Reactive Carbonyl Species Activate Caspase-3-Like Protease to Initiate Programmed Cell Death in Plants

Abstract

Reactive oxygen species (ROS)-triggered programmed cell death (PCD) is a typical plant response to biotic and abiotic stressors. We have recently shown that lipid peroxide-derived reactive carbonyl species (RCS), downstream products of ROS, mediate oxidative signal to initiate PCD. Here we investigated the mechanism by which RCS initiate PCD. Tobacco Bright Yellow-2 cultured cells were treated with acrolein, one of the most potent RCS. Acrolein at 0.2 mM caused PCD in 5 h (i.e. lethal), but at 0.1 mM it did not (sublethal). Specifically, these two doses caused critically different effects on the cells. Both lethal and sublethal doses of acrolein exhausted the cellular glutathione pool in 30 min, while the lethal dose only caused a significant ascorbate decrease and ROS increase in 1–2 h. Prior to such redox changes, we found that acrolein caused significant increases in the activities of caspase-1-like protease (C1LP) and caspase-3-like protease (C3LP), the proteases which trigger PCD. The lethal dose of acrolein increased the C3LP activity 2-fold more than did the sublethal dose. In contrast, C1LP activity increments caused by the two doses were not different. Acrolein and 4-hydroxy-(E)-2-nonenal, another RCS, activated both proteases in a cell-free extract from untreated cells. H₂O₂ at 1 mM added to the cells increased C1LP and C3LP activities and caused PCD, and the RCS scavenger carnosine suppressed their activation and PCD. However, H₂O₂ did not activate the proteases in a cell-free extract. Thus the activation of caspase-like proteases, particularly C3LP, by RCS is an initial biochemical event in oxidative signal-stimulated PCD in plants.

Introduction

The production of reactive oxygen species (ROS), such as superoxide radical (O₂^–), hydrogen peroxide (H₂O₂) and singlet oxygen (¹O₂), is intrinsically associated with photosynthesis, photorespiration and respiration (Foyer and Noctor 2003, Asada 2006). ROS have various biological roles ranging from defense signals to destructive agents, depending on their levels and the manner of their production (Mittler et al. 2011). Plant cells contain abundant antioxidant molecules such as the reduced form of glutathione (GSH) and ascorbic acid (Asc), and an array of ROS-scavenging enzymes such as superoxide dismutase and ascorbate peroxidase (Asada 1999). As a result, the intracellular ROS levels are determined by the balance between their production and scavenging. Under mild stress conditions, the ROS level increases gradually, and O₂^– and H₂O₂ at relatively low concentrations activate a battery of defense genes (Miller et al. 2010). When a plant is exposed to severe and prolonged abiotic stress, its antioxidant capacity is decreased and the ROS levels are further increased, leading to cell death (Mano 2002). In response to attack by pathogens including bacteria, fungi and viruses, the infected cells transiently produce an ‘oxidative burst’ via the activation of the respiratory burst oxidase homologs (Levine et al. 1994, Torres and Dangl 2005). This type of ROS formation leads to death of the original and neighboring cells, and induces defense responses in the surrounding uninfected cells (Torres et al. 2005). These various effects of ROS are collectively designated as oxidative signaling (Mittler et al. 2011).

Programmed cell death (PCD) is one of the typical consequences of oxidative signaling (Van Breusegem and Dat 2006, Petrov et al. 2015). For example, in tobacco (Nicotiana tabacum) Bright Yellow-2 (BY-2) cells, an increase in the O₂^– level is necessary for execution of PCD on salt and sorbitol stress (Monetti et al. 2014). Short-term drought on developing anthers in rice (Oriza sativa) increases the level of H₂O₂ and decreases the level of transcripts of antioxidant enzymes, and thereby leads to PCD (Nguyen et al. 2009). High temperature treatment of tobacco cells increases the H₂O₂ level before PCD occurs (Locato et al. 2008). Hypersensitive response (HR)-like cell death, a typical PCD in tobacco leaves, is also caused by H₂O₂ (Yoda et al. 2003). There is keen interest in the biochemical mechanism by which ROS initiate PCD (Apel and Hirt 2004, de Pinto et al. 2012, Petrov et al. 2015).

To investigate the mechanism of oxidative signal-induced PCD, we previously tested the hypothesis that oxylipin carbonyls are signal mediators. Oxylipin carbonyls collectively designate the aldehydes and ketones produced from lipid peroxides (LOOHs) (Fig. 1). They are produced in abiotic stressed plants and mediate tissue damage (Yin et al. 2010, Mano et al. 2010, Yamauchi et al. 2015) via protein modification (Winger et al. 2007, Mano et al. 2014), and therefore they are strong candidates for oxidative signal mediators (Mano 2012, Farmer and Mueller 2013). We recently examined the roles of short chain (C9 or less) oxylipin carbonyls in PCD in tobacco and Arabidopsis thaliana, and obtained the following results (Biswas and Mano 2015). (i) When tobacco BY-2 cells were treated with H₂O₂ at the level that cause PCD, several species of oxylipin carbonyls such as 4-hydroxy-(E)-2-nonenal (HNE) and acrolein were increased prior to the progression of cell death. (ii) These species of oxylipin carbonyls, when exogenously added, induced PCD in BY-2 cells and roots of tobacco and A. thaliana. (iii) Carbonyl-scavenging compounds suppressed the accumulation of oxylipin carbonyls and PCD in H₂O₂-treated BY-2 cells and A. thaliana roots, without affecting the ROS increase and LOOH accumulation. We further demonstrated that (iv) a transgenic tobacco line that overproduces 2-alkenal reductase, an A. thaliana enzyme which detoxifies α,β-unsaturated carbonyls (Mano et al. 2002), showed a significantly smaller increase in the levels of oxylipin carbonyls and they underwent less PCD in root epidermis after H₂O₂ and salt treatments than the wild type (Biswas and Mano 2015). These results indicate that oxylipin carbonyls, endogenously produced due to an oxidative stimulus, mediated the oxidative signal to induce PCD in plant cells.

