Cadmium is suspected to exert its toxic action on cells through oxidative damage. However, the transition metal is unable to directly generate reactive oxygen species (ROS) via redox reactions with molecular oxygen in a biological environment. Here, we show that bright yellow-2 (BY-2) tobacco cells exposed to millimolar concentrations of CdCl2 developed cell death within 2–3 h. The death process was preceded by two successive waves of ROS differing in their nature and subcellular localization. Firstly, these consisted in the transient NADPH oxidase-dependent accumulation of H2O2 followed by the accumulation of O2−. in mitochondria. A third wave of ROS consisting in fatty acid hydroperoxide accumulation was concomitant with cell death. Accumulation of H2O2 was preceded by an increase in cytosolic free calcium concentration originating from internal pools that was essential to activate the NADPH oxidase. The cell line gp3, impaired in NADPH oxidase activity, and that was unable to accumulate H2O2 in response to Cd2+, was nevertheless poisoned by the metal. Therefore, this first wave of ROS was not sufficient to trigger all the cadmium-dependent deleterious effects. However, we show that the accumulation of O2−. of mitochondrial origin and membrane peroxidation are key players in Cd2+-induced cell death.
Cadmium is a pollutant and its presence in the environment is essentially due to anthropogenic activities (di Toppi & Gabbrielli 1999). Because of its long biological half life, Cd2+, which belongs to the group of non-essential transition metals, is highly toxic. In plants, Cd2+ induces genotoxicity and programmed cell death (Fojtová & Kovarík 2000; Fojtováet al. 2002; Behboodi & Samadi 2004). In cells, the soft metal cation Cd2+ is preferentially bound by thiol-containing biomolecules such as proteins or glutathione (GSH) and phytochelatin (PCs) peptides (Grill, Winnacker & Zenk 1991). Like other divalent ions, Cd2+ is also clearly able to displace essential metal cations from proteins generally leading to inhibition of enzyme activities (Nocentini 1987). In addition to this, because of similar ionic radii and identical charges, Cd2+ can affect Ca2+ homeostasis. In radish for example, Cd2+ competes with Ca2+ for specific ionic binding sites to inhibit calmodulin-dependent phosphodiesterase activity (Rivetta, Negrini & Cocucci 1997). Cd2+ can enter cells through Ca2+ channels (Misra et al. 2002; Perfus-Barbeoch et al. 2002) or through specific transmembrane transporters (Clemens et al. 1998).
A great deal of research has established the ability of cadmium to induce reactive oxygen species (ROS) production in plants (Piqueras et al. 1999; Sandalio et al. 2001; Shah et al. 2001; Boominathan & Doran 2003; Olmos et al. 2003; Romero-Puertas et al. 2004; for reviews, see di Toppi & Gabbrielli (1999), Schützendübel & Polle (2002). However, because of its redox potential (−820 mV), Cd2+ is unable to directly participate in biological redox reactions with oxygen. The inhibition of cadmium-induced H2O2 generation by diphenylene iodonium (DPI), a flavin-containing oxidase inhibitor, has yet suggested the involvement of a NADPH oxidase-like enzyme in the ROS production (Olmos et al. 2003; Romero-Puertas et al. 2004). However, this result is still debatable in that Ranieri et al. (2005) have reported that, in wheat plants, the production of H2O2 upon Cd2+ treatment was insensitive to DPI and was inhibited by KCN, suggesting the activation of a peroxidase-based system. In parallel, the inhibition of antioxidative enzymes is often described and mentioned as a potential mechanism leading to cadmium-mediated increase in cellular ROS levels (Sandalio et al. 2001; Romero-Puertas et al. 2002; Schützendübel & Polle 2002). The mechanisms by which cadmium induces decrease in antioxidative defences are multiple and are mostly related to the ability of Cd2+ to be sequestered by thiol-containing biomolecules (Grill et al. 1991). The consequences of exposure of plant cells to cadmium are the drastic consumption of GSH for both sequestration of the metal and for synthesis of PCs (Zenk 1996). The active transport of both GSH- and PC-metal complexes to the vacuoles (Rea 1999; Clemens, Palmgren & Kramer 2002) make up an efficient protective measure that can subsequently result in the limitation of the GSH level necessary for upholding the redox balance of the cell leading then to ROS accumulation (Mittler 2002). In addition to this, it is also reported that GSH reductase, a key enzyme dedicated to re-reduction of glutathione disulfide (GSSG) at the expense of NADPH, contains highly a conserved disulfide bridge that may undergo cleavage by heavy metals such as cadmium, resulting in inhibition of this activity (Schützendübel & Polle 2002). Mitochondria have been also described as potential targets of cadmium (Koizumi et al. 1994; Tang & Shaikh 2001; Wang et al. 2004), and alteration of the electron transfer chain (ETC) by Cd2+ associated with ROS production has been reported, in mammalian cells, by Wang et al. (2004). Mitochondria had also been reported as the primary target of Cd2+ in photosynthetic and non-photosynthetic strains of Euglena gracilis (Watanabe et al. 2003). In plants, such direct evidence is still missing, but it has recently been described that the gene encoding AtCOX17, an Arabidopsis homologue of the yeast copper chaperone COX17, has been up-regulated in response to stress situations known to interfere with mitochondrial function and with Cd2+ treatment (Balandin & Castresana 2002). Consequently, as in animal cells, the cadmium-mediated ROS production can also originate from mitochondria.
