The FASEB Journal express article 10.1096/fj.02-0259fje. Published online November 15, 2002.

Visualization of the compartmentalization of glutathione and protein-glutathione mixed disulfides in cultured cells Therese Söderdahl,* Mari Enoksson,�,� Mathias Lundberg,� Arne Holmgren,� Ole P Ottersen,§ Sten Orrenius,� George Bolcsfoldi,¶ and Ian A Cotgreave*

*Division of Biochemical Toxicology, �Division of Toxicology, Institute of Environmental Medicine; �Division of Biochemistry, Department of Medical Biochemistry and Biophysics, Karolinska Institutet, Stockholm, Sweden; §Dept of Anatomy, Institute of Basic Medical Sciences, University of Oslo, POB 1105 Blindern, 0317 Oslo, Norway; ¶Department of Genetic Toxicology, AstraZeneca R and D Södertälje, Safety Assessment, S-15185 Södertälje, Sweden Corresponding author: Ian A Cotgreave, Division of Biochemical Toxicology, Institute of Environmental Medicine, Karolinska Institute, Box 210, S-17177, Stockholm, Sweden. E-mail: Ian.Cotgreave@Imm.ki.se ABSTRACT Fluorescence microscopy of A549 cells stained with a glutathione (L-γ-glutamyl-L-cysteinylglycine, GSH)-specific polyclonal antibody displayed uniform staining of the peri-nuclear cytosol, with the nuclear region apparently lacking GSH staining. This discontinuous staining was confirmed in other cell types and also corroborated in A549 cells stained with the thiol-reactive dye mercury orange. The selectivity of antibody binding was confirmed by buthionine sulfoximine (BSO)-dependent inhibition of GSH synthesis. However, confocal visualization of antibody-stained A549 cells in the z-plane revealed the majority of the peri-nuclear staining intensity in the upper half of the cell to be associated with mitochondria, as confirmed by double staining for cytochrome oxidase. Integration of the confocal signals from the nuclear and cytosolic regions halfway down the z-plane showed that the GSH concentrations of these compartments are close to equilibrium. Confirmation of the relatively high levels of mitochondrial glutathione was provided in cells treated with BSO and visualized in z-section, revealing the mitochondrial GSH content of these cells to be well preserved in apposition to near-complete depletion of cytosolic/nuclear GSH. Localized gradients within the cytosolic compartment were also visible, particularly in the z-plane. The antibody also provided initial visualization of the compartmentalization of protein-GSH mixed disulfides formed in A549 cells exposed to diamide. Discontinuous staining was again evident, with heavy staining in membrane blebs and in the nuclear region. Using FACS analysis of anti-GSH antibody-stained Jurkat T lymhocytes, we also demonstrated population variations in the cellular compliment of GSH and protein-GSH mixed disulfides, formed in response to diamide. In addition, we showed cell-cycle variation in GSH content of the cells, with the highest levels of GSH associated with the G2/M mitotic phase of the cell cycle, using double staining with propidium iodide. Similar FACS analyses performed in isolated mitochondria presented a considerable variation in GSH content within mitochondria of uniform granularity from the same preparation. Key words: confocal microscopy • FACS • mitochondria • nucleus • diamide

G lutathione (L-γ-glutamyl-L-cysteinylglycine, GSH) is a ubiquitous tripeptide that possesses several functions that are vital for cellular homeostasis (1). The redox properties of glutathione predisposes its involvement in processes as diverse as protein and DNA synthesis and the transport of amino acids. However, most interest has been

focused on its role in protecting cells from the effects of oxidants and electrophiles, both chemically and through peroxidase- and transferase-dependent mechanisms, respectively. Much has been learned about the biochemistry of glutathione in mammalian cells. Thus, it is well established that GSH is synthesized by a ribosome-independent cytosolic mechanism, the γ-glutamyl cycle, which is highly conserved throughout the eukaryotic kingdom (1). Attempts have been made, largely using broken cell preparations, to determine the intracellular localization of GSH. These studies are essential if one is to correctly understand the importance of GSH in various detoxification processes vital to the protection of functions within the mitochondrion and the nucleus in particular. In view of this, it is established that the mitochondrion possesses a pool of GSH that is separate from that of the cytosol, maintained by a specific transport process, and that is proposed to be conserved in isolated mitochondrial preparations (2). However, it is a current matter of debate whether the nucleus possesses a pool of GSH that is maintained separately from that of the cytosol. Taylor et al. (3) found that subcellular fractions of rat liver nuclei were devoid of GSH, whereas Tiermenstein and Reid (4) demonstrated that such fractions of rat kidney gave values of GSH similar to that of the cytosolic fraction. Several other studies investigating isolated nuclei have, however, given unequivocal results, depending on the experimental conditions used (5, 6). Some studies have attempted to determine the intracellular compartmentalization of GSH by using cytochemical methods. There have been very few efforts to visualize intramitochondrial levels of GSH in intact cells, but some attempts have been made to focus on the nucleus. Early measurements relying on the use of monochlorobimane conjugation with GSH demonstrated the nucleus of cultured cells to contain approximately three times the cytosolic content (7). This was later refuted as an artifact by Ketterer and Sies, who showed that the high nuclear fluorescence is due to an influx of the fluorescent bimane-GSH conjugate into the nucleus (8). However, Hedley et al. (9) reported the use of mercury orange staining of fixed cells to study intracellular GSH compartmentalization and showed a nuclear-cytoplasmic ratio of 0.57. The method is, however, relatively tricky to perform because the reagent reacts readily with protein thiols and therefore requires careful analytical preparation of the samples. Similarly, McConkey et al. (10) have used conjugation between 5-chloromethylfluorescein diacetate (CMFDA) and GSH to show that the nucleus contains less GSH than the cytosol. However, one might question this latter observation based on potential artifacts of conjugate import into the nucleus. In view of the difficulties in using existing cytochemical methods to study intracellular GSH compartmentalization, we have explored the possibility of developing an immunocytochemical method, based on labeling fixed cell preparations with a GSH-specific, polyclonal antibody that has hitherto been largely applied to histochemical analyzes (11, 12). Thus, in the present work, we report on the cellular localization of GSH in intact cells, using a combination of secondary fluorescence staining and confocal microscopy. The data clearly demonstrate that the levels of GSH associated with mitochondria appear to be greater than levels in surrounding cytosolic and nuclear compartments, particularly in the upper levels of the cell. Here, the cytosolic-nuclear GSH compartments seem close to equilibrium with each other. However, the data also reveal

