• Cadmium (Cd2+) is an environmental pollutant that causes increased reactive oxygen species (ROS) production. To determine the site of ROS production, the effect of Cd2+ on ROS production was studied in isolated soybean (Glycine max) plasma membranes, potato (Solanum tuberosum) tuber mitochondria and roots of intact seedlings of soybean or cucumber (Cucumis sativus).
• The effects of Cd2+ on the kinetics of superoxide (), hydrogen peroxide (H2O2) and hydroxyl radical (•OH) generation were followed using absorption, fluorescence and spin-trapping electron paramagnetic resonance spectroscopy.
• In isolated plasma membranes, Cd2+ inhibited production. This inhibition was reversed by calcium (Ca2+) and magnesium (Mg2+). In isolated mitochondria, Cd2+ increased and H2O2 production. In intact roots, Cd2+ stimulated H2O2 production whereas it inhibited and •OH production in a Ca2+-reversible manner.
• Cd2+ can be used to distinguish between ROS originating from mitochondria and from the plasma membrane. This is achieved by measuring different ROS individually. The immediate (≤ 1 h) consequence of exposure to Cd2+ in vivo is stimulation of ROS production in the mitochondrial electron transfer chain and inhibition of NADPH oxidase activity in the plasma membrane.
Life in an oxygen-rich atmosphere has to deal with the danger of oxidative stress. Reactive oxygen species (ROS), such as superoxide (), hydrogen peroxide (H2O2) and hydroxyl radicals (•OH), are produced during normal cell metabolism but their production is drastically enhanced when plants are exposed to natural abiotic stresses such as high light, low temperatures and drought, and to biotic stresses such as attack by pathogens or wounding (e.g. Scandalios, 2002). Environmental pollutants such as heavy metal ions, for example cadmium (Cd2+), are also known to induce oxidative stress (Schützendübel & Polle, 2002). Although ROS are considered to be damaging molecules, it is recognized that they play a major role in defense against pathogens, cellular signaling pathways and regulation of gene expression in a wide range of organisms, including plants (e.g. Apel & Hirt, 2004).
ROS are generated in a variety of reactions. These include the respiratory and photosynthetic electron transport chains and side reactions of enzymes such as peroxidases. In addition, specialized enzymes such as superoxide dismutases, xanthine oxidase and NADPH oxidases (NOXs) produce ROS (Halliwell & Gutteridge, 1999). ROS are not only produced in many different reactions but also in different compartments of the cell, including mitochondria, chloroplasts, peroxisomes, the cytosol and the apoplast.
Plant plasma membranes produce to the apoplastic side in response to different stimuli. This activity has been widely accepted to originate from plasma membrane-localized NOXs that reduce external O2 using cytoplasmic NADPH as the electron source, although other plasma membrane-localized or -associated enzymes may also contribute to production. Ten genes encode NOXs in Arabidopsis thaliana. These genes are termed respiratory burst oxidase homologs A to J (RbohA–J) because of their homology to the catalytic subunit gp91phox (Nox2) of the NOX complex of mammalian phagocytes (Torres & Dangl, 2005). The plant NADPH oxidases are predicted to have six transmembrane helices with two heme-binding sites, cytoplasmic binding sites for NADPH and FAD at the C-terminus and two calcium (Ca2+)-binding EF-hand motifs at the N-terminus (Torres & Dangl, 2005), a similarity with the NOX5 NADPH oxidase in mammals (Banfi et al., 2004). NOX-encoding genes are present in all plant species investigated so far, and they have distinctive expression patterns. Although the involvement of NOXs in important physiological processes has been demonstrated in plants (e.g. Foreman et al., 2003; Kwak et al., 2003; Potockýet al., 2007), their enzymatic properties have rarely been studied. It has been shown that the EF-hand motifs in plant NOXs bind 45Ca2+ (Keller et al., 1998) and that Ca2+ stimulates the NOX activity of isolated plasma membranes (Sagi & Fluhr, 2001). The activation of mammalian NOX5 by Ca2+ in a cell-free system is in the micromolar range (Banfi et al., 2004). Hence, in the plant plasma membrane micromolar Ca2+ concentrations can also be expected to lead to enhanced production.
In most biological systems it is difficult to determine the site of ROS production in vivo. In most studies only H2O2 has been measured. H2O2 is the only ROS that can diffuse easily through aquaporins in the membranes and over larger distances within the cell (Bienert et al., 2007). In this work we used Cd2+ ions as a biochemical tool to distinguish between H2O2 originating from formed by the mitochondria and that originating from formed by the plasma membrane NOX using dyes and electron paramagnetic resonance (EPR) spin traps specific for , H2O2 and •OH. Ca2+ was used as an indicator of NOX activity and a potential competitor for Cd2+. To link the in vitro data to the in vivo situation, we studied the effect of short-term exposure to Cd2+ on ROS production with intact roots of cucumber (Cucumis sativus) and soybean (Glycine max) seedlings by measuring , H2O2 and •OH production.
