Bcl-2 and Mitochondrial Oxygen Radicals

Investigations into the capacity of the Bcl-2 protein to prevent apoptosis have targeted mitochondria as key sites of the preventative action accorded by Bcl-2 to cells. Using novel approaches with fluorescence probes and autofluorescence detection of endogenous NAD(P)H, we have examined the effects of expressing Bcl-2 in the Bcl-2 negative Burkitt’s lymphoma cell line Daudi. We evaluated for the first time the effect of Bcl-2 expression on the intracellular distribution and production of hydrogen peroxide, under basal conditions and after treatment with apoptosis inducing agents, ceramide analogs and tumor necrosis factor (TNF)-α. Increased availability of mitochondrial NAD(P)H was detected in Bcl-2-expressing cells and was correlated with an increased constitutive mitochondrial production of hydrogen peroxide. Although production of hydrogen peroxide was increased by either C6-ceramide or TNF-α in Bcl-2 negative Daudi cells commensurate with the early phases of apoptosis, this increase did not occur in Bcl-2-expressing cells. Thus, Bcl-2 appears to allow cells to adapt to an increased state of oxidative stress, fortifying the cellular anti-oxidant defenses and counteracting the radical overproduction imposed by different cell death stimuli. Furthermore, we report altered cytological features of mitochondria during the early phases of apoptosis induced by C6-ceramide and TNF-α. In particular, mitochondria changed in appearance, clustering in the perinuclear region and Bcl-2 expression prevented these changes from occurring.

Bcl-2 was originally identified as a human lymphoma oncogene (1). It now represents a family of genes that have the fascinating capacity of controlling apoptosis and cell death (see Refs. 2-4 for recent reviews). Despite intense research, we still do not clearly understand the biochemical mechanism whereby Bcl-2 expression prevents apoptosis. Bcl-2 appears to be a multifunctional protein (2,3), and a growing body of evidence indicates that the crucial site of Bcl-2 action resides in mitochondria (4 -6). In particular, Bcl-2 expression appears to affect the transmembrane transport of cations (4, 6), a fundamental aspect of mitochondrial function. One proposal for the action of Bcl-2 is to protect the integrity of mitochondrial oxidative phosphorylation and thus limit the mitochondrial dysfunction that is induced by several stimuli of apoptosis (4, 6 -8).
Whether the protection afforded by Bcl-2 expression derives from a decrease in the production of reactive oxygen species (ROS) 1 is unclear. Reports suggesting an anti-oxidant action of Bcl-2 (9 -12) contrast with results of others (13,14). To further complicate the picture, there is also evidence for a pro-oxidant activity of Bcl-2 (15). However, difficulties in measuring cellular ROS and in interpreting results obtained with cells under conditions of virtual anaerobiosis (13,14,16) may be responsible for the discrepancies and contradictory results that have been reported. To address these issues in depth, we have applied new sensitive approaches to measure the effect of Bcl-2 expression on the mitochondrial production of radicals, especially hydrogen peroxide, which is a key cellular ROS (17). We have found that Bcl-2 expression results in an elevation of the basal levels of hydrogen peroxide but restricts the excessive production of hydrogen peroxide induced by apoptotic stimuli such as TNF-␣ and ceramides. Thus, Bcl-2 allows lymphoma cells to adapt to conditions of increased oxidative stress, fortifying their anti-oxidant defenses against the overproduction of ROS imposed by cell death stimuli. We discuss the biochemical mechanisms that could explain this "homeopathic-like" action of Bcl-2 on mitochondrial ROS.