GSH gradients in the cytosolic compartment, with the immunoreactivity rising to a maximum after the midpoint in the z-plane, falling again toward the basolateral membrane of the cell. Finally, cells and isolated mitochondria stained for either GSH or protein-GSH mixed disulfides proved amenable to FACS analysis, revealing population differences in the GSH redox status of cells undergoing constitutive metabolism or subjected to pro-oxidant stress. MATERIALS AND METHODS Chemicals and biologicals Mercury orange, buthionine sulfoximine (BSO), n-ethylmaleimide (NEM), and 1,4-diazabicyclo [2.2.2] octane were all supplied by Sigma (St. Louis, MO). Monobromobimane was from Calbiochem (Germany). The rabbit anti-glutathione antibody (no744) was raised as previously described (13). Goat anti-rabbit Alexa 568, goat anti-rabbit Alexa 488, goat anti-mouse Alexa 488 antibodies, and Alexa 488-labeled cytochrome oxidase antibodies were purchased from Molecular Probes (Eugene, Oregon). Mouse anti-histone H1 antibodies were from Stressgen (Victoria, Canada). A549, Jurkat T lymphocytes, and ECV304 cells were obtained from the ATCC (Manassas, VA) collection. V79 hamster fibroblasts were the kind gift of Dr. Bengt Jernstöm in the Institute. Human skeletal muscle myoblasts were obtained from Clonetics (Ghent, Belgium). All media and reagents for cell culture were obtained from GIBCO BRL (Middlothian, Scotland). Cell culture The A549 human lung carcinoma cell line, ECV304 human umbilical cord carcinoma cell line, and V79 hamster fibroblasts were all grown at 37°C in 5.0% CO2 on 75-cm2 flasks. The growth medium was Dulbecco�s modified Eagle�s medium (DMEM), supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin, 100 µg/mL streptomycin, and 1 mM sodium pyruvate. Cells were passaged by conventional trypsinization, using 0.25% trypsin-EDTA. Human skeletal muscle myoblasts were maintained as previously described (14). Jurkat cells were grown in logarithmic phase culture in RPMI-1649 medium with 10% heat-inactivated FBS, 2 mM glutamine, 100 U/mL penicillin, and 100 µg/mL streptomycin at 37°C in humidified air containing 5% CO2. Subcellular fractionation Rat liver (Sprague Dawley) mitochondria were isolated by a conventional differential centrifugation method in a buffer containing 210 mM mannitol, 70 mM sucrose, 5 mM HEPES (MSH buffer), and 1 mM ethylenediamine tetraacetic acid (EDTA), pH 7.5 (15). EDTA was omitted from the final washing solution, and the sedimented mitochondria were suspended in the same solution at a protein concentration of 80�100 mg/mL. Mitochondria were incubated in MSH buffer supplemented with 0.8 mM KH2PO4 and 5 mM succinate as the respiratory substrate.

Immunocytochemistry Cells were plated in chamber slides (Costar, 12,000 cells/well) and cultured for 30 h before processing for staining (see below). In some instances, cells were treated with either 100 µM BSO for 24 h or 100 µM NEM for 10 min, before further processing. Cells were washed with phosphate-buffered saline (PBS) containing 0.5 mM MgCl2 and 1.2 mM CaCl2. They were then fixed with 0.1 M phosphate buffer, pH 7.3, containing 2.5% (v/v) glutaraldehyde and 1% (v/v) paraformaldehyde and washed once with PBS for 5 min and then with PBS containing 0.5% (v/v) Triton X-100, 2% BSA, and 0.05% sodium azide for 10 min. The slides were then incubated with anti-GSH and/or anti-histone or anti-cytochrome oxidase antibodies. In brief, the rabbit anti-GSH antibody and the mouse anti-histone H1 were diluted 1:1000 and 1:700, respectively, in PBS containing 0.5% Triton X-100, 2% BSA, and 0.05% sodium azide. Slides were incubated with the primary antibodies overnight at room temperature. The secondary goat anti-rabbit Alexa 568 and goat anti-mouse Alexa 488 antibodies were diluted 1:800 in the same buffer as for the primary antibodies and incubated on the slides for 1 h at room temperature. The Alexa 488-labeled cytochrome oxidase antibody required an antigen retrieval procedure, according to the manufacturer�s procedures. The slides were washed once with PBS before mounting with glycerol:PBS (9:1), containing 0.23% 1,4-diazabicyclo [2.2.2] octane. In another series of cells, nonadherent Jurkat T lymphocytes (10,000 cells per analysis), were used to analyze population variations in cellular GSH content as above. However, the fixed and stained cell populations were analyzed by FACS analysis, as described below. In addition, the population distribution of GSH content in isolated rat liver mitochondria (2 mg per assay) was determined by FACS analysis. Histochemical staining Histochemical analysis of cellular GSH was accomplished by mercury orange staining, according to Hedley et al. (9), with a few modifications. In brief, cells were plated on glass microscope coverslips (Beckman, Fullerton, CA) in six-well plates (120,000 cells/well) and cultured for 12�36 h, depending on the subsequent treatment. In some cases, cells were pretreated with either 100 µM BSO for 24 h or 100 µM NEM for 10 min before staining. After treatment, the cells were washed once with PBS and air dried for 3 h. The cells were then stained with 50 µM mercury orange in acetone:water (9:1) for 30 s at 4°C and washed in acetone:water (9:1) for 5 min and then in water twice for 5 min. The slides were mounted with glycerol:PBS (9:1) containing 0.23% 1,4-diazabicyclo [2.2.2] octane and kept in the dark at 4°C for a maximum of 1 wk before analysis. HPLC analysis of glutathione Cellular GSH contents were determined either in suspension (Jurkat T lymphocytes) or in microwell plates containing either control, BSO-, or NEM-pretreated cells, using in situ derivatization of PBS-washed (two times) intact cells, with the membrane permeant, thiol-reactive fluorigen monobromobimane. Bimane-thiol adducts were subsequently separated by high-performance liquid chromatography (HPLC) and quantified by fluorimetry, essentially as described previously (15). In some experiments, GSH was determined in cells after various forms of fixation (see below).