Because of a very strong capacity of the plants to concentrate cadmium, their tissues, especially the roots, generally have to cope with millimolar concentrations of Cd2+, even when concentrations of the heavy metal do not exceed a few micromolars in the hydroponic solutions (Rivetta et al. 1997; Sandalio et al. 2001; Shah et al. 2001; Perfus-Barbeoch et al. 2002). This work aimed at investigating the molecular mechanisms at the origin of a ROS accumulation in plant cells challenged with cadmium salts. The causal link between the oxidative stress and the metal toxicity has also been extensively examined.
MATERIALS AND METHODS
The following chemicals were prepared in distilled water: stock solutions of horseradish peroxidase, 1 U·µL−1(Sigma–Aldrich, Saint Quentin Fallavier, France); bovine erythrocyte superoxide dismutase (SOD), 50 U·µL−1 (Sigma–Aldrich); Aspergilus Niger glucose oxidase, 0.1 U·µL−1 (Sigma–Aldrich); bovine liver catalase (CAT), 1000 U·µL−1 (Sigma–Aldrich); bovine serum albumin (BSA) fraction V, 80 mg·mL−1(Sigma–Aldrich); D-cis diltiazem, 50 mM (Sigma–Aldrich); L-cis diltiazem, 50 mM (Biomol/TEBU-BIO s.a., Le Perray en Yvelines, France); desferrioxamine mesylate (DFO), 1 M (Sigma–Aldrich); ethylene glycol-bis (2-aminoethylether)-N,N,N′,N′-tetraacetic acid (EGTA), 0.5 M (Sigma–Aldrich); and CdCl2, 1 M (Sigma–Aldrich). The following were prepared in dimethyl sulfoxide (DMSO): propidium iodide, 5 mM (Sigma–Aldrich); SYTO-13 green, 5 mM (Molecular Probes/Interchim, Montluçon, France); hydroethidine (HE), 50 mM (Fluoprobes, Interchim, Montluçon, France); 10-acetyl-3,7-dihydroxyphenoxazine, Amplex Red reagent, 50 mM (Molecular Probes/Interchim); 3-tert-butyl-4-hydroxyanisole (BHA), 250 mM (Fluka, Sigma–Aldrich); W-7, 2 mM (Calbiochem, VWR International, Fontenay sous Bois, France); staurosporine, 1 mM (Calbiochem); 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ), 25 mM (Sigma–Aldrich); rotenone, 50 mM (Sigma–Aldrich); myxothiazol (Myxo), 5 mM (Sigma–Aldrich) and antimycin A (AA), 5 mM (Sigma–Aldrich). Thenoyltrifluoroacetone (TTFA) (100 mM, Sigma–Aldrich) was dissolved in ethanol. Actinomycin D (Act D) (5 mg·mL−1, Sigma–Aldrich) and cycloheximide (CHX) (5 mg·mL−1, Sigma–Aldrich) were dissolved in 80% (v/v) ethanol/H2O. Verapamil (25 mM, Sigma–Aldrich) and coelenterazine (5 mM, Uptima-Interchim, Montluçon, France) were prepared in absolute ethanol and in methanol, respectively. U73122 and U73343 (Sigma–Aldrich) were firstly dissolved in chloroform, aliquoted, and the solutions were evaporated under nitrogen. Stock solutions (10 mM) were made by dissolving compounds in DMSO just before use. Mn(III)tetrakis(4-benzoic acid)porphyrin chloride (MnTBAP) 100 mM (Calbiochem) was dissolved in 0.1 M KOH. Final concentrations of each compound are given in the figure legend. Cryptogein (Cry), prepared according to Bonnet et al. (1996) was kindly provided by Dr M. Ponchet (INRA, Antibes, France).
Plant material and growth conditions
Wild-type tobacco (Nicotiana tabacum) bright yellow-2 (BY-2) cells (J0) and the transgenic lines transformed with either an antisense construct of NtrbohD gene (gp3) or with the empty vector pKy were previously described (Simon-Plas, Elmayan & Blein 2002). Tobacco cell suspensions were grown in Chandler’s medium (Chandler, Tandeau de Marsac & Kouchkovsky 1972) and maintained by dilution, every 2 weeks, of an inoculum into fresh medium at a final concentration of 4.5% packed cell volume (PCV). Cells were agitated on a rotary shaker (125 rpm) at 25 °C with a constant irradiance of 90 µmol·m−2·s−1white light provided by cool white fluorescent tubes (Sylvania, Raunheim, Germany).
