Targeting hypoxia-inducible factors: therapeutic opportunities and challenges

Abstract

Hypoxia-inducible factors (HIFs) are highly conserved transcription factors that are crucial for adaptation of metazoans to limited oxygen availability. Recently, HIF activation and inhibition have emerged as therapeutic targets in various human diseases. Pharmacologically desirable effects of HIF activation include erythropoiesis stimulation, cellular metabolism optimization during hypoxia and adaptive responses during ischaemia and inflammation. By contrast, HIF inhibition has been explored as a therapy for various cancers, retinal neovascularization and pulmonary hypertension. This Review discusses the biochemical mechanisms that control HIF stabilization and the molecular strategies that can be exploited pharmacologically to activate or inhibit HIFs. In addition, we examine medical conditions that benefit from targeting HIFs, the potential side effects of HIF activation or inhibition and future challenges in this field.

Introduction

Fluctuating atmospheric oxygen levels posed an environmental challenge during the evolution of metazoans¹, generating a selection pressure that promoted the development of cellular oxygen-sensing pathways¹. Except for Porifera (sponges) and Ctenophora (comb jellies), all metazoan species use hypoxia-inducible factors (HIFs) to achieve oxygen homeostasis². HIFs are transcription factors that are regulated by the abundance of oxygen. In the presence of oxygen, HIFs are degraded via the proteasomal pathway. During limited oxygen availability (hypoxia), HIFs are stabilized and can directly promote the transcription of target genes, such as those encoding erythropoietin (EPO), members of the glycolytic enzymes³, vascular endothelial growth factors (VEGFs)⁴ or gene products crucial for the generation and signalling of extracellular adenosine^5,6 (Box 1). Systemic reactions to hypoxia include concerted efforts of the respiratory and cardiovascular systems to enhance oxygen delivery by increasing cardiac output and respiration. Adaptive responses on a cellular level include a shift from oxidative metabolic pathways to nonoxidative pathways^7,8, for example, by promoting glycolytic metabolism in ischaemic tissues to enhance cellular survival and ischaemia resistance^9,10. Importantly, both systemic and cellular responses are controlled by HIFs and oxygen-sensing pathways^11,12. In some hypoxic instances, however, HIF promotes the repression of target genes¹³, which most commonly occurs indirectly through the induction of hypoxia-responsive microRNAs (miRNAs)^14,15 (Box 1). For example, liver-specific miRNA miR-122 is a direct HIF target that subsequently promotes repression of its target gene EGLN2 (which encodes HIF prolyl hydroxylase 1, PHD1), thereby promoting ischaemia tolerance of the liver¹⁶.

Box 1. Examples of HIF1α or HIF2α target genes during hypoxia, inflammation and cancer.

Hypoxia-inducible factors (HIFs) are stabilized during conditions of hypoxia¹⁰⁷, inflammation²⁹, metabolic imbalance¹¹⁰ and cancer²⁸. The specific transcriptional response of an individual cell type can vary depending on several factors, including the relative expression or activity of individual HIFα subunits, the expression of different HIF prolyl hydroxylase domain-containing proteins (HIF-PHDs), coactivators or other cell- and tissue-specific modulators²¹⁰. Below are examples of well-established pathways under the transcriptional control of HIFs.

Erythropoiesis

Studies in the early 1990s revealed that the binding of a novel protein to a hypoxia-responsive element (HRE) contained within the erythropoietin promoter stimulated its induction and was subsequently determined to be HIF1α^17–19,297. However, subsequent studies highlighted that HIF2α is the main transcriptional driver of erythropoietin responses during hypoxia. In addition, proteins crucial in iron absorption and metabolism — such as transferrin and ceruloplasmin — have been identified as HIF1α and HIF2α targets²⁹³.

Metabolic reprogramming

Genes encoding most enzymes of glycolysis have been identified as direct HIF1α target genes and are induced during hypoxia^3,298,299 or inflammation¹¹⁰. The HIF1α-dependent induction of the glycolytic system is protective during ischaemia and reperfusion, by promoting metabolic survival of myocytes or through HIF1α interaction with circadian rhythm proteins, such as period 2 (refs. 9,300). By contrast, the induction of glycolytic enzymes during hypoxia has been described as a distinct characteristic of cancer cells, in which HIF1α-dependent induction of glycolytic enzymes controls a metabolic phenotype that promotes glycolysis instead of mitochondrial respiration (Warburg effect)³⁰¹.

Growth factors

Similarly, HIFαs are crucial for the induction of tumour growth factors, such as vascular endothelial growth factor (VEGF)^302,303, via HIF2α in renal cell carcinoma⁵⁸. Another coordinated response by HIFs includes the induction of the growth factor amphiregulin (AREG) and its receptor epidermal growth factor receptor (EGFR) by HIF2α during hypoxia or myocardial injury^105,222. This response has been implicated in cardioprotection from ischaemia and reperfusion injury.

Extracellular adenosine signalling

Extracellular adenosine dampens acute inflammation and promotes tissue adaptation to conditions of hypoxia through the activation of adenosine receptors. HIF1α increases the extracellular production of adenosine through the induction of CD73, a cell-membrane-anchored enzyme responsible for converting precursor nucleotides into adenosine^304,305. Similarly, the ADORA2B adenosine receptor is a direct target gene for HIF1α and is implicated in tissue protection¹⁰⁶. In contrast, the anti-inflammatory ADORA2A is a HIF2α target³⁰⁶ and is crucial for dampening inflammation in vivo³⁰⁷. HIF1α also dampens extracellular adenosine uptake via equilibrative adenosine transporters (ENTs)^294,295 or adenosine metabolism to inosine via the adenosine kinase³⁰⁸. Finally, the alternative ADORA2B ligand netrin 1 (ref. 309) is a direct HIF1α target and is implicated in tissue protection^224,310. Of note, the activation of extracellular adenosine signalling drives cancer progression³¹¹, suggesting another mechanism for HIF-driven cancer development.