Protein-glutathione mixed disulfides: generation and analysis The content of protein-GSH mixed disulfides was assessed in diamide-treated, fixed, and washed cells, using the immunocytochemical procedure outlined above. In order to validate this approach, it was necessary to perform control experiments, demonstrating the efficient removal of unbound GSH from fixed cells. In brief, cells were plated in a 12-well plate (80,000 cells/well) and grown to 95% confluency, and groups were exposed with one of the following treatments: untreated, cells fixed with methanol:acetone (1:1) for 5 min; cells fixed as above and washed with PBS once; and cells fixed and washed as above and then fixed again with methanol:acetone (1:1) for 5 min. Assays of cellular GSH were performed as above. Once the optimal fixation and GSH-removal treatment was determined, cells were exposed to diamide (1 mM) for 5 min, fixed, washed, fixed with aldehyde, stained with antibodies, and analyzed as detailed under Immunocytochemistry above. In another series of experiments, the population distribution of protein-GSH mixed disulfides was monitored by FACS analysis in Jurkat T lymphocytes. Jurkat cells (1 × 106 cells/mL) were exposed to diamide (1 mM) for 5 min and centrifuged at 150g for 5 min, and the pellet was resuspended and fixed in methanol:acetone (1:1). These cells were washed and stained with anti-GSH antibody as described above, but here, a secondary goat anti-rabbit Alexa 488 was used, together with propidium iodide (2.5 µg/mL). Fluorscence microscopy, confocal microscopy and FACS analysis All microscopy was performed using a Leica DM IRBE confocal microscope and Leica TCS NT software version 1.6.587. For the Alexa 568 dye, a 568 krypton laser and LP 590 filter were used, and for the Alexa 488 dye, a 488 argon laser and BP 530/30 filter were used. Stained Jurkat T lymphocytes and mitochondria (both for basal GSH levels and protein-GSH mixed disulfides) were subjected to FACS analysis, using a Becton Dickinson (Franklin Lakes, NJ) FACScan with a 488-nm laser line, and analyzed by Cell Quest software. RESULTS Immunocytochemical staining of A549 cells with an anti-GSH antibody and subsequent visualization by confocal microscopy revealed that the concentration of GSH in the cells is unevenly distributed across the focal plane (Fig. 1, panel 1). Indeed, it appears from this planar view that the peri-nuclear staining of the cells is relatively uniform toward the periphery, whereas the staining associated with the central zone of the cell, probably corresponding to the nucleus, is evidently lower and appears as a �dark hole� in the image. This pattern is irrespective of the cell type under study (Fig. 1, panels 2�5). Similarly, cytochemical staining of the cells with mercury orange showed a low nuclear:cytosolic ratio of GSH staining in A549 cells. The specificity of staining for free GSH in each staining procedure is illustrated in Figure 2, where it is clear that preincubation of A549 cells with BSO drastically diminished the intensity of microscopic images from the subsequently stained cells. When appropriate z-section images were taken through A549 cells stained with anti-GSH antibodies, a more pronounced regional heterogeneity in GSH staining was evident. Thus, as one

