Low cellular oxygenation (hypoxia) represents a significant threat to the viability of affected tissues. Multicellular organisms have evolved a highly conserved signalling pathway that directs many of the changes in gene expression that underpin physiological oxygen homoeostasis. Oxygen-sensing enzymes in this pathway control the activity of the HIF (hypoxia-inducible factor) transcription factor by the direct incorporation of molecular oxygen into the post-translational hydroxylation of specific residues. This represents the canonical hypoxia signalling pathway which regulates a plethora of genes involved in adaptation to hypoxia. The HIF hydroxylases have been identified in other biological contexts, consistent with the possibility that they have other substrates. Furthermore, several intracellular proteins have been demonstrated, directly or indirectly, to be hydroxylated, although the protein hydroxylases responsible have yet to be identified. This chapter will summarize what is currently known about the canonical HIF hydroxylase signalling pathway and will speculate on the existence of other oxygen-sensing enzymes and the role they may play in signalling hypoxia through other pathways.
Molecular oxygen acts as the terminal electron acceptor in mitochondrial oxidative phosphorylation. As such it is required for efficient ATP production, which is essential for most biological processes. Metazoans have evolved specialized systems that respond to oxygen starvation by altering cell metabolism and survival, and rescuing oxygen delivery to the affected tissue. A key player mediating hypoxia-induced responses is the HIF (hypoxia-inducible factor) transcription factor complex (reviewed in [1–8]). Oxygen-dependent regulation of HIF activity controls an array of hypoxia-inducible genes that together programme the cell to respond to changes in oxygen availability.
HIF is a heterodimer consisting of a constitutively expressed β-subunit and an α-subunit which is induced when oxygen levels are low (Figure 1) [4,6,8]. HIF-α and HIF-β are members of the PAS (Per-Arnt-Sim) family of bHLH (basic helix-loop-helix) proteins (Figure 2), each of which are expressed as three proteins from independent genes. Of the α-subunits, HIF-1α and HIF-2α display the greatest similarity both in terms of sequence homology, and function. HIF-3a is more divergent, and its regulation and function are less clearly defined. Under hypoxic conditions, HIF-1α and HIF-2α protein levels increase dramatically. This response is followed by nuclear translocation, association with HIF-β, recruitment of co-activators and activation of transcription, with at least some of these processes constituting independent points of control by oxygen (Figure 1) .
HIF target genes and adaption to hypoxia
Genes that are direct HIF transcriptional targets contain a G/ACGTG motif termed the HRE (hypoxia-responsive element); such HREs being found in genes that regulate both systemic and cellular responses to hypoxia. The HIF system was discovered in studies of the strikingly hypoxia-responsive EPO (erythropoietin) gene, as a DNA-binding complex mediating the activity of an HRE in the EPO 3′-flanking region . In hypoxia, EPO production increases several hundred fold and enhances the oxygen-carrying capacity of the blood by stimulating red blood cell production.
It is now recognized that HIF also promotes delivery of oxygen to hypoxic tissues by stimulating angiogenesis via the upregulation of growth factors such as the VEGF (vascular endothelial growth factor) family and its receptors. Other systemic responses mediated by HIF include the regulation of vascular tone through hypoxic induction of genes encoding vasomotor regulators such as nitric oxide synthases, haem oxygenase, endothelin-1 and adrenomedullin .
Cellular responses to hypoxia controlled by HIF include changes in energy metabolism, cell growth, survival and migration. HIF directs the co-ordinate regulation of many genes in these pathways. For instance HREs are present in the genes encoding glucose transporters and glycolytic enzymes, as well as regulators of aerobic and anaerobic metabolism [5,8]. Together they switch metabolism from oxidative phosphorylation to anaerobic glycolysis in order to maintain ATP production during hypoxia.
Another critical response to chronic hypoxia is cell death, a process in which HIF has also been implicated. Paradoxically, HIF also up-regulates genes involved in cell survival, suggesting that in some circumstances cells may be subjected to conflicting HIF-dependent signals. The outcome may be influenced by the severity and duration of the oxygen starvation or associated stresses such as acidosis. Hypoxic cell death has been associated with transcriptional activation of the Bcl-2 family member, BNIP3 (Bcl-2/adenovirus E1B 19kDa interacting protein 3) . In contrast, VEGF and EPO can have positive effects on cell survival. In addition, IGF (insulin-like growth factor) 2 and TGFα (transforming growth factor α) are HIF targets that activate intracellular survival pathways .
