Evidence of mitochondrial involvement in the transduction of signals required for the induction of genes associated with pathogen attack and senescence


For correspondence (fax +1 517 353 9168; e-mail mcintos1@pilot.msu.edu).


Using the mRNA differential display technique, seven cDNAs have been isolated that are rapidly induced when cultured tobacco (Nicotiana tabacum) cells are treated with the mitochondrial electron transport inhibitor antimycin A (AA). Interestingly, six of the cDNAs show distinct similarity to genes known to be induced by processes that involve programmed cell death (PCD), such as senescence and pathogen attack. All of the cDNAs as well as Aox1, a gene encoding the alternative oxidase, were found to also be strongly induced by H2O2 and salicylic acid (SA). AA, H2O2 and SA treatment of tobacco cells caused a rapid rise in intracellular ROS accumulation that, when prevented by antioxidant treatment, resulted in inhibition of gene induction. Besides AA, both H2O2 and SA were found to disrupt normal mitochondrial function resulting in decreased rates of electron transport and a lowering of cellular ATP levels. Furthermore, the pre-treatment of tobacco cells with bongkrekic acid, a known inhibitor of the mitochondrial permeability transition pore in animal cells, was found to completely block gene induction when AA, H2O2 or SA were subsequently added. These findings suggest that the mitochondrion may serve an important role in conveying intracellular stress signals to the nucleus, leading to alterations in gene expression.


Besides accidental cell death, termed necrosis, all multicellular organisms seem to have in place genetically based programmes that when activated lead to a process of controlled cell execution called programmed cell death (PCD). PCD requires the expression of numerous genes and utilization of specific intracellular pathways and seems to be critical for normal development and the maintenance of tissue homeostasis in all multicellular organisms (Steller, 1995). The most widely studied PCD, animal cell apoptosis, is characterized by a distinct set of morphological features including cell shrinkage, cytoplasmic membrane blebbing, nuclei lobing and DNA fragmentation (Desagher and Martinou, 2000). While it is not known whether plant and animal PCD is initiated and processed through similar mechanisms, changes analogous to animal apoptosis have been reported in plants undergoing various developmental processes including anther development (Wang et al., 1999), xylogenesis (Obara et al., 2001) and both leaf (Yen and Yang, 1998) and petal (Xu and Hanson, 2000) senescence, and pathogen attack (Tronchet et al., 2001). Interestingly, PCD can also be triggered by exposure of plants to abiotic stresses such as salt stress (Katsuhara and Kawasaki, 1996) and heat shock (Balk et al., 1999). Based upon studies of the hypersensitive response (HR), pathogen attack is also considered a form of PCD. The HR occurs at the site of attempted attack by an avirulent pathogen and is characterized by the rapid induction of local cell death around the site of infection (Lamb and Dixon, 1997).

A considerable amount of literature has implicated altered mitochondrial function as a requirement for the initiation of PCD in animal systems (Desagher and Martinou, 2000). Disruption of mitochondrial electron transport has been shown to be an early feature of animal PCD (Susin et al., 1998). One consequence of mitochondrial dysfunction is the increased formation of mitochondrial reactive oxygen species (ROS: H2O2, O2, OH) which is seen as a major trigger in precipitating cell death (Green and Reed, 1998). A defining hallmark of PCD in animal systems is the release of cytochrome c and additional apoptotic factors from the mitochondrion into the cytosol. Cytochrome c acts in the cytosol by recruiting a group of proteases termed caspases, which serve to degrade critical cell components (Susin et al., 1998).

It remains unclear to what extent the mitochondrion is involved in PCD in plant cells. Cytochrome c release has been shown to occur during heat-induced PCD in cucumber (Balk et al., 1999) and menadione-induced cell death in tobacco protoplasts (Sun et al., 1999). However, cytochrome c release was not required for pollination-induced petal senescence in petunia (Xu and Hanson, 2000). A clear role for the mitochondrion in plant PCD is implicit in the finding by Zhao et al. (1999) that animal cytochrome c can activate caspase-like components in isolated carrot cytosol resulting in apoptosis of mouse nuclei. Involvement of caspases in plant PCD has been further strengthened by the finding that peptide inhibitors of animal caspases could abolish HR cell death in tobacco (Del Pozo and Lam, 1998).

