Critical role of mitochondrial glutathione in the survival of hepatocytes during hypoxia.

Hypoxia is known to stimulate reactive oxygen species (ROS) generation. Because reduced glutathione (GSH) is compartmentalized in cytosol and mitochondria, we examined the specific role of mitochondrial GSH (mGSH) in the survival of hepatocytes during hypoxia (5% O2). 5% O2 stimulated ROS in HepG2 cells and cultured rat hepatocytes. Mitochondrial complex I and II inhibitors prevented this effect, whereas inhibition of nitric oxide synthesis with Nomega-nitro-L-arginine methyl ester hydrochloride or the peroxynitrite scavenger uric acid did not. Depletion of GSH stores in both cytosol and mitochondria enhanced the susceptibility of HepG2 cells or primary rat hepatocytes to 5% O2 exposure. However, this sensitization was abrogated by preventing mitochondrial ROS generation by complex I and II inhibition. Moreover, selective mGSH depletion by (R,S)-3-hydroxy-4-pentenoate that spared cytosol GSH levels sensitized rat hepatocytes to hypoxia because of enhanced ROS generation. GSH restoration by GSH ethyl ester or by blocking mitochondrial electron flow at complex I and II rescued (R,S)-3-hydroxy-4-pentenoate-treated hepatocytes to hypoxia-induced cell death. Thus, mGSH controls the survival of hepatocytes during hypoxia through the regulation of mitochondrial generation of oxidative stress.

In conditions of limited O 2 supply cells adapt and survive because of the existence of specific sensors that allow them to acclimate to the deprivation of oxygen and to recover from ischemic conditions. One of the best characterized responses to hypoxia is the activation of hypoxia-inducible factor (HIF), 1 a transcription factor that is central for the cellular adaptation to oxygen limitation as it is known to up-regulate genes involved in angiogenesis and glycolysis (1). HIF is a heterodimer comprising an O 2 -regulated ␣ subunit (HIF1␣) and a constitutively expressed ␤ subunit (2)(3)(4). The O 2 -dependent regulation of HIF1␣ occurs at the level of protein stability. In normoxia, HIF1␣ is targeted for degradation through the hydroxylation of specific proline residues by a family of O 2 -and oxoglutarate-dependent prolyl hydroxylases (5,6). At low O 2 tension, prolyl hydroxylase activity is inhibited, resulting in HIF1␣ protein accumulation.
In addition, another well recognized response to hypoxia is the generation of reactive oxygen species (ROS). In elegant work using hepatoma cells depleted of mitochondrial DNA ( 0 cells), Schumacker and co-workers (7,8) showed that the stimulation of ROS by hypoxia was abrogated, demonstrating that the predominant source of ROS by oxygen limitation originated from mitochondria. Moreover, the mitochondrial ROS stimulation by hypoxia contributed to the HIF1␣ stabilization and subsequent HIF1␣-dependent gene expression and to the enhanced DNA binding of NF-B by a redox-dependent mechanism (7)(8)(9). However, in conditions in which reoxygenation ensues oxygen deprivation (e.g. ischemia/reperfusion), the ROS stimulation then arises from both mitochondrial and extramitochondrial sources (10).
GSH is a major and versatile cellular antioxidant that is found mainly in cytosol where it is synthesized from its constituent amino acids and in mitochondria where it plays a key protective role against oxidant-induced cell death (11,12). Because of its antioxidant function hypoxia would be expected to reduce intracellular GSH stores. For instance, recent findings reported a reduction in cellular GSH in hypoxic human embryonic kidney 293 and Hep3B cells that required mitochondrial generation of ROS (13). Moreover, previous studies reported the decrease of hepatocellular GSH stores by hypoxia through various mechanisms ranging from enhanced GSH efflux to impaired synthesis (14 -17). Furthermore, carbon monoxide has been reported to decrease the GSH/GSSG in rat brain mitochondria (18).