Materials and Methods
Different plant species were chosen for isolation of plasma membranes (soybean) and mitochondria (potato) and for in vivo (cucumber) experiments to obtain material with optimal activity in reasonable quantities. Soybean (Glycine max (L.) Jutro) seedlings were grown in the dark at 25°C in vermiculite for 4 d for plasma membrane preparation. Cucumber (Cucumis sativus L.) and soybean seedlings were grown between damp paper towels in a vertical position (root apex downwards) in rectangular Petri dishes at 25°C for 3 d (soybean) or 4 d (cucumber) in dim white light for in vivo measurements. Under these conditions, seedlings with straight roots of 40–60 mm length were obtained. Arabidopsis thaliana (L.) (background Columbia) wild type and rbohC and D mutants were grown for 20 d in a vertical position on rectangular plates (1% Phytagel and 4.3 g l−1 Murashige–Skoog basal medium without hormones) in short-day conditions.
Potato (Solanum tuberosum L.) tubers were purchased from the local market.
Isolation and solubilization of plasma membrane vesicles
Plasma membrane vesicles from etiolated soybean hypocotyls were isolated according to Michalke & Schmieder (1979). The top 1 cm of the hypocotyls was excised and incubated for 10 min in a fourfold volume (fresh weight/volume) of isolation buffer (10 mM Tris/HCl, pH 8.0, 20 mM Na2EDTA and 300 mM NaCl) to which 0.1% bovine serum albumin (BSA) and 0.5 mM phenylmethylsulfonyl fluoride (PMSF) were added just before use. The hypocotyls were homogenized in a blender (Waring, Torrington, CT, USA) and filtered through a nylon cloth, and the filtrate was centrifuged (30 min at 1300 g and 4°C). The supernatant (raw fraction) was again centrifuged (45 min at 91 000 g and 4°C) to separate soluble proteins and membranes. The pellet was resuspended in isolation buffer and homogenized in a glass potter. The homogenate was diluted twofold (fresh weight/volume) with isolation buffer. Polyethylene glycol (PEG) 6000 was added (4.3 g per 100 ml), and the mixture was stirred on ice for 30 min, followed by centrifugation (30 min at 1300 g and 4°C) to separate the intracellular membranes from plasma membranes. The supernatant, containing the plasma membranes, was centrifuged (45 min at 91 000 g and 4°C). The pellet was resuspended in isolation buffer without Na2EDTA and homogenized in a glass potter. This method resulted in plasma membrane preparations containing 1.5–2 mg protein ml−1.
Plasma membranes were solubilized with 2% Tween-20 (weight/volume) for 30 min at 45°C. The supernatant obtained after centrifugation (12 000 g in a bench centrifuge for 5 min) was stored at –70°C until use.
Isolation of mitochondria from potato tubers
Mitochondria were isolated according to Douce et al. (1987). Peeled tubers were macerated with a Moulinex (type 750) vegetable juice machine. The juice (c. 500 ml) was made up to 1000 ml with cold isolation buffer (0.5 M sorbitol, 8 mM Na2EDTA and 80 mM 3-[N-morpholino] propane-sulfonic acid (MOPS), pH 7.5, with 0.4% BSA, 0.5% polyvinylpyrolidone and 8 mM cysteine added just before use), filtered through a nylon cloth and centrifuged (10 min at 370 g and 4°C). The supernatant was filtered again through a nylon cloth and centrifuged (15 min at 1000 g and 4°C). The supernatant was centrifuged (20 min at 11 700 g and 4°C) and the pellet was resuspended in 40 ml of cold Percoll-gradient buffer (25% Percoll, 0.4 M sorbitol and 100 mM MOPS, pH 6.9) and centrifuged (45 min at 30 000 g and 4°C). The white layer containing the mitochondria was collected and mixed with 8 volumes of cold wash buffer (0.4 M sorbitol and 10 mM MOPS, pH 7.2, and freshly added 0.1% BSA). After centrifugation (20 min at 8000 g and 4°C), the pelleted mitochondria were resuspended in measuring buffer (0.4 M sucrose, 5 mM MgCl2, 30 mM KCl and 20 mM HEPES, pH 7.3) and used for measurements or frozen as droplets in liquid nitrogen and stored at –70°C. The droplets were rapidly thawed in a thin glass tube in 90°C water and immediately cooled in ice-water. This method resulted in basal respiratory activity of 41 ± 7 nmol O2 mg−1 protein min−1 in the presence of 5 mM succinate, resembling the rates reported in Ravanel et al. (1984).
Solubilized plasma membranes were depleted of Ca2+ by incubation with Chelex 100 beads (50–100 dry mesh; Sigma-Aldrich, St. Quertin Fallavier, France) according to Kashino et al. (1986). Membranes (0.5 ml, corresponding to c. 0.75 mg protein ml−1) were incubated in 0.1–0.3 g Chelex 100 for 6 min at room temperature. During the incubation the mixture was vortexed for 10 s every 1 min. The Chelex-free samples were collected using a pipette.
SDS-PAGE and western blotting
Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) was carried out in 8–12% gradient polyacrylamide gels stained with silver. Western blotting was performed using a Multiphor II Novablot unit (Amersham Bioscience, Piscataway, NJ, USA). For detection, the enhanced chemoluminescence (ECL) system (Amersham Bioscience) was used according to the manufacturer's protocol.
Determination of NADH and NADPH oxidation
The kinetics of NADH and NADPH oxidation by solubilized plasma membranes were measured photometrically for 10 min as the decrease in absorbance at 340 nm at 25°C using the molar extinction coefficient ɛ340 = 6.2 × 103 M−1 cm−1. The reaction mixture (1 ml) contained 25 µg of plasma membrane protein and 200 µM NADH or NADPH in 20 mM HEPES, pH 7.8.
