The molecular mechanisms by which cells detect hypoxia (1.5% O2), resulting in the stabilization of hypoxia-inducible factor 1α (HIF-1α) protein remain unclear. One model proposes that mitochondrial generation of reactive oxygen species is required to stabilize HIF-1α protein. Primary evidence for this model comes from the observation that cells treated with complex I inhibitors, such as rotenone, or cells that lack mitochondrial DNA (ρ0-cells) fail to generate reactive oxygen species or stabilize HIF-1α protein in response to hypoxia. In the present study, we investigated the role of mitochondria in regulating HIF-1α protein stabilization under anoxia (0% O2). Wild-type A549 and HT1080 cells stabilized HIF-1α protein in response to hypoxia and anoxia. The ρ0-A549 cells and ρ0-HT1080 cells failed to accumulate HIF-1α protein in response to hypoxia. However, both ρ0-A549 and ρ0-HT1080 were able to stabilize HIF-1α protein levels in response to anoxia. Rotenone inhibited hypoxic, but not anoxic, stabilization of HIF-1α protein. These results indicate that a functional electron transport chain is required for hypoxic but not anoxic stabilization of HIF-1α protein.
- rho zero
- complex I
hypoxia transcriptionally activates a multitude of gene products involved in metabolism, angiogenesis, and erythropoiesis through the activation of the transcription factor hypoxia-inducible factor 1 (HIF-1) (for review, see Ref. 28). HIF-1 is a heterodimer of two basic helix loop-helix/PAS proteins, HIF-1α and the aryl hydrocarbon nuclear translocator HIF-1β (32). Aryl hydrocarbon nuclear translocator protein levels are constitutively expressed and not significantly affected by oxygen, whereas HIF-1α protein is present only in hypoxic cells. During normoxia (21% O2), the von Hippel-Lindau tumor suppressor protein (pVHL) binds to the oxygen-dependent degradation domain located in the central region of HIF-1α. This binding results in the subsequent degradation of HIF-1α protein through the ubiquitin-proteasome pathway (24,26). The targeted degradation of HIF-1α via pVHL binding is dependent on the hydroxylation of proline residues within HIF-1α (15, 16). In contrast, the degradation of HIF-1α is suppressed under hypoxic conditions, and transcription of mRNAs encoding hypoxically responsive genes can occur.
A fundamental question for understanding HIF-1α regulation involves the mechanism by which cells sense the lack of oxygen and initiate a signaling cascade that results in the stabilization of HIF-1α protein. Our laboratory has recently proposed that mitochondrial complex III (Bcl complex) serves as an oxygen sensor by increasing the generation of reactive oxygen species (ROS) during hypoxia that are required for HIF-1α protein stabilization (6). In support of this model, hypoxia does not induce an increase in ROS production or stabilize HIF-1α in cells lacking mitochondrial DNA and electron transport activity (ρ0-cells). Complex I inhibitors, such as rotenone and the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, prevent the hypoxic stabilization of HIF-1α by blocking the electron flow proximally to complex III, thereby ablating the ROS generation (1, 6). In contrast, the complex IV inhibitor cyanide is sufficient to activate HIF-1-dependent transcription in wild-type Chinese hamster ovary cells and HT1080 cells under normoxic conditions (34). Furthermore, hypoxic induction of HIF-1α was severely reduced in human xenomitochondrial cybrids harboring a partial (40%) complex I deficiency (1).
An alternative model for oxygen sensing comes from the observation that the hydroxylation of proline residues within HIF-1α is catalyzed by prolyl hydroxylases and requires molecular oxygen and iron (4,9). In the absence of oxygen, HIF-1α would not undergo proline hydroxylation and subsequent pVHL-mediated ubiquitin-targeted degradation. Thus the prolyl hydroxylases have been suggested to be oxygen sensors that regulate stabilization of HIF-1α protein. Moreover, recent studies have demonstrated that ρ0-cells and Chinese hamster lung fibroblasts that display a significant deficiency in complex I activity are still able to stabilize HIF-1α protein at oxygen levels in the range of 0.1–0.5% O2(30, 31). Depending on the cell confluence and the metabolic rate of the cells, intracellular oxygen tensions might have been close to anaerobic conditions in these latter studies. Therefore, the lack of molecular oxygen would prevent proline hydroxylation, and HIF-1α protein would be stabilized independent of intracellular signaling pathways. Accordingly, in the present study, we examined HIF-1α protein levels under hypoxia (1.5% O2) or anoxia (0% O2) in ρ0-cells, cells with reduced complex I activity, and wild-type cells exposed to rotenone. Our results indicate that mitochondrial-dependent oxidant signaling is required for hypoxic but not anoxic stabilization of HIF-1α protein.
