Now a Nobel gas: oxygen
The recent bestowal of the Nobel Prize 2019 in Physiology or Medicine to Gregg L. Semenza, Sir Peter J. Ratcliffe, and William G. Kaelin Jr. celebrates a series of remarkable discoveries that span from the physiological research question on how oxygen deficiency (hypoxia) induces the red blood cell forming hormone erythropoietin (Epo) to the first clinical application of a novel family of Epo-inducing drugs to treat patients suffering from renal anemia. This review looks back at the most important findings made by the three Nobel laureates, highlights current research trends, and sheds an eye on future perspectives of hypoxia research, including emerging and potential clinical applications.
KeywordsErythropoietin High altitude Hypoxia Oxygen sensing Protein hydroxylation von Hippel-Lindau
The physiological puzzle to solve: Epo induction by hypoxia
One of the classical topics in physiology is the question on how organisms adapt to the ever changing environmental conditions, including to low oxygen (hypoxia). At first glance, this question might only be relevant to high-altitude residents and mountaineers. The decreased oxygen partial pressure in high altitude causes an increase in the blood hormone erythropoietin (Epo) which in turn activates erythropoiesis in the bone marrow . An increased number of red blood cells then compensates for the decreased oxygen content of hemoglobin and normalizes oxygen transport from the lung to the oxygen-consuming tissues of our body. This mechanism became a matter of common knowledge due to the performance enhancing training methods applied by some elite endurance athletes, first at high altitude or in artificially hypoxic environments, but soon also by illegal injections of recombinant Epo protein (Epo doping). However, during evolution, this mechanism did not actually develop to adapt our body to high altitude. On the contrary, populations such as Tibetans, residing for many generations at very high altitude, developed mutations that blunt the hypoxic response . This limited increase in red blood cell concentration lowers the risk for cardiovascular complications in chronic hypoxia. Therefore, the primary reason why we increase Epo in high altitude is not the need for additional erythrocytes but rather a side effect of the fact that Epo-producing cells cannot directly measure the hematocrit. Because these cells only respond to lowered tissue oxygen partial pressure, they cannot distinguish between a decreased hemoglobin content of the blood (anemia) and a decreased oxygen saturation of otherwise normal hemoglobin (high altitude). In either case, Epo production is increased.
The fetal liver and specialized cells in the adult kidney “sense” tissue hypoxia and counteract it by releasing more Epo into the circulation . Following kidney failure, Epo is not produced at sufficient amounts anymore. The reasons for inadequately low Epo production in chronic kidney disease remain poorly understood, but a desensitization of the oxygen sensing mechanism, tissue hyperoxygenation, and/or transdifferentiation of renal Epo-producing (REP) cells may play a role. Dialysis patients hence usually need life-long parenteral treatment with recombinant Epo. Therefore, the basic research question of how hypoxia leads to increased Epo was not only of physiological but also of major medical relevance.
In contrast to other major protein hormones such as insulin, Epo is not stored and released upon the appropriate stimulus, but it is rather de novo transcribed, translated, and immediately released into the blood stream. Concerning the at that time predominant model of oxygen sensing, it was believed that an unknown heme oxygen sensor is responsible for Epo regulation . Many scientists were biased towards this hypothesis and (unsuccessfully) worked on candidate heme proteins.
Gregg L. Semenza
Undoubtedly, Gregg Semenza’s cloning of HIF in 1995  widely opened the door to the hypoxia research field, but the long way to this major discovery was paved with many obstacles. While two hepatoma cell lines capable of oxygen-regulated Epo expression had been reported before , these cell lines were only partly suitable for studying the physiologically more relevant renal expression of Epo. In addition to the lack of a suitable cell model, EPO gene regulation is quite complex and several dozens of kilobases of the EPO gene are required for proper Epo regulation. A simple 2-kb promoter region, like in many other genes, does not recapitulate tissue- and oxygen-regulated gene expression. Thus, Gregg Semenza started to study transcriptional regulation of the EPO gene in the Antonarakis lab by generating transgenic mouse lines that contained long DNA fragments derived from the human EPO gene [120, 121, 124]. Later on, similar experiments were performed by other groups, including Peter Ratcliffe’s team [40, 76, 82]. This tedious approach, which nowadays would probably not be applied anymore, resulted in the identification of a 50-bp DNA fragment close to the 3′ end of the EPO gene. This fragment confers liver-specific hypoxia-inducible human Epo expression in the mouse .
Gregg Semenza then applied extensive scanning mutagenesis of a 50-bp oligonucleotide probe in electrophoretic mobility shift assays (EMSAs) using nuclear protein extracts derived from hypoxic hepatoma cells. These experiments resulted in the identification of three different crucial DNA elements. Two of these sites are mandatory for hypoxic EPO gene induction, but only the first site binds to a hypoxia-inducible nuclear factor which Gregg Semenza called “hypoxia-inducible factor 1” (HIF-1) . A DNA-binding site for HIF is now commonly referred to as “hypoxia response element” (HRE), even though a functional HRE is actually composed of more elements than only one HIF DNA-binding site .
Although at that time, HIF-1 was nothing more than a dark spot on an X-ray film , this paper was the first to visualize the activity of this important transcription factor and hence marks the birth of an entire new research field. Ironically, many years later, it turned out that this 3′ enhancer is not required for renal Epo expression  and that it is actually HIF-2, and not HIF-1, which mediates hypoxic Epo induction in the kidney [105, 152]. Nevertheless, the major conclusions of these experiments provided the robust fundaments on which many other labs, including pharmaceutical industry, built their own research. There is no better way to demonstrate reproducibility of scientific findings.
If HIF was restricted to hypoxic induction of EPO gene expression in hepatoma cells, it would probably still only be known to a small fraction of the scientific community. However, 1 year after the definition of the EPO 3′ HRE Gregg Semenza and Peter Ratcliffe both reported that the EPO 3′ HRE confers hypoxia-inducible reporter gene expression and recruits HIF-1 also in non-Epo-producing cell lines [84, 149]. In fact, hypoxic HIF induction appeared to be a ubiquitous principle of mammalian cells. This next key finding now definitively attracted the attention of life scientists as well as clinicians, because it suggested that genes other than EPO are also HIF targets. In particular, the large community of cancer researchers immediately started to participate in the search for additional HIF targets and the identification of the actual nature of the HIF protein, because it was obvious that tissue hypoxia (intrinsic to each growing solid tumor) is a key factor in metabolic adaptation, new vessel formation, genomic instability, and therapy resistance. A better understanding of the molecular mechanisms involved in adaptation to tumor hypoxia, as cancer researchers thought, would allow for novel treatment options.
It was Gregg Semenza (together with his post-doc Guang L. Wang) who finally managed to clone the hypoxia-inducible factor. It turned out that HIF-1 is actually a heterodimeric transcription factor [148, 150]. This explains why “simple” approaches used by other groups, such as the at-that-time popular expression cloning, did not work. Wang and Semenza rather used classical but tedious biochemistry to enrich for the factor binding to the EPO 3′ HRE using hundreds of liters of cell culture and EMSA to trace the candidate factor through the purification protocols. Protein sequencing revealed that HIF-1 is composed of a known protein (HIF-1β or ARNT) and a novel hypoxia-inducible protein (HIF-1α).
Once more, Gregg Semenza was only a small step ahead of his competitors but, nonetheless, he was the one who finally succeeded. The aryl hydrocarbon receptor (AhR or dioxin receptor) nuclear translocator (ARNT) was known to toxicologists before. They also knew that there are additional members of the bHLH-PAS family of transcription factors able to bind to ARNT . But what they could not know at that time was the fact that HIF-1α cannot be detected in “normoxic” cell cultures as studied by the vast majority of all laboratories worldwide up to this day. If they would have applied more physiological oxygen concentrations to their cell culture experiments, HIF-1α could have been discovered much earlier.