Thus, although the regulation of intracellular GSH levels by hypoxia is known (13)(14)(15)(16)(17)(18), the specific role that mitochondrial GSH (mGSH) plays in the survival of cells during hypoxia has not been reported to the best of our knowledge. Therefore, the aim of the present study was to examine the consequences of mGSH depletion during hypoxia on the regulation of ROS generation and survival of hepatocytes. Although the abrogation of mitochondrial ROS production rescued HepG2 cells or primary rat hepatocytes from hypoxia-induced cell death despite GSH depletion in both cytosol and mitochondria, using (R,S)-3-hydroxy-4-pentenoate (HP), which is transformed in the mitochondria of hepatocytes into a Michael acceptor resulting in selective mGSH depletion with the sparing of cytosol * This work was supported in part by the Research Center for Liver and Pancreatic Diseases, P50 AA 11999, and Grant 1R21 AA014135-01 from the National Institute on Alcohol Abuse and Alcoholism, Plan Nacional de IϩD grants SAF01-2118, SAF2003-04974, and Red Temá tica de Investigación Cooperativa G03/015, and Red de Centros C03/02 supported by Instituto de Salud Carlos III. 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.
GSH (19,20). We provide evidence for a critical role of mGSH in determining the susceptibility of hepatocytes to hypoxia through control of mitochondrial oxidative stress.
Cell Culture and Incubation-The human hepatoblastoma cell line HepG2 was obtained from the European Collection of Animal Cell Cultures (Salisbury, Wilts, UK). Cells were cultured in Dulbecco's modified Eagle's medium containing high glucose levels, supplemented with 10% heat-inactivated (56°C, 30 min) fetal bovine serum, 2 mM Lglutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Rat hepatocytes were prepared from Sprague-Dawley rats (250 g) upon collagenase perfusion and were cultured as described previously (20,21). Cells were maintained at 37°C in a humidified incubator containing 21% oxygen and 5% carbon dioxide (referred to as normoxic conditions). Hypoxic conditions were attained by exposure to 2 or 5% oxygen and 5% carbon dioxide in a humidified incubator (Forma Scientific) in sealed flasks for several periods of time (up to 24 h for rat hepatocytes or 72 h for HepG2 cells). In some instances hypoxic cells were exposed to inhibitors of mitochondrial electron flow, rotenone (complex I), TTFA (complex II), antimycin A (complex III), diphenyleneiodonium, or GSH depletors, DEM and BSO, in the last 4 -8 h of the period of exposure to 5% O 2 . In control experiments we verified that the instantaneous opening of flasks during hypoxia did not result in ROS stimulation or GSH depletion with respect to hypoxic unperturbed flasks.
Mitochondria Isolation-Rat hepatocytes or HepG2 cells were fractionated into cytosol or mitochondria upon digitonin permeabilization of the plasma membrane as described previously (21,22). Mitochondrial enrichment and assessment of contamination by subcellular organelles were performed as described (20,22).
Measurement of ROS and GSH-Intracellular ROS generation was assessed using chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate as described (23). GSH levels in either cytosol or mitochondria were determined by the recycling assay as reported previously (21,22).
Selective Depletion of mGSH-HP was synthesized as described previously (19) and used to selectively deplete mitochondrial GSH as reported previously (20,21). Cultured rat hepatocytes were exposed to 1 mM HP for 5 min, washed thoroughly, and then fractionated by digitonin into cytosol and mitochondria, and GSH content was determined in either fraction as described above (21,22).
Preparation of Cytosolic and Nuclear Fractions and Electromobility Shift Assays-Nuclear and cytosolic extracts of HepG2 cells were prepared by lysing cells with Igepal CA-630 followed by differential centrifugation as described before (24,25). The assay for NF-B activation was performed using nuclear extracts and a consensus oligonucleotide (5Ј-AGTTGAGGGGACTTTCCCAGGC-3Ј) that was labeled as described before (25). In some cases, supershift assays were done using human antibodies for p65, p50, and p52 (25).
Cytotoxicity Assessment-Cytotoxicity was determined by the 3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay as described (26) or through the leakage of lactate dehydrogenase (LDH) into the medium. Viability is expressed as the LDH activity in medium, defined as the percentage of total intracellular LDH activity determined by the lysing of cells with 5% Triton-X100. The viability measurement was verified by trypan blue exclusion method.