Detection of by XTT reduction or cytochrome c reduction
The kinetics of production was measured photometrically as the increase in absorbance of Na, 3′-(1-(phenylaminocarbonyl)-3,4-tetrazolium)-bis(4-methoxy-6-nitro)benzenesulfonic acid hydrate (XTT) at 470 nm. production was calculated using the molar extinction coefficient ɛ470 = 21.6 mM−1 cm−1 (Sutherland & Learmonth, 1997) according to the equation:
production of solubilized plasma membranes was measured in 20 mM phosphate buffer, pH 7.8. Measurements with CdCl2 were performed in 20 mM HEPES buffer, pH 7.8, to avoid the formation of precipitates. The reaction mixture (1 ml) contained 0.5 mM XTT and 25 µg of solubilized membrane protein and was incubated for 5 min in darkness at 25°C. The reaction was started by adding 200 µM NADH. production of roots was measured in 10 mM MES buffer, pH 6.5, containing 0.5 mM XTT. The roots of four intact cucumber seedlings were placed in 5 ml of reaction mixture. Samples (1 ml) were measured at the indicated times. The reaction mixture was bubbled with air during the incubation.
Alternatively, cytochrome c (cyt c) reduction was measured at 550 nm in the presence of 50 µM cyt c (ɛ550 = 21.0 mM−1 cm−1).
Determination of H2O2 production by scopoletin oxidation
The kinetics of H2O2 production were measured as the decrease in scopoletin fluorescence at 25°C at 350 nm (excitation) and 460 nm (emission) (Staniek & Nohl, 1999). A standard curve with known H2O2 concentrations was determined for 5 µM scopoletin in the measuring buffer. H2O2 production in mitochondria was measured after 30 min of incubation (1 mg protein ml−1) in the measuring buffer (see isolation of mitochondria) containing 5 µM scopoletin, 10 U horseradish peroxidase (HRP), 0.5 mM succinate and 2 µM mesoxalonitrile 3-chlorophenylhydrazone (CCCP) as an uncoupler. To avoid perturbations of the fluorescence, mitochondria were sedimented by centrifugation (5 min at 9000 g and 25°C) before measuring the fluorescence of the supernatant. All results were in the linear region of the standard curve.
H2O2 production in roots was measured in 10 mM MES, pH 6.5, containing 5 µM scopoletin. The roots of four intact cucumber seedlings or of three soybean seedlings were placed in 5 ml of assay medium. In case of A. thaliana, 20 seedlings were incubated in 1.5 ml of assay medium. Samples (1 ml) were measured at the indicated times. The assay medium was bubbled with air during the incubation.
EPR spectra were recorded at room temperature with a Bruker 300 X-band spectrometer at 9.69 GHz microwave frequency, 63 mW microwave power and 100 kHz modulation frequency. For detection in solubilized plasma membranes, 200 µg ml−1 protein was incubated for 10 min in 10 mM phosphate buffer, pH 7.5, containing 200 µM NADH and 25 mM 2-ethoxycarbonyl-2-methyl-3,4-dihydro-2H-pyrrole-1-oxide (EMPO) (Olive et al., 2000) in the presence of 50 µg ml−1 superoxide dismutase (SOD) or 1 mM KCN, respectively. For detection in isolated mitochondria, 1 mg protein ml−1 was incubated for 5 min in measuring buffer (see isolation of mitochondria) containing 25 mM EMPO, 0.5 mM succinate, 2 µM CCCP and, if indicated, 30 µM CdCl2 or 2 µM antimycin A. The pure adducts of EMPO were obtained by dissolving KO2 in 10 mM phosphate buffer, pH 7.5 ( adduct), and by the Fenton reaction using 10 µM FeSO4 + 100 µM H2O2 (•OH adduct).
For •OH detection in roots, four intact cucumber or soybean seedlings were incubated for 1 h in 10 mM MES buffer, pH 6.5, 50 mM N-tert-butyl-α-(4-pyridyl)nitrone N′-oxide (4-POBN) and 4% ethanol (Janzen et al., 1978).
The data represent means or representative examples from measurements repeated 3–6 times. Typical standard errors are shown in Figs 4(a) and 6, and in tables, but omitted in other figures for the sake of clarity.
Isolation of plasma membranes from soybean hypocotyls
Plasma membranes were isolated from soybean hypocotyls using a PEG-based membrane fractionation protocol (Michalke & Schmieder, 1979). Proteins at different isolation steps were separated by SDS gel electrophoresis and blotted (Fig. 1). To verify the separation of plasma membranes from other cellular membranes by PEG sedimentation, immunodetection was performed with antibodies directed against mitochondrial Rieske protein and plasma membrane P-ATPase. The Rieske protein was detected only in the total membrane and the intracellular membrane fractions. P-ATPase was detected in the total membrane, weakly in the intracellular membrane and strongly in the plasma membrane fractions. Plasma membrane isolation using a two-phase separation protocol (Larsson et al., 1994) resulted in similar protein compositions (data not shown). The presence of a NOX protein in the plasma membrane fraction was shown by immunodetection using anti-NtrbohD antibodies (Simon-Plas et al., 2002). A weak signal was detected in total membrane and intracellular membrane fractions while a strong signal was found in the plasma membrane fraction.