Human lung epithelial A549 cells, human fibrosarcoma HT1080 cells, human kidney epithelial HEK-293 cells, and the Chinese hamster lung fibroblast lines CCL16-B2 and CCL16-NDI1 were cultured in DMEM containing pyruvate. The medium was supplemented with penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% heat-inactivated fetal calf serum. Dr. Immo Scheffler provided the CCL16-B2 cells (3). Dr. Takao Yagi provided the CCL16-NDI1 cells (29). Wild-type A549 cells and wild-type HT1080 cells were incubated in medium containing ethidium bromide (50 ng/ml), sodium pyruvate (1 mM), and uridine (50 μg/ml) for 4–6 wk to generate ρ0-A549 cells and ρ0-HT1080 cells (20). These cells were not treated with rotenone or antimycin at any point in the preparation of becoming ρ0-cells. The ρ0 status of cells was confirmed by the absence of cytochrome oxidase subunit II by polymerase chain reaction and the failure to grow in the absence of uridine in the medium. In all experiments, cells were plated at 50% confluence to prevent the development of anaerobic conditions at 1.5% O2.
Hypoxic conditions (1.5% O2, 93.5% N2, and 5% CO2) were achieved in a humidified variable aerobic workstation (INVIVO O2, Ruskinn Technologies). The INVIVO O2 contains an oxygen sensor that continuously monitors the chamber oxygen tension. Anoxic conditions (0% O2, 85% N2, 10% H2, and 5% CO2) were achieved in a humidified anaerobic workstation at 37°C (BugBox, Ruskinn Technologies). An anaerobic color indicator (Oxoid) confirms anaerobicity of the chamber. Before experimentation, media were preequilibrated overnight at either oxygen level.
Measurement of ROS.
ROS generation in cells was assessed by using the fluorescent probe 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate acetyl ester (CM-H2DCFDA) (Molecular Probes). Cells were washed with PBS and loaded with CM-H2DCFDA (10 μM) for 30 min in MEM-α without phenol red. CM-H2DCFDA is a cell-permeable indicator for ROS that is nonfluorescent until the removal of the acetate groups by intracellular esterases and oxidation occurs within the cell (23). Cells were exposed to normoxia (21% O2), hypoxia (1.5% O2), or anoxia (0% O2) for 4 h in the presence of the CM-H2DCFDA (10 μM), as described above. All procedures were carried out in the dark. After the exposure, the cells were washed and removed from the plate by using PBS, without calcium or magnesium, containing 1 mM EDTA, centrifuged (200 g for 5 min), and resuspendend in PBS. Fluorescence was measured at the flow cytometry core facility at the Cancer Center of Northwestern University. The graphs were generated by using the Beckman Coulter Epics XL-MCL running System II software (version 3.0).
Analysis of HIF-1α protein by immunoblotting.
HIF-1α protein was analyzed in nuclear extracts prepared from cells, as previously described (6). Nuclear extract (40 μg) was mixed with an equal volume of electrophoresis buffer (1.0 ml glycerol, 0.5 ml β-mercaptoethanol, 3.0 ml 10% SDS, 1.25 ml 1.0 M Tris · HCl, pH 6.7, and 1–2 mg bromophenol blue). After heating, the protein was resolved on a 7.5% polyacrylamide-SDS gel and transferred to Hybond-enhanced chemiluminescence nitrocellulose paper (Amersham). After transfer, the gel was stained with Ponceau stain to verify uniform loading and transfer. Membranes were blocked with 5% milk in TBS-T (10 mM Tris · HCl, 150 mM NaCl, 0.1% Tween-20, pH 8.0) for 3 h at room temperature and subsequently incubated with 2.0 μg/ml of HIF-1α antibody (Novus Biolgical Sciences) overnight at 4°C. The membrane was washed three times with TBS-T and incubated for 1 h at room temperature with 1.0 μg/ml horseradish peroxidase-conjugated secondary antibody (Cell Signaling). Subsequently, the membrane was washed three times with TBS-T and analyzed by electrochemiluminescence (Amersham).