A unique feature of plant mitochondria is that they possess a bifurcated electron transport chain. Besides the ubiquitous cytochrome chain found in all eukaryotes, plants have an additional pathway of electron flow, the alternative pathway (Vanlerberghe and McIntosh, 1997). This pathway is composed of a single homodimeric protein, the alternative oxidase (AOX), which transfers electrons from the ubiquinone pool directly to oxygen. While AOX abundance and alternative pathway activity are low in unstressed plants, both increase when plants are subjected to suboptimal conditions such as chilling (Vanlerberghe and McIntosh, 1997), and pathogen attack (Simons et al., 1999). While the biological function of AOX is not fully understood, it may serve to maintain high rates of respiratory carbon metabolism and electron transport under conditions where the normal cytochrome chain is restricted (Vanlerberghe and McIntosh, 1997). In this regard, we have recently shown using intact plant cells that restriction of the cytochrome chain increases the formation of potentially damaging ROS (Maxwell et al., 1999), which can be at least partially offset by high levels of AOX (Maxwell et al., 1999).

The nuclear gene encoding AOX in tobacco (Aox1) has previously been shown to be rapidly induced in cultured cells by specifically inhibiting the cytochrome pathway of electron transport through the use of antimycin A (AA) (Vanlerberghe and McIntosh, 1994). Using the differential display technique we have isolated seven additional cDNAs which were strongly upregulated following AA treatment of cultured tobacco cells. Interestingly, a number of the cDNAs show distinct similarity to genes known to be induced by processes that involve programmed cell death (PCD), such as senescence and pathogen attack. We present data from a series of experiments using these cDNAs that support a hypothesis that the functional state of the mitochondrion may play a crucial role in conveying intracellular stress signals to the nucleus leading to altered gene expression.


Cloning and identification of AA induced genes

Inhibition of mitochondrial respiration has been shown in many organisms to trigger alterations in the expression of specific nuclear genes. Treatment of human fibroblasts with AA results in the induction of cytochrome c1and cytochrome b transcripts (Suzuki et al., 1998). In Saccharomyces cerevisiae, Kwast et al. (1999) have shown that inhibition of the respiratory chain results in the induction of a subset of hypoxic genes. To date in plants, a single member of the gene family encoding AOX (Tobacco Aox1; Arabidopsis Aox1a) is the only nuclear gene whose expression has been shown to be rapidly modulated by specific inhibition of mitochondrial electron transport (Saisho et al., 1997; Vanlerberghe and McIntosh, 1994).

Through the use of the differential display technique we have cloned seven partial cDNAs (Table 1) that were found to be induced within 4 h after treating cultured tobacco cells with AA (5 µm). We refer to these genes as NtAI for NicotianatabacumAntimycin Induced. These genes were confirmed to be differentially expressed by using the cloned partial cDNAs as probes for RNA slot blot analysis (data not shown). Although four differential display profiles were found showing transcripts that were down-regulated by antimycin A treatment, these cDNAs were found not to be differentially expressed when subsequent slot blot analysis was performed (data not shown).

Table 1.  Sequence similarities identified for the differential display clones
Clone nameAccession no.aSize range (bp)Best homologySignificanceb
  • a

    Sequences deposited in EMBL Nucleotide Sequence Database

  • b

    b Probability that such a match would occur merely by chance as given by BLAST

NtAI1AJ310472388Lycopersicon esculentum 1-aminocyclopropane-1-carboxylate
oxidase homolog (Acc. no. 10967)
3.0 × 10−28
NtAI2AJ310473275N. tabacum GTX1 glutathione S-transferase (Acc. No. Q03662)5.0 × 10−24
NtAI3AJ310474208N. tabacum Sar8.2 (Acc. No. U64811)2.0 × 1064c
NtAI4AJ310475432N. tabacum Cysteine proteinase precursor. (Acc. No. T03941)1.0 × 10−35
NtAI5AJ310476431A. thaliana putative lipase (ACO13354)2.0 × 10−13



N. tabacum salicylate-induced glucosyltransferase
(Horvath and Chua, 1996)
A. thaliana scarecrow-like 11 (Acc. No. ad 24410)
4.5 × 10−29

