The ability to sense and respond to changes in oxygenation represents a fundamental property of all metazoan cells. The discovery of the transcription factor HIF-1 has led to the identification of protein hydroxylation as a mechanism by which changes in Po2 are transduced to effect changes in gene expression.
Multicellular life on Earth is based on the use of O2 for the efficient generation of high-energy compounds, and O2 consumption increases with the mass and metabolic activity of the organism. However, exposure to O2 must be limited due to the damaging effects of reactive oxygen species (ROS) on cellular macromolecules. Thus all of the major physiological systems of mammals participate in complex homeostatic mechanisms that are designed to maintain the O2 concentration to which each cell is exposed within a narrow range (FIGURE 1⇓). The study of these systems has occupied physiologists for centuries. During the course of the past century, these studies have been extended to the cellular level. Finally, research over the past decade has produced dramatic insights into the molecular mechanisms underlying oxygen homeostasis during both prenatal and postnatal life.
Control of Oxygen-Regulated Gene Expression by HIF-1
Physiological responses involve changes in gene expression. The blood O2-carrying capacity is maintained by the O2-regulated production of erythropoietin (EPO), which stimulates the proliferation and survival of red blood cell progenitors. Analysis of cis-acting sequences required for increased transcription of the EPO gene in response to hypoxia led to the identification (70), biochemical purification (81), and molecular cloning (79) of hypoxia-inducible factor-1 (HIF-1). EPO is produced primarily within a rare cell type in the kidney. However, HIF-1 is expressed in all cell types and functions as a master regulator of oxygen homeostasis by playing critical roles in both embryonic development and postnatal physiology. HIF-1 has been identified in all metazoan species that have been analyzed from Caenorhabditis elegans to Homo sapiens (organ-isms whose cell numbers differ by more than 10 orders of magnitude), suggesting that the appearance of HIF-1 represented an adaptation that was essential to metazoan evolution.
The expression of over 70 genes is known to be activated at the transcriptional level by HIF-1, and specific HIF-1 binding sites have been identified for many of these genes. Although the list of HIF-1 target genes is extensive (FIGURE 2⇓), it probably underestimates the total number of genes regulated by HIF-1 by at least an order of magnitude. The battery of genes regulated by HIF-1 is different in each cell type, and, for some genes, expression can be induced or repressed by HIF-1 depending on the cell type (34). Among the critical physiological processes regulated by HIF-1 target genes are erythropoiesis, angiogenesis, and glycolysis, which are examples of systemic, local tissue, and intracellular adaptive responses to hypoxia, respectively (30).
HIF-1 is a heterodimeric protein that is composed of HIF-1α and HIF-1β subunits. The amino-terminal half of each subunit consists of basic helix-loop-helix and Per-ARNT-Sim (PAS) domains that mediate heterodimerization and DNA binding. The carboxy-terminal half of HIF-1α contains two transactivation domains that mediate interactions with coactivators such as CREB binding protein (CBP) and p300 (30, 31, 59). Coactivators interact with both sequence-specific DNA binding proteins such as HIF-1 and with the general transcription factors associated with RNA Polymerase II (reviewed in Ref. 69). Coactivators also have histone acetyltransferase activity that is required to make the DNA embedded in chromatin accessible to the polymerase complex for transcription into RNA.
The HIF-1β subunit is constitutively expressed, whereas the expression and activity of the HIF-1α subunit are precisely regulated by the cellular O2 concentration. HIF-1α accumulates instantaneously under hypoxic conditions and on reoxygenation is rapidly degraded, with a half-life of <5 min in posthypoxia tissue culture cells (28, 79). This represents an overestimate of the half-life, because it includes the time required for O2 to diffuse out of the culture medium. In an isolated, perfused, and ventilated lung preparation subjected to hypoxia and reoxygenation, the half-life of HIF-1α is <1 min (85). No protein has been shown to have a shorter half-life.