HIF target genes have also been implicated in the control of cell migration and invasion . HIF-mediated induction of the receptor tyrosine kinase, Met, promotes cell motility through increased activity of the hepatocyte growth factor signalling pathway. Invasion of cells through the basement membrane is in turn promoted by repression of E-cadherin, an important component of adherens junctions. Such pathways are co-opted by hypoxic cells, and by tumour cells that constitutively express HIF. For instance, HIF-dependent induction of the chemokine receptor CXCR4 promotes the chemotaxis of renal cell carcinoma cells.
Overall, recent microarray analyses have demonstrated that hundreds or even thousands of genes respond to hypoxia in any given type of cell. Somewhat surprisingly, genetic manipulation of the HIF system has indicated that, at least under standard tissue culture conditions, substantially more than 50% of these genes respond directly or indirectly to HIF .
Post-translation hydroxylation of HIF-α
So far two steps in the regulation of HIF by oxygen have been clearly defined. These involve regulation of both the abundance and the transcriptional activity of HIF-α subunits by different types of post-translation hydroxylation.
In the presence of oxygen, cellular levels of HIF-α proteins are suppressed by rapid constitutive degradation via the ubiquitin–proteasome pathway. Two independent domains in HIF-1α and HIF-2α mediate this oxygen-dependent degradation, both positioned around conserved proline residues, Pro402 and Pro564 in HIF-1α (Figure 2) . Hydroxylation of either of these residues creates a binding site for the VHL (von Hippel-Lindau tumour-suppressor) protein (Figure 2), the substrate-docking site of an E3 ubiquitin ligase complex containing Elongin B/C, Cul2 and Rbx1 proteins (Figure 1) . Subsequent ubiquitylation of prolyl-hydroxylated HIF-α signals its rapid degradation by the 26S proteasome.
HIF-α prolyl hydroxylation is catalysed by a family of PHD (prolyl hydroxylase domain) enzymes (PHD1, 2 and 3) (Figure 2) that belong to a larger family of enzymes which require oxygen, 2-OG (2-oxoglutarate) and Fe(II) for activity . Members of this family are characterized by a catalytic core of eight β-strands that fold into a DSBH (double-stranded β-helix) motif. This structural arrangement brings together ligands involved in Fe(II) and 2-OG binding at the PHD active site, which is highly conserved in all three enzymes, both in sequence and predicted tertiary structure. Despite this, there are differences in substrate specificity between the PHDs. PHD1 and 2 hydroxylate Pro402 and Pro564 residues in HIF-1α and the equivalent residues in HIF-2α, whereas PHD3 only targets Pro564 or the equivalent . The PHD2 crystal structure suggests that substrate specificity may be partly explained by variation in the β2-β3 finger motif that is distal to the iron centre . However, there are other differences between the PHDs that could contribute to selectivity.
PHD1 and PHD2 have a similar domain organization with a C-terminal catalytic domain and an N-terminal extension (Figure 2). In the case of PHD2 the N-terminus contains a zinc-finger-like domain that may be involved in autoinhibition of the enzyme . PHD3 is significantly smaller in size because it lacks the N-terminal extension present in the other PHDs. In addition to specific activity, differences in subcellular localization, interactions with other proteins, and relative abundance of the proteins in cells also influence the effective contribution of different HIF-α isoforms to degradation in vivo (reviewed in [2,12]).
Although the three PHDs isoforms are closely related, they do not appear to be redundant. PHD2 is responsible for the majority of prolyl hydroxylase activity under normoxia; and PHD2 RNAi (RNA interference) alone is sufficient to induce HIF-1α protein expression and HIF target gene expression . Further evidence for the importance of PHD2 has come from genetic analyses of inherited erythrocytosis (an increased number of red blood cells). A P317R PHD2 mutation associated with familial erythrocytosis significantly reduced PHD2 activity . Structural studies rationalize this effect, and indicate that P317R most likely affects both Fe(II) and substrate binding . Furthermore, knockout mouse studies have shown that whereas PHD2 null mice are embryonically lethal due to defects in heart and placental development , PHD1 and PHD3 null mice are viable. It is likely that these differences are due in part to the more widespread expression of PHD2 relative to that of PHD1 and PHD3 (see below) .