Transfection and reporter gene assays.
Transfections of A549 and ρ0-A549 cells were carried out on cells plated on 35-mm petri dishes at 30–50% confluence by using LipofectAMINE reagent (Life Technologies,) according to the manufacturer's protocol. A typical transfection was performed by using 0.5 μg of a luciferase reporter driven by a trimer of a hypoxic response element (HRE). The DNA-lipofectamine was incubated with plated cells for 24 h. Subsequently, the media were replaced, and cells were exposed to various conditions. Cell lysis was performed by using a reporter gene lysis buffer from Promega. Luciferase assays were performed by using the Luciferase assay system (Promega). Data were normalized by using total protein concentration as determined by the Bio-Rad protein assay (Bio-Rad).
Measurement of ATP levels and oxygen consumption rates.
ATP levels were measured by the luciferin/luciferase method by using an ATP bioluminescence assay kit HS II (Roche Molecular Biochemicals). Cells were lysed with lysis buffer provided by the manufacturer. Luciferase reagent (50 μl) was manually injected into 50 μl of cell lysate, and luminescence was analyzed after a 30-s delay with a 2-s integration on a SpectraMax Gemini microplate reader (Molecular Devices). A standard curve was generated from known concentrations of ATP and used to calculate the concentration of ATP in each sample. Luminescence increased linearly with the negative log of the ATP concentration in the samples over the range of concentrations measured. Data were normalized by using total protein concentration as determined by the Bio-Rad protein assay (Bio-Rad). Oxygen consumption was measured by using the Oxytherm respirometer (Eurosep instruments). The Oxytherm system provides a simple method for computer-controlled measurement of oxygen consumption from liquid suspensions of 200 μl to 2.5 ml.
Quantitation of apoptosis.
Cells were fixed in 100% methanol for 2 min at −20°C, and nuclei were stained with 1 μg/ml Hoescht no. 33258 for 30 min. The percentage of apoptotic cells was determined by scoring for fragmented and/or condensed nuclei, as visualized by fluorescence microscopy. For each treatment condition, at least 200 nuclei were scored.
Hypoxic but not anoxic stabilization of HIF-1α protein requires a functional electron transport chain.
Previously, our laboratory has shown that a functional electron transport chain is required for the hypoxic (1.5% O2) stabilization of HIF-1α protein in human hepatoma Hep3B cells (6). It is not known whether a functional electron transport chain is required for the anoxic (0% O2) stabilization of HIF-1α protein. To test this hypothesis, wild-type and ρ0-cells of human lung epithelial A549 cells and human fibrosarcoma HT1080 cells were exposed to hypoxic and anoxic conditions for 4 h (Fig. 1,A and B, respectively). Wild-type A549 cells and wild-type HT1080 cells stabilized HIF-1α protein in response to both hypoxia and anoxia. The ρ0-A549 cells and ρ0-HT1080 cells failed to accumulate HIF-1α protein levels in response to hypoxia. However, both ρ0-A549 cells and ρ0-HT1080 cells were able to stabilize HIF-1α protein levels in response to anoxia. These results indicate that hypoxic but not anoxic stabilization of HIF-1α protein requires a functional electron transport chain.