Following the discovery of HIF as a heterodimer, it was evident from sequence comparisons that there are two additional hypoxia-inducible factors able to bind to HIF-1β/ARNT, termed “HIF-2α” and “HIF-3α.” Remarkably, despite three similar family members, there is little if any redundancy among HIFα subunits and the inactivation of each of the three genes results in embryonic lethality [144, 160], demonstrating their physiological relevance. Gregg Semenza was the first to report the phenotype of HIF-1α knock-out mice . This paper already showed that the entire glycolytic pathway is under the control of HIF-1α. This knock-out phenotype was confirmed by others in the same year [12, 109].
It quickly became clear that the crucial step in hypoxia-inducible gene expression is the oxygen-dependent destabilization of the HIFα subunits by proteasomal degradation . But how oxygen interacted with HIFα remained a mystery. Many possible mechanisms were postulated, often involving second messengers or unspecific mediators such as reactive oxygen species. The field somehow got stuck at this time and diverging opinions ranged from the believe in an unknown monolithic oxygen sensor to the sum of all oxygen-dependent processes resulting in the production of reactive oxygen species [118, 153]. The identification of the signaling pathway connecting oxygen availability to HIF-1/2α destruction appeared to become an unsolvable problem. Fortunately, Peter Ratcliffe and Bill Kaelin could later elucidate this mystery.
Another remarkable discovery made by Gregg Semenza was the first identification of the “factor-inhibiting HIF” (FIH) which he correctly recognized to be a HIF-1α inhibitor . The actual function of FIH as an oxygen-dependent asparagine hydroxylase that lowers the transcriptional activity of HIF-1α by preventing the recruitment of the histone acetyl transferases p300/CBP was subsequently made by others [41, 68].
Since these ground-breaking discoveries, Gregg Semenza contributed an impressive wealth of important publications, especially regarding the function of the HIF system in breast cancer and the identification and function of HIF-1α and HIF-2α antagonists. While originally thought to be non-druggable targets, a HIF-2α antagonist has recently been developed . This compound appears to be an optimism-causing agent in the treatment of mutant von Hippel-Lindau (VHL) clear cell renal cell carcinoma as recently shown by Bill Kaelin and others [19, 20].
Sir Peter J. Ratcliffe
As outlined above, Peter Ratcliffe and Gregg Semenza often delivered a head-to-head race in the (reverse) elucidation of the oxygen sensing and signaling pathway. Like Gregg Semenza and several others, Peter Ratcliffe also used transgenic mouse models to identify the crucial regulatory DNA elements involved in hypoxic EPO gene induction, and he made use of these mouse models to characterize the renal cell type that produces Epo [81, 82, 83]. Peter Ratcliffe also immediately recognized the general importance of the HIF system by demonstrating the widespread functionality of the EPO 3′ HRE which is not at all restricted to Epo-expressing cell types .
Following the cloning of HIF-1α by Gregg Semenza, many groups, including the Ratcliffe lab, figured out that this protein is regulated by rapid oxygen-dependent ubiquitinylation followed by proteasomal degradation, targeting a domain within HIF-1/2α which is now commonly referred to as oxygen-dependent degradation (ODD) domain [46, 54, 103, 111, 131, 134]. These findings initiated the search for the ubiquitin E3 ligase that interacts with “oxy-HIF-1α,” an endeavor that took quite a while. The actual breakthrough was made by the Ratcliffe lab when it became clear that the VHL tumor suppressor protein (pVHL) mediates the oxygen sensitivity of the ODD domain by serving as an E3 ubiquitin ligase that interacts with HIFα under well-oxygenated conditions only [24, 85]. This observation was immediately confirmed by several other labs, including the lab of VHL researcher Bill Kaelin [55, 66, 97, 139].
But how did it come that Peter Ratcliffe and his co-workers made this crucial discovery and not for instance Bill Kaelin who discovered the tumor suppressor function of pVHL already back in 1995 ? With the contribution of Bill Kaelin, it had been observed before that VHL loss-of-function results in a hypoxia-like gene expression profile . The important HIF target gene vascular endothelial growth factor (VEGF) was known to be induced in a pVHL-dependent manner and it had been suggested that pVHL is involved in the hypoxic stabilization of VEGF mRNA [70, 72]. However, these authors did not (yet) make the link between pVHL and HIF-1α protein stability. The explanation for this can be found in a technical difficulty: VHL researchers often work with clear cell renal cell carcinoma (ccRCC) cell lines which are usually deficient for functional pVHL. In fact, most members of VHL disease families carrying one mutant VHL allele develop renal cancers following the inactivation of the second VHL allele, and most renal cancers are hence VHL-deficient. But for reasons that were later described by Bill Kaelin (outlined below), the supramaximal stabilization/induction of toxic levels of HIF-1α protein in VHL-deficient ccRCC cell lines leads to a counter-selection against HIF-1α and instead to a positive selection for HIF-2α. Thus, the particular VHL-deficient ccRCC cell lines used at that time happened to express only HIF-2α but no HIF-1α. While antibodies against HIF-1α had been developed by several labs, no antibodies derived against HIF-2α were available—except in the lab of Peter Ratcliffe. Constitutive, oxygen-independent stabilization of HIF-2α in ccRCC cell lines could then “easily” be detected, including the efficient normoxic HIF-2α degradation following re-introduction of wild-type VHL .
The seminal discovery that the E3 ligase pVHL targets HIFα for polyubiquitination and proteasomal degradation finally provided the long-sought-for tool that allowed the differentiation between hypoxic (non-modified) and normoxic (modified) HIFα, because pVHL stably interacts with HIFα only under conditions compatible with the presence of an enzymatic activity that is oxygen- and iron-dependent . Retrospectively, it seemed so obvious that oxygen-dependent protein hydroxylation, as known for a long time from the post-translational modification of collagen fibers, is the “tag” that labels HIFα for normoxic degradation, especially since collagen prolyl-4-hydroxylase was a known HIF target gene itself . But nobody made this educated guess before the oxygen-dependent interaction between pVHL and HIFα had been reported. Remarkably, once the Maxwell 1999 Nature paper was out, it took only 1 year until the Ratcliffe and Kaelin groups reported in two back-to-back publications that it is actually oxygen-dependent prolyl-4-hydroxylation of two proline residues in the ODD domain of HIFα which strongly increases the affinity for pVHL binding [50, 52].
What remained to be done was the identification of the enzyme(s) responsible for the oxidative modification of HIFα, i.e., the actual oxygen sensor(s). It was clear that the known collagen prolyl-4-hydroxylases do not target HIFα and it was again Peter Ratcliffe who was first in reporting a novel family of three different HIFα prolyl-4-hydroxylases which he named PHD1, 2, and 3 . For once, evolutionary conservation helped in the rapid identification of these proteins: a C. elegans researcher working at the same institution had a worm strain that carried a mutation of a protein that resembled the collagen prolyl hydroxylase and its mutation led to the egg-laying (EGL) deficiency phenotype. Fortunately, the HIF system in the nematode worm is conserved but with only one gene encoding for HIFα and only one encoding for the oxygen sensing prolyl hydroxylase named EGL-9. Even more fortunate, Peter Ratcliffe had antibodies that reacted with worm HIFα and he found constitutive oxygen-independent HIFα stabilization in the EGL-9 mutant worm but not in the wild-type worm. In silico homology comparisons then immediately led to the identification of the three homologous mammalian genes . Based on in silico analyses, this family of oxygen sensors has also been reported by another group and was named HIF prolyl hydroxylase (HPH)3, 2, and 1 for PHD1, 2, and 3, respectively . While it had been suggested to use the original mammalian gene nomenclature EGL-Nine (EGLN)2, 1, 3 for PHD1, 2, and 3, respectively , most subsequent publications refer to the Ratcliffe nomenclature.
William G. Kaelin Jr.
Cancer researcher Bill Kaelin discovered the basic functions of pVHL. He could demonstrate that pVHL acts as a tumor suppressor in VHL-deficient kidney cancer . Bill Kaelin was also the first to report that pVHL interacts with elongins B and C . At this time, it was thought that these proteins are part of a transcriptional elongation complex, but nowadays, we know that they form part of the pVHL E3 ubiquitin ligase complex interacting with hydroxylated HIFα. Importantly, he (together with Mark A. Goldberg) obtained the first indications for a role of pVHL in hypoxia signaling by demonstrating that the loss of VHL function results in a hypoxia-like gene expression pattern . While the Goldberg group reported that pVHL is involved in post-transcriptional VEGF mRNA accumulation under hypoxic conditions [70, 71], Bill Kaelin found that pVHL actually interacts via elongin C with the ubiquitin ligase subunit Cul2 and that this interaction is required for pVHL-dependent post-transcriptional accumulation of hypoxia-inducible mRNAs, including VEGF . Obviously, these findings provided the grounds for Peter Ratcliffe’s group to study a direct role of pVHL as oxygen-dependent HIFα E3 ligase .