progresses down through the cell by 0.7-µm increments (Fig. 3, panels 1�16), a sharp band of staining close to the nuclear envelope develops into punctate staining, reminiscent of organellar structure. This is particularly evident in Figure 3, panel 7. In addition, double staining of untreated as well as BSO-treated cells with anti-GSH and anti-cytochrome oxidase antibodies reveals the majority of this punctate GSH-staining in the peri-nuclear cytosol to be coassociated with the mitochondria of the cells (Fig. 4). Furthermore, close inspection of the individual confocal images of the cells, especially comparing the peripheral cytosol with that of the intra-nuclear region, display a uniform staining intensity directed toward GSH, at least in the upper regions of the cell (Fig. 3, panels 5�8). This suggests that the relative concentrations of GSH in these regions are similar, a fact that is further illustrated in Figure 5a, where a cross-sectional analysis of the fluorescence intensity of staining within section 7 of the z-cut is presented. This cross-section intensity plot also shows a semi-quantitative estimate for the mitochondrial:cytosolic/nuclear concentration of GSH of more than twofold in intact A549 cells analyzed in this manner. However, as the confocal images depicted in Figure 3 progress down through the cell, the cytosolic staining becomes enhanced out toward the periphery of the cell, whereas the nuclear region remains poorly stained (Fig. 3, panel 9:16). Analysis of the relative intensities of staining of cytosol:nucleus at the 12th z-section of the cell reveals a clearly lower level within the nucleus, as compared to the cytosol (Fig. 5b). Taken together, the 3-D picture of the regional distribution of GSH in A549 cells displays a complex series of concentration gradients between different compartments of the cell and, indeed, within the cytosol itself. One of the major drawbacks in the study of the GSH-redox biochemistry of intact cells is the relative inability to monitor the levels of oxidized GSH in cells undergoing oxidative stress, for instance. Thus, attempts were made to apply the selectivity of the anti-GSH antibody to the analysis of protein-GSH mixed disulfides, a commonly utilized marker of oxidation in the intracellular GSH pool (15). In order to test the hypothesis that the anti-GSH antibody can be utilized to visualize protein-GSH mixed disulfides in intact cells, it was necessary to validate a washing technique which can be used to remove unbound GSH from fixed cells. Cells were assayed for their GSH content following fixation in ethanol:acetone (1:1) for 5 min alone and following subsequent washing of these fixed cells with PBS once or twice. The results readily indicate that ethanol:acetone fixation does not result in the extraction of cellular GSH, with levels of 83 ± 2 nmol/mg protein and 79 ± 3 nmol/mg protein (n=3 on both), respectively, recorded. However, a single wash with PBS removed in excess of 95% of soluble GSH from the fixed cells, with levels dropping to 2.6 ± 0.2 nmol/mg protein (n=3). Having established this, we then applied the anti-GSH antibody to the analysis of protein-GSH mixed disulfides formed during the metabolism of the thiol oxidant diamide in intact cells. Panel 1 in Figure 6 shows control immunocytochemical staining of A549 cells with the anti-GSH antibody, whereas panel 2 shows the immunoreactivity of control cells after fixation with methanol:acetone 1:1 and washing with PBS. It is evident that no immunoreactivity remains associated with these cells. However, panel 3 presents the results of immunocytochemical staining of A549 cells treated with diamide (1 mM) for 5 min, followed by fixation, washing, and then staining with the anti-GSH antibody. It can be seen that considerable immunoreactivity is revealed, particularly associated with plasma membrane blebs, the nuclear area of the cells and to some extent the peri-nuclear area of the cytosol. Another major drawback in conventional analyzes of the GSH redox balance of cells is that data often reflect an average value obtained over entire populations. Clearly, antibody staining would

offer the opportunity of FACS sorting and facilitate the study of population dynamics. To illustrate this, we did preliminary experiments with adherent A549 cells that unveiled a considerable distribution of red fluorescence intensity (FL-3 channel) against side-scattering (data not shown), but the trypsinization of the adherent cells presented some difficulties for the analysis, particularly due to cell clumping. Thus, Figure 7 details the FACS analysis of human Jurkat T lymphocytes stained with the anti-GSH antibody after diverse treatments. The plot explicitly reveals high GSH levels in untreated cells compared to cells stained with the Alexa-green secondary antibody only. Figure 7 also shows that FACS can be used to analyze populations of protein-GSH mixed disulfides in cells after treatment with diamide. After fixation with methanol and a single PBS wash, it is evident that the intense staining in the FL-1 channel due to antibody labeling of GSH is shifted toward the left of the plot. Note that cells treated in this manner were also shown to be devoid of measurable GSH, using structrual determination by HPLC (data not shown). When Jurkat T cells were treated with diamide and then fixed, washed, and stained with the anti-GSH antibody, a subpopulation of cells was revealed that remained stainable with the antibody and yielded high fluorescence in the FL-1 channel (Fig. 7). In addition to facilitating analysis of GSH and protein-GSH mixed disulfides in mixed populations of cells, combining the use of Alexa 488 secondary staining of bound anti-GSH antibody with propidium iodide staining of cellular DNA demonstrated a strong association of GSH content of individual Jurkat T lymphocytes with its position within the cell cycle (Fig. 8). Thus, cells at the G1 position (Fig. 8a) had the lowest GSH content (Fig. 8b), increasing accordingly through S phase to the G2/M transition into mitosis. To test the applicability of the antibody staining technique to study population dynamics in subcellular fractions, we analyzed isolated rat mitochondria using FACS, the results of which are shown in Figure 9. The data shows a wide variation in the GSH content of mitochondria displaying uniform granularity. DISCUSSION The analysis of cellular GSH has traditionally been performed by chemical derivatization techniques, coupled to some mode of chromatography and spectrophotometric detection, typified by an agent such as monobromobimane (15). These methods, however, fail to reveal any measure of intracellular distribution of the tripeptide, as well as any variation in cellular levels within a given cell population. More recently, methods have been developed to assess intracellular GSH utilizing glutathione S-transferase-mediated conjugation of substrates such as monochlorobimane to yield fluorescent GSH conjugates that are amenable to analysis, either by microscopy or by FACS (16). These methods present several problems to the analyst, however. The first is related to the dynamic nature of the labeling, being enzyme-dependent, and the second is due to the production of potential artifacts due to rapid intracellular distribution of conjugates (8). Both reduce the robustness of the analytical estimates and particularly hamper analysis of the intracellular compartmentalization of the tripeptide. In the present work, we provide, for the first time, a 3-D image of the intracellular distribution of GSH, by combining the specificity of polyclonal antibody labeling for GSH with the ability of confocal microscopy to provide sectional images from stained cells. This represents a development of the previously documented use of the antibody to assay GSH immunohistochemically (11, 12). Figure 1 details the staining pattern obtained by fluorescence