The requirement of a functional electron transport for the hypoxic stabilization of HIF-1α protein in Hep3B cells has been attributed to an increase in mitochondrial ROS (6). To investigate whether A549 cells and HT1080 cells also increase mitochondrial-derived ROS in response to hypoxia, wild-type and ρ0 A549 and HT1080 cells were exposed to hypoxia (1.5% O2) for 4 h by using the fluorescent dye CM-H2DCFDA, a modified version of 2′,7′-dichlorofluorescin (DCFH-DA) dye (23). Previously, our laboratory as well as other investigators have used the DCFH-DA dye and fluorescence microscopy to demonstrate that mitochondrial ROS increase during hypoxia in a variety of cell types and tissues (5, 19, 22,36). The main advantage of CM-H2DCFDA over DCFH-DA is the better retention of the dye within the cells. This allows the use of flow cytometry to examine fluorescence. During hypoxia, both wild-type A549 cells and wild-type HT1080 cells showed an increase in the fluorescence of CM-H2DCFDA, indicating an increase in ROS (Figs. 2 and3). In contrast, ρ0-A549 cells and ρ0-HT1080 cells did not display an increase in CM-H2DCFDA fluorescence in response to hypoxia (Figs. 2 and3). We further examined the relationship of oxygen levels and ROS generation by incubating HT1080 cells and A549 cells under normoxia and anoxia. As expected, in the absence of oxygen, there was no increase in CM-H2DCFDA fluorescence in response to anoxia compared with normoxia (Fig. 4). In fact, there was a decrease in oxidation of the dye under anoxic conditions. Taken together, these results indicate that hypoxia, but not anoxia, increases intracellular generation of ROS from the mitochondria.
Hypoxia stimulates mitochondrial ROS generation in Chinese hamster fibroblasts deficient in complex I activity.
Recently, Vaux et al. (31) have reported that CCL16-B2 cells containing a significant loss of complex I activity (<10%) still exhibit HIF-1α protein stabilization at 0.1% O2, suggesting that complex I inhibition does not affect hypoxic HIF-1α protein stabilization. We investigated HIF-1α protein levels in CCL16-B2 cells and CCL16-NDI1 cells under hypoxic and anoxic conditions. The CCL16-B2 cells require glucose for growth and survival (3, 8). These cells undergo rapid death in media containing galactose instead of glucose. The CCL16-NDI1 cells are CCL16-B2 cells that have been stably transfected with the NDI1gene that encodes the rotenone-insensitive internal NADH-quinone oxidoreductase of Saccharomyces cerevisiae mitochondria (29). The NDI1 gene restores the NADH oxidase activity of complex I. Thus CCL16-NDI1 cells are respiratory competent and can grow in either glucose or galactose media. CCL16-B2 cells and CCL16-NDI1 cells were both able to stabilize HIF-1α protein levels under hypoxic and anoxic conditions (Fig.5 A). A possible explanation for the ability of CCL16-B2 cells to stabilize HIF-1α protein is that these cells retain their ability to generate ROS during hypoxia. Indeed, hypoxia stimulated ROS production, as measured by the oxidation of the fluorescent dye CM-H2DCFDA in both CCL16-B2 cells and CCL16-NDI1 cells (Fig. 5, B and C). These results indicate that cells with decreased complex I activity are still able to generate ROS during hypoxia, thereby stabilizing HIF-1α protein.
Rotenone abolishes hypoxic but not anoxic stabilization of HIF-1α protein.
Complex I inhibitors such as rotenone have been shown to inhibit the stabilization of HIF-1α protein at 1–1.5% O2, presumably by blocking the electron flow proximally to complex III, thereby ablating the ROS generation during hypoxia (6,14). To determine whether rotenone abolishes both the hypoxic and the anoxic stabilization of HIF-1α protein, HEK-293 cells were exposed to rotenone (0.5 μg/ml) for 30 min under normal oxygen conditions followed by exposure to hypoxia (1.5% O2) or anoxia (0% O2). Rotenone abolished the hypoxic stabilization of HIF-1α protein (Fig.6). In contrast, rotenone failed to affect the anoxic stabilization of HIF-1α protein (Fig. 6). To confirm that rotenone prevented hypoxic stabilization of HIF-1α protein by inhibiting complex I, we investigated whether methyl-succinate would restore the ability of HEK-293 cells to generate ROS and stabilize HIF-1α protein in the presence of rotenone. Methyl-succinate is a complex II substrate that can provide electrons to complex III in the presence of rotenone, thus allowing cells to generate ROS at complex III during hypoxia. HEK-293 cells increased ROS production and stabilized HIF-1α protein during hypoxia, as measured by the fluorescent dye CM-H2DCFDA (Fig. 7). Rotenone, but not antimycin, prevented the increase in CM-H2DCFDA fluorescence and HIF-1α protein stabilization during hypoxia (Fig. 7). Antimycin inhibits complex III distally to the site of ROS production, thus maintaining the generation of ROS during hypoxia. Rotenone also abolished the increase in CM-H2DCFDA fluorescence and HIF-1α protein stabilization observed during hypoxia in the presence of antimycin (Fig. 7). Methyl-succinate restored CM-H2DCFDA fluorescence and HIF-1α protein stabilization during hypoxia in the presence of rotenone and antimycin (Fig. 7). The complex II inhibitor thenoyltrifluoroacetone abolished CM-H2DCFDA fluorescence and HIF-1α protein stabilization during hypoxia in the presence of rotenone, antimycin, and methyl-succinate (Fig. 7). These results suggest that rotenone prevents HIF-1α protein stabilization during hypoxia by inhibiting complex I in HEK-293 cells.