The cell suspension cultures of BY-2 tobacco expressing apoaequorin in the cytosol were grown under agitation (130 rpm) at 25 °C in darkness in Linsmaier and Skoog (LS) medium supplemented with 30 g·L−1of sucrose and 1 mg·mL−1 2,4-dichlorophenylacetic acid, pH 5.8. (Pauly et al. 2001). Subculturing was done by inoculating the fresh medium with a 14-day-old culture at a final concentration of 2% PCV.
Before each treatment, the cells were harvested, filtered and rinsed with distilled water before being resuspended, at a concentration of 25 mg·mL−1, into 2 mM methanesulphonic acid (MES)/KOH buffer pH 6.5 containing 30 g·L−1 sucrose. The resuspended cells were then pre-treated for 30 min with the appropriate pharmacological compound before being exposed, under agitation at 25 °C, to CdCl2.
Cytosolic calcium measurement
Two to 10-day-old transgenic BY-2 cells were collected by centifugation (100 g, 5 min), washed with the suspension buffer MES/KOH (10 mM, pH 5.8) containing 30 g·L−1 sucrose and resuspended into the same buffer at a 20% PCV. In vivo reconstitution of aequorin was performed by shaking (150 rpm, 24 °C) in the dark for at least 3 h, an appropriate volume of washed cells with 2.5 µM of coelenterazine. Aequorin light emission was measured using a digital luminometer (Berthold Sirius/Fisher Bioblock, Illkirch, France). Reconstituted cell culture aliquots (100 µL) were carefully transferred into a luminometer tube and luminescence counts were recorded as relative light units per second at 1 s intervals. Twenty seconds after the beginning of the recordings, when the baseline luminescence was reached, 300 µL of cell suspension buffer, complemented or not with cadmium and/or calcium at the indicated concentrations, were added. To calibrate the luminescence counts obtained, at the end of each experiment, the remaining reconstituted aequorin was discharged by adding 300 µL of lysis buffer containing 100 mM CaCl2, 10% ethanol (v/v) and 2% Nonidet P-40 (v/v), and the resulting luminescence increase was monitored until recordings returned to basal levels. Luminescence data transformation into cytosolic Ca2+ concentration was performed using the following equation: pCa = 0.332588 (−log k) + 5.5593, (Knight, Trewavas & Knight 1996), where k is equal to the luminescence intensity per second divided by the total remaining luminescence counts. For the pharmacological experiments, effectors or control solvents were added 10 min before the recordings started.
Extracellular H2O2 concentration was determined using a spectrophotometric method based on a peroxidase-coupled assay. The stoichiometric peroxidase-dependent oxidation of Amplex Red (Molecular Probes/Interchim) by H2O2 leads to the formation of resorufin, the concentration of which can be determined at 570 nm using the molecular extinction coefficient of 67 500 M−1·cm−1. In practice, at different time intervals after treatment, 0.5 mL of cell suspension, at a density of 25 mg fresh weight (FW)·mL−1, was diluted into 0.5 mL of 2 mM MES/KOH buffer, pH 6.5, containing 30 g·L−1 sucrose, 100 µM Amplex Red (Molecular Probes/Interchim) and 2 U·mL−1 of horseradish peroxidase. After 30 s, the cells were discarded by filtration through one layer of a miracloth (Calbiochem) filter paper, and absorbance at 570 nm of the remaining solution was determined.
Superoxide accumulation in cells was estimated by fluorescence microscopy using the probe HE. Briefly, the cells were incubated for 10 min in 50 µM HE before observation under a Leitz dialux 20 microscope (Leica Microsystems SAS, Rueil-Malmaison, France) equipped with a Hg lamp and a standard fluorescein isothiocyanate (FITC) filter set. Images were processed and analysed using a Leica DC100 camera interfaced with the IM50 v1.20 software (Leica Microsystems SAS). The nuclear bright yellow fluorescence was indicative of the ROS accumulation. Quantification of O2.− producing cells, expressed in percent, was estimated by counting at least 30 clusters of cells. Clusters composed of all HE-positive cells (totally stained) scored 100%; those displaying a mixture of HE-positive and HE-negative cells (partially stained) scored 50%, and finally, those showing all HE-negative cells (unstained) scored 0%. To check whether differences in staining reflected actual variations in ROS production rather than modification of the ability of cells to load HE, once incubated with HE, non-treated cells were placed for a few seconds under an ultraviolet (UV) filter (diaminidophenylindol (DAPI), 330–385 nm as the excitation wavelength). After UV irradiation, the cells, immediately observed under fluorescence microscope using the FITC filter, displayed a similar intense red fluorescence. Consequently, it was assumed that, firstly, the UV treatment had led to the photochemical oxidation of HE into ethidium-like compounds and that, secondly, BY-2 cells, unrelated to their physiological status, loaded equivalent and non-limiting quantity of HE.
Cell death determination
Cell death was determined by fluorescence microscopy as previously described using the two fluorescent DNA-intercalating agents, propidium iodide and SYTO-13 green (Molecular Probes/Interchim). After incubation for 10 min in 25 µM propidium iodide and in 0.125 µM SYTO-13 green (Molecular Probes/Interchim), dead cells displayed red fluorescent nuclei, whereas those of the living ones fluoresced in green. For each treatment and time point, at least 400 cells were counted, and cell death was expressed in percent.