EXPERIMENTAL PROCEDURES
Cell Culture-The Daudi human B-lymphoma cell line was obtained from the American Tissue Type Culture Collection (Manassas, VA). The human Bcl-2 gene was transfected into Daudi cells as described previously (18). Daudi cell lines were cultured at 37°C and 5% CO 2 in RPMI medium supplemented with 10% (w/v) fetal calf serum, 1.7 mM glutamine and antibiotics (18). The numbers of cells were quantitated by microcytometry, and cell viability was evaluated by Trypan blue exclusion. To induce apoptosis, cells were resuspended in fresh growth medium at 1-2 ϫ 10 6 /ml and incubated at 37°C for one or more hours in the presence of 10 -20 M ceramides (either C 6 -ceramide from ICN, Sydney, Australia, or C 2 -ceramide from Biomol, Plymouth Meeting, PA) or 20 ng/ml of purified TNF-␣ (from Sigma). Cells were subsequently washed and resuspended in either PBS or growth medium containing fluorescent probes for the functional measurements outlined below. Nuclear apoptosis induced in Daudi cells was measured morphologically using DNA staining with either Hoechst 33342 (5) or 4Ј,6-diamino-2-phenylindole as described previously (18).
Materials and Determinations-Optical and fluorescent probes were purchased from Molecular Probes (Eugene, OR), dissolved in Me 2 SO, and determined in methanol according to the specifications of the manufacturer. Mitochondrial inhibitors were obtained and used as described (19). Other reagents used were from Sigma.
Western Blotting-Samples of cells (5 ϫ 10 7 ) were grown, harvested, washed in PBS, and pelleted in a microfuge tube. The cell pellets were lysed by resuspension in 1 ml of modified RIPA buffer (150 mM NaCl, 50 mM Tris-Cl, pH 7.5, containing 0.25% sodium deoxycholate, 0.1% Nonidet P-40, 1 mM Na 3 VO 4 , 10 g/ml aprotinin, 50 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, and 1 mM NaF (20)) for 10 min at 4°C. The lysates were clarified by microcentrifugation, and the supernatants were added in a 3:1 ration to 4x SDS-polyacrylamide gel electrophoresis * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Measurement of Endogenous NAD(P)H-The blue autofluorescence of Daudi cells was exploited for evaluating the content of mitochondrial NAD(P)H as described previously for other cellular systems (22)(23)(24). Cells were resuspended in glucose-supplemented PBS at a concentration of 1 ϫ 10 6 /ml as described for the ROS assays. Cellular autofluorescence was measured at 37°C under conditions of gentle automatic stirring in a Perkin-Elmer LS50B fluorimeter with filter settings at the maxima in the excitation (358 nm) and emission (443 nm) with a 10-nm bandwidth. Fluorescence at these wavelengths mainly detected mitochondrial NAD(P)H within cells, as indicated by the signal responses to mitochondrial effectors (see Fig. 1B). In particular, addition of carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) induced a rapid decrease in the fluorescence signal because of the stimulation of NAD(P)H oxidation by the respiratory chain (22,23), whereas complex I inhibition by rotenone or rolliniastatin-2 produced the maximal level of reduced NAD(P)H, similar to that detected under anoxic conditions (see Fig. 1C and Refs. [22][23][24]. Measurement of ROS Production-The production of ROS in solution was routinely measured with dichlorodihydrofluorescein diacetate (DCFDA) (25,26) as described previously (16). Briefly, cells (ϳ100,000/ well) were suspended in either growth medium or PBS supplemented with 20 mM glucose and incubated with 2 M DCFDA. The fluorescence increase (mainly because of the production of hydrogen peroxide (10, 16, 24 -26)) was measured with accumulated readings every min at 37°C in a Molecular Dynamics Biolumin 920 plate reader (excitation at 485 nm and emission at 520 nm, with 5-nm bandwidths (16)). The quantitative evaluation of the fluorescence readings was undertaken by comparison with the readings of dichlorofluorescein standards (16,26). Dyhydrotetramethylrosamine was also used for detecting the ROS production in cells using a Hitachi D-4000 spectrofluorimeter. In some experiments, the superoxide-sensitive probe dihydroethidium was used as recently described (27).