3.0 × 10−42

Homology searches using the BLASTN or BLASTP algorithm (Altschul et al., 1990) indicated that these genes all shared significant homology to previously described classes of genes (Table 1). NtAI1 was found to show high homology to the ethylene generating enzyme, 1-aminocyclo-propane-1-carboxylate (ACC) oxidase. ACC oxidase has previously been shown to be strongly induced during pathogen attack, in response to SA treatment (Kim et al., 1998), and during senescence (Jones and Woodson, 1999). NtAI2 was found to share significant similarity to the glutathione S-transferase (GST) family of proteins. GSTs catalyze the conjugation of the tripeptide glutathione to a variety of hydrophobic, electrophilic and usually cytotoxic substrates (Marrs, 1996). GSTs are strongly induced in plants subjected to a wide array of stresses including heavy metals, pathogen attack and senescence (Marrs, 1996). NtAI3 shows clear similarity to the SAR8.2 group of genes, which is strongly induced by pathogen attack, SA and ethylene treatment (Guo et al., 2000). The function of the products of this group of genes remains unknown. The sequence of NtAI4 is very similar to cysteine proteinase-type genes. This group of genes has been shown to be transcriptionally upregulated under a wide range of stress conditions including senescence (Kardailsky and Brewin, 1996). Their induction seems to occur whenever rapid changes in cell metabolism are required and they function presumably to reutilize intracellular resources stored as hydrolysable proteins. NtAI5 shows a high degree of similarity to two Arabidopsis genes directly submitted to GenBank that encode putative lipases. Lipases are hydrolytic enzymes that break down triacylglycerols into fatty acids and glycerol. Interestingly, two Arabidopsis genes encoding probable lipases have been recently cloned, PAD4 (Jirage et al., 1999) and EDS1 (Falk et al., 1999). Both genes have been shown to be required for multiple defence responses after pathogen infection (Falk et al., 1999; Jirage et al., 1999). NtAI6, is a partial cDNA, which corresponds to a glucosyltransferase that has been previously characterized and is rapidly induced by SA (Horvath and Chua, 1996) and pathogen attack (Chong et al., 1999). NtAI7 shows specific similarity to members of the newly delineated GRAS gene family in Arabidopsis (Pysh et al., 1999). Members of the GRAS family have been found to be important in a diverse array of cell functions including phytochrome A signal transduction (Bolle et al., 2000).

Kinetics of differential display transcript accumulation

To initially characterize the NtAI genes, the kinetics of mRNA accumulation following AA treatment was determined. As shown in Figure 1, five out of the seven transcripts showed increased mRNA accumulation starting 90 min after AA addition to tobacco suspension cells. Between 90 min and 4 h, mRNA abundance increased rapidly before declining slowly for the remainder of the 24-h experiment. Interestingly, this general pattern of induction was identical to that seen for Aox1 (Figure 1). The almost identical kinetics of induction are striking, and suggest that for this group of genes, and perhaps others, there exists a common transcription regulatory apparatus that is sensitive to changes in mitochondrial function. An exception to this dominant kinetic profile was found for two genes, NtAI6 and NtAI7, which showed rapid mRNA accumulation starting within 30 min of AA treatment. Since NtAI7 shows similarity to a group of Arabidopsis transcriptional regulators described by Pysh et al., 1999, we are currently isolating a full-length cDNA corresponding to NtAI7 and are undertaking experiments to determine if it may be involved in regulating the expression of Aox1 and the NtAI-genes.

Figure 1.

Gene induction following inhibition of mitochondrial electron transport in cultured tobacco cells.

Time course of mRNA accumulation for Aox1 and the differential display genes (NtAI genes) after addition of AA (5 µm).

Induction of differential display transcripts by H2O2 and SA

Besides AA, other short-term treatments known to increase AOX abundance include H2O2 (Vanlerberghe and McIntosh, 1996) and SA (Rhoads and McIntosh, 1993). SA has been shown to accumulate during the HR and is required for pathogen-resistant (PR) gene expression and the establishment of systemic acquired resistance (SAR) (Guo et al., 2000). To further characterize the NtAI genes, we examined their levels of expression in tobacco cells treated for 4 h with either 5 mm H2O2 or 1 mm SA compared with AA (5 µm). As shown in Figure 2(a), all of the NtAI-genes were found to be strongly induced by both H2O2 and SA to a similar extent as AA. This pattern was identical to that seen with Aox1 (Figure 2a).

