Detecting and responding to hypoxia

Abstract

Adaptation to hypoxia is a topic of considerable clinical relevance, as it influences the pathophysiology of anaemia, polycythaemia, tissue ischaemia and cancer. A growing number of physiologically relevant genes are regulated in response to changes in intracellular oxygen tension. These include genes encoding erythropoietin, vascular endothelial growth factor and tyrosine hydroxylase. Studies on the regulation of the erythropoietin gene have provided insights into the common mechanism of oxygen sensing and signal transduction, leading to activation of the hypoxia-inducible transcription factor 1 (HIF-1). Activation of HIF-1 by hypoxia depends on rescue of its α-subunit from oxygen-dependent degradation in the proteasome, allowing it to form a heterodimer with HIF-1β. This then translocates to the nucleus. There, HIF-1 assembles with a highly conserved orphan nuclear receptor, HNF-4, and a critical transcriptional adaptor, p300. This complex binds to a 3′′enhancer on the erythropoietin gene, enabling transcription of erythropoietin. HIF-1 also activates other genes, the cis-acting elements of which contain cognate hypoxia response elements. There is growing evidence that the oxygen sensor is a flavohaem protein and that the signal transduction pathway involves changes in the level of intracellular reactive oxygen intermediates. We have recently cloned a novel fusion protein called cytochrome b5/b5 reductase, which is a cyanide-insensitive NADPH oxidase and, therefore, a candidate to be the oxygen sensor. This flavohaem protein is widely expressed in cell lines and tissues, with localization in the peri-nuclear space. In the presence of oxygen and iron, it may induce oxidative modifications that target HIF-1α for ubiquitination and degradation.

Introduction

Adaptation to hypoxia is a topic of considerable clinical relevance, as it influences the pathophysiology of anaemia, polycythaemia, tissue ischaemia and cancer [1,2]. Several physiologically relevant genes are now known to be up-regulated in response to changes in intracellular oxygen tension. These include erythropoietin, which regulates red blood cell production; vascular endothelial growth factor (VEGF), which is up-regulated in tumours as well as in ischaemic tissue; tyrosine hydroxylase, an enzyme critical for dopamine synthesis in the carotid body; and glycolytic enzymes, which respond in such a way as to maintain ATP production despite low oxygen availability. These and other biologically important genes have in common either promoter or enhancer elements that respond to hypoxia. Among these, the erythropoietin gene has the most robust response to hypoxia [3].

Erythropoietin acts via a classic physiological feedback loop. It is made primarily in the kidney, in particular in response to hypoxic stress. After release into the circulation, it binds to erythropoietin receptors on erythrocyte progenitor cells and stimulates increased production of red blood cells. An increase in red blood cell mass, in turn, relieves the hypoxia, thereby decreasing erythropoietin production. Additionally, erythropoietin is made in the liver, and although this accounts for less than 10% of the body’s erythropoietin production, it is sufficient to maintain red blood cell production (albeit at a much lower level than normal) even after total nephrectomy.

Regulation of the erythropoietin gene

Regulation of the erythropoietin gene has been a paradigm for understanding the general phenomenon of oxygen-dependent gene regulation [3]. Plasma concentrations of erythropoietin increase markedly with increasing degrees of anaemia. A similar effect can be shown experimentally, e.g. under conditions of hypobaric hypoxia. It thus appears that a common stimulus acting through hypoxia greatly augments the production of erythropoietin.

When the gene for erythropoietin was cloned, it became possible to investigate how regulation occurred, and at what level. In the human hepatoma cell line Hep3B, developed from the human liver, erythropoietin is produced constitutively, with additional marked enhancement of production if the cells are hypoxic. The increase in protein production is reflected in large increases in erythropoietin mRNA on Northern blot analysis as the oxygen tension is reduced (Figure 1). The primary mode of induction thus appears to be increased transcription as oxygen levels decrease. Cobalt (Figure 1) and other transition metals also induce erythropoietin mRNA production in Hep3B cells, in agreement with other data showing that cobalt induces erythropoietin protein production in experimental animals.

Fig. 1.