In addition to HIF-1α, a structurally and functionally related protein designated HIF-2α, which is the product of the EPAS1 gene, can also heterodimerize with HIF-1β (77). HIF-1α:HIF-1β and HIF-2α:HIF-1β heterodimers appear to have overlapping but distinct target gene specificities (22, 73). Unlike HIF-1α, HIF-2α is not expressed in all cell types, and when expressed it can be inactive as a result of cytoplasmic sequestration (56). A third protein, designated HIF-3α, has also been identified (18). Its role has not been well defined, although a splice variant, designated IPAS, has been shown to bind to HIF-1α and inhibit its activity (45, 46).
Molecular Mechanisms of Oxygen Sensing
The mechanism underlying the dramatic regulation of HIF-1α protein expression was a source of great debate, with several models proposed that invoked, for example, the functioning of an O2-binding hemoprotein or an ROS-generating NADPH oxidase as central to the oxygen sensing that determined HIF-1α levels (reviewed in Ref. 66). Among the observations used to support these models was the finding that HIF-1 DNA binding activity and target gene expression were induced in cells exposed to the iron chelator desferrioxamine or to cobalt chloride (80). Remarkably, HIF-1α transactivation domain function is also induced in cells exposed to hypoxia, iron chelation, or cobalt chloride (30, 31, 59), suggesting a common mechanism for regulating both HIF-1α expression and activity.
The O2-dependent degradation of HIF-1α involves ubiquitination and degradation by the 26S proteasome (23, 32, 61). The von Hippel-Lindau tumor suppressor protein (VHL) is required for this process (FIGURE 3⇓), because renal carcinoma cells lacking functional VHL constitutively express HIF-1α and HIF-1 target genes under nonhypoxic conditions (6, 49). VHL forms a complex with elongin B, elongin C, cullin 2, and RBX1 to form an E3 ubiquitin-protein ligase capable of functioning with E1 ubiquitin-activating and E2 ubiquitin-conjugating enzymes to mediate the ubiquitination of HIF-1α (33).
A region of HIF-1α encompassing amino acid residues 400–600 is necessary and sufficient for O2-regulated ubiquitination and degradation (23, 32, 74). VHL interacts, via its β-domain, with amino acid residues 532–585 of HIF-1α (55, 75). Because the ubiquitination and degradation of other key regulatory proteins such as IκB are regulated by phosphorylation, great effort was made to identify phosphorylatable (serine, threonine, tyrosine) residues of HIF-1α that were important for regulation of protein half-life, but to no avail. Instead, Pro-564 is hydroxylated in an O2-dependent manner, and this modification is required for VHL binding (25, 27, 87). Pro-402 represents a second site of hydroxylation and VHL binding (48). Pro-402 and Pro-564 are each contained within a similar amino acid sequence (LXXLAP, where A is alanine, L is leucine, P is proline, and X is any amino acid). HIF-2α and HIF-3α expression are also regulated by prolyl hydroxylation and VHL binding (20, 49, 50).
Three prolyl hydroxylases were identified in mammalian cells and shown to use O2 as a substrate to generate 4-hydroxyproline at residue 402 and/or 564 of HIF-1α (2, 13, 24). These proteins are homologues of EGL-9, which was identified as the HIF-1α prolyl hydroxylase in C. elegans by genetic studies (13). Alternative designations for the three mammalian homologues include EGL-9 homologue (EGLN), prolyl hydroxylase domain protein (PHD), and HIF-1α prolyl hydroxylase (HPH) 1–3. The hydroxylation reaction also requires 2-oxoglutarate (α-ketoglutarate) as a substrate and generates succinate as a side product. Ascorbate is required as a cofactor. The prolyl hydroxylase catalytic site contains an Fe(II) ion that is coordinated by two histidine and one aspartate residue. Unlike heme-containing proteins, the Fe(II) in 2-oxoglutarate-dependent oxygenases can be chelated or substituted by Co(II), rendering the enzyme inactive. Most importantly, these prolyl hydroxylases have a relatively high Km for O2 that is slightly above its atmospheric concentration, such that O2 is rate limiting for enzymatic activity under physiological conditions (13, 20). As a result, changes in the cellular O2 concentration are directly transduced into changes in the rate at which HIF-1α is hydroxylated, ubiquitinated, and degraded. However, a thorough analysis of the relationship between O2 concentration and enzyme activity for each of the PHDs in living cells, and a comparison with the corresponding dose-response curve for HIF-1α expression (29), has not yet been reported. In particular, the plot of HIF-1α protein levels as a function of O2 concentration in HeLa cells yielded a sigmoidal curve suggestive of cooperativity (29), a finding that is not readily explained by the known biochemistry of the HIF-1α prolyl hydroxylases.