Mitochondrial Staining and Microscopy Analysis-To evaluate the direct production of mitochondrial ROS in cells we have complemented the DCFDA measurements with ROS-specific staining by using the reduced Mito Tracker® Red probe (CM-H 2 XRos). Cells suspended in growth medium at 1 ϫ 10 6 /ml were incubated for 15 min at room temperature with freshly prepared CM-H 2 XRos (0.5 M), then washed twice with PBS, and collected on a slide using a Cytospin apparatus (18). The cells were fixed with 3.7% formaldehyde in PBS, followed by washing first with PBS containing 30 mM NH 4 Cl and then with distilled water. In some experiments, the fixed cells were counterstained with Hoechst 33342 before mounting with anti-fade medium, and the slides were stored at 4°C in the dark. In parallel experiments, mitochondrial staining to analyze membrane potential was performed using 100 nM oxidized Mito Tracker® Red (CM-XRos) following the same protocol as outlined above for CM-H 2 XRos. Confocal microscopy was undertaken with a Krypton/Argon Leica instrument, usually with a 40ϫ oil objective and intermediate laser and photomultiplier voltage. The 590-nm bandpass filter was used to detect the red fluorescence of CM-XRos staining (5). Epifluorescence was evaluated with an Olympus microscope using a Texas Red filter and a CCD camera attached to a microcomputer image analyzer (Imaging Research Inc., Brock University, St. Catharines, ON, Canada).
Enzyme and Other Assays-The activities of mitochondrial redox enzymes were determined as described in previous studies (19,28). Mitochondrial membrane potential (⌬ m ) was monitored in living cells with 100 nM tetramethylrhodamine ethyl ester (29) under the same conditions as those used for NAD(P)H autofluorescence.

Bcl-2 Expression Increases Basal Levels of Hydrogen Peroxide and NAD(P)H-Previous studies have described the effects
of Bcl-2 expression on the conditions of oxidative stress induced by treating cells with apoptotic stimuli (9,10,12,30). However, comparison of the basal levels of ROS production in lymphoid cells in the absence and presence of Bcl-2 has not been studied in depth. In Daudi cells expressing Bcl-2 ( Fig. 1A shows the detection of Bcl-2 protein expression), the basal level of ROS was significantly higher than in the parental Bcl-2-negative cell line (Figs. 1B and 2). The rate of oxidation of the DCFDA probe, which predominantly derives from the intracellular production of hydrogen peroxide (16,25,26), was at least 2-fold higher in Bcl-2-expressing cells than in parental Daudi cells (Figs. 1B and 2). Moreover, complex I inhibitors such as rotenone decreased the production of hydrogen peroxide to a greater degree in the parental Daudi cells than in the Bcl-2expressing cells (Fig. 1B).
Rotenone inhibition of hydrogen peroxide production diminishes in proportion to the extent of NADH available to reduce complex I in either isolated mitochondrial preparations or substrate-respiring cells (32)(33)(34). 2 Given that Bcl-2 expression limited the effect of rotenone on the production of hydrogen peroxide production (Fig. 1B) and has been reported to increase the cellular levels of glutathione and NAD(P)H (11), we examined the content of NAD(P)H within Daudi cells. For this evaluation, we exploited the endogenous fluorescence of cellular NAD(P)H (22)(23)(24), which primarily derives from protein-bound NADPH in the mitochondrial compartment of cells (22)(23)(24). Of note, KCl depolarization of the plasmamembrane or cytosolic effectors of cellular NAD(P)H such as pyruvate produced only minor changes in the NAD(P)H autofluorescence signals of Daudi cells. This observation confirmed that the autofluorescence signals derived prevalently from mitochondrial NAD(P)H, in agreement with previous reports in other cell systems (22)(23)(24).
