Decreased oxygen availability (hypoxia) promotes physiological processes such as energy metabolism, angiogenesis, cell proliferation and cell viability through the transcription factor HIF (hypoxia-inducible factor). Activation of HIF can also promote pathophysiological processes such as cancer and pulmonary hypertension. The mechanism(s) by which hypoxia activates HIF are the subject of intensive research. In this chapter we outline the model in which mitochondria regulate the stability of HIF through the increased production of ROS (reactive oxygen species) during hypoxia.
Oxygen is necessary for survival of most eukaryotes. Just as carbon sources are used by glycolysis to generate energy, cells utilize oxygen to generate the majority of ATP required for normal cellular processes through oxidative phosphorylation . In the complete absence of oxygen (anoxia, 0–0.3% oxygen), most organisms and cells cannot survive for long periods of time [2,3]. However, in conditions of limited oxygen (hypoxia, 0.3–5% oxygen), cells activate an adaptive response that is mediated by the transcription factor HIF (hypoxia-inducible factor) . This response facilitates survival through the up-regulation of a number of genes involved in various physiological processes such as glycolysis and angiogenesis that promote survival in hypoxic conditions by increasing the capacity for generating ATP and inhibiting the depletion of oxygen. This adaptation response is not due to compromised bioenergetics during hypoxia. The activity of the terminal complex in oxidative phosphorylation, cytochrome c oxidase, is not significantly decreased, and therefore cellular ATP levels do not change in hypoxic conditions. The Km of cytochrome c oxidase is less than 0.3% oxygen. In contrast with hypoxia, cytochrome c oxidase activity is substantially decreased during anoxia resulting in a decrease in the generation of ATP .
The transcriptional activation of HIF is a stress response developed through evolution to allow for cells to avoid a bioenergetic crisis in low oxygen levels, making HIF a master regulator of oxygen homoeostasis. An example of organismal maintenance of oxygen homoeostasis in higher organisms is their ability to increase their capacity for systemic oxygen delivery. This requires erythropoiesis, the process of increased generation of erythrocytes. Erythropoiesis is increased under conditions of decreased oxygen availability through an increase in expression of the protein erythropoietin, a HIF target gene . In fact early examination of the regulation of erythropoietin in response to oxygen led to the discovery of the transcription factor HIF.
Oxygen homoeostasis is also critical for proper development, making HIF-mediated transcription necessary for normal organismal development. The fact that HIF knockout mice are embryonic lethal at embryonic day 9.5 highlights the importance of HIF-mediated transcription in development [7,8]. The lethality displayed by HIF knockout mice is due to impaired vascular development and abnormal placental development . An important HIF target gene is VEGF (vascular endothelial growth factor), which promotes neovascularization and is required for development [10,11]. Unfortunately, genes such as VEGF and other HIF target genes that are required for normal maintenance of oxygen homoeostasis also promote pathophysiological processes such as pulmonary hypertension and cancer. Studies have indicated that preventing HIF activation can suppress tumorigenesis or hypoxia-induced pulmonary hypertension [12–15]. Therefore defining how a cell senses decreased levels of oxygen to regulate the activity of HIF has broad implications for normal physiology as well as for numerous diseases associated with hypoxia. Genetic studies from our laboratory and others support a model in which hypoxia induces mitochondria to increase the generation of ROS (reactive oxygen species) that are both necessary and sufficient to activate the HIF-mediated transcriptional programme for adaptation to hypoxia [16–18]. This chapter focuses on the role of mitochondrial ROS as a key component of oxygen sensing.