Fig. 1

Generation of RCS under oxidative stress. From various species of LOOHs, dozens of oxylipin carbonyls (LOOH-derived aldehydes and ketones) are formed enzymatically or by redox catalysts such as transition metal ions or free radicals. Among them, those carbonyl species comprising the α,β-unsaturated bond, such as acrolein and HNE, are highly electrophilic and designated as reactive carbonyl species (RCS).

The purpose of this study is to elucidate the mechanism by which oxylipin carbonyls trigger PCD. Here we employed an experimental system of PCD induced in BY-2 cells by reactive carbonyl species (RCS). RCS designates the oxylipin carbonyls comprising the α,β-unsaturated bond (Mano et al. 2012), which are more potent electrophiles than simple aldehydes or ketones (Fig. 1). In H₂O₂-stimulated BY-2 cells, two RCS, acrolein and HNE, were increased in early stages, and they showed stronger effects in causing PCD than any other oxylipin carbonyls found in H₂O₂-stimulated cells (Biswas and Mano 2015). In the present study, we found that acrolein caused depletion of the GSH pool in BY-2 cells, then gradually lowered the ascorbate level and enhanced the ROS level. Importantly, caspase-1-like protease (C1LP) and caspase-3-like protease (C3LP), both of which are involved in triggering PCD, were activated rapidly after acrolein addition. These results reveal the biochemical mechanisms of the RCS-mediated initiation of PCD in plants.

Results

Effects of acrolein on cell viability of cultured tobacco BY-2 cells

To determine the concentration limit of acrolein to trigger PCD, we treated cells with acrolein at various concentrations. Cell viability was determined by fluorescein diacetate (FDA), which is membrane-permeable and fluoresces only in living cells due to cleavage by intracellular esterases. When acrolein at 0.1 mM was added to BY-2 cells, the viability was reduced by 10% in 5 h, not significantly different from that obtained for untreated cells. In 0.2 mM acrolein, BY-2 cells started to die in 30 min and about 75% cells were dead in 5 h (Fig. 2A). Cellular protein content is another marker of cell viability. Acrolein at 0.1 mM did not cause a decrease in the protein level even after 5 h incubation, while the agent at 0.2 mM decreased the level in 30 min, and after 5 h incubation >60% protein was lost (Fig. 2B). Analyses of these two parameters indicate that tobacco BY-2 cells survived in 0.1 mM acrolein, but they suffered irreversible cell death in 0.2 mM. This acrolein-induced cell death is typical PCD characterized by DNA fragmentation, cytoplasm retraction and terminal deoxynuleotidyl transferase dUTP nick end labeling (TUNEL)-positive nuclei, as we have previously shown (Biswas and Mano 2015). Considering that acrolein at 0.2 mM was a lethal dose and that at 0.1 mM it was sublethal, we examined the difference in the effects of these two critical doses on BY-2 cells.

Fig. 2

Effects of acrolein on the viability and protein content of tobacco BY-2 cells. A 30 mg aliquot of cells from 7 d culture were subcultured in 50 ml of fresh culture medum, and after 4 d the culture medium was supplemented with acrolein at 0.1 and 0.2 mM. Cells were harvested at the indicated time point, then living cells (A) and protein content (B) were determined as described in the Materials and Methods. Each point represents the mean of three independent experiments and the error bars of the SEM. Different letters represent significantly different data (P < 0.05 on Tukey test).

Acrolein depletes intracellular glutathione and ascorbate

Thiol compounds such as GSH provide the first defense against electrophiles including RCS (Esterbauer et al. 1975, Mano et al. 2009). Because the intercellular concentration of GSH and the glutathione reduction ratio [GSH/(GSH + GSSG)] determine the redox homeostasis of the cell (Foyer and Noctor 2011), the consumption of GSH would affect the cells considerably. In BY-2 cells treated with a lethal level of acrolein, i.e. 0.2 mM, GSH was decreased by 95% in 30 min (Fig. 3A). The lowered level did not recover afterwards. The glutathione reduction ratio also dropped significantly in 30 min, and the lowered level was retained (Fig. 3B). Interestingly, even a sublethal dose of acrolein, i.e. 0.1 mM, caused very similar effects on the GSH content and the reduction level. In 0.1 mM acrolein, 95% of the cellular GSH pool was lost and the reduction level was lowered to a half in 30 min. These lowered levels did not recover even at 5 h, the time at which >80% of the cells were alive (Fig. 2). These results indicate that the rapid drop of the intracellular GSH pool caused by acrolein, although it is drastic and persistent, does not directly initiate PCD.