Regulation of PHD activity
The physiological challenge of oxygen homoeostasis requires a system capable of flexible responses over a wider range of oxygen tensions. In this respect it is interesting that the total level of PHD activity in a cell is regulated by a number of factors in addition to oxygen availability, including abundance of the PHD enzymes, cofactor availability, and interaction with other proteins.
Abundance of PHDs
The expression level of each PHD is increased by specific signals (see below), including hypoxia. Indeed, PHD2 and PHD3 appear to be direct HIF targets. This provides a negative feedback loop that, in response to temporal changes in oxygenation, may serve to limit HIF-α induction in hypoxia or accentuate the response to reoxygenation. At steady-state it would be predicted to provide a ‘range-extending’ function by adjusting hydroxylase activity in accordance with available oxygen (see below) (reviewed in ). Interestingly, whereas PHD2 appears to be ubiquitously expressed, other isoforms show marked tissue specificity; for instance PHD1 is highly expressed in testes whereas PHD3 is most abundant in the heart . Nevertheless the effects of these differences and whether they alter the oxygen-sensing range in these tissues is unclear.
There are several mechanisms by which PHD expression is regulated, including transcription, translational initiation, mRNA splicing and protein stability. Thus PHD1 is expressed in a variety of cells as two proteins arising from distinct sites of translational initiation . Alternatively spliced transcripts of PHD2 and 3 have also been described, most of which are enzymatically inactive [19,20]. Interestingly, some of these other PHD forms have the potential to exert dominant-negative effects. PHD3 is a highly inducible short-lived protein that is targeted for proteasomal degradation by Siah E3 ubiquitin ligases . The importance of this control has been illustrated in mice bearing targeted inactivation of the Siah2 gene where endogenous PHD3 is elevated and the EPO response to hypoxia is blunted. Though the Siah ubiquitin ligase can also interact with PHD1, its physiological role is less clear in this context and, at least in cell culture, PHD1 levels are not significantly affected by genetic manipulation of Siah1 and 2 .
A number of binding partners have been identified for the PHDs, including those that are likely to promote protein folding and others that modulate activity towards HIF-α. Thus subunits of the cytosolic chaperone TriC were identified by MS in association with over-expressed PHD3 but not other PHDs . OS-9 (amplified in osteosarcoma 9) was identified in a yeast-2-hybrid screen using the C-terminus of HIF-1α as bait, and subsequently shown to interact with PHD2 and PHD3 . OS-9 overexpression may promote HIF-α hydroxylation and degradation under certain conditions, possibly by the formation of a HIF-α–PHD–OS-9 ternary complex. Similarly, PHD3 is reported to interact with the WD-repeat (tryptophan-aspartate repeat) protein Morg-1 [MAPK (mitogen-activated protein kinase) organizer 1], an association that is postulated to suppress HIF-α by promoting PHD3-mediated hydroxylation . It is still unclear what the physiological roles of these processes are, although it is likely that together they contribute to a system that tightly controls HIF-α degradation in a context- and cell type-dependent manner.
The PHDs belong to a large family of 2-OG/Fe(II)-dependent dioxygenases that catalyse oxidative modifications (reviewed in ). Oxygen is split into two atoms during the catalytic cycle. One is incorporated into the prime substrate (e.g. HIF-α) while the other is coupled to the oxidative decarboxylation of 2-OG to yield succinate and carbon dioxide (Figure 3). Fe(II) co-ordinates 2-OG and oxygen towards the prime substrate and forms the reactive intermediate that drives the oxidative process. It is becoming clear that the availability of essential cofactors such as oxygen (see below), Fe(II) and 2-OG are important determinants of PHD activity (Figure 3), and therefore HIF-α expression.
Recent studies suggest that certain TCA (tricarboxylic acid) cycle intermediates can act as natural PHD inhibitors by competing with 2-OG (reviewed in ), and it has been proposed that consequential activation of HIF promotes the growth of tumours associated with genetic defects in the TCA cycle. Thus SDH (succinate dehydrogenase) converts succinate into fumarate, and its inactivation is associated with a predisposition to paraganglioma and phaeochromocytoma. Similarly, fumarate is converted into malate by FH (fumarate hydratase), and inactivating mutations of the FH gene predisposes to a variety of tumour types. Tumours harbouring SDH or FH mutations are generally highly vascularized and express elevated levels of HIF-α. Thus an oncogenic pathway based on accumulation of succinate or fumarate, competitive inhibition of PHD activity , and increased activity of HIF (or other) PHD substrates has been proposed.