Northern blot analysis showing increased mRNA concentration in Hep3B cells for erythropoietin (Epo) in response to a reduction in oxygen tension (left) and in response to the presence of cobalt chloride (right).

The erythropoietin gene is now known to consist of five exons, with important regulatory elements downstream of the gene. The critical element that determines the marked up-regulation of the erythropoietin gene during hypoxia lies in an enhancer region on the 3′ side of the gene. This enhancer region can bind to a transcription factor, hypoxia-inducible transcription factor 1 (HIF-1), which forms a complex with a highly conserved orphan nuclear receptor, HNF-4, and a critical transcriptional adaptor, p300. HIF-1 has been cloned [4], and is a heterodimer that binds to a pentanucleotide (5′-RCGTG-3′) response element within the enhancer only when the cell is hypoxic. HNF-4 is a highly conserved, kidney- and liver-specific orphan nuclear factor that binds downstream. Binding of erythropoietin’s 3′ enhancer to this complex is required for hypoxic induction of the erythropoietin gene.

Structure and function of HIF-1

HIF-1 is a heterodimer, consisting of an α- and a β-subunit. The β-subunit is involved in other functions, including partnering the aryl hydrocarbon receptor in a defence mechanism against xenobiotics (particularly aryl hydrocarbons). The α-subunit, however, is specific for hypoxia. The HIF-1 heterodimer binds not only to the erythropoietin hypoxia response element, but also to response elements of a number of other genes that are up-regulated by hypoxia.

HIF can be found in many species. In Daphnia magna, a small water-living crustacean that turns from colourless in normoxic water to red in deoxygenated water, hypoxia induces expression of the haemoglobin gene. It has been shown recently that globin genes in D. magna have HIF response elements, and that a HIF homologue is involved in gene regulation in Drosophila.

HIF-1 binds to DNA in hypoxic cells but not in normoxic cells. At the mRNA level, both HIF-1α and HIF-1β are constitutively expressed. However, at the protein level, HIF-1α is only found in hypoxic cells, whereas HIF-1β is constitutively expressed [5]. This is due to very rapid turnover of HIF-1α protein in oxygenated cells, which is greatly reduced in hypoxic cells. The basis of this stabilization of HIF-1α appears to be a mechanism whereby at low oxygen tensions, the HIF-1α subunit is protected from chemical modification. In the presence of oxygen, there is a hydroxylation of a proline residue within a highly conserved region in the oxygen-dependent degradation domain of HIF-1α [6,7]. This structural modification of HIF-1α enables it to bind to von Hippel–Lindau protein, an interaction necessary for the ubiquitination of HIF-1α and its degradation within the proteasome [8,9]

Search for the oxygen sensor

The demonstration of an oxygen-dependent transcription factor raises the questions of how oxygen is sensed and what signal transduction cascade leads to the stabilization of HIF-1α degradation. It is known that not only hypoxia but also transition metals (cobalt, nickel and manganese) and iron chelators will induce HIF-1 activation. Conversely, HIF-1 activation by hypoxia can be suppressed by haem ligands such as carbon monoxide, nitric oxide and reactive oxygen intermediates. This suggests that the oxygen sensor may be a haem protein, and this concept is supported by circumstantial evidence. The hypothesis that the oxygen sensor is a haem protein allows a tentative mechanism to be proposed, whereby oxygen would bind to a haem group and be converted to a reactive oxygen intermediate that could then serve as an intracellular signalling molecule. The fact that iodinium compounds inhibit the activation of HIF-1 suggests that the sensor may be a flavohaem protein.

Biochemical attempts to isolate novel haem proteins with the properties of an oxygen sensor (Table 1) proved unsuccessful. A genetic approach, however, identified three novel flavoprotein homologues related to cytochrome b5 reductase [10]. The most interesting of these is a protein containing a cytochrome b5-like haem-binding domain in tandem with a cytochrome b5 reductase domain (Figure 2). This is the first example in animals of a single polypeptide containing these two domains.

Table 1.

Proposed properties for the candidate oxygen sensor

Cytochrome b-type absorption spectrum

Flavoprotein

Physiological oxygen affinity

Reactive oxygen species as intermediate

Localized signalling

Ubiquitous among cells and organisms

Fig. 2.