Fig. 3

Effects of acrolein on the contents and the reduction ratio of GSH and Asc in BY-2 cells. Four day cultured cells were treated with various concentrations of acrolein and then used as in Fig. 1. (A) Contents of GSH, (B) Glutathione reduction ratio (%), (C) reduced ascorbate (Asc) and (D) ascorbate reduction ratio (%). Each point represents the mean of three independent experiments and the error bars of the SEM. Different letters represent significantly different data (P < 0.05 on Tukey test).

Acrolein at 0.05 mM decreased the GSH content to 15% of that of untreated cells in 30 min, and after that the pool was restored to 80% in 5 h (Fig. 3A). The glutathione reduction ratio was dropped to 50%, and then it recovered to 70% (Fig. 3B). A large recovery of the GSH pool is a reflection of active de novo synthesis of GSH in acrolein-treated cells.

We also examined changes in the Asc content (Fig. 3C) and the reduction ratio [Asc/Asc + dehydroascorbate (DHA)] (Fig. 3D). In acrolein at ≥0.1 mM, GSH is extensively consumed (Fig. 3A), and hence the regeneration of Asc from DHA via the DHA reductase reaction (Asada 1999) would be eliminated. Because Asc does not scavenge acrolein efficiently (Mano et al. 2009), the Asc content under such conditions reflects the balance between the Asc consumption by ROS and the de novo synthesis of Asc. Acrolein at the lethal level (0.2 mM) caused a decrease in the Asc content by 30% in 30 min and significantly lowered it in 2 and 5 h (Fig. 3C). The ascorbate reduction ratio was also decreased in 2 h (Fig. 3D). The sublethal level of acrolein (0.1 mM) tended to decrease the Asc level, but the decrease was apparently smaller than that observed for the 0.2 mM acrolein treatment (Fig. 3C). The ascorbate reduction ratio was also decreased by 5% in 0.1 mM acrolein in 30 min, and it was not significantly different from that in 0.2 mM acrolein-treated cells (Fig. 3D). These results suggest that there is a difference between the lethal and sublethal conditions in either the Asc oxidation rate or the Asc synthesis rate, or both.

Acrolein treatment increased the ROS level in BY-2 cells

The loss of GSH and Asc pools in acrolein-treated cells should cause an oxidative load in cells. We thus investigated the effects of acrolein on the ROS level in BY-2 cells with the probe 2′,7′-dihydrodichlorofluoresein-diacetate (H₂DCF-DA). The intracellular oxidation of H₂DCF to the fluorescent dichlorofluorescein (DCF) is a marker of the generation of ROS such as superoxide radical, H₂O₂ and hydroxyl radical. When BY-2 cells were exposed to acrolein at 0.1 mM for 60 min, the level of DCF fluorescence was 1.7-fold higher than in the untreated cells (Fig. 4). In 30 min, acrolein at 0.2 mM resulted in a 1.5-fold higher DCF fluorescence and after 60 min a 2.3-fold higher DCF fluorescence, a significant difference from the untreated control (Fig. 4B). The increase in the ROS was obviously slower than the GSH consumption (Fig. 3A), and appears to reflect the loss of the Asc pool (Fig. 3C). Thus acrolein promoted oxidative stress in an indirect feed-forward manner, but the elevation of ROS level was not so fast and drastic as the loss of the GSH pool.

Fig. 4

Acrolein increases the ROS level in BY-2 cells. (A) Four day cultured cells were incubated with 0.1 mM acrolein for 60 min and with 0.2 mM acrolein for 30 and 60 min. DCF fluorescence was recorded under a fluorescence microscope as described in the Materials and Methods. Typical photographs are shown: untreated cells as control (left), and cells treated with acrolein at 0.1 mM (middle) and 0.2 mM (right). Scale bar = 50 µm. (B) The DCF fluorescence intensity of cells. The fluorescence intensity was integrated per cell with ImageJ software. A total of 200 cells were counted in each treatment. Mean ± SEM of three independent experiments. Differences among treatments were analyzed by Tukey test. P < 0.05.

Activation of caspase-like proteases by acrolein in BY-2 cells

In mammalian cells, the most characterized form of PCD is apoptosis that is executed by highly conserved cysteine-containing aspartate-specific proteases (caspases) (Shi 2002). Caspases are numbered according to their specificity towards artificial peptide substrates. Plants do not have structural homologs of caspases, but they have protease activities with similar substrate specificities to those of animal caspases. In pathogen-induced PCD in plants, the activity of C1LP is increased and the caspase-1 inhibitor suppresses the development of PCD in tobacco (del Pozo and Lam 1998, Hatsugai et al. 2004). The C1LP activity is at least partially attributed to vacuolar processing enzyme (VPE), an ortholog of asparaginyl endopeptidase (legumain). VPEs are localized in the vacuole and responsible for processing vacuolar proteins (Hatsugai et al. 2004). Thus the role of C1LP in plant PCD compares with the executor role of animal caspase-1 in apoptosis (Hatsugai et al. 2015). C3LP activity is involved in HR-induced or abiotic stress-regulated PCD (Fernández et al. 2012, Ye et al. 2013). It can be attributed to the 20S proteasome subunit PBA1 (Hatsugai et al. 2009, Han et al. 2012). Because ROS stimulate plant cells to increase the activities of these proteases (Clarke et al. 2000, Locato et al. 2008), we examined whether or not RCS, as oxidative signal, have similar effects.