microscopy of several cell types. These images demonstrate an apparent discontinuity in staining intensity as one progresses from the periphery into the nuclear region. It is evident from the images that the GSH concentration within the nuclear region lies under that of the cytosolic periphery in each case. Several estimates of nuclear:cytosolic GSH ratios based on cytochemical staining have been made previously. These estimates range from high (3:1) (7), later proven to be flawed by artifacts due to redistribution of conjugates into the nucleus (8), to as low as 0.5 (9). The images in Figure 1 would generally support the idea of a nuclear content of GSH below that of the cytosol, particularly in the case of A549 (panel 1) and skeletal muscle myoblasts (panel 4). Moreover, successful staining of the nucleus by using anti-histone antibodies indicates that this is not the result of low accessibility of the GSH antibody to the nuclear region of the cell (data not shown). The GSH specificity of the staining procedure is illustrated in Figure 2, which shows that the inhibition of GSH synthesis with BSO treatment for 24 h clearly reduces the labeling of GSH seen within the cell (panel 1 vs. panel 2). In addition to GSH, the antibody is known to react with GSSG. However, the GSH/GSSG ratio in untreated cells is so high that this contribution to the staining can be conceived as negligible. Furthermore, the antibody can react with various GSH conjugates, if present. However, in control cells, the concentration of GSH conjugates, if being formed, would be deemed to be very low because these are continuously exported from the cell. The antibody has previously been successfully validated against a variety of amino acids and peptides, including precursor dipeptides to GSH, in order to ensure specificity for the tripeptide (13). Fianlly, the use of another GSH-labeling agent, mercury orange, confirms the results attained by the anti-GSH antibody (panel 4 vs. panel 5). However, use of the antibody continually produced images of superior intensity and quality. BSO depletion of the cells also revealed another interesting feature in A549 cells. Comparisons of panels 1 and 2 and 4 and 5 (Fig. 2) demonstrate a vestige of punctate staining around the nucleus after BSO treatment. Indeed, this staining pattern is also visible in the fluorescence microscopic images of ECV304 cells and V79 cells (Fig. 1, panels 2 and 5, respectively). These images suggest high concentrations of GSH in organellar bodies within the cells, relative to the surrounding cytosol. The areas of high GSH concentration within the cytosol were demonstrated to co-localise with immunohistochemical staining for cytochrome oxidase, an established mitochondrial marker (Fig. 4). Taken together the data suggest that mitochondria constitute an area of high GSH concentration in cells, and that the mitochondrial pool of GSH possesses a half life of turnover in cells, which is very much longer than that of the cytosolic and nuclear pools. The existence of a separate mitochondrial pool of GSH has been demonstrated by classical work from several authors, including Romero et al. (17, 18) Meredith and Reid (19) and Griffith and Meister (2), either through the analysis of mitochondria after their isolation from cells after various treatments, or in detergent-permeabilized cells. Such studies have suggested mitochondrial GSH to be slightly elevated over cytosolic concentrations (17), that the pool has a half-life of turnover one magnitude higher than the cytosolic pool (18, 19) and that it is maintained by specific transport processes (2). However, to our knowledge, the present work provides the first direct visualization of the mitochondrial pool of GSH, confirming some of these concepts. The present investigations add one further dimension to our knowledge by allowing estimates of the local gradients in GSH between mitochondria and the surrounding cytosolic compartment, showing subtle relationships between these compartments in three dimensions. Hence, the intensity of mitochondrial staining at the seventh z-plane of A549 cells

(Fig. 4, panel 7) was twofold that of the surrounding nuclear and cytosolic areas (Fig. 5a). However, in the 12th z-plane of the confocal images (Fig. 4, panel 12), the overall intensity of the mitochondrial staining at this level was only 30-50% higher than the peripheral cytosolic staining. Therefore, in addition to illuminating the subtle relationships between mitochondrial and cytosolic GSH pools in intact cells, the data also provide some of the first evidence for cytosolic GSH concentration gradients. Thus, comparison of the staining intensities of the peripheral cytosolic areas of the 7th and 12th z-planes of analysis of A549 cells (Fig. 5a versus b) indicates an approximately 200% increase in GSH concentration over this distance. The mechanisms underlying the ultra-fine compartmentalization of GSH in the cystosolic compartment remain elusive, as does its importance in the regulation of both GSH-dependent enzyme function and in the general redox regulation of other structural proteins and enzymes. This is the subject of further study in our laboratory. Reduced GSH is in equilibrium with several oxidized disulfide species in cells. These include oxidized GSH itself, GSSG, and protein-GSH mixed disulfides. It is now well established that cells experiencing increased oxidant burden enter a situation of oxidative stress which is readily reflected by a shift in GSH redox to a more oxidized state. Thus, cells experiencing oxidative stress generally show elevated levels of GSSG, as well as marked increases in the binding of GSH to proteins, which has been hitherto detected mostly by bulk chemical analysis (20-24). Over the years an opinion has arisen that reversible S-glutathionylation of cellular proteins, both during constitutive metabolism and during oxidative stress, represents one manner by which cells are able to link both physiological processes and adaptive stress responses to changes in intracellular redox states (25, 26). Therefore, the ability to visualize oxidized forms of GSH in intact cells would represent an important advance in our understanding of the role of GSH in metabolic regulation. However, as the antibody utilized in the present work also reacts with GSSG, the only option open was visualization of the formation of protein-GSH mixed disulfides. In Figure 6, it can be seen that treatment of A549 cells with diamide, an agent known to induce the formation of protein-GSH mixed disulfides in cells (20�22), preserved immunoreactivity toward GSH in cells that had been fixed and washed before staining, thus removing all soluble GSH-containing species. This represents the first reported visualization of intracellular distribution of protein-GSH mixed disulfides in intact cells. Interestingly, most of the immunochemical staining was noted within plasma membrane blebs and in the nuclear/peri-nuclear region. Diamide is known to elicit extensive S-glutathionylation of actin, resulting in depolymerization of this cytoskeletal component (27). Bleb formation in response to treatment of cells with menadione has also been shown to be closely associated with the oxidation of thiol groups on actin (28). Thus, it is possible that one of the components of this immunochemical staining in blebs may be accumulated fragments of disrupted actin filaments. Intense nuclear staining is especially interesting in view of the images presented in Figure 1, in which the nucleus generally exhibits itself as a darker area in the stained cells under constitutive conditions. This may indicate the nucleus as an important area for the accumulation of protein-GSH mixed disulfides at a level that is disproportionate to its content of the reduced tripeptide. As for the peri-nuclear staining, this may in part reflect the high local concentration of GSH in the mitochondria of this area. The ability to assay the intracellular content and distribution of GSH and protein-GSH mixed disulfides in intact cells has opened the door to a deeper understanding of the intracellular

