Report

Redistribution of Intracellular Oxygen in Hypoxia by Nitric Oxide: Effect on HIF1α

See allHide authors and affiliations

Science  12 Dec 2003:
Vol. 302, Issue 5652, pp. 1975-1978
DOI: 10.1126/science.1088805

Abstract

Cells exposed to low oxygen concentrations respond by initiating defense mechanisms, including the stabilization of hypoxia-inducible factor (HIF) 1α, a transcription factor that upregulates genes such as those involved in glycolysis and angiogenesis. Nitric oxide and other inhibitors of mitochondrial respiration prevent the stabilization of HIF1α during hypoxia. In studies of cultured cells, we show that this effect is a result of an increase in prolyl hydroxylase–dependent degradation of HIF1α. With the use of Renilla luciferase to detect intracellular oxygen concentrations, we also demonstrate that, upon inhibition of mitochondrial respiration in hypoxia, oxygen is redistributed toward nonrespiratory oxygen-dependent targets such as prolyl hydroxylases so that they do not register hypoxia. Thus, the signaling consequences of hypoxia may be profoundly modified by nitric oxide.

HIF plays a major role in the response of tissues to low partial pressures of O2 (1). The protein stability of the α subunit (HIF1α) of this heterodimeric transcription factor is regulated in an O2-dependent manner (24) by a family of prolyl hydroxylases (5, 6). At low O2 concentrations, prolyl hydroxylase activity is inhibited, and HIF1α accumulates to heterodimerize with HIF1β and activate the expression of HIF-dependent target genes. In earlier work, we found that inhibition of mitochondrial respiration by low concentrations (< 100 μM) of nitric oxide (NO), the endogenous inhibitor of cytochrome c oxidase (complex IV of the respiratory chain), leads to inhibition of HIF1α stabilization at a low O2 concentration (3%). This effect is mimicked by other inhibitors of the respiratory chain, irrespective of the complex at which they act (7).

To explore the underlying mechanism, we investigated the effect of various mitochondrial respiratory inhibitors, including NO, on HEK293 cells (a human embryonic kidney cell line) grown under hypoxic conditions (1% O2). Consistent with our previous results, this treatment prevented the accumulation and transcriptional activity of HIF1α (Fig. 1) (8). Inhibition of mitochondrial respiration also prevented HIF1α stabilization in a number of other cell lines (Fig. 1C). We next determined whether this effect was a result of decreased synthesis or increased degradation of HIF1α protein. Hypoxia-dependent HIF1α stabilization was measured in the presence of the proteasome inhibitor MG-132, which inhibits the degradation of ubiquitinated HIF1α (Fig. 2A). At 21% O2, ubiquitinated HIF1α accumulated in the presence of MG-132. At 1% O2, the respiratory inhibitor myxothiazol prevented the accumulation of ubiquitinated HIF1α in the absence of MG-132 but not in its presence, indicating that myxothiazol does not inhibit HIF1α synthesis but rather stimulates its degradation. When HIF1α was allowed to accumulate by exposing cells for 4 hours to 1% O2 and then adding myxothiazol to inhibit respiration, HIF1α disappeared completely within 45 min. In contrast, addition of cycloheximide to inhibit protein synthesis led to a much slower decline in HIF1α (Fig. 2B). These results indicate that inhibition of respiration promotes HIF1α degradation in hypoxia. This effect was specific to HIF1α, because myxothiazol did not affect the stability of other proteins which are rapidly degraded (fig. S1).

Fig. 1.

Inhibition of mitochondrial respiration prevents hypoxia-dependent HIF1α stabilization. (A) HEK293 cells were incubated at 1% O2 for 3 hours in the presence of the mitochondrial respiratory inhibitors myxothiazol (1 μM), sodium azide (5 mM), antimycin A (1 μg/ml), and rotenone (0.5 μM), as indicated. HIF1α was detected by Western blotting. (B) HIF1α-dependent reporter gene activity was measured at 21% and 1% O2 in the presence of the same respiratory inhibitors as in (A). (C) Four different human cell types (HEK293, Hep 3B hepatocellular carcinoma cells, 143B osteosarcoma cells, and HeLa cervix carcinoma cells) were incubated at 1% O2 in the absence or presence of 1 μM myxothiazol (myx.). HIF1α was detected by Western blotting. (D) HEK293 cells were incubated at 1% O2 for 3 hours (HIF1α protein) or 5 hours (HIF1α reporter activity), respectively, in the absence or presence of the NO donor DETA-NO (10, 30, and 100 μM). HIF1α protein concentrations and transcriptional activity were determined as above. The results shown are representative of three or more independent experiments (means ± SD).

Fig. 2.

