Increased intracellular H2O2 availability preferentially drives glutathione accumulation in vacuoles and chloroplasts



    1. Institut de Biologie des Plantes, UMR8618 CNRS, Bâtiment 630, Université de Paris sud 11, 91405 Orsay cedex, France
    Search for more papers by this author
    • Present address: Department of Plant Systems Biology, Flanders Institute for Biotechnology and Department of Plant Biotechnology and Genetics, Ghent University, 9052 Gent, Belgium.


    1. Centre Commun de Microscopie Electronique, UMR 8080 CNRS, Bâtiment 440, Université de Paris sud 11, 91405 Orsay cedex, France
    Search for more papers by this author

    1. University of Graz, Institute of Plant Sciences, Schubertstrasse 51, 8010 Graz, Austria
    Search for more papers by this author

    Corresponding author
    1. Institut de Biologie des Plantes, UMR8618 CNRS, Bâtiment 630, Université de Paris sud 11, 91405 Orsay cedex, France
    Search for more papers by this author

G. Noctor. Fax: +33 1 69 15 34 24; e-mail:


One biochemical response to increased H2O2 availability is the accumulation of glutathione disulphide (GSSG), the disulphide form of the key redox buffer glutathione. It remains unclear how this potentially important oxidative stress response impacts on the different sub-cellular glutathione pools. We addressed this question by using two independent in situ glutathione labelling techniques in Arabidopsis wild type (Col-0) and the GSSG-accumulating cat2 mutant. A comparison of in situ labelling with monochlorobimane (MCB) and in vitro labelling with monobromobimane (MBB) revealed that, whereas in situ labelling of Col-0 leaf glutathione was complete within 2 h incubation, about 50% of leaf glutathione remained inaccessible to MCB in cat2. High-performance liquid chromatography (HPLC) and enzymatic assays showed that this correlated tightly with the glutathione redox state, pointing to significant in vivo pools of GSSG in cat2 that were unavailable for MCB labelling. Immunogold labelling of leaf sections to estimate sub-cellular glutathione distribution showed that the accumulated GSSG in cat2 was associated with only a minor increase in cytosolic glutathione but with a 3- and 10-fold increase in plastid and vacuolar pools, respectively. The data are used to estimate compartment-specific glutathione concentrations under optimal and oxidative stress conditions, and the implications for redox homeostasis and signalling are discussed.


Glutathione1 is the major non-protein thiol in most organisms and is a multifunctional metabolite that is involved in various defence reactions and in ensuring appropriate intracellular redox homeostasis. Glutathione concentrations are implicated in the regulation of meristem function and in environmental responses (Vernoux et al. 2000; Ball et al. 2004; Gomez et al. 2004; Frendo et al. 2005; Parisy et al. 2007; Reichheld et al. 2007; Frottin et al. 2009; Bashandy et al. 2010). Various stress conditions drive characteristic changes in the intracellular amount and redox state of glutathione. These dynamic changes are considered to be part of mechanisms acting to maintain redox homeostasis and could also be significant in transducing environmental signals (Foyer et al. 1997; May et al. 1998; Foyer & Noctor 2005; Noctor 2006). Because modifications in the whole tissue glutathione status can be readily and accurately quantified, they can be taken as a marker of the degree of intracellular oxidative stress, also referred to as ‘disulphide stress’.

The close link between increased intracellular H2O2 and changes in glutathione status is underscored by observations in several plant species deficient in catalase (Smith et al. 1985; Willekens et al. 1997; Queval et al. 2007). Increased oxidation and accumulation of glutathione in these systems resemble effects that can occur during exposure of plants to stresses such as ozone, pathogen attack and chilling (e.g. Edwards, Blount & Dixon 1991; Vanacker, Carver & Foyer 2000; Bick et al. 2001; Gomez et al. 2004). Previously, we reported that the catalase-deficient Arabidopsis mutant cat2 shows conditional oxidative stress dependent on photorespiration (Queval et al. 2007). As in barley and tobacco deficient in the major leaf catalase isoform (Smith et al. 1985; Willekens et al. 1997), oxidative stress in cat2 is conditional on H2O2 produced in the peroxisomes through the glycolate oxidase reaction (Queval et al. 2007). When the mutant is growing in air, where photorespiratory H2O2 production is active, leaf glutathione becomes oxidized and accumulates to several-fold wild-type levels, whereas at high CO2, the leaf glutathione contents and redox state in cat2 are similar to wild type (Queval et al. 2007, 2009). The modification of glutathione pools in cat2 that occur in air can be accompanied by the activation of responses to certain pathogens, and some of these responses are modified by secondary mutations in NADP-glutathione systems (Chaouch et al. 2010; Mhamdi et al. 2010a,b). The cat2 mutant is therefore a useful model in which to analyse redox control mechanisms that are triggered by increases in H2O2 and that may be important in stress responses (Mhamdi et al. 2010c). One outstanding question is whether these mechanisms involve changes in the distribution of glutathione between different sub-cellular compartments.