Once the direct link between pVHL and HIFα had been established, Bill Kaelin showed that the cancer mutation-prone pVHL β domain interacts with HIFα and that this interaction is required for ubiquitination and proteasomal destruction . Independently of Peter Ratcliffe’s team, he identified post-translational oxygen-dependent prolyl-4-hydroxylation of HIFα as pre-requirement for avid protein-protein interaction, resulting in two back-to-back publications [50, 52]. Based on structural studies, the increased affinity of pVHL for hydroxylated HIFα could be nicely confirmed 1 year later, in two reports that were co-authored by Peter Ratcliffe and Bill Kaelin, respectively [43, 89]. Following the identification of the HIF prolyl hydroxylases [11, 32], also Bill Kaelin investigated the PHDs in cell-free  as well as in mouse knock-out models [90, 91, 92]. These studies contributed to the functional understanding of the PHD oxygen sensors and to the development of HIF prolyl hydroxylase inhibitors (PHIs) with the aim to treat hypoxia-associated diseases.
Importantly, Bill Kaelin investigated the roles of HIF in VHL-deficient ccRCC tumors and could show that it is mainly HIF-2α (and not HIF-1α) which mediates the tumor suppressive effects of pVHL [62, 63]. He nicely demonstrated that each VHL-deficient kidney tumor finds a different way to get rid of the supramaximal toxic HIF-1α levels and instead selects for HIF-2α expression . This finding led to the somewhat oversimplified view that HIF-1α functions as a tumor suppressor and HIF-2α as an oncogene. However, this is only the case under the exceptional circumstance of constitutive (unphysiological) HIFα stabilization due to the loss of VHL and can only be found in ccRCC as well as in some other quite rare cancers. HIFα mutations do not cause cancer per se. However, Bill Kaelin’s and others’ results made HIF-2α an attractive target for pharmaceutical agents. A recently developed heterodimerization inhibitor appears to be highly specific for HIF-2α and Bill Kaelin obtained very promising results with this antagonist in preclinical mouse models of kidney cancer .
Another hallmark finding of the Kaelin lab concerns the use of unusual substrates of the PHD enzymes. PHDs require oxygen and 2-oxoglutarate as co-substrates to hydroxylate HIFα subunits and to produce succinate by oxidative decarboxylation of 2-oxoglutarate in a manner that depends on iron and reducing agents. Succinate as well as several other intermediary products of the tricarboxylic acid (TCA or Krebs) cycle has been shown to inhibit PHD activity, providing a potential link between mitochondrial metabolism and oxygen sensing. Succinate dehydrogenase (SDH), fumarate hydratase (FH), and isocitrate dehydrogenase (IDH) are enzymes of the TCA cycle whose products can affect PHD activity. Loss-of-function mutations of SDH and FH lead to the accumulation of succinate and fumarate, respectively, two known inhibitors of the PHD enzymes . Intriguingly, mutations of these enzymes have been identified in human cancer which implies a link between cellular metabolism and oncogenic transformation. Bill Kaelin found that cancer-specific mutations of IDH1 and IDH2 caused the accumulation of the unusual (R)-enantiomer of 2-hydroxyglutarate (R-2HG). Unexpectedly, this oncometabolite (but not S-2HG) actually stimulated the PHD enzymatic activity and hence reduced HIF-1α levels . The R-2HG oncometabolite appears to be sufficient to cause leukemia, not only by interfering with PHD enzymes but also with TET2 which epigenetically modifies DNA by oxygen-dependent hydroxymethylation and is involved in cancer transformation .
Widespread transcriptional control by HIF
These experiments also opened the route for detailed analyses of genetic prerequisites of HIF’s action as a transcription factor. Up to then, it was unclear whether all HIF-regulated genes depend on direct HIF-DNA interaction at cis-regulatory elements or whether some genes are indirectly regulated, for example via induction of other trans-acting factors. Again, novel methodologies helped to get detailed insights into HIF-DNA interactions and transcriptional output. For example, the use of chromatin immunoprecipitation of HIF subunits coupled to either microarray analyses or next-generation DNA sequencing permitted to exactly define HIF DNA-binding sites throughout the whole genome. In combination with analyses of transcriptional output by e.g. expression microarray analyses or RNA sequencing, the totality of directly HIF-regulated genes could be described [65, 88, 93, 116, 140, 141, 159].
Using these novel techniques to examine the hypoxic response yielded some important basic findings about HIF transcriptional activity. For example, the repertoire of direct HIF targets comprises all RNA species including mRNAs, miRNAs, and long non-coding RNAs . The number of direct targets is in the order of approximately 1000 genes per any cell type. HIF-1 and HIF-2 have similar DNA-binding motifs and overlapping targets, but the pathways targeted by each of the subunits differ considerably [57, 110, 116]. In general, HIF-DNA interactions lead to activation rather than repression of gene transcription often via preformed long-range enhancer promoter interactions . This finding was somewhat unexpected, since the numbers of genes being up- or downregulated, respectively, by HIF stabilization are similar . Therefore, it is likely that HIF induces repressive factors to inhibit expression of genes which are not necessary or even harmful for a hypoxia adaptation. Some of the aspects of gene repression in hypoxia have been examined recently and the SIN3A histone deacetylase complex and the REST complex have been shown to be involved in hypoxic gene repression, though a direct link to the transcriptional activity of HIF is still missing [13, 145].
Analyzing the DNA sequences bound by HIF allowed for specifying the HIF-binding motif as well as for screening for motifs of other transcription factors potentially interacting with HIF at hypoxia-responsive genes. The core DNA motif responsible for HIF binding was confirmed to be 5′-RCGTG-3′ (R = A or G) which supports findings from earlier work . Because this motif is remarkably similar for HIF-1 and HIF-2 DNA-binding, but the various genetic loci to which HIF-1 and HIF-2 are recruited may differ considerably, there are likely additional factors (e.g., epigenetic or other transcription factors) involved in directing the different HIFs to their target genes. In fact, motifs of other transcription factors are enriched in HIF-binding sites suggesting that HIF cooperates with other factors such as ATF-1/CREB-1 or AP-1 at selected sites which could confer target gene selectivity [30, 67, 110, 116].
The above mentioned studies showed that HIF activates a broad spectrum of pathways. However, the precise relevance of most of these pathways and of the HIF subunits for certain physiological or pathophysiological conditions in humans remains unclear. In this respect, studies on renal cancer, which in most cases display an activated HIF pathway because of loss-of-function of the VHL tumor suppressor gene, shed more light on the consequences of HIF activation [47, 69]. In vitro analyses of cancer cell lines and xenograft experiments suggested differential functions of HIF-1 and HIF-2 for the course renal cancer [62, 63, 106], but only recent genetic and pharmacologic evidence provided more clues about the role of HIF in this malignancy. First, the part of chromosome 14 coding for HIF-1 is frequently deleted in renal tumors . Together with studies showing that re-expression of HIF-1α in cell lines with a mutated endogenous HIF-1α slows xenograft growth, this suggests that HIF-1α acts as a tumor suppressor in renal cancer . Second, in genome-wide association studies single-nucleotide polymorphisms which predispose to renal cancer have been detected in the first intron of the EPAS1 gene, which encodes for HIF-2α . Though, the detailed functions of these SNPs are still not clear, they may relate to a tumor promoting function which is similarly observed in HIF-2α overexpressing xenograft models. Both findings, loss of HIF-1α and SNPs in EPAS1 provide genetic evidence for a functional role of HIF in renal cancer biology. Third, renal cancer associated SNPs cause differential binding of HIF to enhancer sites of key oncogenic drivers such as CCND1 and MYC, suggesting that the transcriptional output of HIF crucially modifies the risk of developing renal cancer [37, 38, 115].