production in solubilized plasma membranes
Solubilized plasma membranes produced when the electron donor NADH or NADPH was present. Solubilization was necessary to allow access to the substrates. Nonsolubilized membranes produced only 10% of the generated by the solubilized membranes. The activity was measured photometrically as NADH oxidation or reduction of either XTT or cytochrome c (Table 1). When NADPH instead of NADH was used as substrate, similar rates for the oxidation were obtained. The activities, measured in the absence of SOD, resembled those reported for plasma membrane preparations in the literature (Brightman et al., 1988; DeHahn et al., 1997; Van Gestelen et al., 1997). In the presence of SOD, the NADH reduction was stimulated almost tenfold, indicating that inhibits NAD(P)H reduction when it is not removed.
Table 1. Activity of solubilized plasma membranes of soybean
Specific activity (nmol mg−1 protein min−1)
Activities were measured in the absence or presence of 50 µg ml−1 superoxide dismutase (SOD). The reaction assay contained 25 µg protein ml−1, 200 µM NADH and either 500 µM Na,3′- (1-(phenylaminocarbonyl)-3,4-tetrazolium)-bis(4-methoxy-6-nitro)benzenesulfonic (XTT) or 50 µM cytochrome c (Cyt c).
2.9 ± 0.20
26.9 ± 0.22
31.0 ± 1.01
1.03 ± 0.02
Cyt c reduction
34.0 ± 0.16
The stoichiometry of NADH oxidation (measured in the presence of SOD) to XTT reduction was almost 1 : 1 (both are two-electron reactions), while the reduction of cyt c (one-electron acceptor) was lower than the theoretically expected 1 : 2 stoichiometry. This might be a result of the relatively low concentration of cyt c used in the assay. Addition of SOD to the XTT or cytochrome c reduction assay inhibited the reduction of the electron acceptors almost completely. Therefore these assays can be used to determine production.
production was shown independently by EPR spectroscopy using the spin trap EMPO (Fig. 2). EMPO reacts with both and •OH and gives specific adducts which can be distinguished by their hyperfine splitting pattern (Olive et al., 2000). Spectrum 1 presents the EMPO/ adduct which was formed by dissolving KO2 in the assay buffer. Measurements of solubilized plasma membranes incubated with NADH showed the typical spectrum of the EMPO/ adduct (spectrum 2), which was almost completely abolished in the presence of the scavenger SOD (spectrum 3). The signal of the EMPO/ adduct was unaffected by KCN (spectrum 4), ruling out plasma membrane-bound peroxidases (Mika & Lüthje, 2003) as the source of . Diphenyleneiodonium chloride (DPI) inhibited production at low concentration both in EPR and in XTT reduction measurements (Fig. 3). Fifty per cent inhibition was observed at 1.5 µM DPI, indicating a specific inhibitory effect. DPI is an inhibitor of the mammalian NOX at low micromolar concentrations (< 10 µM; Doussiere & Vignais, 1992).
Effect of CaCl2 or MgCl2 on production in solubilized plasma membranes
The inhibition of production by DPI (shown in Fig. 3) is in agreement with the assumption that NOX is responsible for the production observed in solubilized plasma membranes. To further test this assumption we investigated the effect of Ca2+ on production. Ca2+ is known to bind to the EF-hand motifs of the NOX encoded by the AtrbohA gene (Keller et al., 1998) and to stimulate the activity of NOX in tobacco (Nicotiana tabacum) and tomato (Lycopersicon esculentum) plasma membranes (Sagi & Fluhr, 2001). Indeed, 5 mM Ca2+ increased the production in solubilized plasma membranes by 30% (Table 2) but it is possible that the plasma membranes used here were not fully depleted of Ca2+. We therefore depleted the solubilized plasma membranes of Ca2+ by preincubating them with the Ca2+ chelator Chelex 100 or ethylene glycol tetraacetic acid (EGTA). Incubation with Chelex 100 or 1 mM EGTA resulted in a 28 or 77% reduction, respectively, in production (Table 2). In the Chelex 100-treated membranes, Ca2+ and magnesium (Mg2+) both stimulated the activity. The initial activity was already reached at a concentration of 10 µM. production was reduced more efficiently by 1 mM EGTA than by Chelex 100, but Ca2+ or Mg2+ in the millimolar range was needed to restore the initial activity.
Table 2. Effect of divalent cations and chelators on -producing activity in solubilized plasma membranes of soybean
production was measured as Na,3′-(1-(phenylaminocarbonyl)-3,4-tetrazolium)-bis(4-methoxy-6-nitro)benzenesulfonic (XTT) reduction in the presence of 200 µM NADH. 100% activity corresponds to 31.6 ± 0.9 nmol XTTH2 mg−1 protein min−1.EGTA, ethylene glycol tetraacetic acid.