Lipid peroxidation analysis
Free and esterified hydroperoxy fatty acids quantified by high-performance liquid chromatography (HPLC) as free hydroxy fatty acids (HFA) along with the enantiomeric composition of each isomer, were determined according to the previously described methods (Rustérucci et al. 1999; Montillet et al. 2004).
Transcript quantification by reverse transcription (RT)-PCR
Quantifications of NtrbohD (accession number AJ309006) transcripts were carried out by duplex relative RT-PCR methods, using the ubiquitously expressed marker gene encoding glyceraldehyde-3-phosphate dehydrogenase (GAPDH) to normalize transcript amounts. The degenerate primers GAPDHf (5′-ATTARGATCGGAATYAA CGG-3′) and GAPDHr (5′-GTAACCCCAYTCRTTGT CRTA-3′) were used to amplify a GAPDH fragment of 950 base pair (bp) between the conserved motifs I(R/K)IGING and YDNEWGY, respectively. A NtrbohD specific fragment of 468 bp was amplified using NtrbohD-5′ (5′-CGGCAAAAAGAGTGCGAGAT-3′) and NtrbohD-3′ (5′-TGGCACGAGGAAGCAAAC-3′) primers. Total RNA was extracted using TRIzol reagent (Invitrogen Life Technologies, Cergy Pontoise, France) according to the manufacturer protocol. RNA (1 µg) was treated with DNAse I (Amp grade, Invitrogen Life Technologies) and was then used to synthesize first-strand cDNAs with a first-strand cDNA synthesis kit (Amersham Biosciences, Orsay, France) and an oligo-(dT)18 as primer. Duplex PCR was performed using 5 µL of cDNA diluted 1/50. The PCR thermocycle profile was 94 °C 5 min (initial denaturation), followed by 34 repetitions of 94 °C for 30 s with GAPDHf/GAPDHr and NtrbohD-5′/NtrbohD-3′ primers, 51 °C for 30 s and 72 °C for 1 min, and one cycle at 65 °C for 10 min (for final elongation). PCR products were separated on 1.5% agarose gels. The fluorescence signal of the amplification products was integrated with ImageQuant 5.2 software (Molecular Dynamics, GE Healthcare Europe GmbH, Saclay, France) to quantify the relative amount of transcript.
Graphical and statistical analyses
Data were analyzed with SigmaPlot (SYSTAT Software GmbH, Erkrath, Germany). Some complementary analyses were performed with Scilab (http://www.scilab.org/), a free and open software similar to Matlab. Data were tested with simple classical models, and curves were fitted with a minimum number of adjustable parameters. The models are described in the figure legends.
Cd2+ induces an accumulation of H2O2 through post-transcriptional activation of the plasma membrane NADPH oxidase
When treated with CdCl2, wild-type BY-2 (J0) cells (Fig. 1a) or control cells (pKy) transfected with empty vector (not shown) responded in a transient accumulation of H2O2, which peaked at 30 min at a value of about 15 µM. Under these conditions of exposure to Cd2+, cell death became significant beyond 2 h. The dose-dependent production of peroxide proceeded maximally at around 1 mM of CdCl2 (Fig. 1b). In contrast, gp3, a BY-2 cell line transformed with an antisense construct of NtrbohD, the respiratory burst plasma membrane NADPH oxidase-encoding gene (Simon-Plas et al. 2002), did not accumulate H2O2 in response to cadmium (Fig. 1c). Consequently, the cadmium-induced oxidative burst (OB) was clearly due to the activity of NADPH oxidase. Cryptogein, a protein of fungal origin, was also demonstrated to elicit such an OB (Simon-Plas et al. 1997). In this case, the transcriptional activation of Ntrboh D was involved (Simon-Plas et al. 2002). Here we show that contrary to CdCl2 (3 mM), cryptogein (50 nM) induced a long-lasting accumulation of H2O2, which was maximum at around 3 h (Fig. 2a) and was inhibited by protein synthesis inhibitors, Act D (66%) and CHX (84%) (Fig. 2b). Moreover, as shown by RT-PCR experiments (Fig. 2c), treatment with CdCl2 did not elicit transcription of the oxidase-encoding gene. Therefore, Cd2+ triggered the accumulation of H2O2 by activation of the constitutive form of the plasma membrane NADPH oxidase.