The C1LP and C3LP reactions were monitored using the fluorogenic substrates N-acetyl-Tyr-Val-Ala-Asp-α-(4-methyl-coumaryl-7-amide) (Ac-YVAD-AMC) and N-acetyl-Asp-Glu-Val-Asp-AMC (Ac-DEVD-AMC), respectively. The activities were determined by subtraction of the fluorescence intensity after a 1 h enzymatic reaction in the presence of the inhibitor from that in its absence. In the untreated cells, there were constitutive activities of C1LP and C3LP (Fig. 5). These activity levels did not change during the experimental period in untreated cells and hence are irrelevant to PCD. In the cells treated with the lethal level of acrolein (0.2 mM), a significant increase in the C1LP activity was detected in 30 min; it reached 2-fold the constitutive level, and the raised activity was maintained up to 60 min (Fig. 5A). This activation of C1LP is not sufficient to cause PCD because a similar extent of activation was observed even in a sublethal concentration (0.1 mM) of acrolein.

Fig. 5

Activation of C1LP (A) and C3LP (B) in BY-2 cells after exposure to acrolein. Four day cultured cells were treated with 0.1 and 0.2 mM acrolein. Protease inhibitors were added to cell extracts to a final concentration of 0.1 mM, and assay was performed using fluorogenic substrates Ac-YVAD-AMC for C1LP (A) and Ac-DEVD-AMC for C3LP (B) as described in the Materials and Methods. The data are the mean ± SEM. Different letters represent significantly different data (P < 0.05 on Tukey test).

The C3LP activity was also increased in the lethal acrolein concentration (Fig. 5B). In 10 min, its activity reached 3,000 pmol min^–1 (g FW)^–1. The increase from the constitutive level [1,200 pmol min^–1 (g FW)^–1] was 1,800 pmol min^–1 (g FW)^–1. Even in 1 min, a significant increase was detected. In the sublethal acrolein concentration, C3LP was also activated, but the increase from the constitutive level was <750 pmol min^–1 (g FW)^–1. Thus the increase in the C3LP activity under the lethal condition was >2-fold than that under the sublethal condition. It appears that the activation of C3LP to such a level is critical to the initiation of PCD.

H₂O₂-induced activation of C1LP and C3LP in BY-2 cells is suppressed by the RCS scavenger carnosine

The C1LP and C3LP activities would also be increased in H₂O₂-stressed cells because the PCD caused by H₂O₂ is ascribed to the action of RCS (Biswas and Mano 2015). Exposure of the cells to 1 mM H₂O₂, a concentration high enough to induce PCD in BY-2 cells, resulted in significant increases of C3LP activity in 1 min, and then they reached a plateau (Fig. 6). The activity levels reached were 2.7-fold higher for C1LP and 1.9-fold higher for C3LP than the corresponding untreated controls. Thus C1LP and C3LP activities were increased in very early stages of the H₂O₂-induced PCD.

To verify the involvement of RCS, we tested the effects of the RCS-scavenging dipeptide carnosine. Carnosine suppresses the increase in RCS levels in H₂O₂-treated BY-2 cells, without affecting the intracellular ROS level (Biswas and Mano 2015). The concentration used here (1 mM) is high enough to eliminate the H₂O₂-induced PCD (data not shown). As expected, carnosine significantly suppressed the increase in the C1LP activity in H₂O₂-stimulated cells (Fig. 6A). The increase in the C1LP activity in H₂O₂-stimulated BY-2 cells is therefore attributed mostly to the endogenously generated RCS. The increase in the C3LP activity was also suppressed, but only partially (Fig. 6B).

Fig. 6

H₂O₂ increased the C1LP (A) and C3LP (B) activities in BY-2 cells, and carnosine suppressed their increase. Four day cultured cells were treated with either 1 mM H₂O₂ or 1 mM H₂O₂ plus 1 mM carnosine at the indicated time. The data are the mean of three independent experiments ± SEM. Different letters represent significantly different data (P < 0.05 on Tukey test).

Thus both acrolein and H₂O₂ caused increases in the C1LP and C3LP activities in BY-2 cells. The C3LP activity increase over the constitutive activity was 2-fold higher under lethal conditions (0.2 mM acrolein or 1 mM H₂O₂) than its increase under sublethal conditions (0.1 mM acrolein or 1 mM H₂O₂ + 1 mM carnosine). Significant increases in activity were observed in 10 min in both treatments. It appears that the increase in the C3LP activity over such a threshold in a very early stage of stress treatment is required for the subsequent progress of PCD.

RCS directly activate C1LP and C3LP

To investigate the mechanism by which acrolein and H₂O₂ increase the C1LP and C3LP activities, proteins were extracted from BY-2 cells and the effects of these agents on the protease activities in vitro were examined. Acrolein at 0.05 mM was added to the cell-free extract prepared from non-stressed cells. After removing acrolein through gel filtration, the C1LP and C3LP activities were determined (Fig. 7). We found that acrolein activated both proteases up to 2-fold as compared with the basal levels. Not only acrolein, but also HNE, another RCS, activated both C1LP and C3LP to the same extents (Supplementary Fig. S1). These results indicate that RCS acted directly on the C1LP and C3LP proteins, and activated them.