Measurements of glutathione in isolated organelles are potentially complicated by exchange and leakage during the purification procedures (Noctor et al. 2002; Krueger et al. 2009). Non-aqueous fractionation was used to separate chloroplastic and extra-chloroplastic compartments in catalase-deficient barley, and H2O2-driven increases in glutathione were reported to be associated with both fractions (Smith et al. 1985). More recently, other techniques have been developed that are able to label glutathione in situ. These include monochlorobimane (MCB) labelling, which can be used to distinguish between different pools on the basis of kinetics (Meyer, May & Fricker 2001), and immunocytochemical analysis of glutathione pools in different sub-cellular compartments using a specific antibody (Zechmann, Mauch & Müller 2008; Zechmann & Müller 2010). In this study, we have used these two in situ labelling techniques, together with analysis of global leaf glutathione concentrations, to investigate whether oxidative stress in the cat2 mutant drives changes in the sub-cellular distribution of glutathione. The data provide further evidence that the chloroplast is an important site of glutathione accumulation during oxidative stress. Most notably, our analysis reveals that glutathione accumulates significantly in the vacuole in cat2, very probably as glutathione disulphide (GSSG). From the data obtained on Col-0 and cat2, we provide estimates of glutathione concentrations in different sub-cellular compartments in the absence and presence of oxidative stress.


Plant material and growth

Two allelic lines of T-DNA knockout mutants in the CATALASE2 (CAT2) coding sequence characterized by Queval et al. (2007) were used. The cat2-1 and cat2-2 lines were named as in Queval et al. (2009). Seeds of cat2 and Col-0 were sown and maintained at a CO2 concentration of 3000 µL L−1 as described in Queval et al. (2009). Nutrient solution was provided every 2 d. After 5 weeks, CO2 enrichment was stopped and plants were grown in air for 4 d.

Bimane labelling and analysis of glutathione

Mixed leaves from two to three plants were vacuum infiltrated twice for 5 min in 25 mm K2HPO4 buffer (pH 5.2) containing 500 µm MCB. A first leaf sample [approximately 50 mg fresh weight (FW)] was rapidly blotted dry and immediately frozen in liquid nitrogen. The remaining leaves were incubated in buffer at 20 °C in the dark for 2 h (unless otherwise stated) then were harvested as above. Labelled and unlabelled thiols were extracted and the solutions neutralized as in Queval & Noctor (2007). For each extract, the neutralized supernatant was separated into three aliquots of 0.2 mL, and these were treated as follows (Fig. 1). To allow detection of in situ labelled thiols, the first aliquot was added to 0.1 mL of 0.2 m Ches (pH 9), 40 µL water and 0.66 mL 10% (v/v) acetic acid (Fig. 1, treatment 1). To enable additional bimane labelling of thiols that were not labelled with MCB in situ, the second aliquot was incubated in 0.1 mL of 0.2 m Ches (pH 9), 20 µL water and 20 µL of 30 mm monobromobimane (MBB) for 15 min in the dark, then the reaction was stopped by the addition of 0.66 mL of 10% (v/v) acetic acid (Fig. 1, treatment 2). The third aliquot was treated like the second, except that an additional reduction step was performed to convert disulphides to thiols. For this, aliquot 3 was incubated in 0.1 mL of 0.2 m Ches (pH 9) and 20 µL of 10 mm dithiothreitol (DTT) for 30 min in the dark; thiols were derivatized by incubation for 15 min after addition of 20 µL of 30 mm MBB, then the reaction was stopped by the addition of 0.66 mL of 10% (v/v) acetic acid (Fig. 1, treatment 3). All samples were filtered into high-performance liquid chromatography (HPLC) vials, and bimane derivatives were separated and quantified by fluorescence detection as described in Queval & Noctor (2007). An enzymatic assay of total glutathione and GSSG was performed using the glutathione reductase (GR) recycling assay on a plate reader, as previously described (Queval & Noctor 2007). GSSG was distinguished from total glutathione by treatment of sample aliquots with 2-vinylpyridine to complex GSH.

Figure 1.

Scheme of the experimental protocol for in situ and in vitro bimane labelling. For explanation, see text. DTT, dithiothreitol; GSH, reduced glutathione; GSSG, glutathione disulphide; MBB, monobromobimane; MCB, monochlorobimane; HPLC, high-performance liquid chromatography.

Cytohistochemical analysis

Sample preparation for cytohistochemical investigations was performed on the wild-type Col-0 and cat2-1. All steps were performed at 4 °C with roller agitation. Small leaf samples from two to three plants were cut into pieces of about 1 mm2, vacuum infiltrated three times for 10 min and left overnight in a 0.1 m cacodylate buffer (pH 7.3) containing 0.5% glutaraldehyde, 4% paraformaldehyde, 2 mm CaCl2, 2% sucrose and 0.1% Brij 35. Samples were rinsed four times in the same buffer, four times in a buffer containing 0.1 m glycine and four times more in the original buffer (15 min each time). Samples were then dehydrated in increasing concentrations of ethanol (10% for 15 min, 20% for 15 min, 30% for 20 min, 50% for 30 min, 70% for 1 h and 90% for 1 h, one time each) and were infiltrated with an increasing content of LR-White resin (Sigma, Saint Quentin Fallavier, France) (33%, one time; 50%, one time; 67%, two times) mixed with pure ethanol for at least 4 h per step. Samples were embedded in pure LR-White resin four times for a minimum of 8 h per step and were polymerized in LR-White resin with accelerator (Sigma) at −20 °C and under ultraviolet (UV) for at least 50 h in small plastic capsules under anaerobic conditions.