Ca2+ calmodulin and kinase activity are required for Cd2+-dependent activation of Ntrboh D
A pharmacological approach was carried out to interfere with the signalling cascade acting upstream from the oxidase activation (Fig. 3). Staurosporine, a broad spectrum kinase inhibitor, abolished the CdCl2-induced H2O2 production indicating that one or several steps of protein phosphorylation were required for the oxidase activation. Moreover, addition of stoichiometric concentration of CaCl2 to the incubation medium led to the severe inhibition of the OB. Among different calcium channel antagonists tested, LaCl3 was the only one active. Verapamil, a putative calcium channel blocker, known to reverse Cd2+-induced stomatal closure in Arabidopsis (Perfus-Barbeoch et al. 2002), had no effect on the activation of the NADPH oxidase upon Cd2+ treatment in BY2 cells. Different inhibitors that potentially interfere with the intracellular calcium mobilization were also active on CdCl2-mediated H2O2 accumulation. Thus, the phospholipase C inhibitor, neomycin sulfate, and the phosphatidylinositol-specific phospholipase C inhibitor, U73122, respectively blocked and reduced by more than 80% the CdCl2 action indicating the likely involvement of inositol-3-phosphate (IP3)-stimulated calcium channels. Additionally, the adenosine diphosphate ribosyl (ADPR) cyclase and guanylate cyclase inhibitors, nicotinamide and ODQ, which reduced about half of the H2O2 production, revealed the probable involvement of cADP-ribose-activated calcium channels in the Ca2+ release from intracellular stores necessary for the Cd2+ signalling. Finally, the potent calmodulin inhibitor, W7, almost abolished the oxidase activation. Collectively, this data provides evidence that Ca2+ calmodulin and protein phosphorylation were crucial actors of the Cd2+ signalling cascade required for the heavy metal-induced OB.
Cd2+ elicits a rapid increase in cytoplasmic Ca2+ concentration
To further investigate the role of Ca2+ in the Cd2+-mediated response, BY-2 cells harbouring the Ca2+ reporter protein, apoaequorin, in the cytoplasm were used to monitor potential variations in free cytosolic calcium concentration ([Ca2+]cyt) in response to CdCl2. After a lag phase of 1–3 min, Cd2+ elicited an increase in [Ca2+]cyt that lasted for a few more minutes (Fig. 4a). The response was dose dependent (Fig. 4b) and reached a maximum for 1 mM CdCl2 that led to a [Ca2+]cyt of about 0.15 µM. The response was completely blocked by adding 25 µM neomycin sulfate or 1 µM staurosporine (not shown) indicating (1) that internal Ca2+ pools were the source of the response and (2) the implication of protein phosphorylation processes in Ca2+ mobilization. Addition of external calcium, per se, had no effect on changes in [Ca2+]cyt. However, as the case of OB, supplementation with calcium before or concomitant with Cd2+ inhibited the Cd2+-induced increase in [Ca2+]cyt (Fig. 4c).
H2O2, generated by the plasma membrane NADPH oxidase, plays a secondary role in the Cd2+-induced toxicity
To assess the potential role of NADPH oxidase-dependent H2O2 production in Cd2+-induced toxicity, we firstly considered the dose effect and kinetics of cell death on J0 and gp3 cell lines. As shown in Fig. 5a, at the optimum dose of 3 mM CdCl2 (Fig. 5b), cell death started after a lag phase of 1–2 h, increased for a few hours and was maximum by 10 h. Differences were observed between the two lines either in the lag phase duration (longer in gp3 than in J0) or in percentage of dead cells for each dose and time point (Fig. 5). As shown in Fig. 5c and in Supplementary Fig. S1, after a 5 h treatment with 3 mM CdCl2, the number of dead cells increased in all three cell lines as a function of age. The computed difference in cell sensitivity clearly showed that NADPH oxidase-dependent events proceeded at early stages of growth, while they were less important when cells became older (Fig. 5c). Such an observation prompted us to gain a deeper insight into the role of H2O2 in the death process. Thus, it clearly appeared that H2O2 accumulation was strictly age dependent in J0 (Fig. 5d) correlating the difference in cell sensitivity between the two lines (Fig. 5c). Moreover, manipulation of extracellular H2O2 concentrations phenocopied the behaviour of J0 and gp3 in response to CdCl2. Treatment of J0 cells with CAT, which dismutates H2O2, made them more resistant (18 ± 7%) to Cd2+ (Fig. 5e). Consistently, addition of the H2O2-generating system glucose oxidase + glucose (GO/G) in the incubation medium enhanced gp3 cell sensitivity to Cd2+ (20 ± 4%), making them approximately as sensitive as J0 (Fig. 5f). Consequently, it may be concluded that the NADPH-oxidase-dependent OB represented, at most, one-fifth of the total effect of Cd2+ on cell death throughout the growth period (Fig. 5c). In addition to this, it appeared that J0 cells challenged with GO/G or with the fungal protein cryptogein (see Supplementary Fig. S2 online), and that were submitted higher levels of H2O2 than cells treated with 3 mM CdCl2, did not display obvious signs of poisoning after a 5 h treatment. Finally, in response to the metal, cell death was maximum 5–6 h after the achievement of the OB (Figs. 5a & 1a), further suggesting that H2O2, per se, was not toxic.
Cd2+ induces a mitochondrial accumulation of O2−., responsible for the main part of the metal toxicity
H2O2 produced during the first OB was not sufficient to explain the overall toxic effects of CdCl2 on BY2 cells. We considered the implication of other potential sources of ROS with special reference to mitochondria, known in mammalian cells to play a role in superoxide anion generation (Wang et al. 2004).