Oxygen regulation of HIF
HIF is a heterodimer of two basic helix-loop-helix/PAS (Per-Arnt-Sim) proteins, HIF-α and the ARNT (aryl hydrocarbon receptor nuclear translocator) or HIF-1β . To date three HIF-α isoforms have been described, HIF-1α, HIF-2α and HIF-3a. HIF-1α and HIF-2α are transcriptionally active [20–21]. Their target genes have been described to be both overlapping and distinct. The function of HIF-3α is less well-defined, however it is thought to be an antagonist for HIF-1a- and HIF-2a-mediated transcription . Both HIF-α and ARNT protein subunits are expressed ubiquitously, but the stability of each protein is differentially regulated by oxygen levels. Initial studies demonstrated that HIF-α protein levels are labile at normal oxygen conditions, whereas HIF-β protein levels are constitutively stable . Recent studies indicate that oxygen levels regulate the hydroxylation of two proline residues, 402 and 564, within the ODDD (oxygen-dependent degradation domain) of HIF-α . This hydroxylation reaction is catalysed by a family of PHDs (proline hydroxylation enzymes) [25,26]. The PHDs require Fe2+, oxygen and 2-oxoglutarate to catalyse the hydroxylation reaction. Hydroxylated proline serves as a binding site for the VHL (von Hippel-Lindau) protein, the substrate recognition component of the VBC-CUL-2 E3 ubiquitin ligase complex [27–29]. Once bound, pVHL tags HIF-α with ubiquitin thereby targeting it for proteasomal degradation . HIF-α also contains two transactivation domains. The NAD (N-terminal transactivation domain) is located within the ODDD. The CAD (C-terminus transactivation domain) contains an asparagine residue that is hydroxylated by FIH (factor inhibiting HIF) . This reaction takes place under normoxic conditions (21% oxygen) and it inhibits the transactivation potential of HIF-α . This is due to the inability of HIF-α to interact with the co-activator CBP/p300 as a result of the hydroxy group on Asn803 .
When oxygen levels decrease below 5%, HIF-α protein is stabilized due to a lack of proline hydroxylation (Figure 1) in the absence of proline hydroxylation, VHL protein cannot bind HIF-α to initiate ubiquitin-proteasomal degradation. When stabilized, HIF-α translocates to the nucleus and dimerizes with HIF-β. Once in the nucleus, the HIF dimer binds to HREs (HIF response elements) located throughout the genome . The absence of a hydroxy group on Asn803 allows HIF to associate with the co-activator CBP/p300 to facilitate the transcription of various target genes. Therefore regulation of HIF activity is tightly controlled by the hydroxylation of various amino acids. Understanding how the hydroxylation enzymes are regulated by hypoxia would provide the next step in elucidating the hypoxic response.
Mitochondrial electron transport and HIF
The mitochondria are responsible for the majority of the oxygen consumed within the cell and evolutionarily poised to be a likely target in the hunt for the cellular oxygen sensor. Mitochondria contain their own DNA (termed mtDNA) which encodes 13 genes that are essential for the assembly of a functional electron transport chain. These genes encode subunits for complexes I, III, IV and V. The major consumer of oxygen is complex IV. However, initial studies indicated that pharmacological inhibition of complex IV by cyanide did not activate or repress HIF . Cells cultured with sub-lethal levels of ethidium bromide, which inhibits the transcription and replication of mtDNA resulting in the loss of a functional electron transport chain, were used to genetically address whether mitochondria are involved in the HIF response . These cells devoid of mtDNA, ρ0 cells, are unable to activate HIF-α in hypoxic conditions [36,37]. However, these cells are still able to stabilize HIF-α in anoxic conditions, indicating that there is a fundamental difference in the mechanism required to stabilize HIF-α protein in hypoxic compared with anoxic conditions . These data indicate that the ability of mitochondria to perform electron transport is necessary for hypoxic stabilization of HIF-α but not anoxic stabilization. The mechanism utilized in anoxic conditions is most likely the direct inhibition of the PHDs due to a lack of oxygen. Since the PHDs require oxygen as a co-substrate, they cannot hydroxylate HIF-α in anoxic conditions to initiate degradation, thereby stabilizing HIF-α protein.