An important finding of these studies was that enhancer accessibility was either restricted to cells of renal tubular origin or gained during renal cancer development. This again implies a key role of cell type specificity for the HIF response and suggests that the VHL-HIF pathway may be able to transform its transcriptional output within one cell type upon prolonged stimulation. Indeed, recent studies have provided evidence that the enhancer landscape changes following VHL loss and HIFα stabilization in renal cancer, which may have a major impact on tumor development, progression, and metastasis [146, 161]. Finally, the most powerful evidence for a functional role of members of the HIF pathway in renal cancer comes from the therapeutic use of agents aiming to modify the HIF signaling pathway. Inhibitors of VEGFA-signaling (a HIF target gene) and the mTOR pathway (a modulator of HIF expression) have long been used to treat patients with renal cancer . Recently, novel HIF-2α selective inhibitors have been introduced to clinical testing and showed promising results at least in a selection of renal cancer patients [19, 20]. These novel compounds expand the possibilities to pharmacologically target the HIF pathway in cancer, and they may also be useful in other HIF-2 associated conditions such as pulmonary hypertension.
Taken together, dissecting the hypoxic transcriptome and genome-wide HIF-DNA interactions has led to important insights into HIF biology downstream of the oxygen sensing mechanism. Through these studies, basic biological principles of the hypoxic response, and genetic and epigenetic dependencies of HIF as well as targets for therapeutic interventions, e.g., in renal cancer, have been identified. However, some important questions about HIF transcriptional activity remain to be answered. For example, what mechanisms determine the tight tissue specificity of HIF-2α expression? Which transcription factors and epigenetic modifiers co-operate at HREs in concert with HIF? Given the broad activity of HIF across the body , is the HIF transcriptional response affected by genetic or epigenetic phenomena in other conditions such as high-altitude adaptation or other cancer types? In addition, what are the long-term transcriptional effects of HIF stabilization, e.g., with respect to the introduction of novel PHIs into clinical routine?
Non-HIF targets of HIF hydroxylases
While PHDs and FIH are essential for HIF regulation, they also have non-HIF targets and there are at least three aspects worth to be considered in the context of HIF regulation:
First, PHDs and FIH are members of the enzyme family of 2-oxoglutarate (2-OG)-dependent dioxygenases to which other enzymes relevant to mammalian physiology belong, such as collagen prolyl and lysyl hydroxylases or the Jumonji C (JmjC) domain histone lysine demethylases . In fact, collagen prolyl hydroxylases have been instrumental to initially understand the biochemistry of HIF-PHDs and to develop PHIs . However, collagen prolyl hydroxylases exhibit high O2-affinity, and are therefore not well suited as oxygen sensors within the cell. This is different for the JmjC demethylases as described earlier this year by two remarkable back-to-back publications [4, 14]. Both groups reported hypoxia-induced decreases in demethylase activity of KDM6A  and KDM5A  which fell into the physiological range of hypoxic pO2 values found in tissues. Moreover, Km values for O2 of the enzymes KDM6A and KDM5A were comparable to PHDs. Interestingly, the chromatin reprogramming effects were oxygen-sensitive but independent of HIF, and significantly affected differentiation  and hypoxia-induced gene expression . Thus, within the physiological range of pO2 values, chromatin appears to be reprogrammed to allow hypoxia-induced gene expression. The question remains on how these enzymes are recruited to the corresponding genomic loci.
Second, looking for non-HIF targets of PHDs relies on the properties of PHDs to act as O2 sensors within the physiological pO2 range. Low affinity for O2 would allow hydroxylation of targets other than HIF at pO2 values relevant to physiology or pathophysiology in cells. Thus, non-HIF-dependent effects could well be controlled by HIF-PHDs if the respective proteins were targets of PHDs.
Third, with the use of PHIs, the question arises whether on-target (i.e., PHDs) or off-target (i.e., other 2-OG and iron-dependent dioxygenases) effects of the PHIs elicit relevant processes in cells or tissue. Initially, with non-selective inhibitors of 2-OG-dependent dioxygenases, one always had to be aware of effects on other members of this enzyme family. In fact, early attempts to inhibit excessive collagen formation by the use of fibrosis inhibitors targeting collagen synthesis, such as in myocardial scarring , may have affected HIF signaling as well. Nowadays, with PHIs more specific not only for PHDs, but also for FIH, it became of special interest to understand potential effects of PHDs on non-HIF targets.
Hydroxylation of non-HIF targets by HIF hydroxylases has been reported for more than two dozens of proteins and has been excellently reviewed very recently [23, 132]. Of note, it has to be critically discriminated between FIH and PHDs targeting non-HIF proteins. From the considerable number of FIH-dependent non-HIF hydroxylation targets identified , at least some have also been linked to biological functions that appear to depend on the enzymatic asparagine hydroxylase activity . However, for most if not all of the potential non-HIF substrates of PHDs, the Ratcliffe group raised considerable doubts whether prolyl hydroxylase activity is relevant . They rigorously tested more than 20 of these potential substrates for hydroxylation, using recombinant PHD enzymes under in vitro conditions which allowed for robust HIF hydroxylation. However, they could not confirm that these candidate substrates were proline hydroxylation targets of HIF-PHDs. One reason for this misconception in earlier studies may have been non-enzymatic oxidation of putative substrates erroneously interpreted as hydroxylation due to an identical mass increase in mass spectrometry . The authors, however, do not exclude that conditions different from the in vitro setting they used may allow for PHD catalyzed prolyl hydroxylation of non-HIF targets. Moreover, protein-protein interaction between PHDs and non-HIF targets, without any enzymatic hydroxylation, may account for previous reports claiming a role of PHDs in non-HIF target regulation. In the following, two examples will illustrate this.
Severe hypoxia causes endoplasmic reticulum (ER) stress, inducing an unfolded protein response (UPR) . Among the proteins involved in UPR is the activation transcription factor 4 (ATF4). ATF4 was reported to be regulated specifically by PHD3 . Hypoxia or PHD inhibition by DMOG increased ATF4 levels in cells. Mechanistically, five prolyl residues were identified in the zipper II-domain of ATF4 to be important for the interaction with PHD3 and to achieve hypoxic stabilization of ATF4. Nevertheless, the authors could not detect direct hydroxylation of prolines in ATF4 and concluded that PHD3 might regulate ATF4 through proline residue-dependent interaction, possibly without proline hydroxylation . This would be in full agreement with the recent in vitro findings by the Ratcliffe group .
Another claimed non-HIF PHD target is the inhibitor of nuclear factor-κB (NF-κB) kinase β (IKKβ), a key kinase within the NF-κB pathway. NF-κB activation by severe hypoxia has been first reported in 1994  and attributed to several NF-κB-dependent responses . Particularly the role for PDH1 to hydroxylate IKKβ at proline 191 inducing NF-κB activity was reported . Hydroxylation of IKKβ was later confirmed by mass spectrometry although not consistently confirmed by other analytical methods. One explanation for this discrepancy may be the special cellular conditions that are required for PHD activity, as reasoned by Cockman et al. . PHD1, responsible for NF-κB-dependent effects such as leukocyte adhesion , is exclusively localized in the cell nucleus . Studies on PHD2 have revealed that nuclear activity differs from cytoplasmic PHD activity . Thus, it is conceivable that factors associated with PHDs in specific subcellular compartments affect hydroxylation of non-HIF targets.
Finally, the synthesis of PHIs has brought HIF signaling into medical use as a therapeutic drug to increase HIF-dependent Epo production . This raises the question whether other PHD substrates such as non-HIF targets may be affected by specific PHIs. A comparison between the readily detectable hydroxylase activities of FIH on non-HIF targets with the lack of hydroxylation by PHDs indicates a significant difference: unlike FIH, PHDs undergo substantial conformational changes to perfectly match the binding of the HIF substrates . This conformational change is not required for FIH which apparently displays an easier accessible catalytic cleft for non-HIF substrates.