Immunolocalization of glutathione was performed according to Zechmann & Müller (2010). Briefly, ultrathin sections (80 nm) of the samples were blocked with 2% bovine serum albumin (BSA) in phosphate buffered saline (PBS, pH 7.2) and then were treated with the primary antibody against GSH (anti-GSH rabbit polyclonal IgG; Millipore Corporation, Billerica, MA, USA) diluted 1:50 in PBS containing 1% goat serum for 2 h at room temperature (RT). After a short rinse, PBS samples were incubated with a 10 nm gold-conjugated secondary antibody (goat anti-rabbit IgG; British BioCell International, Cardiff, UK; diluted 1:50 in PBS for 90 min at RT. After a short wash in PBS, distilled water labelled grids were either immediately observed in a Philips CM10 transmission electron microscope or post-stained with uranyl-acetate (15 s).

The specificity of the immunogold labelling procedure was tested by several negative controls. Negative controls were treated either with (1) pre-immune serum instead of the primary antibody, (2) gold-conjugated secondary antibody (goat anti-rabbit IgG) without the primary antibody, (3) non-specific secondary antibody (goat anti-mouse IgG) and (4) primary antibodies pre-adsorbed with an excess of GSH or GSSG for 2 h at RT prior to labelling of the sections. Available information suggests that the primary antibody binds to the free glycine carboxyl and the non-SH side of the backbone, and to the free γ-glutamyl carboxyl group. The binding site does not include the cysteine SH group or the glutamyl amino group as the latter is conjugated with glutaraldehyde. It is therefore predicted that this antibody should not show a marked difference in the affinity of its binding to GSH and GSSG.

Micrographs of randomly photographed immunogold-labelled sections were digitized, and gold particles were counted automatically using the software package Cell D with the particle analysis tool (Olympus, Life and Material Science Europa GmbH, Hamburg, Germany). For statistical evaluation, at least four different samples were examined. A minimum of 20 (peroxisomes and vacuoles) to 60 (other cell structures) sectioned cell structures of at least 15 different cells were analysed for gold particle density per sample. The obtained data were statistically evaluated using Statistica (Stat-Soft, Tulsa, OK, USA).

Data processing and calculations

To calculate sub-cellular concentrations of glutathione in the mesophyll (Table 2), the mean number of gold particles detected in each compartment per unit leaf area was divided by the mean total number of gold particles to give the fractional contribution of each compartment to the overall glutathione pool. For the wild-type, these values were then converted to quantities of glutathione in each compartment by using a measured global leaf value of 326 nmol g−1 FW (Fig. 5). Concentrations were then calculated by estimating the percentage contribution of each sub-cellular compartment to the total mesophyll cell volume using an approach similar to those previously described in other species (Winter, Robinson & Heldt 1994; Nadwodnik & Lohaus 2008). Images of 50 randomly chosen cells (each on different sections) from eight different plant samples of Col-0 and cat2 were taken with the transmission electron microscope with a magnification of 1650×. The electron imaging film (Maco EM-film EMS, 6.5 × 9.0 cm, resolving power of 300 lines/mm; obtained from Maco Photo Products, Hans O. Mahn & Co. KG, Stapelfeld, Germany) was developed with a high-contrast developer and the negatives were digitized as tif files with an Epson Scanner (Epson Perfection Pro V750; Epson Deutschland GmbH, Klosterneuburg, Austria) at a final resolution of 1200 × 1200 dpi. Measurements of the different cell compartments were performed on digital images with the software programme Cell D with the particle analysis tool (Olympus Soft Imaging Solutions GmbH, Münster, Germany). The percentage volumes for each compartment in Col-0 were the following: cytosol, 6.3; mitochondria, 0.5; nuclei, 0.5; peroxisomes, 0.2; plastids, 12.3; and vacuole, 80.2. From these values, sub-cellular volumes were calculated per gram FW based on a mesophyll volume per leaf mass of 773 µL g−1 FW (Winter et al. 1994). Concentrations in cat2 were obtained as follows. The global content of glutathione was calculated as (gold labelling per cell in cat2/gold labelling per cell in wild type) multiplied by the global leaf content in wild type determined in parallel samples by enzymatic assay (326 nmol g−1 FW). This value was then used to derive concentrations based on the fractional gold labelling in each compartment and the percentage sub-cellular volumes, which in cat2 were determined as cytosol, 6.7; mitochondria, 0.5; nuclei, 0.5; peroxisomes, 0.2; plastids, 13.3; and vacuole, 78.8.