Structure of cytochrome b5 reductase and three novel homologues. The b5/b5R flavohaem protein shown on the bottom is a candidate oxygen sensor.

As the phenomenon of up-regulation of genes in response to hypoxia has been shown to occur in all cell types, one of the criteria for a candidate oxygen sensor is that it is universally expressed. Northern blot analysis shows that the cytochrome b5/b5 reductase (b5/b5R) fusion protein is expressed in a wide variety of human cell lines, and also in a wide variety of organs and tissues. Cytochrome b5/b5R is localized in the cytosol, in the perinuclear space. Some of the properties of this candidate oxygen sensor were determined following expression of the protein in Escherichia coli. It has classic cytochrome b5-type haem spectra in its oxidized and reduced states. It has a relatively low affinity for oxygen, in striking contrast to mitochondrial cytochrome oxidase, which has a much higher oxygen affinity. The oxygen affinity of b5/b5R is in the range that would be expected in order for it to be able to sense oxygen within cells, with a half-maximal velocity at 2% oxygen. It utilizes either NADH or NADPH to convert oxygen to superoxide, and is cyanide insensitive. It can also function as a ferric reductase.

A model for oxygen sensing

From the information presented above, a model can be proposed, shown in Figure 3, encompassing oxygen sensing and subsequent up-regulation of hypoxia-sensitive genes such as erythropoietin. The oxygen sensor acts as the receptor for oxygen in the cell. In the presence of a critical level of oxygen and iron within the cytosol, it activates a proline hydroxylase, which appears to be specific for HIF-1α. Further work is necessary to determine whether b5/b5R participates in this hydroxylation of a specific proline residue in HIF-1α. The modified HIF-1α binds to another protein (von Hippel–Lindau protein), an interaction required for ubiquitination and degradation in the proteasome. When oxygen tension is low, however, HIF-1α is stabilized because its critical proline residue can no longer be hydroxylated. This allows HIF-1α to form a heterodimer with HIF-1β, translocating to the nucleus and up-regulating genes such as the erythropoietin gene.

Fig. 3.

Scheme depicting the mechanism of oxygen sensing and signalling, leading to the hypoxic induction of the erythropoietin gene. Above a critical threshold of intracellular oxygen tension, newly synthesized HIFα subunits (α) are oxidatively modified through interaction with a prolyl hydroxylase (PH). This iron-dependent process results in the hydroxylation of a specific proline residue within a highly conserved region of the HIFα’s internal oxygen-dependent degradation domain. This structural modification is necessary and sufficient for the binding of HIFα to von Hippel–Lindau protein (pVHL), which mediates the assembly of a complex that activates the ubiquitin–E3 ligase (UL). Ubiquitination of HIFα is necessary for uptake and degradation by the proteasome. When cells are hypoxic, proline hydroxylation cannot occur, and therefore HIFα escapes degradation, which enables it to form a stable heterodimer with HIFβ (ARNT). The HIFαβ heterodimer translocates to the nucleus where it binds to cognate hypoxia response elements on genes induced by hypoxia, such as the erythropoietin gene. (Adapted from [12].)

The role of HIF in other pathological conditions

The interest in oxygen sensing and HIF activation extends beyond regulation of red blood cell production, carotid body function and other physiological processes. In particular, oxygen-dependent angiogenesis plays a vital role in common pathological states such as ischaemia and cancer. HIF-1 is a critical inducer of VEGF, thereby enabling growth of new blood vessels needed for tissues to recover from oxygen deprivation. Indeed, forced expression of both VEGF and HIF-1α in situ is being evaluated in the treatment of myocardial infarction. Moreover, HIF-1-dependent angiogenesis plays an equally important role in tumour biology. The von Hippel–Lindau protein is a tumour suppressor. Inactivating mutations of the von Hippel–Lindau protein result in highly vascular tumours, particularly nephromas and angiomas. However, the role of HIF extends far beyond the rare von Hippel–Lindau syndrome. There is a surprisingly high incidence of overexpression of HIF-1α in the majority of common human malignancies [11]. Further research into the role of HIF in these various pathological conditions might eventually lead to new therapeutic interventions.