After initial studies demonstrating that ρ0 cells fail to activate HIF during hypoxia, there were reports with conflicting data indicating that hypoxic activation of HIF is intact in ρ0 cells [38–40]. However these data were most like due to the use of oxygen levels closer to anoxic conditions than hypoxic conditions. Alternatively, these data could be due to a difference in ρ0 cells generated by ethidium bromide. Recently three independent reports utilized rigorous genetic methods to manipulate mitochondrial electron transport system to determine the requirement of a functional mitochondrial electron transport in hypoxic stabilization of HIF-α protein. The first study used mouse genetics to knockout cytochrome c . In the electron transport chain, cytochrome c accepts electrons from complex III and transfers them to complex IV to be used to reduce molecular oxygen to water. Murine embryonic cells lacking cytochrome c failed to stabilize HIF-α during hypoxic conditions. The other two studies demonstrated that inhibiting complex III function by knocking down the Rieske Fe-S protein inhibits the ability of multiple cell lines to stabilize HIF-α in hypoxic conditions [16,17]. However, cells that displayed knockdown of the Rieske Fe-S protein were able to stabilize HIF-α in anoxic conditions, further supporting the notion that different mechanisms allow for the stabilization of HIF-α protein in hypoxic and anoxic conditions.
These data provide conclusive evidence that a functional electron transport chain is required for the hypoxic stabilization of HIF-α protein. However, the manner by which a functional electron transport chain activates HIF during hypoxia remains controversial. The loss of a functional electron transport chain prevents ROS generation as well as oxygen consumption and both have been proposed as potential mechanisms by which the mitochondrial electron transport chain activates HIF during hypoxia. Both of these mechanisms are discussed below.
Mitochondria regulate HIF dependent on respiration
Molecular oxygen is used as the terminal electron acceptor in the mitochondrial electron transport chain when cytochrome c oxidase converts oxygen into water . This property of mitochondria combined with the requirement of the PHDs for molecular oxygen as a co-substrate led some groups to propose a model in which mitochondrial oxygen consumption is the regulator of HIF activation via the PHDs. These groups propose that during conditions of limited oxygen, mitochondria create an oxygen gradient within the cells as a result of their ability to consume oxygen [41,42]. This gradient would effectively sequester molecular oxygen away from the cytosolic PHDs, thus inhibiting their ability to hydroxylate HIF-α. In contrast, our studies have provided genetic experiments that demonstrate that HIF-α protein stability is independent of respiration in hypoxic conditions. Cells lacking a functional complex IV due to a mutation in the surf gene which encodes for a protein required for the proper assembly of complex IV, retain the ability to stabilize HIF-α protein in hypoxic conditions . These cells that lack complex IV are respiratory deficient and cannot generate a cytosolic oxygen gradient to sequester molecular oxygen from the PHDs. Thus presently there is equivocal data on the role of mitochondrial respiration in the regulation of HIF during hypoxia.
Mitochondria regulate HIF dependent on ROS
Initial studies in exploring the mechanism by which mitochondria regulate the stability of HIF-α demonstrated that ROS levels paradoxically increase in hypoxic conditions . The increase in ROS production during hypoxia is required and sufficient to stabilize HIF-α protein . Cells deficient in cytochrome c or Rieske Fe-S protein which are unable to stabilize HIF during hypoxia also do not display an increase in ROS generation during hypoxia. Incubating cells with pharmacological antioxidants such as ebselen and MitoQ attenuates HIF activation in hypoxic conditions [17,43]. Expression of protein antioxidants such as GPX (glutathione peroxidase) and catalase also attenuates HIF-α protein stabilization and activation, but expression of SOD (superoxide dismutase) has no effect . SOD converts superoxide into H2O2, while GPX and catalase converts H2O2 into water. The specificities of these antioxidants for different forms of ROS tends toward the conclusion that H2O2 is the ROS moiety required for stabilization of HIF-α protein. In fact HIF-α protein is stabilized when cells are pulsed with 25μ t-butyl H2O2, a more stable form of H2O2, in normal oxygen conditions, indicating that H2O2 is also sufficient to activate HIF-mediated transcription . ROS are a normal by-product of electron transport within the mitochondria, therefore the ability of cells to increase the generation of ROS in hypoxic conditions provides the link between mitochondrial electron transport and HIF activation.