Flow Cytometry Analysis-Necrosis and apoptosis were evaluated using an annexin V-fluorescein isothiocyanate apoptosis detection kit (Oncogene Research Products, Boston, MA). Briefly, after treatment, cells were collected, washed twice in cold phosphate-buffered saline, and then resuspended in binding buffer at a density of 1 ϫ 10 5 cells/ml. Fluorescein-labeled annexin V and propidium iodide were added to the cells, and the samples were incubated for 15 min before being analyzed with FACScan (FACSCalibur, Becton Dickinson). Annexin V-fluorescein isothiocyanate generated signals were detected with a fluorescein isothiocyanate signal detector (FL1).
Statistical Analysis-Statistical analyses of mean values for multiple comparisons were made by one-way analysis of variance followed by the Fisher's test.

Hypoxia Stimulates Mitochondrial ROS Generation in
HepG2 Cells and Rat Hepatocytes-Although the stimulation of ROS during hypoxia predominantly from mitochondria has been already described, we used this paradigm to specifically address the role of mGSH in the survival of hepatocytes to hypoxia. Furthermore, because previous findings showed a differential tolerance between hepatoma cells and primary hepatocytes to anoxia (27), we compared the ROS stimulation and survival of HepG2 cells and rat hepatocytes cultured under 2% and 5% O 2 . First we validated the source of ROS generated by hypoxia in hepatocytes. As seen, hypoxia stimulated a ROS generation in both HepG2 cells and primary rat hepatocytes with respect to normoxic incubation in a time-dependent fashion that increased with the severity of oxygen deprivation (Fig.  1, A and B). Furthermore, primary rat hepatocytes were more sensitive to hypoxia than HepG2 cells as reflected by the ROS stimulation and loss of survival observed at 2% O 2 exposure (Fig. 1, C and D). Moreover, because previous studies demonstrated that the predominant source of ROS during hypoxia was the mitochondrial electron transport chain (7-9, 13), we examined the effect of inhibition of mitochondrial electron flow at distinct respiratory complexes on the stimulation of ROS in both cell types during 5% O 2 exposure rather than 2% O 2 as this was lethal for rat hepatocytes. Rotenone, a complex I inhibitor, or TTFA, a complex II blocker, decreased ROS formation in a similar fashion and their combination suppressed the stimulation of ROS by 5% O 2 in both HepG2 cells and primary hepatocytes (Fig. 1, E and F). Furthermore diphenyleneiodonium, which blocks flavin-linked enzymes including respiratory complex I (28), reduced the generation of ROS in hypoxic cells, whereas antimycin A, which blocks electron flow at the Q cycle within complex III, enhanced ROS formation caused by 5% O 2 (Fig. 1E), consistent with previous findings (8). Because it was recently reported that rat liver mitochondria generated NO contributing to hypoxia-mediated oxidative stress (29), we examined the effect of L-NAME or uric acid, a peroxynitrite scavenger (30), on the stimulation of ROS, as DCF fluorescence is also indicative of peroxynitrite (31). L-NAME failed to prevent ROS stimulation by 5% O 2 in primary hepatocytes and uric acid did not prevent the ROS stimulation caused by 5% O 2 . In contrast, MnTBAP, a superoxide anion scavenger, abrogated ROS generation during hypoxia (Fig. 1F). Furthermore, allopurinol (200 M), a xanthine oxidase inhibitor, did not prevent the ROS stimulation caused by 5% O 2 (not shown). Thus, these findings established that 5% O 2 causes a burst of superoxide anion generated mainly from the mitochondria in both HepG2 cells or rat hepatocytes and that hepatoma HepG2 cells appear more resistant than primary cells to the oxidative stress induced by hypoxia. Moreover, we observed increased nuclear levels of HIF1␣ and enhanced DNA binding of NF-B in HepG2 or primary hepatocytes subjected to 5% O 2 incubation by ROS-dependent and independent mechanisms (not shown) thus further validating published observations (7-9) on the role of ROS in hypoxia-induced gene expression.