Fig. 7

Activation of C1LP (A) and C3LP (B) by acrolein in a cell-free extract. Protein was extracted from the 4 d cultured cells and desalted against 50 mM sodium acetate, pH 5.5, by gel filtration as desribed in the Materials and Methods (cell-free extract). Acrolein was added to the cell-free extract (1.06 mg protein ml^–1), and C1LP (A) and C3LP (B) activity were measured as described in the Materials and Methods. Data are the mean ± SEM. Different letters represent significantly different data (P < 0.05 on Tukey test).

As shown in Fig. 5, the C3LP activity in BY-2 cells was increased 2-fold in 0.2 mM acrolein. In cell-free extract, acrolein at 0.05 mM activated C3LP, but it did not do so at 0.2 mM (Fig. 8). Acrolein at ≥0.1 mM also inactivated C1LP (data not shown). These different concentration dependencies of the activation of C1LP and C3LP between the in vivo and in vitro conditions were most probably due to the difference in the effective acrolein concentrations between two experimental systems (discussed below).

Fig. 8

Effects of different concentrations of acrolein and H₂O₂ on the activation of C3LP in cell-free extract. Protein was extracted from the 4 d cultured cells. H₂O₂ and acrolein were added at the indicated concentration in 1 ml of protein extract (1.06 mg ml^–1) and desalted as described in the Materials and Methods. The data are the mean ± SEM of three independent experiments.

On the other hand, H₂O₂ was unable to activate C3LP (Fig. 8) and C1LP (data not shown) in a cell-free extract, excluding the possibility that H₂O₂ directly activated these proteases. The activation of C1LP and C3LP in the H₂O₂-treated BY-2 cells (Fig. 6) is therefore attributed solely to the action of RCS that are generated downstream of ROS.

Acrolein enhances the expression of VPE genes in BY-2 cells

VPE genes are up-regulated in the vacuolar cell death process in rice after exposure to H₂O₂ (Deng et al. 2011) and in Al-stressed BY-2 cells (Kariya et al. 2013). We thus investigated whether acrolein induces the expression of VPE genes. Acrolein at 0.2 mM was added to BY-2 cells, and the expression of the four tobacco VPE genes (VPE1a, VPE1b, VPE2 and VPE3) was assessed. We found that all the four VPE genes were up-regulated slightly but insignificantly after 30 min exposure to acrolein (Fig. 9A). After 60 min exposure, the expression of VPE1a and VPE1b was up-regulated by 3- and 4-fold compared with the untreated cells, respectively (Fig. 9B). The expression of VPE2 and VPE3 was also up-regulated, but the enhancement was insignificant (Fig. 9B). These results indicate that acrolein induces the expression of VPE1a and VPE1b genes. Their up-regulation was apparently slower than the increase in C1LP activity, which occurred within 30 min (Fig. 5). The rapid activation of C1LP in acrolein-treated cells is therefore primarily accounted for by the direct biochemical activation of the C1LP protein by acrolein (Fig. 7).

Fig. 9

Acrolein enhanced the expression of VPE genes in BY-2 cells. Four day cultured cells were treated with 0.2 mM acrolein for 30 min (A) and 60 min (B), then total RNA was isolated. The expression level of VPE1a, VPE1b, VPE2 and VPE3 was determined by quantitative RT–PCR as described in the Materials and Methods. Expression of the ACT9 gene was used as an internal standard. The data are the mean ± SEM of three independent experiments. The asterisks indicate a significance difference between untreated and acrolein-treated samples by Student’s t-test, **P < 0.01.

Discussion

Activation of C3LP is the earliest event in acrolein-induced PCD

Based on our recent finding that RCS mediate the oxidative signal to induce PCD in plants (Biswas and Mano 2015), in this study, we aimed at investigating the mechanism by which RCS initiate PCD in tobacco BY-2 cells. Acrolein, one of the most potent RCS, had multiple effects on the cells, i.e. decreasing the glutathione and ascorbate levels, increasing the activity of two caspase-like proteases, C1LP and C3LP, and inducing the expression of VPE1a and VPE1b genes. We consider that the C3LP activity increase was the most critical event for the initiation of PCD, for the following reason. To BY-2 cells, 0.2 mM acrolein was lethal; it caused death to 95% of the population of cells in 5 h. The treatment with 0.1 mM acrolein was sublethal; 80% of the population was alive after 5 h, a viability comparable with that of untreated control cells. At the lethal concentration of acrolein, the following biochemical events were observed in the cells. (i) GSH was exhausted in 30 min, and the lowered level did not recover. (ii) The Asc level was decreased to a half in 2 h. (iii) C1LP activity was increased to double its constitutive level in 30 min. (iv) C3LP activity was increased to double its constitutive level in 1 min, and reached a 2.7-fold increase in 10 min. The sublethal concentration of acrolein also caused events (i)–(iii) to the same extents. Specifically, the GSH consumption and the increase in the C1LP activity are not sufficient conditions for the initiation of PCD. On the other hand, the Asc decrease was obviously smaller, and the C3LP activity increase was a half-maximal in the sublethal acrolein concentration, as compared with those in the cells under PCD. Thus the increase in the C3LP activity was more closely associated, than that in the C1LP activity, with the initiation of PCD. When the lethal level of acrolein was added, the C3LP activity was increased significantly much faster than the Asc pool was decreased. Taking these results together, we conclude that the early increase in the C3LP activity determined the cell fate.