Table 2.  Estimates of sub-cellular glutathione concentrations from immunogold labelling densities
nmol g−1 FW% cellular contentConcentration (mm)
  1. Wild-type concentrations were based on a global leaf GSH + GSSG content of 326 nmol g−1 FW, measured in vitro in parallel samples of the leaves used for immunolabelling. Leaf contents in cat2 were estimated by multiplying this value by the increase in total gold labelling in cat2 relative to Col-0. This procedure gave a value of 743 nmol g−1 FW in cat2, which was close to the value measured in vitro on parallel leaf samples (706 nmol g−1 FW). The amount of glutathione in each compartment (nmol g−1 FW, left column) was obtained by multiplying 326 (Col-0) or 743 (cat2) by the measured fractional contribution of each compartment to the overall gold label. From these values, concentrations were calculated using sub-cellular volumes estimated in leaf sections of each genotype (see Materials and Methods). Similar results were obtained in two independent experiments using different batches of plants.

  2. GSSG, glutathione disulphide; FW, fresh weight.

Figure 5.

Compartment-specific quantification of gold labelling in Col-0 and cat2. (a) Comparison of relative increase in glutathione measured by in situ immunogold labelling and by enzymatic assay of extracts. Total gold labelling (left) was calculated by multiplying labelling density in each compartment (shown in part b) by the fractional contribution of each compartment to mesophyll cell volume and then by summing the values for all compartments (values are divided by 100 for ease of y-axis labelling). (b) Values are the number of gold particles bound to GSH per unit area in different cell structures of leaf cells of Col-0 (black columns) and the cat2 mutant (white columns). Data are means ± SE of 20 (peroxisomes) and 60 (other compartments). Significant differences between Col-0 and cat2 were calculated with the Mann–Whitney U-test at *P < 0.05, or ***P < 0.001. P > 0.05 was considered as not significant (ns); FW, fresh weight.


In situ and in vitro fluorescence labelling of glutathione

As a first step to analysing possible oxidative stress-induced differences in glutathione compartmentation, we compared in situ and in vitro labelling of the thiol group with the fluorescent bimane group. In vitro labelling of thiols for HPLC analysis commonly uses MBB (Fahey et al. 1981). In this reaction, thiolate anions that are favoured at alkaline pH rapidly displace the MBB bromine atom to form a stable fluorescent glutathione–bimane (GS–bimane) derivative (Kosower & Kosower 1987). Compared to MBB, the chemical reaction of MCB with GSH is slow and is accelerated by glutathione S-transferases (GSTs). Consequently, only some of the different cellular pools are rapidly labelled by MCB in situ (Meyer et al. 2001). To compare MCB-available glutathione pools in Col-0 and cat2, we used the protocol shown in Fig. 1.

Preliminary experiments with Col-0 revealed that a large proportion of the total leaf glutathione could be labelled within 15 min incubation, including vacuum infiltration time, with MCB followed by immediate freezing of samples (Fig. 1). This reflects rapid labelling of available GSH through GS–bimane conjugation catalysed by GSTs (Meyer et al. 2001), which are abundant in the cytosol but are also found in other compartments (Dixon et al. 2009). Subsequent incubations of 60–120 min were required for complete labelling, that is, to reach levels that were not further increased by subsequent in vitro labelling of extract aliquots with MBB. This time-dependent effect can be attributed to the slow labelling of GSH that is present in less available compartments or that is present as GSSG in vivo and is converted to GSH by GR as GSH is removed by conjugation (Meyer et al. 2001). Accordingly, when GSH was quantified in extracts of Col-0 leaf material frozen directly after vacuum infiltration (0 h), the three treatments shown in Fig. 1 produced different values. Both for plants growing in air and at high CO2, in situ MCB treatment alone (treatment 1) labelled about 60–70% of the glutathione that could be recovered by subsequent in vitro incubation with MBB (treatment 3; Fig. 2; Col-0, 0 h). For Col-0 leaves, treatment 2, in which in situ MCB and in vitro MBB labelling were combined without an in vitro DTT treatment to reduce GSSG present in the extract, produced similar values to treatment 3 (Fig. 2). This is consistent with the highly reduced state of glutathione in wild-type plants in the absence of stress. When the three treatments were applied to leaf tissue incubated for 2 h following the vacuum infiltration of MCB, subsequent MBB treatment had no further effect, showing that this time was sufficient to allow complete labelling of glutathione in situ.

Figure 2.

Comparison of bimane labelling of glutathione in Col-0 and cat2 leaves. The treatment numbers at the bottom correspond to the procedures shown in Fig. 1 on leaf samples treated immediately after vacuum infiltration with monochlorobimane (MCB) (0 h) or following a further 2 h incubation (2 h). Data are means ± SE of two independent samples and are expressed in GSH equivalents. The results show a representative example of several experiments. For a statistical analysis of data from six experiments performed on plants growing in air, see Table 1.