biochemistry of GSH. The methods employed also, however, facilitate the analysis of the contents of these redox states of GSH in populations of cells and isolated organelles. Thus, the data in Figure 7 clearly demonstrate population variations in the content of GSH in Jurkat T cells under exponential growth conditions. Lymphocytes have been previously sorted according to their GSH content using a dynamic, glutathione transferase-dependent staining of cells with monochlorobimane (16), but the present method offers several advantages. First, the cell labeling is not dependent on the presence of an enzyme. This negates any artifacts due to variations (or even absence) in glutathione transferase activities in particular cellular populations. Second, the use of fixed cells allows for multiple staining of internal cellular components, as illustrated by propidium iodide staining of cellular DNA, thus facilitating refined cell sorting. Hence, one possible explanation for this population variation could lie in the asynchronous nature of the cell culture, in terms of cell cycle. This is clearly illustrated in Figure 8, which shows double staining with a green secondary antibody for GSH, and propidium iodide clearly revealed cell cycle variations in total cellular GSH. When cells were gated in FL-3 for their propidium iodide staining, three populations of DNA content were isolated, representing G1, S, and G2/M phases of the cycle (Fig. 8a). Analysis of the GSH content of these gated cells clearly showed that the GSH content of the cells rose steadily, reaching a maximum at premitotic G2/M checkpoint (Fig. 8b). The origins of this increase in intracellular GSH are obscure. Previous work measuring bulk contents of GSH in synchronized cell cultures has demonstrated periodic fluctuations in GSH associated with entry into exponential growth (29). However, to our knowledge, the present report is the first time direct evidence for variations in GSH during the cell cycle has been provided. It seems reasonable to assume that these fluctuations arise from either an increased capacity to synthesize GSH and or decreased metabolic turnover of the tripeptide. These matters are the focus of further work in our laboratory. The functional significance of elevated GSH contents in prelude to mitosis may be speculated to be related to adaptations to the redox regulation of the cellular machinery (cytoskeleton etc) during spindle formation and chromosome segregation. Similarly, one might speculate that this serves a generalized requirement for increased antioxidative cyto-protection during karyo-kinesis. Studies of fluctuations in intracellular distribution of GSH during cell cycle in adherent cells may provide clues as the functional significance of these cell-cycle variations in GSH content. In addition to revealing a population distribution in cellular GSH content, the present work also reveals variation in the protein-GSH mixed disulfides formed in cells during the metabolism of diamide. Comparisons of the area under the curve of the intensity plots of protein-GSH mixed disulfides and control GSH contents can also be used to estimate the degree of S-glutathionylation of cellular protein. The origins of the variation in mixed disulfide formation may lie in the variation of cellular GSH in the cell cycle, with the highest levels residing in cells at the G2/M checkpoint. It will be interesting to develop this analytical approach further, in order to detect population variations in protein-GSH mixed disulfides residing in cells under constitutive metabolism. This is envisaged to require a degree of signal amplification, particularly at the level of antibody labeling. Finally, the data generated from the analysis of mitochondrial GSH in intact cells clearly suggested that there is a wide variation in GSH contents within a population of mitochondria displaying similar granularity. This may be interpreted in one of two ways. First, it may be that some organelles undergo artifactual losses of the tripeptide during the isolation procedure. This might then support our proposal that mitochondrial GSH contents visualized in intact cells in the