RCS-mediated activation of caspase-like proteases is a mechanism of ROS-induced PCD

We also found that a lethal level of H₂O₂ increased the C1LP and C3LP activities in the cells. Carnosine, an RCS scavenger, suppressed these increases and also PCD. Because carnosine does not scavenge H₂O₂ and other ROS in the cells (Biswas and Mano 2015), it was unlikely that H₂O₂ had a direct effect to activate these proteases. This is supported by the finding that H₂O₂ did not increase the activity of C1LP and C3LP in a cell-free extract (Fig. 8), whereas acrolein and HNE did (Fig. 7; Supplementary Fig. S1). These results indicate that RCS mediated the oxidative signal to cause PCD by activating C1LP and C3LP.

The increase in the C3LP activity is associated with PCD in heat-shocked BY-2 cells (Vacca et al. 2007) and the cadmium-induced PCD in A. thaliana (Ye et al. 2013). Chilling and starvation stress in the microspore of barley led to accumulation of H₂O₂ and an increase in the C3LP activity, resulting to PCD (Rodriguez-Serrano et al. 2011). Bacteria-induced PCD in tobacco leaves was abolished by a C3LP inhibitor (del Pozo and Lam 1998). C3LP activity was increased in bacterial pathogen-induced HR in A. thaliana (Hatsugai et al. 2009). ROS are involved in many cases of such plant PCD caused by biotic and abiotic stressors (Petrov et al. 2015). The megagametophyte death during post-germinative seedling growth of white spruce depends on ROS (He and Kermode 2010) and requires an increase in the C3LP activity (He and Kermode 2003). Because RCS are formed inevitably downstream of ROS (Mano 2012), it is highly likely that they cause the increase in the C3LP activity also in such developmental types of ROS-induced PCD.

The critical involvement of the C1LP activity in many types of PCD has been demonstrated by the inhibition of PCD by a C1LP inhibitor or by the deficiency of the VPE gene (Hatsugai et al. 2015). In the current study, we judged that the increase in the C1LP activity was not sufficient to cause PCD because it was also observed under a sublethal condition (Fig. 5A). Under lethal conditions, VPE1a and VPE1b genes were up-regulated in 60 min, but in 30 min their up-regulation was insignificant (Fig. 9). Although they are not sufficient, the activation of C1LP and up-regulation of VPE genes may be necessary for PCD, in combination with the C3LP activation.

C3LP and C1LP are activated in vitro by RCS

In apoptosis of mammalian cells, caspases are activated from latent forms and exert their protease activities. For example, various stimuli such as oxidative agents and Cyt c released from mitochondria activate caspase-9, which in turn activates caspase-3 via the cleavage of its inactive pre-protein procaspase-3. The active caspase-3 then acts as an effector caspase to degrade various target proteins and facilitate the death program (Shi 2004). Because plant caspase-like proteases are structurally unrelated to mammalian caspases, they would be activated in a different way from mammalian caspases. In this study, we found that RCS rapidly increased the C3LP and C1LP activities in a cell-free extract, indicative of the activation without gene expression/de novo protein synthesis. Such activation of caspase-like proteases by the action of RCS will provide a key to the mechanism of oxidative signal-induced PCD.

For BY-2 cells, the acrolein concentration necessary to increase the C1LP and C3LP activities fully was 0.2 mM, but in the cell-free extract, the optimal concentration to activate these activities was 0.05 mM. This difference in the concentration dependencies between in vivo and in vitro conditions was probably due to the difference in the actual concentration of acrolein, as follows. Because exogenously added acrolein is scavenged by GSH in cells, its intracellular concentration should be much lower than that outside in the medium. A large recovery of the GSH pool in the cells treated with 0.05 mM acrolein (Fig. 2A) indicates a continuous GSH supplementation via de novo synthesis; the lowered GSH level was restored after all acrolein was scavenged. In other words, the intracellular acrolein concentration is determined by the balance between the supply of acrolein from outside and the de novo synthesis of GSH. In 0.1 mM acrolein treatment, continuous de novo synthesis of GSH probably contributed to suppressing the intracellular acrolein concentration to a level just lower than that required for triggering PCD. In 0.2 mM acrolein, the GSH biosynthesis flux was probably overwhelmed by the consumption flux. As a result, the intracellular acrolein concentration would reach the level to activate C1LP and C3LP.

Signaling roles of carbonyls

Recently, evidence is accumulating that various carbonyl compounds act as signals in plants. For example, acrolein can inhibit stomatal opening at physiological concentrations (Islam et al. 2015). Because ROS are generated in guard cells by the ABA stimulus (Pei et al. 2000), it is likely that acrolein and other RCS are produced and inhibit stomatal opening, thereby acting as a signal for stomatal closure. RCS are also involved in the heat shock response (Yamauchi et al. 2015). Interestingly, the species which mediate the heat shock response are limited to RCS of carbon chain length 4–9. Acrolein, the C3 RCS, is unable to act as a heat shock signal. This implies that, in distinct carbonyl signaling responses, putative carbonyl receptors have specificities to distinguish various carbonyl species.