Daudi cells overexpressing Bcl-2 showed an enhanced level of NAD(P)H fluorescence (Fig. 1C). The difference between the maximal level of NAD(P)H reduction obtained in the presence of complex I inhibitors such as rolliniastatin-2 (Fig. 1C) and the minimal level obtained after FCCP-stimulated oxidation of mitochondrial NAD(P)H was 59 Ϯ 5% higher with Bcl-2 expression (Fig. 1D). The results also indicated that expression of Bcl-2 in Daudi cells led to an increased ratio of reduced to oxidized nicotinamide adenine dinucleotides compared with the parental cells. This increased ratio was indicated by the commensurable reduced effect of complex I inhibitors on the basal signal obtained with Bcl-2 positive cells ( Fig. 1C and Refs. 23 and 24).
Bcl-2 Diminishes the Oxidative Stress Induced by Ceramide Analogs-That Bcl-2 expression leads to an increased basal level of hydrogen peroxide production in Daudi cells (Fig. 1B) would appear to contrast with the anti-oxidant effects reported in other cells (4, 8 -12, 30). To investigate the effects of Bcl-2 expression on the oxidative stress associated with apoptosis in Daudi cells, we have used cell-permeant ceramide analogs that have been previously established as inducers of apoptosis that increase mitochondrial ROS production (16,30,31,(35)(36)(37)(38). Because we were interested in the early phases of apoptosis, we investigated the effects of ceramide treatment over 1-or 2-h periods. Both C 2 -and C 6 -ceramide increased the rates of DCFDA oxidation in parental Bcl-2 negative Daudi cells (Fig.  2) concomitant with onset of observable apoptotic alterations (cell shrinkage, externalization of phosphatidylserine, and chromatin condensation; results not shown). The biologically inactive dihydro-C 2 -ceramide was not as effective (results not shown). Surprisingly, in Bcl-2-expressing Daudi cells addition of the ceramide analogs caused a significant decrease in the basal rate of DCFDA oxidation (Fig. 2). This decrease was particularly noticeable with C 6 -ceramide (Fig. 2B), the analog that seemed to be most effective in increasing ROS production during early apoptosis (16).
Bcl-2 Action on the ROS-sensitive Staining of Mitochondria-Because DCFDA monitors production of hydrogen peroxide in all cellular compartments (24,26), it was important to locate The probe then covalently binds to mitochondrial proteins and forms a permanent red stain of these organelles (39,40). To assess whether cellular staining with CM-H 2 XRos would distinguish between ROS produced outside of mitochondria from ROS specifically produced within mitochondria (26,39,41,42), we have also used the DTMR probe (41), which, contrary to its MitoTracker® derivatives, freely diffuses among the different cellular compartments (39). As shown in Fig. 3, the cellular fluorescence of DTMR was only slightly affected by disrupting ⌬ m with FCCP. At times longer than 30 min, FCCP marginally increased DTMR fluorescence (Fig. 3), presumably because of the diminished anti-oxidant capacity resulting from the oxidation of mitochondrial NAD(P)H affected by the uncoupler (Fig. 1C). Depolarization of ⌬ m produced little effect, not only on the detection of intracellular ROS with DTMR ( Fig. 3) but also on the red staining obtained using CM-H 2 XRos (results not shown). Hence, it can be concluded that the predominant source of red fluorescence in cells treated with CM-H 2 XRos resulted from the production of radicals localized within the mitochondria.
The confocal images of Fig. 4 show that treatment with C 6 -ceramide strongly enhanced the ROS-sensitive staining of mitochondria with CM-H 2 XRos, consistent with the results obtained using DCFDA ( Fig. 2A and Refs. 16 and 36) and those results reported using other techniques (30,35). Interestingly, C 6 -ceramide also induced perinuclear clustering of the enhanced red staining (Fig. 4A), which was particularly evident in cells exhibiting nuclear condensation and other changes typical of the onset of apoptosis ( Fig. 4C; see also below). A similar perinuclear clustering of mitochondria has been observed in the human histiocytic lymphoma U937 and in other cell lines treated with TNF-␣ for 1 h (42). Conversely, in Daudi cells expressing Bcl-2, C 6 -ceramide treatment significantly decreased the intensity of but did not significantly alter the cellular distribution of CM-H 2 XRos staining (Fig. 4B). This effect occurred in concert with Bcl-2-mediated protection from the morphological alterations characteristic of apoptosis (Figs. 4 and 5).