present work may be higher than those predicted from broken cell preparations. The second explanation may lie in an innate variation in the GSH contents of individual mitochondria. This might be envisaged in terms of mitochondria in different stages of maturation, or from different cellular locations, containing differing amounts of GSH. Glutathione has been implicated centrally in many mitochondrial processes. For instance, mitochondrial redox states are centrally involved in the regulation of inner membrane potential and its dependent processes. Therefore the present finding suggests that caution must be used in the interpretation of data related to redox-dependent processes in populations of isolated mitochondria prepared by these standard procedures. In conclusion, the methods described allow us to illuminate more closely the organization of intracellular GSH in intact cells, the variation in levels of GSH in homogeneous and heterogeneous cells and organelle populations and, importantly, the disposition of GSH in such systems with respect to reversible S-glutathionylation of protein. The application of the techniques may therefore add new dimensions to our understanding of the biochemistry of this important cellular mediator. ACKNOWLEDGMENTS We thank Marcello Toro at Astrazeneca R&D Södertälje, Bioscience, for expert technical assistance. We acknowledge the financial support of the Swedish Medical Research Council (VR, to IC, SO and AH) and the Karolinska Institute. REFERENCES 1. Meister, A., and Anderson, M. E. (1983) Glutathione. Ann. Rev. Biochem. 52, 711-760 2. Griffith, O. W., and Meister, A. (1985) Origin and turnover of mitochondrial glutathione. Proc. Natl. Acad. Sci. USA 82(14), 4668�4672 3. Taylor, C. W., Yeoman, L. C., et al. (1973) Two-dimensional electrophoresis of proteins of citric acid nuclei prepared with aid of a Tissumizer. Exp. Cell Res. 82(1), 215�226 4. Tirmenstein, M. A., and Reed, D. J. (1988) The glutathione status of rat kidney nuclei following administration of buthionine sulfoximine. Biochem. Biophys. Res. Commun. 155(2), 956�961 5. Britten, R. A., Green, J. A., et al.. (1991) The relationship between nuclear glutathione levels and resistance to melphalan in human ovarian tumour cells. Biochem. Pharmacol. 41(4), 647�649 6. Jevtovic-Todorovic, V., and Guenthner, T. M. (1992) Depletion of a discrete nuclear glutathione pool by oxidative stress, but not by buthionine sulfoximine. Correlation with enhanced alkylating agent cytotoxicity to human melanoma cells in vitro. Biochem. Pharmacol. 44(7), 1383�1393

7. Bellomo, G., Vairetti, M., et al.. (1992) Demonstration of nuclear compartmentalization of glutathione in hepatocytes. Proc. Natl. Acad. Sci. USA 89(10), 4412�4416 8. Briviba, K. Fraser, G. et al.. (1993) Distribution of the monochlorobimane-glutathione conjugate between nucleus and cytosol in isolated hepatocytes. Biochem. J. 294(Pt 3), 631�633 9. Thomas, M., Nicklee, T., et al.. (1995) Differential effects of depleting agents on cytoplasmic and nuclear non- protein sulphydryls: a fluorescence image cytometry study. Br. J. Cancer 72(1), 45�50 10. Voehringer, D. W., McConkey, D. J., et al. (1998) Bcl-2 expression causes redistribution of glutathione to the nucleus. Proc. Natl. Acad. Sci. USA 95(6), 2956�2960 11. Huster, D. Hjelle, O. P. et al.. (1998) Subcellular compartmentation of glutathione and glutathione precursors. A high resolution immunogold analysis of the outer retina of guinea pig. Anat. Embryol. (Berl) 198(4), 277�287 12. Ramirez-Leon, V., Kullberg, S., et al.. (1999) Increased glutathione levels in neurochemically identified fibre systems in the aged rat lumbar motor nuclei. Eur. J. Neurosci. 11(8), 2935�2948 13. Hjelle, O. P., Chaudhry, F. A., et al. (1994) Antisera to glutathione: characterization and immunocytochemical application to the rat cerebellum. Eur. J. Neurosci. 6(5), 793�804 14. Cotgreave, I. A., Goldschmidt, L., Tonkonogi, M., and Svensson, M. (2002) Differentiation-specific alterations to glutathione synthesis and hormonally-stimulated release from human skeletal muscle cells. FASEB J. 16, 435�437 15. Cotgreave, I. A., and Moldeus, P. (1986) Methodologies for the application of monobromobimane to the simultaneous analysis of soluble and protein thiol components of biological systems. J. Biochem. Biophys. Methods 13(4�5), 231�249 16. Herzenberg, L. A., De Rosa, S. C., Dubs, J. G., Roederer, M., Anderson, M. T., Ela, S.W., Deresinski, S. C., and Herzenberg, L. A. (1997) Glutathione deficiency is associated with impaired survival in HIV disease. Proc. Natl. Acad. Sci. USA 94(5), 1967�1972 17. Romero, F. J., and Sies, H. Subcellular glutathione contents in isolated hepatocytes treated with L-buthionine sulfoximine (1984) Biochemical & Biophysical Research Communications 123(3), 1116�1121 18. Romero, F. J., Soboll, S., and Sies, H. (1984) Mitochondrial and cytosolic glutathione after depletion by phorone in isolated hepatocytes. Experientia 40(4), 365�367 19. Meredith, M. J. Reed, D. J. (1982) Status of the mitochondrial pool of glutathione in the isolated hepatocyte. J. Biol. Chem. 257(7), 3747�3753