The importance of Fe(II) for PHD activity is well demonstrated by the classical characteristics of induction of the HIF system by either divalent cations, which substitute for Fe(II) in the active site, or by compounds that chelate the cellular iron pool . However, it is becoming clear that other more physiological stimuli may also affect PHD activity by actions on Fe(II) at the active site. Rapidly growing cells are often iron deficient and supplementation with iron and/or ascorbate is sufficient to down-regulate normoxic HIF-α, most likely by enhancing PHD activity. The exact role of ascorbate in promoting PHD activity remains unclear, but may involve the reduction of Fe(III) at the active site or an action to increase the availability of Fe(II) . Interestingly, ascorbate is sufficient to rescue PHD activity in cells that overproduce ROS (reactive oxygen species). For example, JunD regulates genes involved in antioxidant defence and JunD-deficient cells manifest activation of HIF in association with accumulation of hydrogen peroxide and ROS. It has been proposed that the increase in ROS leads to the oxidation of Fe(II) to Fe(III) at the PHD2 catalytic centre thus inducing HIF-α even in the presence of oxygen .
Other investigators have proposed that production of mitochondrial ROS plays a role in PHD inactivation under hypoxia. In support of this, mitochondrial defects that reduce ROS production are associated with reduced induction of HIF in moderate hypoxia (reviewed in ). However an alternative explanation is that these defects simply reduce the severity of intracellular hypoxia by reducing oxygen consumption. It should also be noted that the PHDs are able to rapidly hydroxylate HIF-α upon reoxygenation, suggesting that either they are not in fact damaged in hypoxia, or that they are very rapidly repaired in the presence of oxygen.
In addition to the oxygen-dependent degradation domains, the HIF-α C-terminus mediates a second oxygen sensitivity control of HIF activity. This region contains the CAD (C-terminal transactivation domain) (Figure 2) and is responsible for recruiting the p300/CBP (cAMP-response-element-binding protein-binding protein) transcriptional co-activator to HIF target gene promoters (reviewed in ). HIF-α CAD activity is suppressed in normoxia due to the incorporation of oxygen into a conserved asparagine residue by FIH (factor inhibiting HIF) which, like the PHDs, is also a 2-OG/Fe(II)-dependent dioxygenase (Figure 2) . FIH-mediated HIF-α CAD asparaginyl hydroxylation prevents the interaction with the p300 CH1 domain. Unlike the PHDs, FIH appears to be abundant and ubiquitously expressed, with protein levels relatively constant between cell types and conditions . Although FIH targets the CADs of both HIF-1α and HIF-2α, it does so to differing extents. HIF-2α is less responsive to FIH than HIF-1α due to a conserved amino acid substitution in HIF-2α that reduces the efficiency of asparagine hydroxylation .
Genetic analyses confirm that FIH is an important component of the oxygen-dependent suppression of HIF; FIH RNAi is sufficient to up-regulate the HIF target genes GLUT1 (glucose tranporter 1) and CA9 (carbonic anhydrase 9) in a variety of cell lines under normoxia . However, neither PHD2 nor FIH RNAi elevates HIF target genes to the same extent as that observed under hypoxia. However, the combined effect of FIH and PHD2 RNAi is sufficient to induce maximal HIF activity at least in certain cells, suggesting that these enzymes are the dominant HIF hydroxylases . These studies also revealed that FIH differentially regulates specific HIF target genes, perhaps reflecting target gene specific differences in the use of either of two transactivation domains that are (CAD) or are not [NAD (N-terminal transactivation domain)] affected by asparaginyl hydroxylation (Figure 2).
HIF hydroxylases as ‘oxygen sensors’
Since molecular oxygen is absolutely required as co-substrate for the HIF hydroxylation reactions catalysed by both the PHDs and FIH, these enzymes are well placed to mediate an oxygen-sensing function. Nevertheless certain conditions must be fulfilled for a sensing role in physiological oxygen homoeostasis. First, the affinity for oxygen must be lower than the physiological concentration of oxygen; a drop in oxygen concentration would therefore lead to a concomitant decrease in hydroxylase activity, with a consequent increase in HIF expression. Secondly the rate of HIF hydroxylation must be rate-limiting for the overall inactivation process .