Besides RCS, which are derived from lipid peroxides, dicarbonyl species such as methylglyoxal and glyoxal are formed as inevitable by-products in sugar metabolism including the Calvin cycle (Takagi et al. 2014). Methylglyoxal, when produced in chloroplasts, can mediate the photoreduction of O₂ to form superoxide radical, and thereby enhances oxidative stress (Saito et al. 2011). Thus, in chloroplasts, the production of methylglyoxal, an upstream event of ROS production, may be connected via ROS to the formation of RCS.

There are more kinds of carbonyl species that have potential signaling roles in plants (Mueller and Berger 2009, Farmer and Mueller 2013). It appears that each carbonyl species has a distinct metabolism, i.e. formation and scavenging, and physiological actions. A larger number of experimental facts in broader aspects in plant physiology should be accumulated to build a consistent overview of the signaling action of carbonyls.

Conclusion

We here demonstrate that RCS, downstream products of ROS, can directly activate C1LP and C3LP, and thereby initiate PCD in plant cells, and that they affect the redox homeostasis of cells greatly by consuming GSH. These results provide a specific biochemical explanation for ‘plant oxidative injury’, the importance of which is widely accepted but the process remains to be clearly described.

Materials and Methods

Culture of cells

Tobacco BY-2 (Nicotiana tabacum L. cv. Bright Yellow-2) cell suspension was cultured in Murashige and Skoog medium supplemented with sucrose (30 g l ^{− 1}), myo-inositol (100 mg l ^{− 1}), KH₂PO₄ (200 mg l ^{− 1}), thiamine HCl (0.5 mg l ^{− 1}) and 2,4-dichlorophenoxyacetic acid (0.2 mg l ^{− 1}), pH 5.6, with continuous rotation at 120 r.p.m. in darkness at 25°C. The cells were subcultured every 7 d; approximately 0.5 ml of cell suspension was transferred to fresh medium (50 ml). Cells in the exponential growth phase, which is established on the fourth day, were used for the experiment. Cells were collected by filtration and washed once with distilled water for analyses.

Cell viability assay and protein determination

FDA staining is specific to determine viable cells because only viable cells are able to cleave FDA to form fluorescein and fluoresce with excitation at 485 nm and emission at 515 nm. Dead cells do not fluoresce. Cells were incubated in a solution of FDA (1 µg ml^–1) for 5 min and then observed under a fluorescence microscope (Leica AF6000, Wetzlar) under white and fluorescent light. Protein content was determined with Protein Assay CBB solution (Nacalai Tesque) with bovine serum albumin as the standard. Briefly, about 0.4 g of cells were harvested and homogenized with extraction buffer (50 mM KH₂PO₄, 1 mM EDTA, 1× Protease Inhibitor Cocktail; pH 7.4). Cell extract was centrifuged at 1,000 × g for 5 min at 4°C and 0.5 ml of supernatant was passed through the equilibrated PD MiniTrap G-25 column (GE Healthcare) to remove small molecules.

Extraction and analysis of glutathione

The glutathione pool was assayed according to de Pinto et al. (1999) with a slight modification. The cells were harvested and washed with distilled water. After chilling with liquid nitrogen, cells (approximately 0.35 g) were ground with a mortar and pestle and 2 vols. of cold 5% sulfosalicylic acid were added. The homogenate was centrifuged at 20,000 × g for 15 min at 4°C, and the supernatant (cell extract) was collected for analysis. A 0.1 ml aliquot of cell extract was mixed with 0.9 ml of 0.1 M HEPES-KOH, pH 7.4, containing 5 mM EDTA (neutralized cell extract), and divided into two. One fraction was used for determining the total glutathione [GSH + GSSG] as GSH. The other fraction was used for determining GSSG. The concentration of GSH was determined in 1 ml of reaction mixture containing 0.1 M HEPES-KOH, pH 7.4, 5 mM EDTA, 10 mM 5,5′-dithiobis-2-nitrobenzoic acid, 0.5 U of glutathione reductase and 0.3 ml of neutralized cell extract. After addition of 0.2 mM NADPH to the reaction mixture, the increased rate in A₄₁₂ was recorded for 1 min. A standard curve for GSH in the range of 0–0.1 mM was prepared. For GSSG determination, 20 µl of 2-vinylpyridine was added to the neutralized cell extract and mixed well until an emulsion was formed, to mask GSH. After removal of the residual 2-vinylpyridine by centrifugation, the GSH concentration was determined as above. The GSH content in the cell was determined as the difference between the amount of total glutathione and that of GSSG.