20. Chai, Y. C., Jung, C. H., Lii, C. K., Ashraf, S. S., Hendrich, S., Wolf, B., Sies, H., and Thomas, J. A. (1991) Identification of an abundant S-thiolated rat liver protein as carbonic anhydrase III; characterization of S-thiolation and dethiolation reactions. Arch. Biochem. Biophys. 284(2), 270�278 21. Schuppe, I., Moldeus, P., and Cotgreave, I. A. (1992) Protein-specific S-thiolation in human endothelial cells during oxidative stress. Biochem. Pharmacol. 44, 1757�1764 22. Rokutan, K., Johnston, Jr. R. B., et al. (1994) Oxidative stress induces S-thiolation of specific proteins in cultured gastric mucosal cells. Am. J. Physiol. 266(2 Pt 1), G247�G254 23. Chai, Y. C., Hendrich, S., et al. (1994) Protein S-thiolation in hepatocytes stimulated by t-butyl hydroperoxide, menadione, and neutrophils. Arch. Biochem. Biophys. 310(1), 264�272 24. Schuppe-Koistinen, I., Moldeus, P., Bergman, T., and Cotgreave, I. A. (1994) S-thiolation of human endothelial cell glyceraldehyde-3-phosphate dehydrogenase after hydrogen peroxide treatment. Eur. J. Biochem. 221, 1033�1037 25. Cotgreave, I. A., and Gerdes, G. R. (1998) Recent trends in glutathione biochemitry: Glutathione�protein interactions: A molecular link between oxidatiev stress and cell proliferation? Biochem. Biophys. Res. Comms. 242, 1�9 26. Klatt, P., and Lamas, S. (2000) Regulation of protein function by S-glutathiolation in response to oxidative and nitrosative stress. Eur. J. Biochem. 267(16), 4928�4944 27. Schuppe-Koistinen, I., Moldeus, P., Bergman, T., and Cotgreave, I. A. (1995) Reversible S-glutathionylation of human endothelial cell actin accompanies a structural rearrangement of the cytoskeleton. Endothelium 3, 301�308 28. Mirabelli, F., Salis, A., Marinoni, G., Finardi, G., Bellomo, G., Thor, H., and Orrenius, S. (1988) Menadione-induced bleb formation in hepatocytes is associated with the oxidation of thiol groups in actin. Arch. Biochem. Biophys. 264, 261�269 29. Atzori, L., Dypbukt, J. M., Sundqvist, K., Cotgreave, I. A., Edman, C. C., Moldeus, P., and Grafstrom, R. C. (1990) Growth-associated modifications of low-molecular-weight thiols and protein sulfhydryls in human bronchial fibroblasts. J. Cell. Physiol. 143(1), 165�171

Received May 29, 2002; accepted September 26, 2002.

Fig. 1

Figure 1. Fluorescence microscopic images of a variety of cultured cells stained with an anti-GSH antibody. Various cultured cell lines and primary cells (1=A549, 2=ECV304, 3=Hela, 4=human skeletal muscle myoblasts, and 5=V79) were fixed and stained with a polyclonal anti-GSH antibody and visualized by fluorescence microscopy, as described in Materials and Methods. Images were obtained with the ×63 objective in each case.

Fig. 2

Figure 2. The glutathione specificity of labeling of intact cells with the polyclonal anti-GSH antibody. Control A549 cells (panel 1), cells pretreated with BSO (100 µM), for 24 h (panel 2), and cells pretreated with NEM (100 µM) for 10 min (panel 3) were stained with the anti-GSH antibody and visualized as in Materials and Methods. Panels 4, 5, and 6 represent similar cellular groups to above but are stained with mercury orange, as described in Materials and Methods. All images were obtained with the ×63 objective.

Fig. 3

Figure 3. Laser confocal imaging of cells stained with an anti-GSH antibody. A549 cells were fixed and stained as in Figure 1, and the cells were imaged using laser confocal microscopy, as described in Materials and Methods. The images are obtained in 0.7-µm cuts of the z-plane, starting from the apex of the cell (top left image), moving right toward the baso-lateral surface (bottom right-hand image).

Fig. 4

Figure 4. The colocalization of mitochondrial GSH and cytochrome oxidase. A549 cells were stained with both anti-GSH antibody and a red secondary antibody, and with anti-cytochrome oxidase antibody conjugated to Alexa green fluorochrome, as described in Materials and Methods. Confocal images depict untreated cells (panels 1 and 2) and BSO-treated cells (panels 3 and 4).

Fig. 5

Figure 5. Cross-sectional fluorescence intensity of anti-GSH staining at different levels of the z-plane of the cell. The data depicted in Figure 3 on the distribution of anti-GSH staining at different levels in the cell were integrated across the x/y-planes. A) Rrelative GSH concentration across the cell at the 7th z-plane (Figure 3, panel 7). B) A similar trace at the 12th plane (Figure 3, panel 12). Note the arbitrary nature of the fluorescence intensity scale.

Fig. 6

Figure 6. Immunocytochemical localization of protein-GSH mixed disulfides formed in cells in response to diamide metabolism. A549 cells were fixed and stained with anti-GSH antibody either before (panel 1) or after methanol:acetone fixation and a single wash with phosphate-buffered saline (PBS) (panel 2) or after treatment with diamide (1 mM, for 5 min), methanol:acetone fixation and single wash with PBS (panel 3). All images were obtained by fluorescence microscopy with a ×63 objective.

Fig. 7

Figure 7. Population distributions in Jurkat T-lymphocyte GSH levels and degree of S-glutathionylation of cellular protein during diamide metabolism. Human Jurkat T lymphocytes in exponential growth phase were subjected to either no staining (dashed line) or staining with anti-GSH antibodies and Alexa 488 secondary antibodies after no treatment (thick line), fixation and washing (dotted line), or treatment with diamide (1 mM for 5 min) followed by methanol:acetone fixation and a single wash with PBS (thin line). The cells were then subjected to FACS analysis in the FL-1 mode.

Fig. 8

Figure 8. Cell cycle variations in the constitutive GSH levels of Jurkat T lymphocytes. Human Jurkat T lymphocytes in exponential growth phase were double-stained with PI and the anti-GSH antibody, as described in Materials and Methods. B) GSH immunoreactivity (FL-1) associated with the different stages of the cell cycle in untreated cells, as defined by the PI staining (FL-3) in A. Population 1 is presented as a thin line, population 2 as a dashed line, and population 3 as a thick line.

Fig. 9

Figure 9. Population distribution in the constitutive GSH level in rat liver mitochondria. Rat liver mitochondria were either stained with anti-GSH antibody and Alexa 488 secondary antibodies (thick line) or subjected to fixation alone (dotted line), as described in Materials and Methods. The mitochondria were then subjected to FACS analysis in the FL-1 mode.