Several studies have sought to measure reaction kinetics of the HIF hydroxylases as a function of oxygen concentration. Initial in vitro studies using recombinant PHD enzymes and short HIF-1α peptides suggested a very high Km for oxygen of approximately 250 μM. Though these early results suggested an unusually low affinity for oxygen, other work suggests that the true affinity of the PHDs for oxygen in vivo may be substantially higher. Thus similar experiments using a longer HIF-1α polypeptide have yielded substantially lower apparent Km values for oxygen [32,33]. Though the order of co-substrate and substrate-binding has not yet been defined for the HIF hydroxylases, these results are compatible with studies of other enzymes in this class that indicate that it is the enzyme–substrate complex that binds molecular oxygen, allowing both components to influence affinity. Though the affinity of FIH for oxygen has been reported to be higher than the PHDs at approx. 90 μM, these results were also obtained with short peptide substrates that may behave differently from the intact HIF-α protein. Thus given the increased oxygen affinity that is observed when longer (and more physiological) HIF-α polypeptides are used in kinetic studies, it is not yet clear that the HIF hydroxylases have kinetic properties that are specialized for an oxygen-sensing role. Nevertheless given that physiological tissue oxygen concentrations are likely to be in the range of 10–30 μM, even much lower Km values would be quite compatible with an oxygen-sensing function provided that the overall rates of HIF hydroxylation were limiting for HIF inactivation. That this is indeed the case is supported by the observation that modest changes in enzyme abundance affect the oxygen dependence of HIF activation. Thus it is important to recognize that the actual oxygen-dependence of HIF regulation is dependent on the characteristics of the system as a whole and will be determined by all the factors (see above) that influence the rate of HIF hydroxylation, as well as the affinity for molecular oxygen itself.
HIF hydroxylase substrates
Since the discovery of oxygen-sensitive signalling through protein hydroxylation in the context of HIF, attention has focused on whether a similar mechanism might operate on other systems. For instance the HIF hydroxylases themselves might have other substrates, or other enzymes in this family might regulate different systems in an oxygen-sensitive manner.
Several lines of evidence have suggested that the PHDs might have other substrates, though clear proof is lacking at the present time. First, the encoding genes have been identified in screens that are not readily connected with known functions of HIF. Thus PHD1 has been identified as a target of the oestrogen receptor and promotes the colony growth of breast cancer cells . The Drosophila homologue of PHD2 has been implicated downstream of cyclin D/Cdk4 in a pathway that regulates cell growth . PHD3 was independently cloned as a transcript induced by the differentiation of smooth muscle cells, by the activation of p53, and following nerve growth factor withdrawal in sympathetic neurones (reviewed in ). Further studies have indicated that PHD3 expression is necessary and sufficient for neuronal apoptosis under these conditions [36,37], and that the property is oxygen-sensitive and dependent on hydroxylase activity . However, as yet it is unclear whether any of these activities derive from enzymatic activities on substrates other than HIF [36,38].
Other transcription factors are responsive to hypoxia; NF-κB (nuclear factor κB) signalling being one which has attracted considerable interest. However, compared with the HIF system, regulation by hypoxia is less robust and the pathways remain poorly understood. Cummins et al.  have recently suggested that NF-κB activation in hypoxic cells involves the inhibition of PHD1, which they propose normally suppresses IKK (IκB kinase) activity by hydroxylating IKKb Pro191. PHD1 overexpression was sufficient to inhibit TNFα-induced NF-κB activity, although the effect of a hydroxylation-defective PHD1 mutant was not reported, and no direct evidence for IKK hydroxylation was provided.
In contrast with current uncertainties about the nature of other substrates for the PHDs, clear evidence of alternative substrates for the asparaginyl hydroxylase FIH has been obtained. Initially a yeast-two-hybrid screen identified members of the IκB family as novel FIH-interacting proteins . Conserved asparagine residues in the ARDs (ankyrin repeat domains) of p105 and IκBα are FIH substrates. Unlike HIF-α however, asparaginyl hydroxylation of IκBα does not appear to affect its activity dramatically. Recent studies on FIH-mediated hydroxylation of analogous asparagine residues in the Notch transmembrane receptor have shown that, like IκBα, the activity of the canonical Notch signalling pathway does not appear to be regulated by FIH. Nevertheless given the prevalence of the ARD domain in the mammalian proteome, these results strongly suggest that protein hydroxylation will turn out to be a common post-translational modification. Interestingly the affinity of FIH for ARDs is much higher than that for HIF so that these proteins are effective competitive inhibitors of FIH-mediated HIF-α CAD hydroxylation thereby providing an additional mechanism of modulating HIF signalling (M.L. Coleman, unpublished work).