Extraction and analysis of ascorbate

Asc and DHA were measured according to Kampfenkel et al. (1995) with minor modifications. Briefly, total ascorbate (Asc + DHA) was determined after reduction of DHA to Asc with dithiothreitol (DTT), and the concentration of DHA was estimated from the difference between total ascorbate and Asc. Cell extract was prepared in the same way as for GSH determination above. The 1 ml reaction mixture for total ascorbate contained a 0.1 ml aliquot of cell extract, 0.25 ml of 150 mM phosphate, pH 7.4, containing 5 mM EDTA, and 0.05 ml of 10 mM DTT. After incubation for 10 min at room temperature, 0.05 ml of 0.5% N-ethylmaleimide was added to remove excess DTT. Asc was determined in the same reaction mixture, except that 0.1 ml of H₂O was added rather than DTT and N-ethylmaleimide. Color was developed in both reaction mixtures after addition of the following reagents: 0.2 ml of 10% trichloroacetic acid, 0.2 ml of 44% ortho-phosphoric acid, 0.2 ml of 4% α,α-dipyridyl in 70% ethanol and 0.3% (w/v) FeCl₃. After thorough mixing, the mixture was incubated at 40°C for 40 min and then the A₅₂₅ was determined. A standard curve was developed based on Asc in the range of 0–250 µM.

Determination of C1LP and C3LP activities

Preparation of cell extract and assay of C1LP and C3LP activities were performed as described by Hatsugai et al. (2004). Briefly, 4 d cultured BY-2 cells were harvested, frozen in liquid nitrogen and homogenized with a mortar and pestle in 50 mM sodium acetate, pH 5.5, containing 50 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride and 0.1 mM E64-d (a thiol protease inhibitor). After centrifugation at 14,000 × g for 30 min at 4°C, the supernatant was collected (cell extract).

C1LP and C3LP activities were measured with synthetic tetrapeptide fluorogenic substrates Ac-YVAD-AMC and Ac-DEVD-AMC (both from the Peptide Institute), respectively. Fifty µl of cell extract (or cell-free extract containing 50 µg of protein for Figs. 7 and 8) was mixed with 0.2 ml of reaction mixture containing 20 mM Na-acetate, pH 5.5, 0.1 M DTT, 0.1 mM EDTA and 1 mM phenylmethylsulfonyl fluoride, and incubated at 37°C for 1 h (Hatsugai et al. 2004). Fluorescence of AMC (excitation 380 nm; emission 445 nm) was determined with a spectrofluorometer (FP-8300, JASCO). C1LP inhibitor and C3LP inhibitor (Ac-YVAD-CHO and Ac-DEVD-CHO, respectively; Peptide Institute) were added at 0.1 mM to the cell extract 1 h before addition of the substrate, and incubated at 37°C. The fluorescence intensity difference between the absence and the presence of inhibitor was considered as the activity of the protease. For the detection of in vitro activation of C1LP and C3LP, acrolein or H₂O₂ was added to the cell extract and at each time point passed through a PD MiniTrap G-25 column (GE Healthcare) to remove small molecules. The eluted extracts were added to 0.2 ml of reaction mixture as described above. A standard curve was prepared with AMC (Peptide Institute) in the range of 0–200 nM.

ROS detection with H₂DCF-DA

BY-2 cells were incubated for 1 h in 0.2 mM acrolein or 0.2 mM acrolein plus 1 mM carnosine, collected by a brief centrifugation and washed with distilled water. They were then incubated with 20 µM H₂DCF-DA (Wako Pure Chemical) in phosphate-buffered saline (PBS) at 37°C for 30 min in darkness and washed twice with PBS. The fluorescence was monitored under a fluorescence microscope (Leica LED3000; Leica Microsystems IR GmbH) with excitation at 488 nm and emission at 530 nm.

RNA extraction and real-time RT–PCR

Total RNA was isolated from the 4 d cultured acrolein-treated and untreated BY-2 cells. Cells were harvested and homogenized with a mortar and pestle in liquid nitrogen and RNA was prepared using an RNasey Plant Mini Kit (Qiagen). Contaminating genomic DNA was digested with RNase-free DNase (Ambion RNA, Life Technologies). First-strand cDNA was synthesized from 2 µg of total RNA with a ReverTra Ace qPCR kit (Toyobo). Real-time quantitative reverse transcription–PCR was carried out using Light Cycler (Roche) using Thunderbird SYBR qPCR mix (Toyobo). The gene-specific primers and the PCR conditions are given in Supplementry Table S1. ACT9 was used to normalize the amount of total transcripts in each sample (Kariya et al. 2013).

Funding

This was work supported by the Japan Society for the Promotion of Science [Grant-In-Aid for Scientific Research (C) No. 26440149].

Disclosures

The authors have no conflicts of interest to declare.

Abbreviations

Abbreviations
 
• AMC

  α-(4-methyl-coumaryl-7-amide)

 
• Asc

  ascorbic acid

 
• BY-2

  Bright Yellow-2

 
• C1LP

  capase-1-like protease

 
• C3LP

  caspase-3-like protease

 
• DCF

  dichlorofluorescein

 
• DHA

  dehydroascorbate

 
• DTT

  dithiothreitol

 
• FDA

  fluorescein diacetate

 
• GSH

  reduced glutathione

 
• HNE

  4-hydroxy-(E)-2-nonenal

 
• HR

  hypersensitive response

 
• LOOH

  lipid peroxide, PCD, programmed cell death

 
• RCS

  reactive carbonyl species

 
• ROS

  reactive oxygen species

 
• VPE

  vacuolar processing enzyme