Inhibition of mitochondrial respiration in HEK293 cells prevents HIF1α stabilization by promoting HIF1α degradation in a ROS-independent manner. (A) High-molecular-weight ubiquitinated HIF1α protein species were detected by Western blotting with a monoclonal antibody to HIF1α (8) after incubation of cells at 21% or 1% O2 for 3 hours in the absence or presence of 1 μM myxothiazol and 25 μM MG-132, as indicated. (B) HIF1α protein was allowed to accumulate to levels shown at time zero by incubating cells for 4 hours at 1% O2 in a hypoxia chamber. Cells were then treated with 1 μM myxothiazol or 17.5 μM cycloheximide in the hypoxia chamber (without exposing the cells to atmospheric O2 tension). Cell lysates were obtained after 20, 45, and 90 min and analyzed by Western blotting with HIF1α antiserum. (C and D) Cells were incubated at 1% O2 for 3 hours in the presence (C) or the absence (D) of 1 μM myxothiazol. The antioxidants ascorbate (Asc., 2.5 mM), glutathione ethylester (GSH, 5 mM), and N-acetylcysteine (NAC, 5 mM) were included as indicated. HIF1α protein abundance was determined by Western blotting. The results shown are representative of two or more independent experiments.

Reactive oxygen species (ROS) generated by mitochondria have been implicated in the regulation of HIF1α stability (914). If ROS generation were to account for HIF1α destabilization upon inhibition of respiration (1114), antioxidants should reverse this effect, at least in part. However, the presence of the antioxidants ascorbate, glutathione, or N-acetylcysteine did not reverse the inhibition of hypoxia-dependent HIF1α stabilization by myxothiazol (Fig. 2C). Mitochondrial generation of ROS varies according to the point in the respiratory chain where inhibition occurs (1517). However, inhibition of respiration at any site within the respiratory chain completely prevented HIF1α stabilization (Fig. 1) (7). Furthermore, if HIF1α stabilization was mediated through increased mitochondrial ROS production in hypoxia (9, 10), antioxidants should attenuate HIF1α accumulation. However, in our system antioxidants had no effect on hypoxia-dependent HIF1α stabilization (Fig. 2D).

To test whether HIF1α destabilization upon mitochondrial inhibition is dependent on the O2-dependent degradation domain of HIF1α (18), we investigated whether part of this domain is sufficient to regulate the stability of green fluorescent protein (GFP) in a mitochondria-dependent way. Residues 521 to 652 of HIF1α, previously shown to be sufficient to confer O2-dependence (19), were fused to the N-terminus of GFP [HIF1α (521-652)-GFP-V5]. As shown in Fig. 3A, incubation of transfected HEK293 cells at 1% O2 resulted in HIF1α(521-652)-GFP-V5 protein stabilization, an effect that was prevented by the NO donor (Z)-1-[2-aminoethyl-N-(2-ammonioethyl)amino] diazen-1-ium-1, 2-diolate (DETA-NO) and by myxothiazol. Thus, the destabilization of HIF1α upon inhibition of mitochondrial respiration is dependent on the O2-dependent degradation domain of HIF1α.

Fig. 3.

HIF1α destabilization in hypoxia upon inhibition of mitochondrial respiration is dependent on prolyl hydroxylase activity. (A and B) Cells were transfected with the HIF1α (521-652)-GFP-V5 (A) or wild-type and P402A/P564A mutant HIF1α-V5 (B) constructs for 1.5 days and then incubated at 1% O2 for 3 hours in the absence or presence of 1 μM myxothiazol. Protein concentrations were determined by Western blotting with V5 antiserum. (C) Cells were transfected with 0.2 μg pGL3-HRE, 0.2 μg wild-type or P402A/P564A mutant HIF1α-pcDNA3, and 0.4 μg von Hippel-Lindau(VHL)-pcDNA3 plasmid for 24 hours and then incubated for 5 hours at 21% or 1% O2 in the absence or presence of 1 μM myxothiazol. The results shown are representative of three or more independent experiments (means ± SD).

HIF1α is targeted for ubiquitination and proteasomal degradation through hydroxylation of Pro402 and Pro564 in HIF1α by the prolyl hydroxylases. We generated an expression construct of wild-type HIF1α (HIF1α-V5) and a mutant construct (P402A/P564A) in which both Pro402 and Pro564 were replaced by alanines, thereby rendering HIF1α resistant to prolyl hydroxylase–dependent degradation. At 21% O2, wild-type HIF1α-V5 was barely detectable, whereas the P402A/P564A mutant was abundantly expressed (Fig. 3B). At 1% O2, wild-type HIF1α-V5 accumulated, but no further increase in the concentrations of the P402A/P564A mutant was observed, consistent with its lack of regulation by O2. Inhibition of mitochondrial respiration with DETA-NO or myxothiazol reduced accumulation of wild-type but not mutant HIF1α-V5 in hypoxia. Transfection of wild-type HIF1α-V5 markedly increased reporter activity in normoxic cells and to an even greater degree in hypoxic cells (Fig. 3C). This increase was partially reversed when mitochondrial respiration was inhibited with myxothiazol. In contrast, the reporter activity observed with the P402A/P564A mutant was unaffected by respiratory inhibitors at both 21% and 1% O2. These results indicate that the effect of mitochondrial respiratory inhibition on HIF1α stability is dependent on prolyl hydroxylation of Pro402 and Pro564.