Dihydroethidium, also called HE, probes superoxide radicals. The radical specifically oxidizes HE into a DNA-intercalating compound (Zhao et al. 2003), which stains nuclei with a bright yellow fluorescence. Figure 6a shows a typical fluorescence microscopy observation of BY-2 J0 cells exposed to 3 mM CdCl2. After a 15 min treatment, all microscopic fields observed fell into three different classes consisting in unstained, partially and totally stained clusters, respectively (Figs. 6a & b). The time course of changes in the percentage of HE-positive clusters (Fig. 6b) revealed a rather slow intracellular ROS production (compared with the extracellular one; see Fig. 1) that was detected after a lag period of 10–15 min at 3 mM of CdCl2 and that could not be detected before 24 h at 20 µM (see Supplementary Fig. S3 online). At an identical growth stage, the control line pKy displayed similar kinetics of staining (data not shown). In addition to this, treatment of cells with the SOD mimetic MnTBAP before exposure to Cd2+ significantly decreased the number of HE-positive clusters indicating that HE staining was specific of O2−. (Fig. 6c). While ageing was characterized by a progressive loss of cells to accumulate extracellular H2O2 in response to Cd2+ (Fig. 5d), an increased ability to accumulate superoxide radicals was observed in parallel (Fig. 6d). When treated with 3 mM CdCl2, the superoxide accumulation in gp3 cell line was half of the wild-type cells (Fig. 7a). Accordingly, it may be assumed that H2O2 accumulation through activation of the plasma membrane oxidase might regulate the intracellular synthesis of superoxide radicals. As shown in Fig. 7b, the exogenous application of CAT partially inhibited (30%) the O2−. accumulation of CdCl2-treated J0 cells without modifying the response of gp3. Moreover, the artificial extracellular production of H2O2 by means of the enzymatic generator GO/G stimulated the intracellular OB of about 50% above that of CdCl2-treated gp3 cells. Because H2O2 alone (application of GO/G) was unable to trigger intracellular O2−. accumulation on either gp3 nor J0 cell lines (not shown), it was concluded that H2O2, in synergy with Cd2+, amplified the superoxide radical generation. However, because aged cells had only a limited capacity to accumulate H2O2, when cells became older, the production of O2−. was unlikely linked to H2O2-dependent regulations.
A pharmacological approach was then carried out (1) to locate the intracellular source of superoxide anion production, with particular reference to mitochondrial ETC, and (2) to assess the role of O2−. on cadmium-induced cell toxicity. As shown in Fig. 8, the simultaneous impairment of complexes I and II of ETC, through addition of rotenone + TTFA, or disturbance of the functioning of complex III by means of myxothiazol or AA, drastically inhibited (70–90%) the superoxide radical production and, to a similar extent, the cell death process. Together, our data are consistent with a mitochondrial origin for the intracellular Cd2+-induced ROS production. In this way, transcripts encoding the tobacco alternative oxidase NtAOX1 were shown to increase as soon as 6 h after a treatment with 20 µM CdCl2 (see Supplementary Fig. S3a online), indicating that the additional pathway of electron flow was most likely required to prevent formation of damaging ROS in mitochondria (Maxwell, Nickels & McIntosh 2002). Nevertheless, at 24 h, the intracellular ROS production could be observed (see Supplementary Fig. S3b online).
Hydroxy radical-dependent membrane peroxidation parallels Cd2+-induced cell death
Polyunsaturated fatty acids (PUFAs) are particularly susceptible to the ROS-mediated oxidation. As hydrophobic components of lipid bilayers, their chemical modification may lead to the impairment of functions of membranes and may subsequently evolve into their degradation. Because lipid peroxidation actually represents another source of ROS, we examined whether cadmium was able to trigger this oxidative stress in BY-2 cells. In response to CdCl2 treatment, after a lag phase of 1–2 h, a progressive rise in HFA levels was observed (Fig. 9a). Again, J0 and gp3 lines displayed a marked difference. At the same growth stage, lipid peroxidation in J0, which was comparable to pKy (not shown), was about three times higher than in gp3. Interestingly, the kinetics of the oxidative process fit the cell death pattern (Figs 5a & 9a). Typical chromatograms (Fig. 9b) showed a clear predominance of metabolites derived from linoleic acid (13 and 9 hydroxy-octadecadienoic acid (HODE), which represented about 80% of the PUFAs present in BY-2 cells (not shown). Considering that (1) 13 and 9 isomers of HODE, unrelated to the time point, were present in a ratio of 0.998 ± 0.1 (n = 20) and that (2) all HFAs derived from both PUFAs were racemic (the percentage of the S enantiomer was 50.3 ± 0.28 for 13 HODE and 50.1 ± 0.15 for 9 HODE (n = 3), respectively), it was concluded that membrane peroxidation that developed in BY-2 cells, in response to CdCl2, was exclusively derived from the action of ROS.