When grown at high CO2, where photorespiratory H2O2 production and oxidative stress in cat2 are suppressed, labelling of both allelic cat2 lines produced a very similar pattern to Col-0. About two-thirds of the glutathione pool was labelled in situ by the infiltration alone, while labelling was complete within 2 h further incubation (Fig. 2; cat2, high CO2; compare treatments 1 and 3 at 0 and 2 h). As for Col-0 in both growth conditions, treatments 2 and 3 produced similar results for cat2 at high CO2, consistent with the highly reduced glutathione pool in these mutants when photorespiratory H2O2 production is not active (Queval et al. 2007). Very different effects were observed when H2O2-triggered glutathione accumulation was activated in cat2 by transferring plants from high CO2 to air. In this condition, total glutathione (detected by treatment 3) increased to about four times that detected in Col-0 (Fig. 2, right). Compared to Col-0, a much smaller fraction of this glutathione could be labelled in situ (0 h, treatment 1). This proportion was somewhat increased by subsequent in vitro treatment with MBB (0 h, treatment 2) or by subsequent incubation prior to sampling (MCB, 2 h). However, at both 0 and 2 h, the glutathione that could be detected by treatment 2 did not exceed 50% of that detectable using treatment 3. Thus, unlike Col-0 and cat2 at high CO2, an in vitro treatment with DTT was absolutely required to label a large part of leaf glutathione in cat2 exposed to air.

The fraction of glutathione detected in situ in the three genotypes at the two time points can be calculated as contents obtained using treatment 1 divided by those obtained with treatment 3. The results of six experiments performed in Col-0 and cat2 lines grown in air are summarized in Table 1. When tissue was sampled immediately after the infiltration, 69% of the total pool was MCB available in Col-0 but only about 30% in cat2. Whereas leaf glutathione could be completely labelled by a further 2 h incubation of Col-0 leaves prior to sampling, about 50% of the glutathione in cat2 remained unavailable (Table 1). To investigate whether labelling was slower in cat2 because of the greater global leaf contents, prolonged incubation with MCB was performed. However, no further increase was observed between 2 and 8 h (data not shown).

Table 1.  Summary of percentage leaf glutathione available for in situ labelling with monochlorobimane (MCB) in Col-0 and cat2 mutants
Incubation time0 h2 h0 h2 h0 h2 h
  1. Incubated leaf material was frozen in liquid nitrogen immediately following vacuum infiltration (0 h) with MCB or after a further 2 h incubation with the reagent. The numbers give percentage glutathione contents relative to values detected after subsequent in vitro disulphide reduction and labelling with MBB [100 (data from treatment 1/data from treatment 3); Fig. 1]. Six independent experiments were performed, each involving duplicate leaf samples. Different letters indicate significant differences (P < 0.05) between the mean values. The two cat2 lines were significantly different from Col-0 at P < 0.00001 and were not significantly different from each other at P < 0.6. The two time points were significantly different from each other at P < 0.05 for all three genotypes.

  2. MBB, monobromobrimane.


Analysis of the leaf glutathione reduction state was performed by an in vitro enzymatic assay of extracts of parallel samples of Col-0 and cat2 lines growing in air. These measurements revealed that the percentage reduction states of glutathione (Fig. 3, black bars) were very similar to the percentage glutathione labelled in situ after 2 h incubation with MCB (Table 1). Both values were close to 100% for Col-0 and to 50% for both cat2 lines. The percentage glutathione that was available for bimane labelling in situ and then in vitro without a DTT treatment was quite close to the whole tissue glutathione reduction state in all three lines (Fig. 3; compare white and black bars). This last observation suggests that the MCB labelling procedure did not greatly alter the in vivo GSH : GSSG ratio within the timescale of the experiment, and therefore that much of the GSSG detected in cat2 remained unavailable for in situ conjugation with MCB.

Figure 3.

Percentage reduction states of global leaf glutathione pools in Col-0 and cat2 mutants (black bars) compared to the percentage glutathione available for bimane labelling in the absence of in vitro treatment with reductant (white bars). For 1, glutathione was assayed in neutralized acid extracts without or with 2-vinylyridine (VPD) pre-treatment, and the values show 100 [(total glutathione − 2 GSSG)/total glutathione]. For 2, labelled glutathione was measured by high-performance liquid chromatography (HPLC) with or without treatment of extracts with dithiothreitol (DTT) and the values show 100 [GSH peak after treatment 2/GSH peak after treatment 3]; Fig. 1). Data are means ± SE of three independent samples. GSSG, glutathione disulphide.

Immunolocalization of GSH

To investigate whether the unavailable glutathione pools in cat2 were linked to sub-cellular compartmentation, Col-0 and cat2 leaf sections were prepared from plants placed in air and labelled with an anti-glutathione antibody. As CAT2 is the major catalase isoform in Arabidopsis leaves, where it is required for metabolism of H2O2 produced in photorespiration, we focused on photosynthetic mesophyll cells. Because the glutathione status is very similar in cat2 and Col-0 grown at high CO2 (Fig. 2; Queval et al. 2007; Mhamdi et al. 2010a), the analysis was restricted to samples taken from plants growing in air. As previously reported for Arabidopsis (Zechmann et al. 2008), the label could be readily detected in Col-0 within chloroplasts, mitochondria, peroxisomes, nuclei and cytosol (Fig. 4). A label was also detected in these compartments in cat2. The most striking qualitative difference was the observation of gold particles in the vacuole in cat2 (Fig. 4b,c), a compartment that showed very little labelling in Col-0 (Fig. 4a). No labelling was evident in the cell wall, intrathylakoid space or mitochondrial intermembrane space in either genotype.