No treatment (control)
+ 5 mM CaCl2
125 ± 6
+ 5 mM MgCl2
126 ± 4
1 mM EGTA
23 ± 7
72 ± 7
+ 10 µM CaCl2
95 ± 4
+ 100 µM CaCl2
97 ± 7
+ 500 µM CaCl2
103 ± 5
+ 10 µM MgCl2
91 ± 9
+ 100 µM MgCl2
104 ± 6
+ 500 µM MgCl2
108 ± 5
Effect of CdCl2 and CaCl2 on production in solubilized plasma membranes
Cd2+ is known to cause oxidative stress in the cell and NOX was reported to be involved in this process in vivo (Olmos et al., 2003; Garnier et al., 2006). As a potent competitor for Ca2+ and an enzyme inactivator (Rivetta et al., 1997; Faller et al., 2005), Cd2+ may bind to the EF-hand motifs of the NOX and affect the enzyme activity. We tested the effect of Cd2+ on the production of plasma membranes. Cd2+ inhibited production even at the lowest concentration used (100 µM). Concentrations > 1 mM resulted in 85% inhibition (Fig. 4a).
Given that Cd2+ often acts on Ca2+-binding sites, we tested the effect of Ca2+ on the Cd2+-induced inhibition of production. Fig. 4(a) shows that Ca2+ and Mg2+ competed efficiently with Cd2+: when 5 mM Ca2+ or Mg2+ was present 300 µM Cd2+ had no inhibitory effect. To show the Ca2+ dependence, the activity of samples containing 0.3–1 mM Cd2+ was measured in the presence of different Ca2+ concentrations (Fig. 4b). To determine the inhibition constant (Ki) for Cd2+ the data of Fig. 4(b) are presented as a Dixon plot, including the data for 100 µM Cd2+ (Fig. 4c). The straight lines intersect at the same value, indicative of Cd2+ acting as a competitive inhibitor at the Ca2+-binding site. The intersection of the lines, that is, Ki, is approx. 180 µM. A plot of the slopes obtained from the Dixon plot against the Ca2+ concentration (inset, Fig. 4c) resulted in a straight line, a further indication of competitive inhibition. The intercept is not at 0, because the samples were not depleted of Ca2+ before the addition of Cd2+. This may explain why the Ki value is relatively high; Ca2+ in its binding site has first to be displaced by Cd2+.
Effect of CdCl2 on and H2O2 production in isolated potato tuber mitochondria
Cd2+ stimulates production in isolated mitochondria from animal tissue (Wang et al., 2004) by inhibiting the electron transfer at respiratory complex III (Miccadei & Floridi, 1993). Antimycin A is a well-known inhibitor of complex III. The binding of antimycin A to complex III results in increased ROS formation (e.g. Boveris et al., 1972) and this is thought to be attributable to the accumulation of semiubiquinone radicals that react with O2 to form (e.g. Cape et al., 2007; Zhang et al., 2007).
To investigate plant mitochondria as a potential source of ROS in response to Cd2+, we monitored generation indirectly using the oxygen electrode, measuring the catalase-induced decrease in oxygen consumption. We found that 10–30 µM Cd2+ decreased succinate-driven oxygen consumption by 75%, while 2 µM antimycin A decreased it by 50%. (Two molecules of are transformed by mitochondrial SOD into H2O2 and O2. Catalase spilts two molecules of H2O2 into H2O and O2. The evolved O2 is detected as a further decrease in O2 consumption.)
production could not be measured by XTT. The most likely explanation for this is that it simply does not enter the mitochondria. Instead, we measured production in isolated mitochondria by EPR spectroscopy using the EMPO spin trap. Spectra a and f (Fig. 5) show the EMPO/ and EMPO/•OH adducts from KO2 and the Fenton reaction, respectively. The dashed lines represent the hyperfine splitting pattern of the EMPO/ adduct. Because of the short lifetime of the EMPO/ adduct, mitochondria were incubated with the spin trap for only 5 min rather than the 30 min used for H2O2 measurements (Fig. 6). In the presence of 30 µM Cd2+ or 2 µM antimycin A (spectra b and c), significant production of the EMPO/ adduct was seen. In the absence of the inhibitors, the EMPO/ adduct was significantly smaller (spectrum c). SOD abolished the antimycin A-induced EMPO/ signal (spectrum 5e), indicating that caused the hyperfine splitting seen in spectra 5b, c and d. In the presence of SOD, a pure signal of the EMPO/•OH adduct was obtained. This signal is also present in spectra b and c, indicating that part of the generated is immediately transferred by mitochondrial SOD to H2O2. H2O2 reacts in the presence of transition metals, which are present in the mitochondria, to form •OH. In the presence of SOD and catalase, no EMPO spin trap adduct was detected (not shown). The spectra obtained with mitochondria are much noisier than those obtained with plasma membranes (Fig. 2) because of the more complex composition of the assay medium. High sorbitol concentrations, HEPES and traces of Percoll in the measuring buffer interfere with the spin trapping assay and induce changes in the hyperfine splitting.
H2O2 production of the succinate-respiring mitochondria was measured by fluorescence spectroscopy using the scopoletin oxidation assay (Staniek & Noll, 1999) (Fig. 6). The control rate of H2O2 production in mitochondria was 0.02 nmol min−1 mg−1 protein with 0.5 mM succinate as electron donor. The control values were increased twofold by 10 µM Cd2+ and almost threefold by 30 µM Cd2+. A fourfold increase in H2O2 production was induced by 2 µM antimycin A. Both methods of measuring H2O2 (EMPO and scopoletin) showed that ROS production was succinante-dependent and that it could be stimulated by antimycin A. This behavior is specific for mitochondrially generated ROS and would not occur if the ROS originated from any potential contamination from peroxisomes which might be present in our mitochondrial preparation (see Corpas et al., 2001; del Río et al., 2002; Hänsch et al., 2006; Nyathi & Baker, 2006 for evidence of ROS production in peroxisomens).