Figure 2.

Link between stress-induced gene expression and intracellular ROS formation.

(a) Tobacco cells were either left untreated or incubated in the presence of AA (5 µm), H2O2 (5 mm) or SA (1 mm) for 4 h prior to RNA isolation.

(b) Stress treatments increase intracellular ROS abundance. ROS was measured after treatment of tobacco suspension cells with either AA (5 µm), H2O2 (5 mm), SA (1 mm), monofluoroacetate (20 mm), or l-cysteine (1 mm) for 4 h, or after a 12-h cold treatment at 13°C.

Data represent means ± SD for 3–6 separate experiments.

Treatments that induce Aox1 increase ROS

It has been widely suggested that a cellular response which underlies many stress conditions, and which may alter gene expression, is an increase in intracellular reactive oxygen levels (Jabs, 1999). In both yeast (Kwast et al., 1999) and mammalian cells (Suzuki et al., 1998) inhibition of mitochondrial electron transport by AA results in increased ROS formation. We have recently shown that this is the case with plant cells (Maxwell et al., 1999). SA treatment has been shown to cause a rapid increase in intracellular ROS levels, specifically H2O2 (Chen et al., 1993). The SA-dependent increase in ROS is important for the activation of certain pathogen defence responses (Wendehenne et al., 1997), with evidence suggesting that this occurs via the binding of SA to catalase (Chen et al., 1993) as well as other iron-containing enzymes such as aconitase (Ruffer et al., 1995).

Aox1 induction has been shown to occur when plants are subjected to a range of stress conditions. Besides AA, H2O2 and SA, Aox1 is also strongly induced by chilling (Vanlerberghe and McIntosh, 1992a), monofluoroacetate, which is a potent inhibitor of aconitase, and l-cysteine (Vanlerberghe and McIntosh, 1996), which has been shown to induce oxidative stress (Reiter et al., 1998). To ascertain whether increased oxidative stress may be an underlying common feature of these treatments, intracellular ROS levels were estimated using dichlorofluorescein diacetate (Maxwell et al., 1999). As shown in Figure 2(b), treatment of cultured tobacco cells with either H2O2 (5 mm), AA (5 µm), SA (1 mm), monofluoroacetate (20 mm), or l-cysteine (1 mm) for 4 h, or a 12-h cold treatment at 13°C, caused a dramatic increase in cellular ROS levels compared with the levels seen in untreated cells.

Antioxidants inhibit ROS accumulation and gene induction

The results presented in Figure 2 suggest that treatments that cause the induction of Aox1 and the NtAI genes may act through raising intracellular ROS levels. We investigated this possibility further by examining whether gene induction could be blocked by pre-treating cells with commonly used antioxidants prior to the addition of AA, H2O2 or SA. In plants, antioxidants have been used to implicate H2O2 in the plant hypersensitive response (Levine et al., 1994), ROS involvement in the SA-dependent induction of PR-1 expression (Wendehenne et al., 1997), as well as H2O2 involvement in ozone-induced MAP kinase signalling (Samuel et al., 2000). We tested two common antioxidants, N-acetylcysteine (25 mm) (Wendehenne et al., 1997) and flavone (1 mm) (Minagawa et al., 1992), for their efficacy in scavenging intracellular ROS produced as a consequence of AA, H2O2, and SA treatment, as well as their ability to inhibit the subsequent induction of Aox1 and the NtAI genes. As shown in Figure 3(a), both of these antioxidants were found to be effective at decreasing the levels of intracellular ROS accumulation. Furthermore, both NAC and flavone could inhibit, to a significant extent, the induction of both Aox1 and the NtAI genes when added to cells 45 min prior to the addition of AA, SA or H2O2 (Figure 3b). As seen in Figure 3(b) the extent of inhibition seemed to vary slightly among the different genes and the different treatments.

Figure 3.

Antioxidants lower intracellular ROS levels and inhibit gene induction.