GSH Depletion Sensitizes HepG2 Cells to 5% O 2 -To evaluate the role of GSH on survival during exposure to 5% O 2 , GSH levels were depleted in HepG2 cells during hypoxia. Initially we verified that, consistent with the effect on ROS generation, 5% O 2 depleted GSH in both cytosol and mitochondrial compartments of HepG2 cells (Fig. 2A). However, when cells were incubated with DEM (an ␣-␤ unsaturated carbonyl electrophile, which is conjugated with GSH by GSH S-transferases) during the hypoxic period followed by BSO (a potent and specific inhibitor of the rate-limiting enzyme in GSH biosynthesis, ␥-glutamylcysteine synthetase) to prevent any potential GSH recovery caused by the rapid induction of ␥-glutamylcysteine synthetase by DEM (32), GSH levels in either compartment were severely depleted compared with the levels found in normoxic or hypoxic untreated cells ( Fig. 2A). Although this approach caused a significant increase in ROS formation (Fig.  2B), the survival of normoxic cells did not decrease (Fig. 2C). However, when this strategy was used during the exposure of cells to 5% O 2 this translated into enhanced ROS formation and a significant loss of viability (Fig. 2, B and C). The role of GSH in the control of cell survival was further verified by treatment of cells with GSHEE during exposure of cells to 5% O 2 that caused the replenishment of GSH stores in both cytosol and mitochondria ( Fig. 2A). GSHEE attenuated ROS generation and protected DEM plus BSO-treated HepG2 cells from 5% O 2 -induced cell death (Fig. 2, B and C). In addition, rotenone plus TTFA diminished the enhanced ROS generation in DEM plus BSO-treated HepG2 cells, thus protecting them from 5% O 2 -induced cell death (Fig. 3, B and C). In addition to these results in HepG2 cells, similar findings were observed with primary rat hepatocytes. DEM plus BSO depleted GSH in both cytosol and mitochondrial compartments, which enhanced the susceptibility of hepatocytes to hypoxia (Fig. 3). Moreover, rotenone plus TTFA, which decreased ROS generation during hypoxia (Fig. 1F), protected rat hepatocytes against hypoxia-mediated cell death (Fig. 3). Furthermore, GSH replenishment with GSHEE protected DEM plus BSO-treated rat hepatocytes from 5% O 2 -induced cell death (Fig. 3).
Using flow cytometry we evaluated the percentage of propidium iodide-and/or annexin V-positive HepG2 cell populations exposed to 5% O 2 . Most cells subjected to 5% O 2 were annexin V-and propidium iodide-negative (Fig. 4A). However, DEM plus BSO treatment increased the population of cells with enhanced propidium iodide uptake and annexin V externalization, as well as those positive for both markers (Fig. 4, B  and C). Moreover, consistent with the findings on cell survival (Fig. 2), rotenone plus TTFA decreased the population of DEM plus BSO-treated HepG2 cells positive for propidium iodide uptake and annexin V externalization (Fig. 4, D compared with C). Thus, GSH depletion sensitizes HepG2 cells to 5% O 2induced necrotic and apoptotic death.
