a protease inhibitor cocktail (Roche Diagnostics, Brussels, Belgium). Homogenates were centrifuged at 1500 g for 15 min and the supernatant at 16 000 g for 15 min. Protein concentration of the obtained supernatant was determined by using a protein assay (Bio-Rad) with bovine serum albumin as a reference. Five micrograms of the cytosolic fraction was separated on a 12.5% SDS–PAGE, blotted to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Bedford, MA, USA). Membranes were blocked with 1% skimmed milk (Difco, Detroit, MI, USA) and 0.05% Tween-20 in PBS. Primary antibody used was a mouse monoclonal cytochrome c antibody (BD Biosciences Pharmingen, San Diego, CA, USA), diluted 1 : 2500 in blocking buffer. After washing three times in PBS + 0.005% Tween-20, secondary antibody (horseradish peroxidase-conjugated sheep antimouse Ig; Amersham Biosciences, Little Chalfont, UK) was hybridized at a dilution of 1 : 10 000 in blocking buffer. After three washes, labeling was detected by chemiluminescence (Western Lightning™ Chemiluminescence Reagent Plus; Perkin Elmer, Norwalk, CT, USA) and exposed to Hyperfilm™ ECL™ (Amersham Biosciences). Films were scanned and analyzed using the ImageMaster VDS (Amersham Biosciences).

RNA extraction and hybridization

RNA was extracted from leaf tissue with TriZol Reagent according to the manufacturer's instructions (Invitrogen, Gaithersburg, MD, USA). Gel blotting and hybridization were performed as described previously (Chamnongpol et al., 1996; Willekens et al., 1994). The alternative oxidase (aox) probe was a 700-bp EcoRI–XhoI fragment identical to the 3′ part of the aox cDNA described by Vanlerberghe and McIntosh (1994). The manganese superoxide dismutase (mnsod) probe was a 483-bp HpaI–HindIII fragment from tobacco mnsod (Bowler et al., 1989). The rbohD probe was a 415-bp PCR fragment containing part of the NtrbohD gene (this fragment is identical to the corresponding 415-bp fragment of the tobacco NADPH oxidase cDNA present in GenBank under accession number AF506374, nucleotide 977–1391). Equal loading of samples was checked by staining the membranes with methylene blue before hybridization.


We would like to thank the Electron Microscopy Unit of the Institute of Biotechnology (University of Helsinki) for providing laboratory facilities, Martine De Bleeckere, Freya De Winter, and Stijn Morsa for excellent technical assistance, Dr Gerrit Beemster for assistance with the time-lapse video experiments, Dr Wim Van Camp and Dr Wim De Clercq for helpful discussions, Dr Danny Geelen and Dr Frédérique Van Gijsegem for critical reading of the manuscript, and Karel Spruyt, Rebecca Verbanck, and Martine De Cock for help in preparing it. The PR antibodies were a kind gift of R. Fluhr. This work was supported by grants from the Finnish Centre of Excellence Programme (2000–2005) to J.K.; J.F.D. and F.V.B. are indebted to the European Science Foundation for a long-term fellowship, the European Union for an Individual Marie Curie Fellowship, and to the Vlaams Instituut voor de Bevordering van het Wetenschappelijk-Technologisch Onderzoek in de Industrie for a postdoctoral fellowship, respectively. R.P. was supported by the ENSTE Graduate Programme.