Effect of CdCl2 on and H2O2 production in intact roots
It has been reported that Cd2+ stimulates extracellular H2O2 generation in tobacco cell cultures upon short-term exposure to the cation (Olmos et al., 2003; Garnier et al., 2006). The question arises as to whether Cd2+ affects the activity of the NOX in the plasma membrane or rather stimulates /H2O2 production in the mitochondria. We measured the production of and H2O2 by intact roots of cucumber and soybean seedlings for 1 h using XTT reduction as a measurement for and scopoletin oxidation as a measurement for H2O2 (Fig. 7a–d). The rates of and H2O2 production were uncorrelated. production was inhibited by Cd2+ (Fig. 7a,c) while H2O2 production was stimulated (Fig. 7b,d). As Ca2+ reversed the Cd2+-induced inhibition of production in vitro (Fig. 4), we investigated the effect of Ca2+ and Cd2+ in vivo. When both Cd2+ and Ca2+ were present, the inhibitory effect of Cd2+ on production was overridden (Fig. 7a,c). As antimycin A and Cd2+ stimulated ROS production in isolated mitochondria, we investigated the effect of antimycin A on H2O2 production. In the presence of antimycin A the extracellular concentration of H2O2 was increased (Fig. 7b) while production was slightly inhibited (c. 15%; data not shown). The increase in H2O2 production by antimycin A shows that the H2O2 originating from mitochondrial electron transport can travel long distances out of the cell to the apoplast and/or out of the roots.
In absolute terms, the effects of Cd2+ on and H2O2 production in cucumber and soybean roots are different but the qualitative effects are the same: production is inhibited and H2O2 production is stimulated. In soybean the rate of production measured by XTT was approximately four to five times lower than in cucumber roots. This may be a result of restricted diffusion of XTT and Cd2+ into and out of the apoplast in soybean compared with cucumber roots (see next section). In the absence of Cd2+ the rate of H2O2 production in soybean was higher than in cucumber, while the effects of Cd2+ on both and H2O2 production were lower in soybean than in cucumber.
Effect of CdCl2 on •OH production in intact roots
It has been reported previously that ROS, especially •OH, are involved in extension growth (Schopfer et al., 2002). As •OH production in roots has been shown to require both and H2O2 (Liszkay et al., 2004) the question arises as to whether Cd2+ stimulates or inhibits •OH production. We measured •OH production in roots in response to Cd2+ and Ca2+ by spin trapping EPR spectroscopy using EtOH/POBN as a spin trap (Janzen et al., 1978). After 1 h of incubation without added cations a characteristic EPR spectrum of the stable nitroxide radical was detected (Fig. 8, spectrum 1). The presence of 5 mM Ca2+ in the reaction medium increased the signal size by 30% (Fig. 8, spectrum 2). The presence of 300 µM Cd2+ decreased the signal size drastically (Fig. 8, spectrum 3). In the presence of both Ca2+ (5 mM) and Cd2+ (300 µM) the signal size was restored to the same level as seen in the presence of only Ca2+ (Fig. 8, spectrum 4).
In Fig. 8, •OH generation was measured with cucumber and soybean roots. Qualitatively the effect of the cations was the same in both species. However, quantitatively the effect differed significantly, being more pronounced in cucumber. This was probably caused by lower penetration of the spin traps and the cations into the soybean roots. (We observed significant differences in the amount of •OH formation in roots from several plant species. The largest signals compared to fresh weights were obtained from A. thaliana and cucumber, followed by cress (Lepidium sativum), pea (Pisum sativum), soybean and maize (Zea mays). The access of the spin trap to the apoplast seems to be more restricted in species with more robust roots.)
Comparing the data in Figs 7 and 8, it can be seen that the effects of Cd2+ and Ca2+ on •OH resemble those observed for production. production by the Cd2+-sensitive NAD(P)H oxidase seems to be required for this process. H2O2 alone, in the absence of , is not sufficient to generate •OH in the apoplast as H2O2 production increases after the addition of Cd2+ (Fig. 7b,d) while •OH production decreases (Fig. 8).
In the present study, Cd2+-dependent ROS production at the level of the plasma membrane and mitochondrial electron transport chain was investigated in vitro and in vivo. and H2O2 production was followed after exposure of isolated plasma membranes, mitochondria or intact roots to Cd2+. production by isolated plasma membranes was inhibited by Cd2+ competitively to Ca2+ (Fig. 4) while the addition of Cd2+ stimulated and H2O2 generation by mitochondria (Figs 5, 6). Upon exposure of intact roots to Cd2+, formation was inhibited (Fig. 7a,c) while H2O2 formation increased (Fig. 7b,d). Therefore, we conclude that the Cd2+-induced generation of ROS, namely H2O2, originates from inside the root cells, mainly from mitochondrial electron transport.