Selective mGSH Depletion Determines the Survival of Rat Hepatocytes to 5% O 2 -To examine specifically the role of mGSH on the sensitivity of hepatocytes to hypoxia, we selectively depleted the mitochondrial pool of GSH by HP during the exposure of rat hepatocytes to 5% O 2 . HP is transformed in mitochondria into a Michael acceptor (␣-␤ unsaturated carbonyl) that reacts with GSH in the matrix resulting in its depletion (19,20). The strategy for mGSH depletion by HP is shown in Fig. 5A. 18 -22 h after rat hepatocytes were subjected to hypoxia, HP was added for 5 min for examining, during the last 2-6 h the generation of ROS formation and cell survival. As seen, HP depleted mGSH in hypoxic hepatocytes by 60 -70%, whereas cytosol GSH was unaffected (Fig. 5B). The pool of mGSH remained depleted for up to 8 -10 h following HP exposure (not shown). This treatment during normoxia was accompanied by an increased generation of ROS that was insufficient to cause cell death, indicating that mGSH depletion per se is not cytotoxic as reported previously (19 -21). However, when hepatocytes were treated with HP during 5% O 2 , ROS generation was enhanced (2-5-fold) over time and was correlated with a progressive loss of viability compared with normoxic HP-treated cells or hypoxic untreated hepatocytes (Fig. 5, C  and D). In contrast to this effect in hepatocytes, the levels of mGSH from HepG2 cells following exposure to HP (1-5 mM) were 85-93% of untreated cells, indicating that HP failed to deplete the mitochondrial pool of GSH in this cell line (not shown). In addition, consistent with the suppression of ROS stimulation, rotenone plus TTFA protected rat hepatocytes from hypoxia despite HP treatment (Fig. 6, A and B). Moreover, restoration of GSH by GSHEE rescued mGSH-depleted hepatocytes from hypoxia-induced oxidative stress and cell death (Fig. 6, A and B). Thus, these findings clearly demonstrated the relevance of mGSH in the control of survival during hypoxia rendering hepatocytes susceptible to 5% O 2 after its depletion. DISCUSSION It is known that hypoxia stimulates ROS generation, and the contribution of mitochondria to this process may vary depending on whether reoxygenation follows the hypoxic phase. Although hypoxia/reoxygenation has been shown to stimulate a burst of ROS of mitochondrial and extramitochondrial origin (e.g. xanthine oxidase) (10), hypoxia is known to stimulate mitochondria to generate ROS, as reported, in different cell types cultured under 1.5-5% O 2 (7-9, 13, 33, 34). We showed in this study that in both HepG2 cells and primary rat hepatocytes the main source of superoxide anion during 5% O 2 incubation is the mitochondrial electron transport chain. Thus, inhibition of mitochondrial electron flow at complexes I and II significantly ameliorated the stimulation of ROS by hypoxia in both cell types, whereas MnTBAP suppressed it. Moreover, diphenyleneiodonium, known to block electron flow at complex I (28), prevented ROS generation by 5% O 2 . Previous observations, however, indicated that NO produced in rat liver mitochondria contributed to oxidative stress after hypoxia (29), and the generation of a superoxide anion in the presence of NO is known to yield peroxynitrite. However, we observed that treatment of hepatocytes with L-NAME or uric acid, a peroxynitrite  (30), did not prevent ROS generation caused by 5% O 2 . Furthermore, consistent with these findings, we did not detect changes in the synthesis of NO determined as the release of NO 2 Ϫ , the stable breakdown product of NO (35), in normoxic versus hypoxic (5% O 2 ) rat hepatocytes (not shown), thus further discarding the contribution of stimulated NO in the oxidative stress induced by hypoxia. Thus, it seems that the bulk of superoxide anion generated by 5% O 2 in the absence of reoxygenation derives from the flavin mononucleotide group of complex I from reversed electron transfer as suggested recently (33). In addition, complex III of the respiratory chain has been confirmed as a major source of ROS generation, and the inhibition of mitochondrial respiration is known to stimulate ROS production due to the increased half-life of the intermediates of the electron transport chain at complex I and III that are capable of generating superoxide anion (36,37).
Our data indicated that the inhibition of electron transfer at complexes I/II with rotenone plus TTFA did not completely block the rise in ROS during hypoxia, suggesting alternative ROS sources. For instance, complex III has been reported to be responsible for the ROS generation during hypoxia in Hep3B cells and isolated rat liver mitochondria (7,8), and hence electrons released from the ubisemiquinone site of complex III may contribute to the generation of ROS during hypoxia. In addition, Walford et al. (30) observed that ROS generation in bovine aortic endothelial cells by NO donors was potentiated by hypoxia in wild or 0 cells, indicating the contribution of extramitochondrial sources. Indeed, these authors observed that indomethacin, which in addition to blocking cyclooxygenases has been shown to decrease NAD(P)H oxidase activity (38), reduced hypoxia potentiation of ROS by NO donors (30).