Functional consequences of hypoxia and anoxia.
We further examined the difference between hypoxia and anoxia by determining the effects of oxygen concentration on respiratory rates, ATP levels, HIF-1-dependent transcription, and apoptosis. To investigate the changes in respiratory rate as a function of oxygen levels, fifteen million A549 cells were placed in a respirometer. As seen in Fig. 8 A, the oxygen consumption rate of A549 cells was independent of oxygen levels until the oxygen levels reached ∼0.5%. Below 0.5% oxygen levels, the respiratory rate was dependent on oxygen levels. Furthermore, ATP levels in A549 cells did not change between 21 and 1.5% O2over 8 h (Fig. 8 B). In contrast, there was a ∼60% decrease in ATP levels at 0% O2 compared with 21% O2. To examine HIF-1-dependent transcription, A549 cells were transiently transfected with a luciferase reporter construct driven by a trimer of the HRE and exposed to hypoxia or anoxia for 8 h (Fig. 9 A). Both hypoxia and anoxia were able to activate HIF-1-dependent transcription, as measured by an increase in HRE-luciferase expression. To address whether a functional electron transport chain was required for hypoxic or anoxic stimulation of HIF-1-dependent expression, ρ0-A549 cells were transfected with a luciferase reporter construct driven by the HRE and exposed to hypoxia or anoxia for 8 h. Hypoxia-induced HRE-luciferase expression was not observed in ρ0-A549 cells. However, anoxia induced HRE-luciferase (Fig. 9 A). Although both hypoxia and anoxia can activate HIF-1-dependent transcription, the long-term consequences of hypoxia and anoxia are different with regard to apoptosis. A549 cells exposed to anoxia underwent apoptosis after 48 h, whereas cells exposed to hypoxia did not (Fig. 9 B). Collectively, these findings indicate that there are physiologically different outcomes in cells exposed to hypoxia compared with anoxia.
The molecular events regulating HIF-1α protein stabilization during hypoxia are important for understanding the mechanisms of cellular oxygen sensing. Our laboratory has previously proposed a model in which the increased generation of ROS at complex III of the mitochondrial electron transport chain serves as the oxygen sensor for HIF-1α protein stabilization during hypoxia (6). We demonstrated that the stabilization of HIF-1α protein at oxygen concentrations of 1–2% required a functional electron transport chain. In the present study, we confirm these findings and also demonstrate that anoxic stabilization of HIF-1α protein does not require a functional electron transport chain. This observation is consistent with the requirement of proline hydroxylation as a mechanism for HIF-1α protein degradation under normal oxygen conditions. In the absence of oxygen, hydroxylation of proline residues within HIF-1α by prolyl hydroxylases cannot occur, and intracellular signaling events are not required for the stabilization of HIF-1α protein. Thus prolyl hydroxylases would effectively serve directly as the oxygen sensors during anoxia.