As shown in Fig. 9c, inhibition of CdCl2-mediated ROS production by AA or CAT resulted in a fairly proportional decrease in both lipid peroxidation and cell death. In contrast, DFO, a ferric iron-chelating agent that prevents the Fenton reaction-derived hydroxy radical (OH) formation, did not affect O2−. production, although it drastically inhibited the oxidative process and efficiently protected cells from cadmium toxicity. Finally, BHA, a free radical scavenger, severely inhibited the three analyzed parameters. It appeared from this pharmacological study that oxidation of membranes and cell death were always similarly altered by the selected compounds, strengthening the assumed causal link between degradation of membranes and death.
In the literature, a great deal of data indicates that plants are able to concentrate cadmium such that tissues and organs, especially the roots, are rapidly in contact with millimolar concentrations of Cd2+ (Rivetta et al. 1997; Sandalio et al. 2001; Shah et al. 2001; Perfus-Barbeoch et al. 2002). Here we report that treatment of tobacco cells with such concentrations of cadmium results in the induction of cell death within a few hours. We further show that ROS, differing in chemical nature and subcellular localization, contribute to cytotoxicity of cadmium.
Disturbance of Ca2+ homeostasis by Cd2+ leads to the first OB
The earliest OB induced by cadmium consisted in the transitory accumulation of H2O2, and the use of an antisense strategy has enabled us to identify the intrinsic plasma membrane NADPH oxidase as the sole source of this accumulation (Fig. 1). These data corroborate the pharmacological approach previously described by Olmos et al. (2003). Additionally, by analogy with the animal models, one can assume that H2O2 is produced at the external side of the cell membrane. In contrast to the biotic elicitor, cryptogein, that required de novo protein synthesis (Simon-Plas et al. 2002; this work, Fig. 2), the elicitation of the OB with CdCl2 proceeds by activating the oxidase present in resting state in plasma membrane of BY-2 cells.
Because of identical charges and similar ionic radii, Cd2+ may interfere with the biological activity of Ca2+. Cd2 may enter cells through Ca2+-selective channels (Misra et al. 2002; Perfus-Barbeoch et al. 2002) or via specific transmembrane transporters (Clemens et al. 1998), and is also able to compete with Ca2+ for the calmodulin binding sites (Rivetta et al. 1997). Clemens et al. (1998) reported that Cd2+ uptake activity of the wheat transporter LTC1 was blocked by La3+ and Ca2+. Similarly, our current results provide evidence that Cd2+-induced OB is impaired by these cations (Fig. 3). Nevertheless, with the exception of La3+, none of the potential Ca2+ channel inhibitors tested were able to inhibit the Cd2+-induced H2O2 production. One may assume that, in BY2 cells, Cd2+ enters cells via La3+-sensitive transporters or channels to activate the plasma membrane NADPH oxidase. Preceding the OB, Cd2+ induced a rapid and transient [Ca2+]cyt increase requiring protein phosphorylation and IP3-mediated release of calcium from internal stores (Fig. 4). Furthermore, inhibition of the Cd2+-induced [Ca2+]cyt increase by exogenous application of Ca2+ strengthened the previously mentioned assumption that the two cations competed for the same way of entry. In addition to this, it was demonstrated that [Ca2+]cyt increase relayed the Cd2+ signal that led to NADPH oxidase activation and H2O2 accumulation. Indeed, neomycin sulfate and another phospholipase C inhibitor, U73122, abolished or strongly decreased the Cd2+-induced H2O2 production (Fig. 3). Moreover, the ability of ODQ or nicotinamide, the respective inhibitors of guanylate and ADP-ribose-cyclases, to partially inhibit the OB led us to assume that oxidase activation might also involve Ca2+ release via cyclic ADP-ribose-gated channels, known to be present in various plant endomembranes (Knight, Trewavas & Knight 1997; MacRobbie 1998; Navazio, Mariani & Sanders 2001). Downstream from the cytoplasmic calcium mobilization, protein phosphorylation and Ca2+ might also directly regulate NtrbohD activity. Finally, because of a block of H2O2 production by W7, the putative signalling network might involve calmodulin/Ca2+-dependent steps in the Cd2+-induced Ca2+ wave through nitric oxide synthase (NOS) (Guo, Okamoto & Crawford 2003) and/or oxidase activations. From this body of data, the signalling network that goes into action a few minutes after the cadmium exposure and ends in the accumulation of H2O2 is sketched out in Fig. 10.
Although our data clearly demonstrated that difference in sensitivity to CdCl2 between gp3 and J0 cells was correlated with the ability of the latter to trigger an extracellular OB, exposure of J0 cells with either GO/G or cryptogein, which resulted in doses of H2O2 about 10 times higher than those produced by cells in response to Cd2+ (see Supplementary Fig. S2 online), did not lead to significant cell poisoning. More than a direct element of cadmium toxicity, the extracellular OB would then act as an amplifier. Despite the absence of experimental evidence demonstrating higher uptake of Cd2+ in J0 than in gp3, it is nevertheless possible to postulate that H2O2 produced by J0 cells exposed to Cd2+ could activate H2O2-regulated Ca2+ channels present in various plant plasma membranes (Pei et al. 2000; Murata et al. 2001; Demidchik, Davenport & Tester 2002; Köhler, Hills & Blatt 2003),leading then to a more important entry of Cd2+ in J0 than in gp3 (Fig. 10).