In a number of other systems indirect evidence for protein hydroxylation or action of a 2-OG oxygenase has suggested existence of hydroxylase pathways but the enzyme(s) involved are unknown (reviewed in ). Thus the hyperphosphorylated form of the Rpb1 subunit of RNA polymerase II has been proposed to be prolyl hydroxylated and targeted for proteasomal degradation by VHL. Iron-regulatory protein 2 is involved in cellular iron sensing and, based on regulation by inhibitors of 2-OG/Fe(II)-dependent enzymes, has also been postulated to be a target of post-translational hydroxylation. Hypoxia leads to an increase in the synthesis of phosphatidic acid by diacylglycerol kinase, and again regulation by inhibitors of this family of enzymes has suggested the involvement of an as yet unknown 2-OG oxygenase.
2-OG dioxygenases as general oxygen sensors
Sequence analyses predict that the human genome encodes over 60 2-OG/Fe(II)-dependent dioxygenases , most with unknown functions. As discussed above, since it is unlikely that the HIF hydroxylases require special biochemical properties to function as oxygen sensors, other family members might also signal oxygen availability if appropriately coupled to a regulatory process. Since these enzymes are not all protein hydroxylases and in fact oxidize a variety of substrates they might signal oxygen availability to a variety of cellular processes.
For instance, DNA is damaged by a variety of alkylating agents that are toxic and mutagenic. Human homologues of Escherichia coli AlkB are 2-OG/Fe(II)-dependent dioxygenases that repair DNA methylated at 1-methyladenine and 3-methylcytosine (reviewed in ). Hydroxylation of the methyl group produces a hydroxyl intermediate that decomposes to the repaired base plus formaldehyde (Figure 3). This class of enzymes may therefore represent a novel interface between hypoxia and DNA repair.
Post-translational modifications of histones (H) are important regulators of gene expression. Methyltransferases catalyse the methylation of H3 and H4 at specific lysine residues; the site of modification determines whether methylation results in transcriptional activation or silencing. Recent studies suggest that 2-OG/Fe(II)-dependent dioxygenases with similarity to FIH can reverse this process (reviewed in ). These enzymes regulate gene expression and the proliferation of tumour cells. They possess what has previously been termed a JmjC (Jumonji C) domain, which is a DSBH motif with significant similarity to that of FIH. Several JmjC containing histone demethylases have recently been identified that demethylate H3 Lys36 and/or H3 Lys9 . Demethylation is thought to result from hydroxylation of the methyl group which forms a highly unstable intermediate that decomposes into formaldehyde plus a lysine residue (Figure 3). JmjC histone demethylases may therefore represent an even more direct link between oxygen availability and gene expression than the HIF-dependent regulation of HRE-containing promoters.
Studies into the regulation of EPO expression have led to the discovery of an oxygen-sensing system based on the activity of a set of 2-OG- and Fe(II)-dependent oxygenases and a transcriptional complex termed HIF that is central to gene regulation by hypoxia. The recognition that this enzyme family has many members, the discovery of other hydroxylated proteins and other substrates of the HIF hydroxylases, has raised the possibility that similar oxygen-sensitive signalling systems may operate on other processes. An important challenge will now be to define these pathways.
• HIF is a major component of the hypoxia-induced response.
• HIF-α stability and activity are regulated by prolyl and asparaginyl hydroxylases respectively.
• The HIF hydroxylases belong to a large family of 2-OG/Fe(II)-dependent dioxygenases.
• Other substrates of the HIF hydroxylases have been identified, some of which may play a role in hypoxia-induced responses.
• Several other intracellular proteins are suggested to be, or have been shown to be, hydroxylated. In most cases the hydroxylases responsible are yet to be identified.
• 2-OG/Fe(II)-dependent dioxygenases regulate diverse processes including DNA repair and histone demethylation, which may be hypoxia-responsive.
- Biochemical Society