(a) Effects of antioxidant addition on AA, H2O2, and SA-dependent accumulation of intracellular ROS in tobacco suspension cells. ROS levels were measured 4 h after AA (5 µm), H2O2 (5 mm) and SA (1 mm) addition with and without preincubation for 45 min with N-acetylcysteine (25 mm) or flavone (1 mm). Data represent means ± SD for three experiments.

(b) Effect of the antioxidant treatment described above on the AA-, H2O2-and SA-dependent expression of Aox1 and the NtAI genes.

Kinetics of ROS accumulation compared with Aox1 induction

To further characterize the link between intracellular ROS accumulation and gene induction, we monitored the increase in intracellular ROS formation concurrently with Aox1 expression as a function of time after the addition of either AA, H2O2 or SA to tobacco cells. As shown in Figure 4(a), H2O2 treatment caused a very rapid increase in intracellular ROS accumulation, while AA-dependent ROS accumulation was significantly slower. Interestingly, while AA showed the slowest induction of intracellular ROS accumulation, it showed the fastest induction of Aox1 (Figure 4b), with appreciable transcript accumulation detected 60 min faster than either the H2O2 or SA treatment. It may be argued that the delay in gene induction by H2O2 is possibly due to cell damage resulting from such a large oxidative insult. However, this hypothesis is not supported by previous work with both soybean (Levine et al., 1994) and Arabidopsis cell cultures (Desikan et al., 2000), where 2–20 mm H2O2 was found to cause very rapid changes in gene expression within 30 min of treatment. What the data presented in Figure 4 may indicate is that while ROS formation is a common feature of stresses that induce Aox1, an increase in intracellular ROS alone may not be sufficient to activate the transduction pathway leading to changes in gene expression. An alternative hypothesis that could explain the differential kinetics of Aox1 induction is that the gene-inducing signal, regardless of whether AA, SA or H2O2 is used, is derived from the mitochondrion itself. These results show why both SA and H2O2 induce the NtAI genes and Aox1 to a similar extent as inhibiting mitochondrial function with AA (Figure 2a).

Figure 4.

Relationship between the kinetics of intracellular ROS accumulation and kinetics of Aox1 gene induction.

Representative time-course of intracellular ROS accumulation (a) and Aox1 transcript induction (b) following addition of AA (5 µm), H2O2 (5 mm) or SA (1 mm) to tobacco suspension cells.

AA, H2O2 and SA inhibit respiration and lower ATP levels

Because both SA and H2O2 were as effective as AA in inducing the NtAI genes, we investigated whether inhibition of mitochondrial function may be a common consequence of all three stress treatments. Rates of electron transport, including cytochrome and alternative pathway capacities, were measured as a function of time following the addition of AA, H2O2 and SA to intact tobacco cells. As shown in Figure 5, each of these treatments caused a rapid decrease in overall respiration rate, measured starting 30 min after their addition. The decrease in total respiration was found to be 75%, 70% and 47% for AA, H2O2 and SA, respectively. After the initial drop, overall respiration rates recovered over the 8-h duration of the experiment. In all cases the initial inhibition of the total respiration rate was followed by an increase in alternative pathway capacity. The rapid recovery of total respiration in cells treated with AA was due solely to induction of alternative pathway respiration (Figure 5). At the end of the 8-h experiment, cells incubated with AA had respiration rates that were higher than the initial control values, a finding that has been shown previously (Vanlerberghe and McIntosh, 1992b). By comparison, cells treated with SA showed levels of respiration that recovered to near control values after 8 h. H2O2 treatment had the most deleterious effect on mitochondrial respiration, with overall rates, recovering to only 65% of the initial value after 8 h. As an independent measure of respiratory function we measured cellular ATP levels following AA, SA, or H2O2 addition. As shown in Figure 6, the initial decreases in total respiration rate were mirrored by a decrease in total ATP level, which stabilized over the remainder of the experiment.

Figure 5.

AA, H2O2 and SA inhibit mitochondrial function. Measurement of total respiration and cytochrome (CP) and alternative pathway (AP) capacities following addition of AA (5 µm), H2O2 (5 mm) or SA (1 mm) to tobacco suspension cells. Units for total respiration and respiratory capacity are nmol O2 mg DW−1 min ñ1. Data represent means ± SD of 3–5 separate experiments.

Figure 6.