Remarkably, HIF-1α transactivation domain function is regulated by O2-dependent hydroxylation of Asn-803, which blocks the binding of the coactivators CBP and p300 (41). Factor inhibiting HIF-1 (FIH-1), which was identified in a yeast two-hybrid screen as a protein that interacts with and inhibits the activity of the HIF-1α transactivation domain (44), functions as the asparaginyl hydroxylase (19, 40). As in the case of the prolyl hydroxylases, FIH-1 appears to use O2 and 2-oxoglutarate and contain Fe(II) in its active site (11, 42, 51), although it has a Km for O2 that is three times lower than the prolyl hydroxylases (37).
Spectroscopic analyses of a peptide from the HIF-1α transactivation domain complexed with the interacting domain of CBP or p300 revealed that Asn-803 is present within an α-helix that is buried deep within the protein interface, where it participates in multiple hydrogen-bonding interactions that are predicted to stabilize the complex (9, 12, 15). Hydroxylation of Asn-803 is predicted to disrupt these protein-protein interactions. Similarly, hydroxylation of Pro-564 has been shown to also function as a molecular switch to positively regulate the interaction of HIF-1α and VHL (21, 53). Thus hydroxylation provides a mechanism for regulating protein-protein interactions, similar to the effect of phosphorylation and other posttranslational modifications. However, what sets hydroxylation apart is that the modification occurs in an O2-dependent manner, thus establishing a direct link between cellular oxygenation and HIF-1 activity.
One remarkable aspect of the O2-sensing system described above is its plasticity. Although O2 may be the limiting substrate for hydroxylation under physiological conditions, it appears that under pathophysiological conditions iron or ascorbate may also be limiting (36). Furthermore, the expression of the PHDs varies from one cell type to another as well as in response to various physiological stimuli, including hypoxia (1, 10, 13, 52). Thus the O2 dose-response curve may be shifted to the left or right under different developmental or physiological conditions. Alternative splicing of the primary RNA transcripts for two of the PHDs provides yet another mechanism for modulating prolyl hydroxylase activity (20). Finally, the transcriptional response elicited by a hypoxic stimulus also demonstrates a remarkable degree of plasticity, because the battery of target genes that is regulated by HIF-1 is unique to each cell type (34). Thus the identification of the molecular components of the O2-sensing system represents a milestone, rather than a finish line, on the course to defining the physiology of oxygen homeostasis.