Figure 4.

Transmission electron micrographs showing the sub-cellular distribution of gold-labelled glutathione in leaf cells from Col-0 (a) and cat2 (b–e). Upper case letters within the frames indicate cell walls (CWs), mitochondria (M), nuclei (N), peroxisomes (Px), plastids (P) and starch grains (St). The arrow in panel (d) indicates thylakoid lumen, while those in panel (e) indicate the lumen of cristae in a mitochondrion. Bars = 1 µm.

A quantitative analysis of glutathione immunolabelling was performed. First, we compared total increases in glutathione labelling in cat2 detected by the antibody with those detected by plate-reader assays. If the antibody detects GSH and GSSG with equal sensitivity (see Materials and Methods), the increases in gold labelling will be less than increases in glutathione expressed as GSH equivalents. For instance, if all the extra glutathione that accumulates in cat2 is GSSG, and GSSG is very low in Col-0 (Fig. 3), then an increase in gold labelling of two- or threefold in cat2 will represent a three- and fivefold increase, respectively, in total glutathione expressed as GSH equivalents (GSH + 2GSSG). Thus, for the present comparison, the total glutathione detected in vitro by the enzymatic assay of extracts was expressed as GSH + GSSG. This comparison revealed a close correspondence between cat2/Col-0 total leaf glutathione ratios detected by immunostaining and by in vitro assay of homogenized leaf extracts (Fig. 5a). Quantification of labelling across numerous mesophyll sections revealed that the cat2 mutant had more intense labelling than Col-0 in the cytosol, though this increase was only about 20% (Fig. 5b). In contrast, labelling was increased relative to Col-0 by about threefold in the cat2 chloroplast and about 10-fold in the cat2 vacuole.

To analyse whether the in situ immunolabelling technique detects GS conjugates, analysis was performed on Col-0 incubated with or without MCB for 30 min prior to embedding. Compared to the control condition, gold particle density was reduced to background levels in cells from leaves that were incubated with MCB and no increase in vacuolar labelling was observed (Fig. 6).

Figure 6.

Effect of monochlorobimane (MCB)-dependent conjugation of glutathione on immunogold labelling. Transmission electron micrographs show the sub-cellular distribution of glutathione in leaf cells from Arabidopsis thaliana Col-0 plants after incubation with (a) or without MCB (b) for 30 min prior to embedding. Bars = 1 µm. CW, cell wall; P, plastids; M, mitochondria; N, nuclei; Px, peroxisomes; V, vacuoles.

We used the immunogold labelling data to estimate glutathione concentrations in the different compartments (for details, see Materials and Methods). In Col-0, concentrations in the vacuole were about 100-fold lower than in other intracellular compartments, where concentrations ranged from 1 to 6 mm (Table 2). In agreement with previous studies (Zechmann et al. 2008), the highest concentrations were detected in the mitochondria, while concentrations in plastids were significantly lower (Table 2). Oxidative stress in cat2 caused an increase in plastid and vacuolar concentrations to about 3.0 and 0.3 mm, while the cytosolic concentration increased from 3.5 to 4.4 mm. Nuclear and peroxisomal concentrations were similar to Col-0, while the mitochondrial concentration was slightly decreased (Table 2).


Plants deficient in catalase are well suited to the analysis of glutathione metabolism during oxidative stress. Catalase deficiency in barley, tobacco and Arabidopsis has been shown to drive photorespiration-dependent oxidation and accumulation of tissue glutathione (Smith et al. 1985; Willekens et al. 1997; Queval et al. 2007, 2009). These changes qualitatively mimic those that occur in response to stresses such as pathogens, ozone and cold (Vanacker et al. 2000; Bick et al. 2001; Gomez et al. 2004). Apart from one study in a barley mutant that analysed the distribution of the accumulated glutathione between chloroplastic and extra-chloroplastic fractions (Smith et al. 1985), little is known about how the sub-cellular compartmentation of glutathione changes in response to oxidative stress.