Collectively, these findings may preclude proline hydroxylation of non-HIF targets and confirm the selectivity of PHDs for HIF, at least until now . Clearly, clinical application of PHIs will provide the data whether other targets than HIF are affected by these compounds in a biologically relevant way.
Development of PHIs for the treatment of anemia
Almost all patients with chronic kidney disease develop a normochromic and normocytic anemia, largely irrespective of the etiology and other characteristics of their kidney disease. The interindividual variability of hemoglobin concentrations for a given decrease in glomerular filtration rate is large, but average hemoglobin levels start to decline when the glomerular filtration rate falls to about 60 ml/min per 1.73 m2 . When left untreated, renal anemia contributes significantly to impaired well-being and reduced quality of live. In the absence of alternative therapy, patients on dialysis long suffered from the need for regular blood cell transfusions with the associated consequences of immunization and iron overload. The pathogenesis of renal anemia is multifactorial, with increased blood loss, reduced red cell life span, and impaired erythropoiesis . Serum Epo levels do not increase with decreasing hemoglobin concentrations, as in any other types of anemia occurring in the presence of normal kidney function. The role of inappropriately low Epo production remained somewhat unclear, however, until recombinant human Epo became available in the late 1980s and turned out to be surprisingly effective. Treatment with recombinant human Epo or its derivatives has since become the widely established standard of care, in particular in patients on dialysis, and represents a major progress in dialysis care. The intended increase in hematocrit may facilitate thrombosis formation and the occlusion of dialysis shunts. Otherwise, recombinant human Epo is generally safe, with the exception of primarily two safety issues that have arisen. First, treatment with recombinant Epo can induce antibodies that cross-react with residual endogenous Epo and cause pure red cell aplasia; a serious but fortunately very rare complication . Second, attempts to increase the hemoglobin concentration with recombinant Epo to the (sub)normal range, as compared with lower target concentrations, have revealed increased risks of stroke  and mortality , which raised concerns about Epo therapy in general . Of note, the amounts of recombinant Epo used routinely in patients with CKD, in particular when aiming for high hemoglobin concentrations or in the presence of Epo resistance due to inflammation, is much higher than the amount endogenously produced Epo in individuals with intact kidneys. In addition, following i.v. Epo injections “unphysiologically,” high plasma concentrations are achieved. It has been suggested that such high Epo concentrations may induce adverse off-target effects , but the underlying mechanisms have not been resolved.
Soon after the discovery of the mechanisms that regulate HIF and in particular of the role of oxoglutarate-dependent dioxygenases for the destabilization of HIFα subunits, the concept arose that oxoglutarate analogues may be used to enhance endogenous Epo formation and thereby treat the anemia associated with chronic kidney disease. In fact, several clinical observations, obtained before recombinant human Epo became available, had indicated that the inappropriately low Epo production in patients with kidney disease may not be due to a total failure of its production capacity, opening the possibility that production of endogenous Epo might be enhanced. For example, an increase in hemoglobin was observed when dialysis patients were exposed to high-altitude conditions  or experienced acute episodes of hypoxia . Similiarly, in some patients with kidney failure anemia improved during episodes of hepatitis . While this could easily be explained by increased hepatic Epo production, case reports in patients improving their anemia after the development of cysts in otherwise non-functioning kidneys provided early evidence also for retained renal production capacity of Epo . This was subsequently confirmed by a proof-of-concept study in which a single dose of a PHI was given to healthy volunteers and dialysis patients, some of them being anephric after bilateral nephrectomy for various reasons . These anephric individuals revealed an increase in their plasma Epo concentration that appeared only slightly lower than in healthy individuals. Most remarkably, in some dialysis patients who still had their “non-functioning” kidneys in situ, Epo levels increased much higher than in anephric patients. Although the response was more variable, it clearly indicated a preserved and quantitatively very significant production capacity for Epo .
Meanwhile, at least six different companies have initiated clinical development programs for orally active PHIs [80, 112, 133], with the programs of three compounds being most advanced: roxadustat, vadadustat, and daprodustat. Phase 2 studies in patients with kidney disease, with and without need for dialysis, have confirmed the ability of these PHIs to stimulate erythropoiesis and increase the hemoglobin concentration in treatment-naïve patients, or maintain the hemoglobin concentration in patients previously treated with recombinant human Epo . The response was demonstrated to be dose-dependent and depending on dose and dosing intervals, gradual increases in hemoglobin concentrations of varying speed could be achieved . These studies were usually short, lasting for up to 20 weeks, and most of them enrolled no more than 100 patients. There were no apparent safety signals, so that each compound moved into phase 3 programs, aiming to demonstrate non-inferiority in comparison to recombinant human Epo with respect to hemoglobin concentrations and at least similar if not improved cardiovascular safety. The results of two small phase 3 studies conducted in China have recently been published [17, 18] and resulted in regulatory approval of the first PHI for renal anemia management in China .
One of the fascinating aspects of this approach is that despite the widespread range of HIF target genes, the use of PHIs for anemia management relies primarily on HIF-dependent stimulation of a single gene, i.e., the EPO gene that is among the earliest and most sensitive responders to HIF activation. Why the EPO gene is so much more sensitive to HIF than several hundred other genes is not fully understood. Moreover, there is some evidence that this sensitivity is relative rather than absolute, and that other pathways are simultaneously activated. For example, in anemia PHI trials, both roxadustat and dapradustat were found to lower total and LDL cholesterol levels [17, 18], a potentially beneficial side effect. It has also been suggested that HIF stabilization enhances enteral iron absorption, in part by increasing HIF-2-dependent intestinal iron transport [79, 126], and in part by decreasing the concentrations of hepcidin, a systemic negative regulator of iron uptake [44, 158]. To which extent the effects observed in clinical PHI trials are direct or indirect, e.g., due to increased iron demand by the bone marrow and hepcidin inhibition induced by stimulated erythropoiesis , requires further study. In any case, the confirmation in humans that PHIs do not exclusively stimulate Epo expression underlines that the use of these drugs needs to be monitored with a high level of vigilance for potentially adverse effects. On the other hand, using PHIs intentionally to induce a broad hypoxia response has a huge potential beyond anemia treatment.
Use of PHIs beyond anemia treatment
Major diseases like ischemic insults, including coronary artery, cerebrovascular and peripheral vascular disease, but also traumatic and infectious tissue injury as well as inflammatory diseases, are associated with altered tissue oxygen availability [9, 28, 142]. Numerous studies, for which a comprehensive citation is beyond the limits of this review, have shown that non-specific 2-oxoglutarate-dependent hydroxylase inhibitors like dimethyloxalylglycine (DMOG), but also more specific PHIs like roxadustat protect ischemic tissues when applied to rodent models. This includes disease models of myocardial infarction , cerebral ischemia , kidney ischemia  or organ transplantation , and inflammatory colitis  or pulmonary disease . In these preclinical studies, inhibition of PHDs has been shown to stabilize HIFα, conferring hypoxia-adaptive pathways including angiogenesis, metabolic reprogramming, and anti-apoptotic effects. The important questions that need to be answered in the future are in which scenario, when and for how long is a treatment with a PHI useful? Moreover, which PHD isoform should ideally be trageted for tissue protection? Since PHD1-3 and FIH have different functions in normoxia and hypoxia, inhibition of a specific PHD isoform in comparison with non-selective pan-inhibition of all PHDs and FIH might rather induce a mild than a full activation of the HIF system. Roxadustat is known to inhibit the PHD enzymes, whereas FIH is only marginally affected. In line with this, the relative specificity of roxadustat towards PHDs results in an incomplete mimicry of the hypoxic response .