This study addressed the role of GSH, particularly in mitochondria, on the susceptibility of hepatocytes to hypoxia-induced oxidative stress. Consistent with the burst of ROS generation, it has been shown that hypoxia depletes GSH stores (14 -16) and that carbon monoxide decreases GSH/GSSG in rat FIG. 2. GSH depletion sensitizes HepG2 cells to hypoxia. A, GSH was depleted by incubation of hypoxic or normoxic cells with DEM 0.8 mM for 15 min and BSO (1 mM) for the last 8 h of hypoxic incubation, and cells were fractionated into cytosol and mitochondria to determine GSH levels. In some cases GSHEE (2 mM) was added along with BSO to replenish GSH in both compartments. B, in other instances rotenone plus TTFA (2.5 M ϩ 2.0 M) or GSHEE (2 mM) were added during the last 4 h of the 72-h hypoxic period, and ROS production was determined from DCF fluorescence under the various experimental conditions. C, viability was assessed by measuring the leakage of lactate dehydrogenase from HepG2 cells into the culture medium and confirmed by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay. Results are the mean Ϯ S.D. of five independent experiments. Control cytosol GSH and mGSH levels were 28 Ϯ 5 and 6.7 Ϯ 0.8 nmol/mg of protein, respectively. *, p Ͻ 0.05 versus normoxia; **, p Ͻ 0.05 versus hypoxia (72 h, 5% O 2 ). brain mitochondria (18). In agreement with these findings we showed that 5% O 2 results in GSH depletion in both cytosol and mitochondrial compartments. However, the depletion seen during 5% O 2 exposure was insufficient to cause cell death in either HepG2 cells or primary hepatocytes in agreement with previous findings. For instance, hepatoma cell lines have been reported to resist the exposure to various ranges of hypoxia (1.5-5% O 2 ) (7-9), and previous studies showed that 3% O 2 did not cause toxicity in cultured rat hepatocytes despite GSH depletion because of inactivation of methionine adenosyltransferase activity (15). The present study examined for the first time the impact of mGSH depletion on the survival of hepatocytes during exposure to hypoxia. This aim was addressed by two approaches. The abrogation of mitochondrial ROS generation by rotenone plus TTFA rescued both HepG2 cells and primary rat hepatocytes from hypoxia-induced oxidant cell FIG. 3. GSH depletion by DEM plus BSO sensitizes rat hepatocytes to hypoxia. Rat hepatocytes were treated under hypoxia with DEM plus BSO as described in Fig. 2 for HepG2 cells, and then hepatocytes were fractionated into cytosol and mitochondria to determine GSH levels (A). In some cases, rotenone plus TTFA (15 mM ϩ 20 mM) or GSHEE (2 mM) were added during the last 4 h of the 24-h hypoxic period to measure cell viability by leakage of lactate dehydrogenase into the culture medium (B). Control cytosol GSH and mGSH of rat hepatocytes were 33.7 Ϯ 6 and 7.5 Ϯ 0.9 nmol/mg of protein, respectively. death despite severe GSH depletion in both cytosol and mitochondria. Although this approach suggested the relevance of mGSH rather than that of cytosol GSH in the susceptibility of HepG2 cells to hypoxia-induced mitochondrial ROS generation, it was not definitive proof for the vital role of mGSH. Using HP to selectively deplete mGSH in primary rat hepatocytes with the sparing of cytosol GSH (19 -21), we showed that HP-treated hepatocytes became susceptible to 5% O 2 -induced ROS generation, and this sensitivity was prevented if mitochondrial ROS generation was abolished by rotenone plus TTFA or upon GSH replenishment by GSHEE. However, as opposed to primary hepatocytes, HP did not decrease the levels of GSH in mitochondria in HepG2 cells. Although the reason for this differential outcome between primary hepatocytes and HepG2 cells is not fully understood, it may be because of the inability of mitochondria from hepatoma cells to convert HP into the Michael acceptor 3-oxo-4-pentenoate that is then conjugated with GSH (19 -21). Interestingly, lower hydroxybutyrate dehydrogenase activity has been reported in hepatoma cells (39), and whether or not the hydroxybutanoate NAD ϩ oxidoreductase responsible for the mitochondrial biotransformation of HP into 3-oxo-4-pentenoate is functional in HepG2 remains to be established. Thus, the present study widens the protective role of mGSH against oxidant cell death induced by tumor necrosis factor-␣ or sphingolipids (12,20,21,40).