Our present data are not consistent with the hypothesis that prolyl hydroxylases serve as the primary oxygen sensor regulating the hypoxic stabilization of HIF-1α protein. Rather, our data suggest that the stabilization of HIF-1α protein under 1–2% O2levels requires the activation of intracellular signaling pathways by a mitochondrial-dependent ROS signal. Recently, Hirota and Semenza (14) have shown that hypoxia (1% O2) stimulates Rac1 activity, and Rac1 is required for the hypoxic stabilization of HIF-1α protein. Both the hypoxic activation of Rac1 and the stabilization of HIF-1α protein were abolished by the complex I inhibitor rotenone. These results indicate that Rac1 is downstream of mitochondrial signaling. Additional intracellular signaling systems, including phosphatidylinositol 3-kinase and diacylglycerol kinase, have been shown to be required for hypoxic (1% O2) stabilization of HIF-1α protein (2, 37). Moreover, mitochondrial-dependent oxidant signaling has been shown to regulate HIF-1α protein accumulation after exposure to tumor necrosis factor-α (13). Nonmitochondrial-dependent oxidant signaling has also been shown to stabilize HIF-1α protein under normoxia. For example, thrombin or angiotensin II stabilizes HIF-1α under normoxia through an increase in ROS generation from nonmitochondrial sources (12, 27). Therefore, if prolyl hydroxylase is by itself the oxygen sensor for both hypoxic and anoxia-induced HIF-1, then there would be no signaling required upstream of prolyl hydroxylase. There should be no requirement for activation of kinases or oxidant-dependent signaling upstream of prolyl hydroxylase. We speculate that the ultimate target of the oxidant-dependent signaling pathway originating from mitochondria during hypoxia or nonmitochondrial sources such as angiotensin II or tumor necrosis factor-α during normoxia is to inhibit proline hydroxylation (Fig. 10).
Two previous observations further support the premise that hypoxic signaling is distinct from anoxia with respect to HIF-1 activation. First, Gleadle et al. (10) have demonstrated that diphenylene iodonium (DPI), an inhibitor of a wide range of flavoproteins, including complex I, prevents stabilization of HIF-1α protein and HIF-1 target genes at oxygen levels of 1%. However, DPI fails to affect stabilization of HIF-1 in response to the iron chelator desferrioxamine (DFO). Iron and oxygen are required for prolyl hydroxylases to be catalytically active. Thus DFO or anoxia directly inhibit prolyl hydroxylase activity because of substrate limitations and stabilize HIF-1α protein. Interestingly, DPI can prevent a variety of other hypoxic responses, such as pulmonary vasoconstriction and carotid body nerve firing. Second, Jiang et al. (17a) have demonstrated that HIF-1α protein levels stabilize half-maximally between 1.5 and 2% O2 and are maximally stabilized at 0.5%. If prolyl hydroxylase were the oxygen sensor over a wide range of oxygen tensions, then HIF-1α protein levels should be maximally stabilized at 0% O2. Taken together, these previous findings and our present observations suggest distinct pathways of initiating the hypoxic response compared with DFO or anoxia.
A more fundamental difference between oxygen levels of 1–2% and 0–0.5% is the metabolic state of the cells. Below an oxygen concentration of 0.5%, molecular oxygen begins to limit the respiratory rate, causing ATP levels to decrease (18, 35). Exposure of cells to 1–2% O2 compared with 0% O2 can have varying consequences for intracellular signaling, cellular metabolism, and survival. For example, cells do not undergo apoptosis or growth arrest at 1–2% O2(7). However, oxygen concentrations closer to 0% cause cells to undergo apoptosis that requires a functional electron transport chain and the Bcl-2 family members Bax or Bak (25). Gleadle and Ratcliffe (11) have also demonstrated that src kinase is activated at oxygen levels of 0.1 or 0% O2 but not at 1% O2 (11).
Our results also provide an explanation for the differing results obtained from using ρ0-cells to examine mitochondrial regulation. Previously Vaux et al. (31) and Srinivas et al. (30) have demonstrated that ρ0-cells are still able to stabilize HIF-1α protein at oxygen levels of 0.1 or 0.5% O2. The use of antimycin and rotenone as a selection procedure in the generation of ρ0-cells has been cited as a potential explanation for the difference in our previous results compared with those of Vaux et al. and Srinivas et al. In the present study, we generated ρ0-cells by exposing cells to ethidium bromide for 4–6 wk in the absence of any mitochondrial inhibitors. A possible explanation for the conflicting results is that cells exposed to oxygen levels of 0.1 or 0.5% are likely to experience anaerobic conditions within the cytosol, because mitochondrial cytochrome-c oxidase will consume residual molecular oxygen. Taken together, these findings suggest that intracellular signaling pathways are likely to be important for hypoxic but not anoxic stabilization of HIF-1α protein.