In our work with isolated plasma membranes, at least part of the was produced by a plasma membrane NADPH oxidase as this activity was inhibited by low concentrations of DPI (Ki = 1.5 µM) whereas KCN, an inhibitor of peroxidases, had no effect on the activity (Figs 2, 3). Potential -producing enzymes in the plasma membrane are NOXs (Torres & Dangl, 2005) and quinone oxidoreductases (Schopfer et al., 2008). Furthermore, the activity was stimulated by Ca2+ (Table 2) and inhibited by Cd2+ in a competitive manner (Fig. 4). The inhibition by low concentrations of DPI and the antagonistic effects of Ca2+ and Cd2+ indicate that the NADPH oxidase activity investigated in the present study can be attributed to a NOX activity. Sagi & Fluhr (2001) observed a twofold increase of production in isolated tobacco and tomato plasma membranes in response to millimolar Ca2+ concentrations, while in the present study production was only slightly (up to 30%) increased by 1–5 mM Ca2+. The maximum production activity obtained in soybean plasma membranes was similar to that reported previously for tobacco plasma membranes (Sagi & Fluhr, 2001). In contrast to the data reported by Sagi & Fluhr (2001), after depletion of the samples of cations using Chelex 100, the activity could be restored to the control level even with micromolar Ca2+ concentrations (Table 2). The human EF-hand motif-containing NOX5 was previously shown to be activated at low micromolar Ca2+ concentrations (Banfi et al., 2004), in accordance with the data presented here. In the present work, Mg2+ had similar effects to Ca2+ on production. Binding of Mg2+ and manganese (Mn2+) to the EF-hand motif of 13-4-4 proteins was reported previously (Athwal & Huber, 2002). It remains to be investigated whether these cations bind to the EF-hand motifs of NADPH oxidase and further whether the effect of Mg2+ and Mn2+ is restricted to plants, as the EF-hand motifs of the human NOX5 do not bind Mg2+ (Banfi et al., 2004).
It has been demonstrated that Cd2+ increases ROS formation in mitochondria from animals (Wang et al., 2004). Here, Cd2+ and antimycin A stimulated and H2O2 production in plant mitochondria respiring on succinate (Figs 5, 6). In animal tissue mitochondria are thought to be among the major targets of Cd2+ toxicity (Martel et al., 1990). Cd2+ and zinc (Zn2+) are well-known inhibitors of electron transport in mitochondria (Skulachev et al., 1967). Cd2+ blocks electron transfer between semiquinone and cytochrome b in respiratory complex III (Miccadei & Floridi, 1993) and causes the accumulation of semiquinone radicals, leading to the formation of (Wang et al., 2004). The binding site of Cd2+ in complex III is likely to be the same as that of Zn2+, that is, blocking a protonable group which is thought to be associated with deprotonation reactions of the quinol oxidation site in complex III (Link & von Jagow, 1995; Giachini et al., 2007).
We attribute most of the production measured in vivo to NOX activity because it was inhibited by Cd2+ and this inhibition was reversed by Ca2+ both in vitro and in vivo (Figs 4, 7a,c, 8). We can exclude the possibility that peroxidases are the source of production, because we found that isolated horseradish peroxidase was insensitive to Cd2+ at the concentration used here (data not shown). It has been proposed previously that the production of •OH involved in extension growth is initiated by NAD(P)H oxidase-catalyzed formation of at the plasma membrane (Schopfer et al., 2002; Liszkay et al., 2004) and involves peroxidases, which are abundant in the cell wall (Dunand et al., 2007). In the presence of the catalytic heme in peroxidases is converted into its so-called ‘compound III’. In this state peroxidases transform H2O2 into −OH and •OH (Chen & Schopfer, 1999; Schopfer et al., 2002). Thus, in the presence of H2O2, a measure of the production of •OH (Fig. 8) reflects the amount of produced in the apoplast in vivo. The , generated at the plasma membrane, is needed for the activation of the peroxidase to compound III. seems to be necessary in more than just catalytic amounts, because in the presence of Cd2+ production (Fig. 7a,c) and •OH production (Fig. 8) are strongly inhibited while H2O2 increases with time (Fig. 7b,d).
Cd2+-induced H2O2 production in vivo is likely to originate largely from mitochondrial electron transfer as Cd2+ increased ROS production in isolated mitochondria (Figs 5, 6). According to Varga et al. (2002), in cucumber, c. 70% of the extracellular Cd2+ can reach the cytoplasm, implying that a high percentage of the added Cd2+ will reach the mitochondria. H2O2 produced inside the mitochondria seems to be able to diffuse out of the cells as in the presence of antimycin A the amount of extracellular H2O2 was increased (Fig. 7b). In addition to H2O2 of mitochondrial origin, a basal level of peroxisomal H2O2 production (Corpas et al., 2001; del Río et al., 2002; Hänsch et al., 2006; Nyathi & Baker, 2006) may affect the total amount of H2O2 produced by intact roots.