Consistent with previous findings (27), we show that HepG2 cells appear to be more resistant than primary rat hepatocytes to the oxidative stress induced by hypoxia. Although as discussed above the mitochondrial pool of GSH is of relevance in this response, the resistance of HepG2 to hypoxia is not determined by higher mGSH levels in these cells, as the mGSH levels in both cell types are similar (6 -7 nmol/mg of protein).
Rather, it appears that mitochondria from hepatoma cells are less capable of generating a superoxide anion than primary cells in response to hypoxia (Fig. 1). In line with this, many tumor cells including hepatoma cell lines display a lower state 3 respiratory rate and oxidative phosphorylation than normal liver, which is accounted for by their dependence on glycolysis for energy production (27,41,42).
Our data indicate a threshold for ROS increase to inflict cell death in both HepG2 cells and rat hepatocytes. Because the major source of ROS under hypoxia was mitochondria, this FIG. 5. Selective mGSH depletion sensitizes rat hepatocytes to hypoxia. A, primary cultured rat hepatocytes were subjected to 21 or 5% O 2 for 18, 20, or 22 h after which HP (1 mM) was added for 5 min, washed to remove excess HP, and left for the end of the 24-h incubation as shown. B, cells during 5% O 2 were taken after HP addition and fractionated into cytosol and mitochondria to measure the GSH in both compartments, which is expressed as percentage of hypoxia values. Aliquots of hepatocytes incubated under 5% O 2 were taken after HP addition and DCF fluorescence (C), as a measure of ROS generation, and viability (D) by LDH release were measured at the indicated periods of time after HP addition. Data are the mean Ϯ S.D. of three individual experiments. *, p Ͻ 0.05 versus normoxic cells; **, p Ͻ 0.05 versus hypoxic conditions in the absence of HP addition. threshold may relate to the threshold for mGSH depletion to stimulate ROS generation and cell death (43). Although mGSH concentration is high, and moderate mGSH decrease may not impact negatively on ROS generation and cell death, the depletion of mGSH below a critical level would compromise adequate elimination of reactive species and cell survival, particularly in conditions of stimulated ROS generation from the mitochondrial electron transport chain. For instance, under complex III inhibition by antimycin A, stimulated hydrogen peroxide formation increased exponentially when GSH was depleted to below 2 nmol/mg of protein (24), which may correspond to the K m (3 mM) of GSH peroxidase for GSH (44). Thus, although the depletion of mGSH below a critical level may translate to life-threatening accumulation of ROS during hypoxia, a minor contribution of cytosol GSH in the susceptibility to hypoxia cannot be completely ruled out.
Finally, our present findings, although expected, may have important pathological implications. For instance, a hypoxic environment is known to promote tumor growth and survival (1,2), and because mGSH is shown here to be important for the susceptibility of hepatoma cells to oxygen deprivation, its depletion may constitute a promising strategy to sensitize tumor cells to hypoxia. Furthermore, hypoxia is known to contribute to alcohol-mediated hepatocellular injury (45). Under physiological conditions, the oxygen tension within the liver is ϳ65 mm Hg in the periportal area and falls to ϳ35 mm Hg in the perivenous zone (46,47), and the oxidative metabolism of alcohol increases the rate of oxygen uptake accentuating the oxygen gradient existing in the portal to the central venous end of the liver sinusoid. In addition, chronic alcohol feeding is known to deplete mGSH particularly in the perivenous zone (48) that would be expected to contribute to the sensitization of hepatocytes to tumor necrosis factor-␣ (20, 21, 49), as well as to hypoxia. Thus, our data illustrate the synergism between mGSH depletion and hypoxia contributing to the recognized susceptibility of perivenous hepatocytes to the damaging effects of alcohol.