Exogenous ceramides have been reported to decrease ⌬ m in lymphocytes (43). To evaluate ⌬ m in Daudi cells, we have used both confocal imaging with MitoTracker Red® (5, 40) and fluorescence measurements with tetramethylrhodamine ethyl ester (24). The results showed that 1 h of treatment with C 6ceramide produced a small decrease in the magnitude of ⌬ m in parental cells but essentially no effect in the Bcl-2-expressing cells ( Fig. 5; see also Ref. 37). However, C 6 -ceramide altered the morphology and cytoplasmic distribution of mitochondrial organelles in the Bcl-2 negative Daudi cells as indicated by confocal analysis (Fig. 5). Bcl-2 expression, in parallel with its effect on reducing ROS production (Fig. 4), prevented the alterations in the spatial distribution of mitochondria induced by ceramides in the parental Daudi cells (Fig. 5).
Bcl-2 and the Early Effects of TNF-␣ Treatment-We also studied the mitochondrially specific production of ROS in Daudi cells exposed to TNF-␣, a physiological inducer of cell death (36 -38). We focused on the early phase of the cytotoxic action of TNF-␣ (42,44) because it seemed to be associated with mitochondrial dysfunction (31,36,37,42) and also with an increased production of hydrogen peroxide in mitochondria (Fig. 6A). Bcl-2 expression protected Daudi cells from TNF-␣induced apoptosis and the concomitant increase of ROS production. In fact, TNF-␣ treatment increased the ROS-sensitive staining of mitochondria in parental Bcl-2-negative cells compared with Bcl-2-expressing cells (Fig. 6). Results with the DCFDA probe and microplate fluorescence confirmed the findings of confocal staining, albeit that the quantitative differences detected were less pronounced using TNF-␣ compared with C 6 -ceramide (result not shown).
Tetramethylrhodamine ethyl ester measurements of ⌬ m revealed no significant difference in cells treated with TNF-␣ for up to 2 h, whether or not they expressed Bcl-2 (results not shown). By contrast, confocal microscopy studies with Mito-Tracker Red® showed alterations in the cytoplasmic localization and morphology of mitochondria in Daudi parental cells treated with TNF-␣ (Fig. 6B). In particular, cell staining with either the reduced or the oxidized form of MitoTracker Red® exhibited clustering of mitochondria around the nucleus (Figs. 6 and 7), in agreement with the findings of De Vos et al. (42). The clustered perinuclear distribution of mitochondria was more pronounced in cells undergoing nuclear fragmentation (Fig. 7, arrows). Bcl-2 expression prevented these cytological changes, as shown in Fig. 6B. DISCUSSION We present here a detailed study of the effect of Bcl-2 expression on the oxidative state of lymphoid cells. For the first time we have applied ROS-sensitive probes to measure the intracellular production of hydrogen peroxide by mitochondria FIG. 4. Comparison of ROS detection with DCFDA and CM-H 2 XRos. Cells (10 6 /ml) were treated with C 6 -ceramide for 1 h as in Fig.  2. After washing, one aliquot of the cells was suspended in glucosesupplemented PBS for the DCFDA measurements as in Fig. 2, whereas another aliquot was incubated with CM-H 2 XRos and subsequently fixed for confocal staining as described under "Experimental Procedures." DNA counterstaining revealed that C 6 -ceramide induced chromatin condensation and nuclear fragmentation in 5% of parental Daudi cells. Less than 0.5% of control Daudi cells or Bcl-2-expressing cells, with or without C 6 -ceramide, exhibited similar changes in nuclear morphology. The confocal images of ceramide-treated and control samples were obtained with the same instrument settings. However, the photomultiplier sensitivity used was in part reduced for the analysis of Bcl-2expressing cells (B) compared with parental cells (A) to obtain images of similar resolution. Note the hemispheric thread-like distribution of the staining around the nucleus, which is particularly evident in Bcl-2expressing cells under the conditions used (see Fig. 5 for colocalization). The red fluorescence recorded in the confocal instrument was converted into a linear grayscale palette using the program ImageTool (University of Texas Health Science Center of San Antonio). Images in C were obtained in slides double stained with CM-H 2 XRos and Hoechst 33342 for DNA using an epifluorescence microscope and were manipulated with the CCD camera software to highlight the perinuclear clustering of organelles induced by C 6 -ceramide (A, right panel) in an apoptotic cell exhibiting chromatin condensation (arrow).