AA, H2O2 and SA treatment lower cellular ATP levels. As a function of time following addition of AA (5 µm), H2O2 (5 mm) or SA (1 mm), aliquots of cells were removed for the determination of total cellular ATP levels. Data represent means ± SD of three independent experiments.

Our finding that SA inhibits mitochondrial function is in agreement with a recent study by Xie and Chen (1999), who showed that treatment of cultured tobacco cells with 500 µm SA caused a 70% decrease in respiration over a 4-h period and an 85% drop in ATP levels over 8 h. This is significantly greater inhibition, using half the SA concentration, than was observed in the current study. While the basis of this discrepancy is not known, we did observe that the inhibitory effect of SA on respiration was much greater using day-2 cells (as were used by Xie and Chen) when compared with cells 3 days after subculture, which we routinely employed.

Inhibitor of animal PT pore blocks gene induction

An event now recognized to be important in animal apoptosis is the opening of the mitochondrial permeability transition (PT) pore. The PT pore, which is formed by association of the inner membrane adenine nucleotide transporter and different outer membrane proteins, such as porin, can be triggered to open by oxidative stress and changes in Ca2+ (Jones, 2000). Current research suggests that the opening of the PT pore is a likely mechanism whereby cytochrome c and perhaps other PCD-triggering factors are released into the cytosol (Desagher and Martinou, 2000). To further investigate the possibility of mitochondrial involvement in the induction of the NtAI genes upon AA, SA and H2O2 treatment we examined whether the opening of the mitochondrial PT pore may play a role in relaying changes in the functional state of the mitochondrion to the nucleus. For this experiment we employed bongkrekic acid, which is known to bind to the PT pore in animal systems and prevent its opening (Narita et al., 1998). Bongkrekic acid (50 µm) has recently been shown to prevent H2O2-induced apoptosis in cultured human cells, thus implicating mitochondrial involvement in the subsequent H2O2-dependent cell death (Dumont et al., 1999). As shown in Figure 7, treatment of tobacco cells with 50 µm bongkrekic acid for 4 h did not change the expression of any of the genes when simply added to control cells. However, preincubating cells with bongkrekic acid for 45 min prior to the addition of AA, H2O2 or SA was found to completely abolish the induction of both Aox1 and the NtAI genes.

Figure 7.

Treatment of cells with inhibitor of mitochondrial permeability transition blocks gene induction.

Cells were pre-treated with bongkrekic acid (BA) (50 µm) for 30 min prior to the addition of AA (5 µm), H2O2 (5 mm) or SA (1 mm) to tobacco suspension cells. RNA was isolated after 4 h.


A number of genes that were previously identified to be induced in plants subjected to conditions that trigger PCD, such as pathogen attack and senescence, were found in this study to be induced in cells treated with AA, a specific inhibitor of mitochondrial function. Unlike other stresses, the specificity of AA for inhibiting the cytochrome respiratory chain has made it ideal for investigating how changes in mitochondrial function lead to alterations in nuclear gene expression (Poyon and McEwen, 1996). Besides AA, the NtAI genes are equally induced by H2O2 and SA, two oxidative stress-causing agents not readily associated with inhibiting mitochondrial function. Based on the data presented in Figures 2 and 3 one could readily conclude that ROS, produced upon inhibition of mitochondrial electron transport (AA), arise within the cell through inhibition of certain enzymes (SA), or (H2O2) added directly acts as an intracellular signalling molecule which activates the expression of both Aox1 and the NtAI genes. While potentially harmful ROS formation is exacerbated when plants are subjected to unfavourable environmental conditions, there is growing evidence that ROS formation may also regulate numerous signal transduction pathways in both plants and animals (Jabs, 1999). It has recently been shown in Arabidopsis that H2O2 activates a specific MAPK kinase, ANP1, which in turn initiates a phosphorylation cascade involved in stress signalling (Kovtun et al., 2000).