The mitochondrial electron transport chain disturbance by cadmium induces a second wave of ROS production responsible for cell toxicity
In the literature, several lines of evidence indicate that mitochondria are a prominent target of Cd2+, and numerous results report the organelle as a potential and important source of oxidative stress (Møller 2001). This work demonstrates that BY-2 cells exposed to Cd2+ mounted a second wave of ROS accumulation (Fig. 6). Consistent with the mitochondrial origin of this accumulation, it appeared that all respiration inhibitors tested in this work, significantly decreased the Cd2+-induced superoxide accumulation (Fig. 8). Recent evidence indicates that Cd2+ likely binds between semi-ubiquinone and cytochrome b566 of the Q0 site of cytochrome b of complex III. Assumingly, the resulting increase of the semi-ubiquinone concentration facilitated electron transfer to molecular oxygen to form superoxide (Wang et al. 2004). Consequently, it can be suggested from our data that the action of the complexes I and II inhibitors (rotenone + TTFA), which are able to drastically limit electron flow to ubiquinone, or inhibition of the Q cycle by complex III inhibitors (AA or myxothiazol), might both prevent excessive accumulation of the semi-ubiquinone radical, limiting the Cd2+-induced ROS accumulation. The phenolic free radical scavenger, BHA, appeared to prevent in vivo O2−. accumulation efficiently (Fig. 9) but was unable to scavenge directly superoxide radical produced in vitro by the enzymatic generator xanthine oxidase (not shown). This strongly suggested that BHA acted upstream from the superoxide formation, strengthening the assumption that Cd2+ action resulted in the rise in semi-ubiquinone radical that could be efficiently scavenged by the phenolic antioxidant. Our data suggest that Cd2+ induces the synthesis of superoxide rather than a decrease in its rate of degradation. The accumulation of superoxide anion increased with the age of the culture (Fig. 6), and interestingly, these changes are fairly correlated with cell sensitivity to Cd2+ (Fig. 5). Consistent with this, the gp3 line, more resistant to Cd2+ than J0, displayed a lower ability to accumulate O2−. (Fig. 7), and all pharmacological compounds able to inhibit the superoxide accumulation efficiently protected cells from Cd2+ toxicity (Figs. 8 and 9). Our data also provided evidence that exogenous application of H2O2 stimulated the superoxide accumulation (Fig. 7) and enhanced cell toxicity (Fig. 5), strengthening the idea of a causal link between O2−. production and cell death. In the presence of the iron redox recycling inhibitor DFO (Fig. 9), although O2−. accumulation remained unchanged, the cells were efficiently protected, suggesting that steps downstream from this production were inhibited.
The third wave of ROS concomitant with cell death, consisted in membrane peroxidation
The detailed analysis of lipid peroxidation revealed a ROS-mediated process (Fig. 9) that followed the mitochondrial O2−. production (Fig. 6) and appeared in parallel with cell death (Fig. 5), leading us to assume that membrane degradation resulting from lipid peroxidation certainly promotes the loss of cell integrity. The ability of both AA and BHA to inhibit superoxide production, lipid peroxidation and cell death (Fig. 9) suggests a sequence involving the mitochondrial production of O2−., followed by the oxidation and collapse of membranes, and ending in cell death. Finally, lipid peroxidation inhibition by DFO and its protective action (Fig. 9) were symptomatic of the involvement of Fenton-type reactions in initiations of both the oxidative process and cell death. Although we did not search for a Cd2+-induced iron release, such an increase in cellular free iron concentrations might happen either through the direct competition between the two cations for some cellular binding sites or through mediation by superoxide ions that have also been demonstrated to provoke protein demetallation (Keyer & Imlay 1996; Varghese, Tang & Imlay 2003).
In conclusion, these results indicate that, in BY-2 tobacco cells, Cd2+ induces, within minutes, a transient increase in [Ca2+]cyt that appears to regulate the extracellular NADPH oxidase-dependent generation of H2O2 (Fig. 10). The involvement of the first OB in Cd2+-induced toxicity is minor and may contribute to further penetration of the heavy metal via H2O2-activated calcium (cadmium)-permeable channels. Above a threshold cellular concentration of cadmium, mitochondria are altered and superoxide anion is produced. Finally, analysis of the Cd2+-induced lipid peroxidation reveals that the final outcome of the superoxide accumulation likely consists in a general and non-specific hydroxy radical-mediated oxidation of the cell content. Although membrane integrity is a deciding factor for cell survival, oxidation and degradation of proteins and nucleic acids have already been described (Sandalio et al. 2001; Romero-Puertas et al. 2002) and could also contribute to cell death (Fig. 10).
This study was supported by Grant ‘Toxicologie Nucléaire’ (L. Garnier) from the French Atomic Energy Commission. The authors wish to express their warmest thanks to Dr J. Thièry (http://ModLibre.org) for his assistance in the mathematical treatment of their data.