The preclinical studies performed with either PHD knock-out mouse models or with in vivo applications mentioned above indicate that ischemic and inflammatory diseases might primarily benefit from a respective treatment strategy, which leaves the questions of when and how. The right timing of inhibiting the PHD enzymes is important and must be addressed since the onset of ischemic insults or the development of inflammatory diseases is usually not predictable. The physiological tissue function of HIF should be mimicked at best. Therefore, it needs to be analyzed if a post-insult treatment can still mediate protection for each potential indication. Reperfusion after revascularization of an acute myocardial infarction for example can contribute up to 50% of the resulting infarct damage. Interestingly, in this setting, tissue protection is not only mediated when PHIs are given pre-insult, but also when given immediately post-insult, which is much more relevant from a clinical perspective . Protection from reperfusion injury is therefore an important strategy to improve tissue integrity. Likewise, organ and tissue tropism are important. Systemic inhibition of the PHD enzymes will induce a widespread response. In case of anemia management, liver and kidneys would be the ideal target organs. To target other tissues, spatiotemporally restricted PHI applications need to be considered in addition to pharmacokinetics to limit both side-effects (e.g., overstimulation of the HIF pathway or dysregulation of non-HIF pathways) as well as off-target effects (inhibition of non-HIF hydroxylases). Local application in case of inflammatory bowel disease and intracoronary injection during reperfusion in case of myocardial infarction are just two examples of how a PHI could be applied in a tissue-restricted manner.
One of the unwanted side and/or off-target effects of PHIs is a putative function as tumor promoter, which needs to be ruled out especially in view of the function of HIFα in tumor progression and therapy resistance, as discussed above. So far, there is no indication that roxadustat or other PHIs promote tumor progression [1, 5, 117]. PHD inhibition or PHD2 loss-of-function even seems to inactivate stromal fibroblasts, leading to decreased tumor stiffness and metastasis .
In summary, a better understanding of the side and off-target effects as well as an increased insight into the HIF-dependent tissue-protective pathways will help to design PHIs with improved on-target and tissue-protective features.
RHW was supported by the National Centre of Competence in Research “Kidney.CH” and the Swiss National Science Foundation (310030_184813); and JS received funding from the German Research Foundation (387509280-SFB 1350 C5).
- 1.Adams DF, Watkins MS, Durette L, Laliberte J, Goulet F, Debien E, Frazier KS, Mellal N, Chen L, Shi W, Thomas R, Hu E (2019) Carcinogenicity assessment of daprodustat (GSK1278863), a hypoxia-inducible factor (HIF)-prolyl hydroxylase inhibitor. Toxicol Pathol:192623319880445. https://doi.org/10.1177/0192623319880445
- 5.Beck J, Henschel C, Chou J, Lin A, Del Balzo U (2017) Evaluation of the carcinogenic potential of roxadustat (FG-4592), a small molecule inhibitor of hypoxia-inducible factor prolyl hydroxylase in CD-1 mice and Sprague Dawley rats. Int J Toxicol 36:427–439. https://doi.org/10.1177/1091581817737232 CrossRefPubMedGoogle Scholar
- 6.Berchner-Pfannschmidt U, Tug S, Trinidad B, Oehme F, Yamac H, Wotzlaw C, Flamme I, Fandrey J (2008) Nuclear oxygen sensing: induction of endogenous prolyl-hydroxylase 2 activity by hypoxia and nitric oxide. J Biol Chem 283:31745–31753. https://doi.org/10.1074/jbc.M804390200 CrossRefPubMedGoogle Scholar
- 7.Bernhardt WM, Gottmann U, Doyon F, Buchholz B, Campean V, Schödel J, Reisenbuechler A, Klaus S, Arend M, Flippin L, Willam C, Wiesener MS, Yard B, Warnecke C, Eckardt KU (2009) Donor treatment with a PHD-inhibitor activating HIFs prevents graft injury and prolongs survival in an allogenic kidney transplant model. Proc Natl Acad Sci USA 106:21276–21281PubMedGoogle Scholar
- 10.Blumberg A, Keller H, Marti HR (1973) Effect of altitude on erythropoiesis and oxygen affinity in anaemic patients on maintenance dialysis. Eur J Clin Invest 3:93–97. https://doi.org/10.1111/j.1365-2362.1973.tb00334.x CrossRefPubMedGoogle Scholar
- 12.Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans K, Dewerchin M, Neeman M, Bono F, Abramovitch R, Maxwell P, Koch CJ, Ratcliffe P, Moons L, Jain RK, Collen D, Keshet E (1998) Role of HIF-1α in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature 394:485–490PubMedGoogle Scholar
- 13.Cavadas MA, Mesnieres M, Crifo B, Manresa MC, Selfridge AC, Keogh CE, Fabian Z, Scholz CC, Nolan KA, Rocha LM, Tambuwala MM, Brown S, Wdowicz A, Corbett D, Murphy KJ, Godson C, Cummins EP, Taylor CT, Cheong A (2016) REST is a hypoxia-responsive transcriptional repressor. Sci Rep 6:31355. https://doi.org/10.1038/srep31355 CrossRefPubMedPubMedCentralGoogle Scholar
- 14.Chakraborty AA, Laukka T, Myllykoski M, Ringel AE, Booker MA, Tolstorukov MY, Meng YJ, Meier SR, Jennings RB, Creech AL, Herbert ZT, McBrayer SK, Olenchock BA, Jaffe JD, Haigis MC, Beroukhim R, Signoretti S, Koivunen P, Kaelin WG Jr (2019) Histone demethylase KDM6A directly senses oxygen to control chromatin and cell fate. Science 363:1217–1222. https://doi.org/10.1126/science.aaw1026 CrossRefPubMedGoogle Scholar
- 15.Chan MC, Ilott NE, Schodel J, Sims D, Tumber A, Lippl K, Mole DR, Pugh CW, Ratcliffe PJ, Ponting CP, Schofield CJ (2016) Tuning the transcriptional response to hypoxia by inhibiting hypoxia-inducible factor (HIF) prolyl and asparaginyl hydroxylases. J Biol Chem 291:20661–20673. https://doi.org/10.1074/jbc.M116.749291 CrossRefPubMedPubMedCentralGoogle Scholar
- 17.Chen N, Hao C, Liu BC, Lin H, Wang C, Xing C, Liang X, Jiang G, Liu Z, Li X, Zuo L, Luo L, Wang J, Zhao MH, Liu Z, Cai GY, Hao L, Leong R, Wang C, Liu C et al (2019) Roxadustat treatment for anemia in patients undergoing long-term dialysis. N Engl J Med 381:1011–1022. https://doi.org/10.1056/NEJMoa1901713 CrossRefPubMedGoogle Scholar
- 18.Chen N, Hao C, Peng X, Lin H, Yin A, Hao L, Tao Y, Liang X, Liu Z, Xing C, Chen J, Luo L, Zuo L, Liao Y, Liu BC, Leong R, Wang C, Liu C, Neff T, Szczech L et al (2019) Roxadustat for anemia in patients with kidney disease not receiving dialysis. N Engl J Med 381:1001–1010. https://doi.org/10.1056/NEJMoa1813599 CrossRefPubMedGoogle Scholar
- 19.Chen W, Hill H, Christie A, Kim MS, Holloman E, Pavia-Jimenez A, Homayoun F, Ma Y, Patel N, Yell P, Hao G, Yousuf Q, Joyce A, Pedrosa I, Geiger H, Zhang H, Chang J, Gardner KH, Bruick RK, Reeves C et al (2016) Targeting renal cell carcinoma with a HIF-2 antagonist. Nature 539:112–117. https://doi.org/10.1038/nature19796 CrossRefPubMedPubMedCentralGoogle Scholar