It is possible that increases in prolyl hydroxylase activity are a result of increased availability of its substrates, 2-oxyglutarate and O2, or cofactors, e.g., ascorbate. However, addition of ascorbate (Fig. 2C) or 2-oxoglutarate to cells incubated under hypoxic conditions did not affect HIF1α accumulation, even when digitonin was included to facilitate uptake of 2-oxoglutarate (20).

To investigate whether mitochondrial respiratory inhibitors increase intracellular O2 availability, we targeted Renilla luciferase to the mitochondria of HeLa cells as a monitor of available O2. Renilla luciferase activity is solely dependent on its substrates, coelenterazine and O2. The mitochondrial targeting sequence of MnSOD was fused to the N-terminus of the luciferase, and mitochondrial localization of the transfected protein was confirmed by immunostaining with MitoTracker Red (Molecular Probes, Eugene, OR) (20). Renilla luciferase activity was found to correlate directly with the O2 concentration both in immunoprecipitated protein and in intact cells (Fig. 4, A and B). We next equilibrated transfected cells for 1 hour in a hypoxic chamber (1% O2), then treated them with 1 to 100 μM DETA-NO. After the addition of coelenterazine, luminescence was measured in hypoxia. Addition of the NO donor resulted in a concentration-dependent increase in luciferase activity (Fig. 4C). Similar results were obtained when respiration was inhibited by myxothiazol (1 μM) or sodium azide (5 mM) (20). When cytosolic Renilla luciferase was used, 100 μM DETA-NO resulted in a 2.4-fold increase in luminescence (20). When the cells were reequilibrated to atmospheric O2 concentrations (21%), DETA-NO had no effect on the luciferase activity. Furthermore, myxothiazol prevented the HIF1α activity induced by high-density culture (fig. S2). Thus, the effect of NO and O2 distribution is observed only when intracellular O2 concentrations are low.

Fig. 4.

NO increases O2 availability in hypoxia (n = 3; single asterisk indicates P < 0.01; double asterisks, P < 0.005; means ± SEM). (A) Immunoprecipitated mitochondrial Renilla luciferase activity is O2-dependent in vitro. (B) Mitochondrial Renilla luciferase transfected into HeLa cells demonstrates O2-dependence. (C) HeLa cells expressing mitochondrially targeted Renilla luciferase were incubated with increasing concentrations of DETA-NO (0 to 100 μM) in a hypoxia chamber at 1% O2. After the addition of the substrate coelenterazine (5 μg/ml), luminescence was determined in the hypoxia chamber. NO caused a concentration-dependent increase in luciferase activity. When reequilibrated to 21% O2, the effect of DETA-NO on luciferase activity was lost.

We have shown that the destabilization of HIF1α upon inhibition of mitochondrial respiration in hypoxia is dependent on prolyl hydroxylase activity. Because inactivation of prolyl hydroxylases in hypoxia is a result of limited O2 availability, we reasoned that respiratory inhibitors might increase the availability of nonrespiratory O2 and consequently reactivate the enzymes. The experiments with Renilla luciferase indicate that inhibition of mitochondrial respiration can indeed increase cellular O2 availability. This increase would be significant in hypoxia when the cellular O2 concentration becomes limiting for enzymes such as prolyl hydroxylases, which have a higher Km for O2 than does cytochrome c oxidase (21, 22). Our results suggest that NO acts as an endogenous regulator of intracellular O2 availability in mammalian cells (23, 24). Supporting the physiological relevance of the NO-dependent increase in O2 availability is a study showing that bioluminescence in fireflies is dependent on NO release (25).

We have recently discussed metabolic hypoxia, a pathophysiological state in which, although O2 is present, its use in mitochondrial respiration is prevented by occupation of cytochrome c oxidase by NO (23). We now demonstrate that inhibition of mitochondrial O2 consumption creates the paradox of increased O2 availability for prolyl hydroxylation of HIF1α, leading to a situation in which the cell may fail to register hypoxia. It is possible that NO-dependent diversion of O2 may reactivate other enzymes whose activities are reduced in hypoxia. The physiological and pathological implications of these observations remain to be investigated.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5652/1975/DC1

Materials and Methods

Figs. S1 and S2

References and Notes

View Abstract

Navigate This Article