A major controversy with the ROS model with respect to hypoxic activation of HIF is whether hypoxia increases ROS. There have been multiple studies that have demonstrated both an increase and a decrease in ROS production during hypoxia [36,38,44–48]. All of these studies utilized dyes which have limitations in their measurements of intracellular oxidative stress . To resolve whether hypoxia increases or decreases ROS, Guzy et al.  have utilized a sensitive FRET (fluorescence resonance energy transfer) probe to assess redox status in the cytosol during hypoxia. This probe consists of a fusion protein containing two fluorescent peptides (CFP and YFP) linked by a redox-sensitive hinge that contains cysteine thiols that become cross-linked by oxidant stress. When expressed in cells the redox-sensitive FRET probe responds to hypoxia by producing a dose-dependent increase in FRET ratio. FRET can be difficult to use, thus the recent development of redox-sensitive GFP (green fluorescent protein) probes will help resolve whether hypoxia increases or decreases ROS. Furthermore, Pan et al.  have shown that MitoQ can inhibit respiration thus making it difficult to discern whether MitoQ works as a respiratory inhibitor or an antioxidant. Future genetic studies need to uncouple ROS generation from oxygen consumption to address which function of the mitochondrial electron transport chain is the major regulator of HIF during hypoxia.
Mitochondria and PHDs
The most proximal regulators of HIF activation are the PHDs. How mitochondrial-generated ROS would cross-talk to the PHDs is not presently known. The addition of exogenous H2O2 to cells under normal oxygen conditions stabilizes the HIF-α protein, implying that ROS inhibit the hydroxylation of the HIF-α protein. Presently there are two mechanisms by which ROS could prevent hydroxylation of HIF-1α protein. One possibility is that low levels of oxygen decrease PHD activity while the ROS decrease the availability of the co-factor Fe2+. Indeed Gerald et al.  found that loss of JunD caused ROS to accumulate which decreased the availability of Fe2+ and reduced the activity of HIF-PHDs . A second possibility is that ROS activate signalling pathways that catalytically make PHDs inactive. Multiple signalling pathways have been implicated in hypoxic stabilization of HIF [52–55]. Finally, it could be that the low oxygen levels decrease both PHD activity and increase the ROS produced during hypoxia. These two events coupled together decrease PHD activity to levels that prevent hydroxylation of HIF-α protein during hypoxia (Figure 2). Further studies need to address how the mitochondria and PHDs crosstalk.
HIF is the master regulator of oxygen homoeostasis, with key roles in various physiological processes especially during development. Aberrant HIF activation leads to pathophysiological processes, most notably cancer, thereby demonstrating the importance of understanding how HIF is regulated by oxygen. Existing literature clearly demonstrates that mitochondria regulate HIF activation during hypoxia. However, it is not clear whether it is the ability of the mitochondria to generate ROS or simply consume oxygen that is the major requirement in regulation of HIF. Furthermore, additional data are required to understand how to couple mitochondria to regulation of PHD activity during hypoxia. Understanding this pathway will provide insight into how hypoxic signalling can be modified in disease to promote physiological processes as well as how to treat disease by hindering the pathophysiological processes governed by HIF.
• Decreased oxygen availability (hypoxia) promotes physiological processes such as energy metabolism, angiogenesis, cell proliferation and cell viability through the transcription factor HIF.
• Activation of HIF can also promote pathophysiological processes like cancer and pulmonary hypertension.
• HIF-1 is composed of two subunits, oxygen sensitive HIF-1α and HIF-β. HIF-1α is hydroxylated at two different proline residues and an asparagine residue under normoxia. The hydroxylation of proline residues serves as a recognition motif for VHL protein. The binding of VHL protein targets the HIF-1α protein for ubiquitin-mediated degradation under hypoxia, the hydroxylation of proline and asparagine is diminished, which allows for the protein to be stabilized and bind to HIF-1β as well as p300/CBP to allow HIF-1-dependent gene transcription.
• Multiple studies have demonstrated that the mitochondrial electron transport chain regulates the activation of HIF during hypoxia. Presently, there are two models to explain the mitochondrial regulation of HIF.
• The first model states that mitochondria create an oxygen gradient within the cells as a result of their ability to consume oxygen. This gradient effectively sequesters molecular oxygen away from the cytosolic PHDs, thus inhibiting their ability to hydroxylate HIF-α.
• The second model postulates that mitochondria generate ROS which are required and sufficient to activate HIF during hypoxia.
• Future genetic studies need to uncouple ROS generation from oxygen consumption to address which function of the mitochondrial electron transport chain is the major regulator of HIF during hypoxia.
- © The Authors Journal compilation © 2007 Biochemical Society