- 20.Cho H, Du X, Rizzi JP, Liberzon E, Chakraborty AA, Gao W, Carvo I, Signoretti S, Bruick RK, Josey JA, Wallace EM, Kaelin WG (2016) On-target efficacy of a HIF-2α antagonist in preclinical kidney cancer models. Nature 539:107–111. https://doi.org/10.1038/nature19795 CrossRefPubMedPubMedCentralGoogle Scholar
- 22.Chowdhury R, Leung IK, Tian YM, Abboud MI, Ge W, Domene C, Cantrelle FX, Landrieu I, Hardy AP, Pugh CW, Ratcliffe PJ, Claridge TD, Schofield CJ (2016) Structural basis for oxygen degradation domain selectivity of the HIF prolyl hydroxylases. Nat Commun 7:12673. https://doi.org/10.1038/ncomms12673 CrossRefPubMedPubMedCentralGoogle Scholar
- 23.Cockman ME, Lippl K, Tian YM, Pegg HB, Figg WDJ, Abboud MI, Heilig R, Fischer R, Myllyharju J, Schofield CJ, Ratcliffe PJ (2019) Lack of activity of recombinant HIF prolyl hydroxylases (PHDs) on reported non-HIF substrates. eLife 8. https://doi.org/10.7554/eLife.46490
- 25.Cockman ME, Webb JD, Kramer HB, Kessler BM, Ratcliffe PJ (2009) Proteomics-based identification of novel factor inhibiting hypoxia-inducible factor (FIH) substrates indicates widespread asparaginyl hydroxylation of ankyrin repeat domain-containing proteins. Mol Cell Proteomics 8:535–546PubMedPubMedCentralGoogle Scholar
- 26.Cummins EP, Berra E, Comerford KM, Ginouves A, Fitzgerald KT, Seeballuck F, Godson C, Nielsen JE, Moynagh P, Pouysségur J, Taylor CT (2006) Prolyl hydroxylase-1 negatively regulates IκB kinase-β, giving insight into hypoxia-induced NFκB activity. Proc Natl Acad Sci USA 103:18154–18159PubMedGoogle Scholar
- 32.Epstein AC, Gleadle JM, McNeill LA, Hewitson KS, O’Rourke J, Mole DR, Mukherji M, Metzen E, Wilson MI, Dhanda A, Tian YM, Masson N, Hamilton DL, Jaakkola P, Barstead R, Hodgkin J, Maxwell PH, Pugh CW, Schofield CJ, Ratcliffe PJ (2001) C. elegans EGL-9 and mammalian homologs define a family of dioxygenases that regulate HIF by prolyl hydroxylation. Cell 107:43–54PubMedGoogle Scholar
- 37.Grampp S, Platt JL, Lauer V, Salama R, Kranz F, Neumann VK, Wach S, Stohr C, Hartmann A, Eckardt KU, Ratcliffe PJ, Mole DR, Schodel J (2016) Genetic variation at the 8q24.21 renal cancer susceptibility locus affects HIF binding to a MYC enhancer. Nat Commun 7:13183. https://doi.org/10.1038/ncomms13183 CrossRefPubMedPubMedCentralGoogle Scholar
- 38.Grampp S, Schmid V, Salama R, Lauer V, Kranz F, Platt JL, Smythies J, Choudhry H, Goppelt-Struebe M, Ratcliffe PJ, Mole DR, Schödel J (2017) Multiple renal cancer susceptibility polymorphisms modulate the HIF pathway. PLoS Genet 13:e1006872. https://doi.org/10.1371/journal.pgen.1006872 CrossRefPubMedPubMedCentralGoogle Scholar
- 40.Haidar MA, Loya F, Yang Y, Lin H, Glassman A, Keating MJ, Goldwasser E, Albitar M (1996) Differential expression of lacZ in the liver and kidney of transgenic mice carrying chimeric lacZ-erythropoietin gene constructs with or without its 1.2 kb 3'-flanking sequence. Nucl Acids Res 24:3621–3628PubMedGoogle Scholar
- 41.Hewitson KS, McNeill LA, Riordan MV, Tian YM, Bullock AN, Welford RW, Elkins JM, Oldham NJ, Bhattacharya S, Gleadle JM, Ratcliffe PJ, Pugh CW, Schofield CJ (2002) Hypoxia-inducible factor (HIF) asparagine hydroxylase is identical to factor inhibiting HIF (FIH) and is related to the cupin structural family. J Biol Chem 277:26351–26355PubMedGoogle Scholar
- 49.Ivan M, Haberberger T, Gervasi DC, Michelson KS, Günzler V, Kondo K, Yang H, Sorokina I, Conaway RC, Conaway JW, Kaelin WG Jr (2002) Biochemical purification and pharmacological inhibition of a mammalian prolyl hydroxylase acting on hypoxia-inducible factor. Proc Natl Acad Sci USA 99:13459–13464PubMedGoogle Scholar
- 52.Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, von Kriegsheim A, Hebestreit HF, Mukherji M, Schofield CJ, Maxwell PH, Pugh CW, Ratcliffe PJ (2001) Targeting of HIF-α to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science 292:468–472Google Scholar
- 56.KDIGO Clinical Practice Guideline for Anemia in Chronic Kidney Disease (2012) Kidney Int Suppl 2:279–335Google Scholar
- 60.Koivunen P, Lee S, Duncan CG, Lopez G, Lu G, Ramkissoon S, Losman JA, Joensuu P, Bergmann U, Gross S, Travins J, Weiss S, Looper R, Ligon KL, Verhaak RG, Yan H, Kaelin WG Jr (2012) Transformation by the (R)-enantiomer of 2-hydroxyglutarate linked to EGLN activation. Nature 483:484–488PubMedPubMedCentralGoogle Scholar
- 75.Losman JA, Looper RE, Koivunen P, Lee S, Schneider RK, McMahon C, Cowley GS, Root DE, Ebert BL, Kaelin WG Jr (2013) (R)-2-hydroxyglutarate is sufficient to promote leukemogenesis and its effects are reversible. Science 339:1621–1625. https://doi.org/10.1126/science.1231677 CrossRefPubMedGoogle Scholar
- 77.Madsen CD, Pedersen JT, Venning FA, Singh LB, Moeendarbary E, Charras G, Cox TR, Sahai E, Erler JT (2015) Hypoxia and loss of PHD2 inactivate stromal fibroblasts to decrease tumour stiffness and metastasis. EMBO Rep 16:1394–1408. https://doi.org/10.15252/embr.201540107 CrossRefPubMedPubMedCentralGoogle Scholar
- 85.Maxwell PH, Wiesener MS, Chang GW, Clifford SC, Vaux EC, Cockman ME, Wykoff CC, Pugh CW, Maher ER, Ratcliffe PJ (1999) The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature 399:271–275Google Scholar
- 87.Metzen E, Berchner-Pfannschmidt U, Stengel P, Marxsen JH, Stolze I, Klinger M, Huang WQ, Wotzlaw C, Hellwig-Burgel T, Jelkmann W, Acker H, Fandrey J (2003) Intracellular localisation of human HIF-1 α hydroxylases: implications for oxygen sensing. J Cell Sci 116:1319–1326. https://doi.org/10.1242/jcs.00318 CrossRefPubMedGoogle Scholar
- 88.Mimura I, Nangaku M, Kanki Y, Tsutsumi S, Inoue T, Kohro T, Yamamoto S, Fujita T, Shimamura T, Suehiro J, Taguchi A, Kobayashi M, Tanimura K, Inagaki T, Tanaka T, Hamakubo T, Sakai J, Aburatani H, Kodama T, Wada Y (2012) Dynamic change of chromatin conformation in response to hypoxia enhances the expression of GLUT3 (SLC2A3) by cooperative interaction of hypoxia-inducible factor 1 and KDM3A. Mol Cell Biol 32:3018–3032PubMedPubMedCentralGoogle Scholar
- 96.Nwogu JI, Geenen D, Bean M, Brenner MC, Huang X, Buttrick PM (2001) Inhibition of collagen synthesis with prolyl 4-hydroxylase inhibitor improves left ventricular function and alters the pattern of left ventricular dilatation after myocardial infarction. Circulation 104:2216–2221. https://doi.org/10.1161/hc4301.097193 CrossRefPubMedGoogle Scholar
- 98.Ortiz-Barahona A, Villar D, Pescador N, Amigo J, del Peso L (2010) Genome-wide identification of hypoxia-inducible factor binding sites and target genes by a probabilistic model integrating transcription-profiling data and in silico binding site prediction. Nucl Acids Res 38:2332–2345PubMedGoogle Scholar