Increasing ROS levels play a key role in the current study. Treatments that trigger Aox1 and NtAI-gene expression not only result in a rapid increase in intracellular ROS abundance, but both gene expression and ROS accumulation can be effectively inhibited by pre-treating cells with antioxidants. As previously discussed (Wendehenne et al., 1997), results of experiments using antioxidants, in both animal and plant systems, have led to inconsistent results and need to be assessed with caution. Since chemical antioxidants are reducing agents, their interaction with cellular constituents and their effects on cellular redox status are poorly understood. In spite of this concern, our finding that the AA-dependent induction of Aox1 and the NtAI genes could be effectively inhibited through the use of antioxidants supports previous work. With both the yeast Hansenula anomala (Minagawa et al., 1992), and cultured human cells (Suzuki et al., 1998) AA-dependent induction of nuclear gene expression could be blocked by using either flavone or NAC, respectively. Taken together with these findings, our data support a role for mitochondrial ROS formation as a component required for stress-induced mitochondria-to-nucleus signalling.

While the findings presented in Figures 2 and 3 suggest a role for increasing intracellular ROS formation in triggering alterations in gene expression, such a scenario is hard to reconcile with the data from additional experiments. First, while H2O2 treatment caused the fastest increase in intracellular ROS accumulation, it showed the slowest induction of Aox1 (Figure 4). AA, which causes increased ROS formation only indirectly by inhibiting mitochondrial electron transport, showed the fastest induction of Aox1 (Figure 4). Second, while AA is the only treatment we used that specifically inhibits mitochondrial function, both H2O2 and SA caused significant inhibition in the rate of mitochondrial electron transport as well as causing a reduction in cellular ATP levels (Figures 5 and 6). Thirdly, the AA-, H2O2- and SA-dependent induction of Aox1 and the NtAI genes can be effectively blocked by pre-treating the cells with bongkrekic acid, an inhibitor of the mitochondrial PT pore (Figure 7). Based upon these findings we propose the following scenario: both H2O2 and SA treatment lead to a rapid rise in intracellular ROS levels, which cause a non-specific disruption of normal mitochondrial function. This disruption mimics, to some extent, the specific inhibition of mitochondrial electron transport caused by AA (Figure 5). We suggest that, for the genes discussed in this study, H2O2 and SA do not alter gene expression directly, but rather both of these stresses act through disruption of normal mitochondrial function. Our findings support the notion that the mitochondrion may act as a central depot within the cell where diverse stress stimuli are integrated (Lam et al., 1999). Once interpreted by the mitochondria, these stresses may be relayed elsewhere in the cell, including the nucleus, to bring about alterations in gene expression.

While our data using bongkrekic acid suggest that the components required to form the animal PT pore may also exist in plant mitochondria, there is no clear biochemical evidence to support this. Opening of the PT pore is a possible mechanism whereby cytochrome c and other apoptotic factors are released into the cytosol, triggering PCD (Desagher and Martinou, 2000). It is clear that cytochrome c release from mitochondria also occurs in plant cells undergoing cell death (Balk et al., 1999; Stein and Hansen, 1999; Sun et al., 1999). While the mechanism of release in plants has not been specifically identified, it seems unlikely that the movement of cytochrome c into the cytosol during PCD would occur through a fundamentally different mechanism in plants than it does in animals. Suggestive evidence for the existence of the PT pore in plant mitochondria has come from work on Bax, a death-promoting member of the Bcl-2 family in animals (Lacomme and Santa Cruz, 1999). Bax has been shown in animal cells to interact with the PT pore, specifically the adenine nucleotide translocator (ANT). Furthermore, in a number of animal systems, Bax has been shown to trigger cytochrome c release from the mitochondrion that can be effectively blocked using bongkrekic acid (Narita et al., 1998). Work with Bax in plants includes the finding that overexpression of mouse Bax in tobacco plants triggers a form of cell death which closely resembles the HR (Lacomme and Santa Cruz, 1999). In the same study, through the use of green fluorescence protein-tagged constructs, BAX was specifically localized to the mitochondrion in tobacco. These data suggest that BAX may operate in plant cells in a similar fashion as in animal cells by interacting with the PT pore.

The opening of the mitochondrial PT pore results in the collapse of the electrochemical transmembrane potential and uncouples oxidative phosphorylation, resulting in a drop in ATP formation leading to cell death (Desagher and Martinou, 2000). Besides occupying this so-called high-conductance state, there is an emerging view that the PT pore can open transiently, allowing for the movement of a number of small molecules in and out of the mitochondrion and, more importantly, enabling normal signalling events to occur between the mitochondrion and the cytosol (Miller, 1998). Our findings using bongkrekic acid suggest that transient opening of the PT pore may be necessary for changes in mitochondrial function to be communicated to the nucleus.