Developmental and Physiological Consequences of HIF-1 Activity
The identification of HIF-1, VHL, FIH-1, and the PHDs over the past decade has delineated a pathway by which cells sense O2 and respond to changes in oxygenation with changes in gene expression, a property that is fundamental to the cells of all metazoan species. Coincident with these dramatic molecular discoveries have been equally dramatic discoveries regarding the remarkable variety of biological processes in which HIF-1 plays an important role. Analyses of mice, in which expression of HIF-1α has been lost either in all cells (germline knockout) or a single cell lineage (conditional knockout), have identified multiple aspects of development and physiology that are dependent on HIF-1 (TABLE 1⇓). Indeed, the study of HIF-1’s role in development and physiology provides a basis for unifying these two central areas of biology. O2 delivery to cells of the developing embryo becomes limited by diffusion such that establishment of a functioning circulatory system is required for embryonic survival by embryonic day 9 (E9) in the mouse. In wild-type mouse embryos, HIF-1α expression increases dramatically between E8.5 and E9.5, whereas embryos that lack HIF-1α expression die between E9.5 and E10.5 and show cardiac malformations, vascular regression, and massive cell death (7, 26, 39, 60). Complete HIF-2α deficiency is also associated with embryonic lethality (58, 76), and because the embryos survive longer than Hif1a−/− mice, effects on multiple organ systems can be demonstrated (63).
Whereas complete HIF-1α deficiency results in developmental defects, partial HIF-1α deficiency is sufficient to result in impaired responses to physiological stimuli. A particularly dramatic example is the loss of O2 sensing in the carotid body of Hif1a+/− mice (35). Although the carotid bodies are anatomically and histologically normal and depolarize normally in response to cyanide application, they show essentially no response to hypoxia. Thus partial HIF-1α deficiency in the carotid body results in a complete loss of the ability to sense and/or respond to changes in the arterial Po2 by stimulation of the central nervous system cardiorespiratory centers. HIF-2α is also expressed in the mouse carotid body (76), which suggests that HIF-1α and HIF-2α play distinct roles in this organ. The HIF-1 target genes that are critical for O2 sensing and/or efferent responses by the carotid body have not been identified. Remarkably, in the intact animal, other chemoreceptors are less sensitive to Hif1a gene dosage and compensate for the loss of carotid body activity in Hif1a+/− mice (35).
Another dramatic phenotype is the complete inability of Hif1a−/− myeloid cells (granulocytes and macrophages) to respond to inflammatory stimuli (8). Myeloid cells are dependent on glycolysis for ATP generation, perhaps reflecting the hypoxic microenvironment that is often associated with inflammation and infection. HIF-1α deficiency results in ATP deficiency, which impairs critical myeloid cell functions such as aggregation, motility, invasion, and bacterial killing. The role of HIF-1 in immunity is not restricted to myeloid cells, because HIF-1 also plays critical roles in B lymphocyte development (38) and T lymphocyte activation (47). The ability to create mice in which HIF-1α deficiency is restricted to a limited number of cell types (8, 62, 64) is likely to result in the identification of an increasing number of developmental and physiological processes that are regulated by HIF-1.
Medical Consequences of HIF-1 Activity
The preceding sections provide a brief summary of the critical role of HIF-1 in understanding oxygen sensing, development, and physiology. HIF-1 plays an equally important role in disease pathophysiology, including ischemic cardio-vascular disease (3, 67) and cancer (68, 82), the most common causes of mortality in the US population. As a result, there is considerable interest in HIF-1 as a therapeutic target in these disorders (16, 68, 82). In the case of cardiovascular disease, increased HIF-1 activity induced as a result of HIF-1α gene therapy (34, 72, 78), small molecule inhibitors of prolyl hydroxylase activity (20, 24, 43), or inhibitors of HIF-1α-VHL interaction (83) may provide a means to stimulate neovascularization of ischemic tissue. In contrast, small-molecule inhibitors of HIF-1 activity may be useful as anticancer agents (84). However, because HIF-1 functions as a global regulator of oxygen homeostasis, it may not be a useful therapeutic target if the treatment results in unintended and undesirable side effects. An alternative approach may be to focus on the products of HIF-1 target genes. For example, erythropoietin administration may reduce ischemia-induced apoptosis in patients presenting with acute cerebral or myocardial infarction (4, 14, 54, 57). The translation of a rapidly growing body of basic science data into clinical applications looms as the most challenging and most important goal in this exciting field.
- © 2004 Int. Union Physiol. Sci./Am. Physiol. Soc.