Leaf sub-cellular glutathione concentrations in wild-type plants

In wild-type Arabidopsis leaves, glutathione concentrations in the cytosol, nuclei and mitochondria were between 3 and 6 mm, with the highest concentrations found in mitochondria, as reported previously (Zechmann et al. 2008). Our estimates of sub-cellular glutathione concentrations are approximate though likely no more so than other values that have been generated using techniques that provide less sub-cellular resolution (i.e. concentrations measured in extracted organelles or estimated in situ using MCB labelling). The cytosolic concentration of about 3 mm obtained from our approach is in good agreement with data obtained by MCB labelling of cultured Arabidopsis cells (Meyer and Fricker 2002). Chloroplast GSH concentrations of around 1 mm are at the low end of previous estimates, but the finding that the plastids account for about 30% of total leaf cell glutathione (Table 2) is consistent with a previous study using non-aqueous fractionation of pea leaves (Klapheck, Latus & Bergmann 1987). A vacuolar glutathione concentration in the micromolar range is also consistent with earlier estimates (Rennenberg 1982). Recently, non-aqueous fractionation of Arabidopsis leaves was used to estimate glutathione concentrations in different compartments (Krueger et al. 2009). Three compartments were considered to be sufficiently resolved using this method, and glutathione concentrations were estimated to be 2.5–3.0 mm (plastid), 3.0–3.5 mm (cytosol) and 0.6–0.7 mm (vacuole). The higher plastidial and vacuolar concentrations estimated by this method might reflect effects of different nutrition regimes on global leaf glutathione contents, which were about twofold higher in the study of Krueger et al. (2009) than in the present analysis of wild-type plants.

H2O2-induced accumulation of glutathione involves pools that are not available for in situ labelling with MCB

Several factors may explain the unavailability of glutathione for MCB labelling in situ. Firstly, glutathione may be localized in compartments with low GST activity (Meyer et al. 2001). As MCB can react chemically with GSH, this should only slow the rate of labelling. This is likely one factor underlying the labelling kinetics of Col-0 leaves, where 100% of the MBB-sensitive glutathione can be labelled by MCB in situ within 2 h (Fig. 2), in agreement with previous observations on Arabidopsis cell cultures (Meyer et al. 2001). The difference in MCB labelling between 0 and 2 h, which was also observed in cat2 at high CO2, could therefore be due to differences in GST capacities between compartments. A second explanation of inaccessibility to MCB is that glutathione is present as GSSG and that these GSSG pools are not reduced to GSH during the time of the incubation. As only about 50% of the MBB-available pool could be labelled in cat2, even after prolonged incubation (up to 8 h), it is likely that a substantial part of the leaf glutathione pool in cat2 in air is composed of GSSG that accumulates at sites where conversion to glutathione does not occur during the period of incubation with MCB. This is consistent with the finding that the percentage GSH found as glutathione in cat2 did not change significantly within the time course of the incubation (Fig. 3) and that DTT treatment of extracts from MCB-treated leaves approximately doubled the detected glutathione pool in cat2 (Fig. 2). This in vitro reduction step allowed recovery of the high glutathione contents routinely measured in cat2 extracts using enzymatic techniques that do not distinguish between GSH and GSSG (e.g. Fig. 5a, right).

Oxidative stress drives a substantial increase in chloroplastic glutathione

A previous study of barley catalase-deficient plants reported a significant accumulation of GSSG in both chloroplast and extra-chloroplast compartments (Smith et al. 1985). Our results using immunolabelling in Arabidopsis are consistent with this earlier study. They suggest that chloroplast accumulation of GSSG contributes somewhat to the MCB-inaccessible pools of glutathione that are triggered by H2O2 produced in the peroxisomes. Isolated wheat chloroplasts can import radiolabelled GSH (Noctor et al. 2002), consistent with the viability of Arabidopsis engineered to synthesize glutathione exclusively in the cytosol (Pasternak et al. 2008). Recently, specific Arabidopsis chloroplast envelope glutathione transporters have been identified (Maughan et al. 2010). However, whether GSSG accumulation in the chloroplast results from import is not clear. It could also reflect oxidation of GSH already present within the chloroplast following entry of H2O2 or derived oxidants.

Whatever the mechanism by which GSSG accumulates in the chloroplast, it could have important consequences for redox regulation in this compartment. If all or most of the chloroplast-accumulated glutathione is GSSG, a conclusion supported by an analysis of the barley catalase mutant (Smith et al. 1985), the GSSG concentration would likely exceed 0.5 mm (Table 2). As well as other effects on chloroplast redox balance, this may be sufficient to contribute to protein S-glutathionylation reactions through a thiol-disulphide exchange mechanism. Protein S-glutathionylation may regulate photosynthetic and respiratory enzymes and provide an interface between glutathione and thioredoxin redox systems during oxidative stress conditions (Ito, Iwabuchi & Ogawa 2003; Dixon et al. 2005; Michelet et al. 2005; Zaffagnini et al. 2007; Holtgrefe et al. 2008). In terms of the regulation of glutathione accumulation, it is interesting to note that the first enzyme of the committed glutathione synthesis pathway, γ-glutamylcysteine synthetase (γ-ECS), is located in the plastid in Arabidopsis (Wachter et al. 2005). This enzyme is subject to feedback inhibition by glutathione with Ki values of approximately 1–2 mm (Jez, Cahoon & Chen 2004). The accumulation of glutathione in cat2 presumably involves de novo synthesis, consistent with 35S-sulphate labelling studies in barley catalase-deficient mutants (Smith et al. 1985). The threefold accumulation of chloroplast glutathione in cat2 shows that other factors are able to drive increased glutathione synthesis against the background of what should be more stringent feedback inhibition caused by increasing glutathione concentrations (Queval et al. 2009).