It is becoming clear that the mitochondrion, besides being the powerhouse of the eukaryotic cell, also serves as an active participant in a number of intracellular signalling pathways, including playing a role in animal PCD. Additional work is necessary to elucidate the molecular mechanism of PCD in plants and the role the mitochondrion may play in the process. We have presented data that suggest the mitochondrion may serve as an intermediary in intracellular stress signalling, by both interpreting stress signals as well as initiating an appropriate response. Such a response may include subtle adjustments in cell function, alterations in nuclear gene expression, as well more severe responses such as triggering PCD.

Experimental procedures

Growth conditions and chemicals

Cultured tobacco (Nicotiana tabacum cv Petit Havana SR1) cells were grown heterotrophically as previously described (Maxwell et al., 1999). All experiments were conducted using cells 3 days after a weekly subculture. Chemicals were purchased from Sigma–Aldrich (St Louis, MO, USA), except N-acetylcysteine (Calbiochem, San Diego, CA, USA), and H2O2, which was provided from JT Bakter (Phillipsburg, NJ, USA) as a 30% solution. Salicylic acid was made up as a 10-mm stock in growth medium (pH 5.8–6.0).

Differential display and RNA gel blot analysis

RNA was isolated from tobacco cells using a hot phenol method followed by LiCl precipitation (Maxwell et al., 1999). Prior to RNA isolation cells were either incubated for 4 h with AA (2 µm) or left untreated. To remove contaminating DNA, isolated RNA was treated with RNase-free DNase (Roche Inc., Nutley, NJ, USA). Differential display was carried out essentially as described by Liang et al. (1994) using the Genhunter RNAimage system (GenHunter Corporation, Nashville, TN, USA) with no modifications to the provided protocol. PCR was conducted in the presence of α-[33P] dATP (2000 Ci mmol−1), with labelled PCR products separated on a 6% sequencing gel. Tentatively identified differentially expressed cDNAís were excised from the gel and reamplified prior to being cloned into the pGEM-T vector (Promega Corporation, Madison, WI, USA). To minimize the cloning of aberrant bands differential display was conducted in duplicate. Cloned products were sequenced at the Michigan State University DNA Sequencing Facility using an Applied Biosystems Prism 377 DNA sequencer. RNA for gel blot analysis was isolated as described above, separated on formaldehyde containing gels and transferred to Hybond N membrane (Amersham Pharmacia Biotech Inc., Piscataway, NJ, USA) as previously described (Maxwell et al., 1999). Radiolabelled DNA probes were made using the random prime labelling kit (Roche Inc.). Blots were routinely stripped with boiling 0.1% SDS and re-probed.

Measurement of reactive oxygen species

Intracellular production of ROS was measured using 2′,7′-dichlorofluorescein diacetate (H2DCF-DA) (Maxwell et al., 1999). Dichlorofluorescein (DCF) fluorescence was measured using a Hitachi F2000 fluorescence spectrophotometer, except for data presented in Figure 4(a), which was collected using a Molecular Devices Spectra Max Gemini microplate spectrofluorometer. With both instruments, excitation and emission wavelengths were set at 488 nm and 520 nm, respectively.

Measurement of whole-cell respiration and cellular ATP

Respiration was measured using intact cells in culture medium using a Rank Brothers O2 electrode cuvette at 28°C. Cytochrome pathway (CP) capacity was taken to be that portion of the O2 consumption that was inhibited by 1 mm KCN in the presence of 2 mm SHAM, and alternative pathway (AP) capacity was taken to be the portion of O2 consumption inhibited by 2 mm SHAM in the presence of 1 mm KCN (Vanlerberghe and McIntosh, 1992b). Residual respiration (in the presence of both SHAM and KCN) was subtracted from all measurements. ATP was measured using a luciferin-luciferase system (Sigma). Cell extracts for ATP measurements were prepared as previously described (Xie and Chen, 1999).

EMBL accession numbers: AJ310472; AJ310473; AJ310474; AJ310475; AJ310476; AJ310477; AJ310478.