FIG. 5. Effect of C 6 -ceramide on ⌬ m staining of Daudi cells.
Parallel samples from the same experiments shown in Fig. 4 were stained with 100 nM of MitoTracker Red® for mitochondrial colocalization and membrane potential staining. All confocal images were obtained under the same instrumental settings for direct quantitative comparison. These settings appeared to be more effective in resolving the morphological features of mitochondria in parental cells than in Bcl-2-expressing cells, which were generally larger in size. Hence, the qualitative differences among control cells are only apparent and do not reflect general morphological differences because of Bcl-2 expression in Daudi cells. See Figs. 4 and 6 for more resolved images of mitochondrial staining in Bcl-2-expressing cells. The arrow indicates a parental cell undergoing fragmentation into apoptotic bodies. (Figs. 4 and 6). Our results indicate that Bcl-2 expression in Daudi lymphoma cells exerts a protective action against the excessive production of hydrogen peroxide that is associated with ceramide and TNF-induced apoptosis (Figs. 2, 4, and 6). This fundamentally agrees with results of several other reports (9 -12, 30, 45) and raises doubts on the significance of opposing reports (13,14,46). For the first time, we report that Bcl-2 expression also prevents the alterations in the cellular distribution of mitochondria induced by different stimuli such as C 6 -ceramide and TNF-␣ (Figs. [4][5][6]cf. Ref. 42). The similarities we found in the mitochondrial effects of ceramide analogs and TNF-␣ do not necessarily imply a direct involvement of ceramide signaling in TNF-induced death. We believe that these similarities reflect general alterations in mitochondria occurring in the early phase of apoptosis induced by different stimuli.
Intriguingly, we have found that Bcl-2 expression in Daudi cells is associated with an increased mitochondrial production of hydrogen peroxide (Figs. 1 and 4). This is in agreement with the results of Steinman (15), which also indicated a pro-oxidant effect of Bcl-2 expression. The enhanced level of ROS could be responsible for the reported cytotoxic (15,47) and cytostatic (46,48) action of Bcl-2 expression. Based on these results and our own experience, we propose that in other studies where increased hydrogen peroxide levels were not detected in Bcl-2expressing cells, the technology was inadequate for this measurement. For instance, Hockenbery et al. (9) used a high concentration of DCFDA and incubated the treated cells for several hours in growth medium. The excessive amounts of DCFDA probe applied would be nonspecific, resulting in an excessive background signal associated with poor sensitivity to changes in production of mitochondrial ROS in cells (22)(23)(24), explaining why no apparent change was found even upon induction of apoptosis (Refs. 9, 12, 30, and 46 and Fig. 2). The recent data of Cai and Jones (12) also showed little difference in the basal level of hydrogen peroxide in HL-60 cells expressing Bcl-2. However, the dihydrorhodamine probe used, when converted into rhodamine123 by ROS reactions, undergoes fluorescence quenching upon accumulation into mitochondria (23,42,49,50). Therefore, it is likely that differences in the basal production of hydrogen peroxide would not be determined accurately by this method.