Potential significance of vacuolar accumulation of GSH in response to oxidative stress

Increased H2O2 availability in cat2 triggers induction of certain GSTs at the transcript level (Queval et al. 2007; Chaouch et al. 2010; Mhamdi et al. 2010a). These enzymes may have a peroxidase and/or conjugase function (Dixon et al. 2009). Conjugates formed by GSTs are notably transported into the vacuole. However, the analysis shown in Fig. 6 reveals that the antibody does not readily detect glutathione in S-conjugate form, at least for the GS–bimane conjugate. Additionally, it has been shown recently that cadmium treatment, which also leads to the formation of GS conjugates, resulted in strongly decreased glutathione labelling density and lack of labelling in vacuoles (Kolb et al. 2010). Thus, it is unlikely that the increased vacuolar signal in cat2 can be attributed to conjugates. Some of the tonoplast transporters of the multi-drug resistance-associated protein (MRP) type that import GS-conjugates can also transport GSSG (Lu et al. 1998), and studies in vitro have shown that GSSG can be taken up by isolated barley vacuoles (Tommasini et al. 1993). The present report provides in vivo evidence that vacuolar accumulation of glutathione is part of the response to oxidative stress. As GR is absent from the vacuole, GSSG accumulation in this compartment, which is estimated to account for 26% of the total leaf glutathione pool in cat2 (Table 2), probably makes a major contribution to the MCB-inaccessible leaf pool. As previously noted, sequestration of GSSG in the vacuole may function to maintain a reduced environment in the cytosol (Tommasini et al. 1993). In agreement with this notion, we suggest that the vacuolar glutathione signal observed in cat2 mainly, if not exclusively, reflects GSSG and that GSSG accumulation in this compartment acts to mitigate what could otherwise be an excessively positive shift in cytosolic glutathione redox potential. Given the observed accumulation of glutathione in the chloroplast, likely also mainly as GSSG, our data imply that the plant cell is configured to limit large swings in the cytosolic (and potentially nuclear) glutathione redox potentials.

While GSSG accumulation has been implicated in lesion formation and dormancy (Creissen et al. 1999; Kranner et al. 2006), phenotypic analysis of cat2 and derived double mutants reveals that leaves can tolerate marked increases in GSSG without engagement of cell death (Queval et al. 2007; Chaouch et al. 2010; Mhamdi et al. 2010a). Vacuolar accumulation of GSSG may provide a partial explanation of this observation. However, tissue GSSG accumulation is still likely to be a useful indicator of oxidative stress. Its accumulation in the vacuole would presumably reflect a process initiated by increased concentration in the cytosol. This conclusion is supported by the relatively small but significant increase in the cytosolic pool in cat2 (Fig. 5). Studies of GSSG transport into purified vacuoles reported an apparent KM value of 0.4 to 0.6 mm (Tommasini et al. 1993). Such concentrations are likely well above those that occur in wild-type plants in optimal conditions, where recent reports situate cytosolic glutathione redox potentials at below −300 mV (Meyer et al. 2007; Jubany-Mari et al. 2010). If, as the present and previous data suggest, the cytosolic glutathione concentration is around 3–4 mm, redox potentials below 300 mV imply that cytosolic GSSG must be in the nanomolar range. In this case, H2O2-triggered increases in cytosolic GSSG that are important in relative terms but minor in terms of absolute concentrations (e.g. from nanomolar to the micromolar range) would cause the glutathione redox potential to become significantly more positive, possibly acting to relay part of the oxidative signal while at the same driving transport of GSSG into the vacuole as a homeostatic signal-damping mechanism. It remains to be established whether specific transporters are important in transporting GSSG into the vacuole in response to oxidative stress. Likewise, the fate of the GSSG that accumulates there remains unclear. Vacuolar GSSG could be partly catabolized by γ-glutamyl transpeptidases (Grzam et al. 2006; Ohkamu-Ohtsu et al. 2007) or by carboxypeptidases (Wolf, Dietz & Schröder 1996). Finally, certain aspects of the cat2 phenotype, such as lesion formation, can be abolished by blocking salicylic acid (SA) synthesis or by supplying myo-inositol, and in both cases, these effects were linked to a less oxidized glutathione status (Chaouch & Noctor 2010; Chaouch et al. 2010). It could therefore be interesting to investigate the influence of SA and other molecules in determining the intracellular distribution of glutathione that accumulates in response to H2O2.


We thank the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis T-DNA insertion mutants and the Nottingham Arabidopsis Stock Centre, UK, for the supply of seed stocks. B.Z. is supported by the Austrian Science Fund (FWF, P20619). We thank Andreas Meyer (Heidelberg, Germany) for discussions and advice on MCB labelling.


  • 1

    GSH is used here to indicate the thiol (reduced) form of glutathione while GSSG denotes the disulphide form. The term ‘glutathione’ is used where no distinction is drawn or both forms may be concerned.