An alternative genetic strategy to using ρ0-cells in examining the role of mitochondria in the regulation of HIF-1α is to use cells with diminished complex I activity. Agani et al. (1) have demonstrated that cybrid cells containing a partial defect in complex I activity have reduced HIF-1α protein levels at 1% O2. Succinate, a mitochondrial complex II substrate, restored the hypoxic response in cybrid cells, suggesting that electron transport chain activity is required for the stabilization of HIF-1α protein. In contrast, Vaux et al. (31) have shown that Chinese hamster fibroblasts CCL16-B2 which contain complex I activity <10% of control cells are still able to stabilize HIF-1α protein at 0.1% O2. In the present study, we found that CCL16-B2 cells were still able to generate ROS during hypoxia and stabilize HIF-1α protein under both hypoxic and anoxic conditions. Mitochondrial generation of ROS has been estimated to be 1–2% of the total electron flux through the respiratory chain (17). The electron flux required for ROS generation at complex III is likely to be significantly less than the electron flux required to reduce molecular oxygen to H2O at cytochrome-c oxidase. Thus a significant loss of complex I activity might significantly inhibit oxygen consumption at cytochrome-c oxidase but still provide sufficient electrons to generate ROS at complex III, thus stabilizing HIF-1α protein during hypoxia.
Our present observations are also consistent with the role of complex I inhibitors in preventing the stabilization of HIF-1α protein. Previously, our laboratory reported that a 1 μg/ml concentration of rotenone inhibits the stabilization of HIF-1α protein at 1.5% O2 in Hep3B cells (6). These results have been corroborated by Hirota and Semenza (14), who reported that rotenone at a concentration of 1.0 μg/ml inhibited the stabilization of HIF-1α protein at 1.0% O2 in HEK-293 cells. The neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, another complex I inhibitor, prevents hypoxic accumulation of HIF-1α protein in dopaminergic cell lines (1). In the present study, we found that rotenone (0.5 μg/ml) abolished ROS generation and the stabilization of HIF-1α protein during hypoxia. The ability of succinate to restore ROS generation and to stabilize HIF-1α protein in the presence of rotenone during hypoxia and the abolishment of this effect by complex II inhibitor suggest that rotenone exerts its effects specifically by inhibiting mitochondrial complex I. Rotenone did not abolish HIF-1α protein stabilization during anoxia. These results are consistent with the hypothesis that rotenone affects hypoxic but not anoxic stabilization of HIF-1α protein. Furthermore, these results potentially explain why rotenone failed to inhibit HIF-1α protein stabilization in previous studies that examined HIF-1α protein levels at 0.1% O2 (31).
In summary, our present results demonstrate mitochondrial-dependent signaling for the hypoxic stabilization of HIF-1α. Data from a variety of other investigators suggest that other signaling elements are also likely to be required for the hypoxic stabilization of HIF-1α. In contrast, the anoxic stabilization of HIF-1α does not require mitochondrial-dependent signaling. We would predict that HIF-1α stabilization under anoxic conditions occurs in the absence of intracellular signaling upstream of proline hydroxylation. Mitochondria might also serve as oxygen sensors for a variety of other hypoxic events. For example, we and other investigators have demonstrated that similar mitochondrial-dependent signaling is required for hypoxic pulmonary vasoconstriction (21, 33). Future experiments will have to address how hypoxia stimulates the production of mitochondrial ROS and the downstream targets of ROS that result in a diverse activation of responses to hypoxia.
We thank Mary Paniagua and Mehrnoosh Abshari at the Flow Cytometry Facility for technical assistance.
This study was supported by the Crane Asthma Center and National Institute of General Medical Sciences Grant GM-60472-03 (to N. S. Chandel).
Address for reprint requests and other correspondence: N. S. Chandel, Division of Pulmonary & Critical Care, Tarry Bldg. 14–707, 300 East Superior St., Chicago, IL 60611–3010 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
May 24, 2002;10.1152/ajplung.00014.2002
- Copyright © 2002 the American Physiological Society