- 99.Pfeffer MA, Burdmann EA, Chen CY, Cooper ME, de Zeeuw D, Eckardt KU, Feyzi JM, Ivanovich P, Kewalramani R, Levey AS, Lewis EF, McGill JB, McMurray JJ, Parfrey P, Parving HH, Remuzzi G, Singh AK, Solomon SD, Toto R (2009) A trial of darbepoetin alfa in type 2 diabetes and chronic kidney disease. N Engl J Med 361:2019–2032. https://doi.org/10.1056/NEJMoa0907845 CrossRefPubMedGoogle Scholar
- 101.Pollard PJ, Brière JJ, Alam NA, Barwell J, Barclay E, Wortham NC, Hunt T, Mitchell M, Olpin S, Moat SJ, Hargreaves IP, Heales SJ, Chung YL, Griffiths JR, Dalgleish A, McGrath JA, Gleeson MJ, Hodgson SV, Poulsom R, Rustin P et al (2005) Accumulation of Krebs cycle intermediates and over-expression of HIF1α in tumours which result from germline FH and SDH mutations. Hum Mol Genet 14:2231–2239PubMedGoogle Scholar
- 102.Provenzano R, Besarab A, Sun CH, Diamond SA, Durham JH, Cangiano JL, Aiello JR, Novak JE, Lee T, Leong R, Roberts BK, Saikali KG, Hemmerich S, Szczech LA, Yu KH, Neff TB (2016) Oral hypoxia-inducible factor prolyl hydroxylase inhibitor roxadustat (FG-4592) for the treatment of anemia in patients with CKD. Clinical journal of the American Society of Nephrology : CJASN 11:982–991. https://doi.org/10.2215/cjn.06890615 CrossRefPubMedGoogle Scholar
- 104.Purdue MP, Johansson M, Zelenika D, Toro JR, Scelo G, Moore LE, Prokhortchouk E, Wu X, Kiemeney LA, Gaborieau V, Jacobs KB, Chow WH, Zaridze D, Matveev V, Lubinski J, Trubicka J, Szeszenia-Dabrowska N, Lissowska J, Rudnai P, Fabianova E et al (2011) Genome-wide association study of renal cell carcinoma identifies two susceptibility loci on 2p21 and 11q13.3. Nat Genet 43:60–65. https://doi.org/10.1038/ng.723 CrossRefPubMedGoogle Scholar
- 108.Rossert J, Casadevall N, Eckardt KU (2004) Anti-erythropoietin antibodies and pure red cell aplasia. J Am Soc Nephrol 15:398–406. https://doi.org/10.1097/01.asn.0000107561.59698.42 CrossRefPubMedGoogle Scholar
- 113.Scheuermann TH, Li Q, Ma HW, Key J, Zhang L, Chen R, Garcia JA, Naidoo J, Longgood J, Frantz DE, Tambar UK, Gardner KH, Bruick RK (2013) Allosteric inhibition of hypoxia inducible factor-2 with small molecules. Nat Chem Biol 9:271–276. https://doi.org/10.1038/nchembio.1185 CrossRefPubMedPubMedCentralGoogle Scholar
- 114.Schley G, Klanke B, Kalucka J, Schatz V, Daniel C, Mayer M, Goppelt-Struebe M, Herrmann M, Thorsteinsdottir M, Palsson R, Beneke A, Katschinski DM, Burzlaff N, Eckardt KU, Weidemann A, Jantsch J, Willam C (2019) Mononuclear phagocytes orchestrate prolyl hydroxylase inhibition-mediated renoprotection in chronic tubulointerstitial nephritis. Kidney Int 96:378–396. https://doi.org/10.1016/j.kint.2019.02.016 CrossRefPubMedGoogle Scholar
- 115.Schödel J, Bardella C, Sciesielski LK, Brown JM, Pugh CW, Buckle V, Tomlinson IP, Ratcliffe PJ, Mole DR (2012) Common genetic variants at the 11q13.3 renal cancer susceptibility locus influence binding of HIF to an enhancer of cyclin D1 expression. Nat Genet 44(420-425):S421–S422Google Scholar
- 117.Seeley TW, Sternlicht MD, Klaus SJ, Neff TB, Liu DY (2017) Induction of erythropoiesis by hypoxia-inducible factor prolyl hydroxylase inhibitors without promotion of tumor initiation, progression, or metastasis in a VEGF-sensitive model of spontaneous breast cancer. Hypoxia 5:1–9. https://doi.org/10.2147/hp.s130526 CrossRefPubMedPubMedCentralGoogle Scholar
- 138.Tambuwala MM, Manresa MC, Cummins EP, Aversa V, Coulter IS, Taylor CT (2015) Targeted delivery of the hydroxylase inhibitor DMOG provides enhanced efficacy with reduced systemic exposure in a murine model of colitis. J Control Release 217:221–227. https://doi.org/10.1016/j.jconrel.2015.09.022 CrossRefPubMedGoogle Scholar
- 141.Tausendschön M, Rehli M, Dehne N, Schmidl C, Doring C, Hansmann ML, Brüne B (2015) Genome-wide identification of hypoxia-inducible factor-1 and -2 binding sites in hypoxic human macrophages alternatively activated by IL-10. Biochim Biophys Acta 1849:10–22. https://doi.org/10.1016/j.bbagrm.2014.10.006 CrossRefPubMedGoogle Scholar
- 145.Tiana M, Acosta-Iborra B, Puente-Santamaria L, Hernansanz-Agustin P, Worsley-Hunt R, Masson N, Garcia-Rio F, Mole D, Ratcliffe P, Wasserman WW, Jimenez B, Del Peso L (2018) The SIN3A histone deacetylase complex is required for a complete transcriptional response to hypoxia. Nucl Acids Res 46:120–133. https://doi.org/10.1093/nar/gkx951 CrossRefPubMedGoogle Scholar
- 147.Vogler M, Zieseniss A, Hesse AR, Levent E, Tiburcy M, Heinze E, Burzlaff N, Schley G, Eckardt KU, Willam C, Katschinski DM (2015) Pre- and post-conditional inhibition of prolyl-4-hydroxylase domain enzymes protects the heart from an ischemic insult. Eur J Physiol 467:2141–2149. https://doi.org/10.1007/s00424-014-1667-z CrossRefGoogle Scholar
- 151.Warnecke C, Weidemann A, Volke M, Schietke R, Wu X, Knaup KX, Hackenbeck T, Bernhardt W, Willam C, Eckardt KU, Wiesener MS (2008) The specific contribution of hypoxia-inducible factor-2α to hypoxic gene expression in vitro is limited and modulated by cell type-specific and exogenous factors. Exp Cell Res 314:2016–2027. https://doi.org/10.1016/j.yexcr.2008.03.003 CrossRefPubMedGoogle Scholar
- 152.Warnecke C, Zaborowska Z, Kurreck J, Erdmann VA, Frei U, Wiesener M, Eckardt KU (2004) Differentiating the functional role of hypoxia-inducible factor (HIF)-1α and HIF-2α (EPAS-1) by the use of RNA interference: erythropoietin is a HIF-2α target gene in Hep3B and Kelly cells. FASEB J 18:1462–1464PubMedGoogle Scholar
- 157.Winning S, Splettstoesser F, Fandrey J, Frede S (2010) Acute hypoxia induces HIF-independent monocyte adhesion to endothelial cells through increased intercellular adhesion molecule-1 expression: the role of hypoxic inhibition of prolyl hydroxylase activity for the induction of NF-κ B. J Immunol 185:1786–1793. https://doi.org/10.4049/jimmunol.0903244 CrossRefPubMedGoogle Scholar
- 160.Yamashita T, Ohneda O, Nagano M, Iemitsu M, Makino Y, Tanaka H, Miyauchi T, Goto K, Ohneda K, Fujii-Kuriyama Y, Poellinger L, Yamamoto M (2008) Abnormal heart development and lung remodeling in mice lacking the hypoxia-inducible factor-related basic helix-loop-helix PAS protein NEPAS. Mol Cell Biol 28:1285–1297. https://doi.org/10.1128/mcb.01332-07 CrossRefPubMedGoogle Scholar
- 161.Yao X, Tan J, Lim KJ, Koh J, Ooi WF, Li Z, Huang D, Xing M, Chan YS, Qu JZ, Tay ST, Wijaya G, Lam YN, Hong JH, Lee-Lim AP, Guan P, Ng MSW, He CZ, Lin JS, Nandi T et al (2017) VHL deficiency drives enhancer activation of oncogenes in clear cell renal cell carcinoma. Cancer Discov 7:1284–1305. https://doi.org/10.1158/2159-8290.cd-17-0375 CrossRefPubMedGoogle Scholar