Increased H2O2 production during short-term exposure of tobacco cell cultures to Cd2+ has been reported previously (Olmos et al., 2003; Garnier et al., 2006; Ortega-Villasante et al., 2007). Using video microscopy Ortega-Villasante et al. (2007) recently observed Cd2+-induced H2O2 generation inside and outside of roots, in agreement with our present conclusions. Olmos et al. (2003) and Garnier et al. (2006), however, concluded that NOX activity is responsible for Cd2+-induced H2O2 production, at least during the initial phase of the oxidative burst. Using tobacco cell cultures Olmos et al. (2003) measured Cd2+-induced H2O2 generation which was strongly inhibited by DPI and interpreted this as evidence for the involvement of a NOX-like enzyme in ROS production. Addition of DPI abolishes the Cd2+-induced generation of H2O2 in vivo and this is often taken as a sign of the involvement of a plasma membrane NOX-like enzyme in the Cd2+-induced oxidative burst (Olmos et al., 2003; Romero-Puertas et al., 2004; Rodriguez-Serrano et al., 2006). DPI is often regarded as a specific inhibitor of NOX at low micromolar concentrations (< 10 µM; Doussiere & Vignais, 1992) although it also inhibits complex I of the respiratory chain at equally low concentrations (Ragan & Bloxham, 1977). Therefore, DPI also inhibits mitochondrial ROS production in vivo. In addition, DPI inhibits other flavin-containing enzymes at higher concentrations (≥ 10 µM; Doussiere et al., 1992) and peroxidases at even higher concentrations (Frahry & Schopfer, 1998). In our hands, DPI inhibited Cd2+-induced H2O2 production in roots (data not shown) but because of the different sites of action of this inhibitor it seems to us to be impossible to draw conclusions from this fact.
Garnier et al. (2006) identified different phases of Cd2+-induced ROS production using tobacco cell cultures. A transient first wave was linked to the activity of plasma membrane NOX and a second longer lasting wave to production in mitochondria, and a third wave was characterized by lipid hydroperoxide accumulation concomitant with cell death. The first wave was completely abolished in an antisense construct of NtrbohD, implying that the NOXs of the plasma membrane were responsible for the first wave. We measured Cd2+-induced H2O2 production using mutants of atrbohC and atrbohD defective in NOX activity (Torres et al., 2005). ROS production is decreased in atrbohC (Foreman et al., 2003; Renew et al., 2005). In the atrbohC mutant, Cd2+ stimulated H2O2 production to the same extent as in wild type (Supplementary Material Fig. S1), indicating that the Cd2+-dependent H2O2 generation was not impaired in this mutant and was therefore not linked to NOX activity. In atrbohD, Cd2+-induced H2O2 production was abolished, in agreement with Garnier et al. (2006). This observation is in contradiction to the results presented here, which clearly indicate that Cd2+ induces mitochrondrial ROS production. This seems to indicate that some other processes involved in Cd2+ uptake or toxicity are impaired in the atrbohD mutant and in NtrbohD antisense strains (Garnier et al., 2006). Sagi et al. (2004) investigated Rboh antisense lines in tomato and reported major pleiotrophic effects on the phenotype of the plant and its reproductive organs. A total of 384 expressed sequence tags (ESTs) were down-regulated and 485 ESTs were up-regulated in the Rboh antisense line. According to Sagi et al. (2004), Rbohs play a role in redox-related cellular activities and they affect the level of the expression of ROS-dependent genes, signal transduction and developmental processes. Therefore, it seems to be difficult to draw conclusions from Rboh antisense lines. Further investigations must be performed on antisense lines to investigate in detail whether the lack of the Cd2+-induced ROS generation is attributable to alterations in the amount of ROS-detoxifying enzymes or to alteration of other metabolic processes.
A short time of exposure to the metal is crucial to study the primary effect of Cd2+ toxicity. After a long-term exposure of plants to Cd2+ many physiological processes are affected. Higher expression levels of Rbohs have been reported after long-term exposure of plants to Cd2+ (Van Belleghem, 2007) and chelation of Cd2+ takes place in the cytosol (Schützendübel & Polle, 2002). Furthermore, Cd2+ increases the concentration of abscisic acid (ABA) (Hsu & Kao, 2003), which is known to lead to the activation of the NOX (Kwak et al., 2003). Long-term exposure of plants to Cd2+ has been reported to activate and H2O2 production both in the cytoplasm and in the plasma membrane, where NOX is thought to be responsible for the activity (Romero-Puertas et al., 2004; Rodriguez-Serrano et al., 2006).
One problem limiting progress in understanding the biological role of ROS is that it is difficult to attribute an oxidative burst to a well-defined ROS-producing reaction. A new view of the primary sources of ROS produced in response to signaling is developing. Ashtamker et al. (2007) using tobacco cell cultures reported a first cryptogein-induced H2O2 burst in mitochondria, the endoplasmic reticulum and the nucleus followed by a second burst at the plasma membrane after a few seconds. In a mammalian system, mitochondrial H2O2 production in response to a stress signal partially mediated the activation of Nox1 after a lag time of a few hours (Lee et al., 2006). The availability of inhibitors of enzymatic reactions producing ROS in specific cell compartments, as shown here for Cd2+, may elucidate the complex interaction of ROS-producing reactions in vivo.
We would like to thank Peter Schopfer, Universität Freiburg, and Bill Rutherford, CEA Saclay, for stimulating discussions and critical reading of the manuscript, Katharina Kienzler, Universität Freiburg, for excellent technical assistance and Francis Haraux, CEA Saclay, and Sébastien Thomine, CNRS Gif sur Yvette, for stimulating scientific discussions. We are grateful to Wolfgang Michalke, Universität Freiburg, Ulrich Schulte, Universität Düsseldorf, and Françoise Simon-Plas, Laboratoire de Phytopharmacie Dijon, for providing us with antibodies against P-ATPase, Rieske and NtrbohD proteins, respectively.