Dihydroethidine, the ROS probe routinely used by Kroemer and co-workers (8,30,51), also has the disadvantage of undergoing quenching like dihydrorhodamine. Dihydroethidine fluorescence is sensitive to both an increase in intracellular superoxide and a decrease in ⌬ m (27,39,50). Nevertheless, dihydroethidine fluorescence was not significantly altered by Bcl-2 expression in cells (30,51). Moreover, using conditions that minimized ⌬ m interferences (27), CEM and other lymphoma cells expressing Bcl-2 showed a slightly decreased fluorescence of dihydroethidine, whereas CM-H 2 XRos fluorescence was significantly increased. 3 Therefore, it is likely that Bcl-2 expression enhanced the level of hydrogen peroxide, as detected by CM-H 2 XRos, but not that of superoxide radicals.
The simplest explanation for the differential effect of Bcl-2 expression on hydrogen peroxide and superoxide is that the presence of Bcl-2 leads to an enhanced expression of superoxide dismutase. Superoxide dismutase catalyzes the conversion of superoxide to hydrogen peroxide and thus may lead to accumulation of the latter. Of note, superoxide dismutase activity is increased in cells expressing Bcl-2 (11,15). We also propose that increased production of hydrogen peroxide derives from the increased levels of mitochondrial NAD(P)H present in Bcl-2-expressing cells (Fig. 1C and Ref. 11). In particular, high concentrations of NADH will maintain the iron-sulfur clusters of mitochondrial redox enzymes such as complex I in a more reduced steady state. This situation will favor auto-oxidation of the iron-sulfur clusters with molecular oxygen, thereby producing ROS at sites distant from the ubiquinone substrate site of mitochondrial dehydrogenases. The limited effect of ubiquinone antagonist inhibitors such as rotenone on the hydrogen peroxide produced in Bcl-2-expressing cells (Fig. 1B) lends support to this interpretation. Additionally, the enhanced level of unsaturated lipids in Bcl-2-expressing cells (52) may further contribute to the increased oxidation of probes via reaction with lipid peroxides.
We have discussed at length the possible reasons for the increased levels of hydrogen peroxide detected in Bcl-2-expressing cells because we believe that this observation is important for the anti-apoptotic mechanism of Bcl-2. First, it helps to rationalize previous conflicting reports, including the pro-oxidant action of Bcl-2 (15). Second, it reflects the increased availability of reduced nicotinamide adenine nucleotides in mitochondria (Fig. 1, C and D, and Ref. 11). This enables Bcl-2 positive cells to have more anti-oxidant capacity to counteract the excessive ROS production induced by stress stimuli such as ceramide ( Fig. 4 and Ref. 16). In particular, the increase in NADPH will fortify the ROS-scavenging defense of mitochondria via the coupled system of glutathione reductase and peroxidase (53).
How does Bcl-2 expression induce the increased levels of mitochondrial NAD(P)H that reinforce the anti-oxidant defense of cells? One possible explanation is that there is an enhanced activity of NADP ϩ transhydrogenase, an energy-consuming enzyme of the inner mitochondrial membrane that physiologically produces NADPH at the expense of NADH (53,54). The resulting increase in NADPH would act to reinforce the anti-oxidant defenses in the mitochondria of Bcl-2 positive cells. In fact, the activity of NADP ϩ transhydrogenase is considered to be crucial in reducing states of mitochondrial oxidative stress (53). Unfortunately, the lack of specific inhibitors prevents a direct test of the role of NADP ϩ transhydrogenase during apoptosis. We are currently exploring the alternative possibility that Bcl-2 expression may alter key metabolic pathways involved in the homeostasis of intramitochondrial NAD(P)H. For instance, a decreased activity of mitochondrial respiratory enzymes could favor an increased availability of NAD(P)H. However, preliminary measurements in subcellular fractions have failed to detect changes in the activity of respiratory enzymes such as complex III, in agreement with previous reports (6,12). Studies are under way to test whether lipid catabolism or the malate-aspartate shuttle may be responsible for the metabolic and stress related changes elicited by Bcl-2.