MicroRNAs

In many instances, repression of HIF targets includes the direct induction of HIF-dependent microRNAs (miRNAs) that subsequently promote repression of an indirect HIF target gene³¹². miR-210 is one of the best-studied hypoxia-regulated miRNAs (referred to as ‘hypoximiRs’) during cancer. HIF-dependent miR-210 induction functions as a positive feedback loop to enhance HIF signalling while providing key gene modification to drive cell cycle progression and metastasis³¹².

HIFs were originally discovered by Gregg Semenza in the early 1990s as a newly identified protein that binds to the EPO promoter and confers its transcriptional induction during hypoxia^17–19. This discovery was awarded the Nobel Prize for physiology or medicine in 2019 to Semenza, who was joined in receiving the award by William Kaelin and Peter Ratcliffe²⁰. Kaelin identified the crucial connection between Semenza’s discovery of HIF and von Hippel–Lindau (VHL) disease²¹. Ratcliffe discovered key regulatory components involving oxygen-sensing HIF prolyl hydroxylase domain-containing proteins (HIF-PHDs), a set of enzymes that function as oxygen sensors and regulate HIF stabilization during hypoxia^22,23.

Most of the earlier studies of HIFs were carried out by exposing cell lines or experimental animals to ambient hypoxia^24–27. However, it became apparent that HIFs have functional roles beyond hypoxia adaptation. HIFs are also stabilized during inflammatory conditions or cancer^28–32, partly owing to imbalances in metabolic supply and demand (including oxygen)^33,34. For example, activated inflammatory cells such as polymorphonuclear neutrophils (PMNs) are among the cells with the highest oxygen consumption. When PMNs are activated and undergo respiratory bursts, they cause hypoxia imprinting on their microenvironment and promote stabilization of HIFs³⁵. Notably, many of these studies revealed important functional roles of HIFs during inflammation and cancer by attenuating inflammation or dampening immune responses^29,36,37. Therefore, it is unsurprising that targeting HIFs for therapy has been evolving for inflammatory disease or cancer^30,38.

Over the past decade, pharmacological approaches that effectively modulate the stabilization or functions of HIFs have been developed and tested in clinical trials. Pharmacological approaches to promote the stabilization of HIFs in the absence of hypoxia have focused on small-molecule inhibitors of HIF-PHDs. Promotion of HIF stabilization has been implicated in benefiting ischaemia–reperfusion injury of the heart³⁹ and liver^40,41, acute kidney injury^42,43, acute respiratory distress syndrome (ARDS)^31,44, inflammatory bowel disease^45,46, infections with pathogens^47,48, haemorrhagic shock⁴⁹ and renal anaemia⁵⁰. Importantly, these pharmacological interventions have been tested in several phase III clinical trials reporting the efficacy of orally available HIF-PHD inhibitors (HIF-PHDis) for treatment of renal anaemia, as an alternative to recombinant EPO^51–54. Moreover, there are ongoing phase II clinical trials on the use of the HIF-PHDis to treat ARDS in patients with coronavirus disease 2019 (COVID-19) (vadadustat, NCT04478071, and desidustat, NCT04463602), as well as acute kidney injury (roxadustat, NCT05010460) and acute myocardial infarction (roxadustat, NCT04803864).

Similar advancements have also been made in developing novel pharmacological compounds that prevent the activation of HIF-dependent gene transcription. For example, PT2977 (belzutifan)^55–57 prevents the transcriptional activity of the HIF2α subunit by inhibiting the formation of its active complex with the HIF1β subunit⁵⁸. Phase III clinical trials have implicated belzutifan in treating renal cell carcinoma (RCC) in VHL disease^55,56. Moreover, other emerging compounds that effectively target HIF1α function have been assessed in combination with immune checkpoint blockade for treating solid tumours (such as hepatocellular carcinoma (HCC), 32–134D)⁵⁹. Other examples of therapeutic opportunities to target HIF stabilization include Pacak–Zhuang syndrome⁵⁷, pulmonary hypertension⁶⁰ and ocular neovascularization⁶¹.

In this Review, we describe the molecular mechanisms that control the transcriptional activity of HIFs and highlight pharmacological opportunities that promote the stabilization of HIFs or inhibit their functional activity. We discuss the pharmacological stabilization of HIFs to treat renal anaemia, inflammatory diseases or ischaemia–reperfusion injury. In addition, we focus on HIF inhibition for the treatment of cancer and other emerging indications. Finally, we review specific challenges to using HIF activators or inhibitors in patients and provide an outlook on the future opportunities and potential limitations for targeting hypoxia signalling pharmacologically.

HIFs and PHDs in oxygen sensing

Oxygen-dependent regulation of the HIF pathway is central for cellular and systemic adaptation to low oxygen levels. The tight regulation of HIFs is mediated by several key enzymes, including factor inhibiting HIF (FIH), HIF-PHDs and von Hippel–Lindau protein (pVHL).

Molecular structure of HIFs

The active transcriptional form of HIFs consists of an α-unit (HIFα) and a β-unit (HIF1β). HIFα is crucial for regulation of the transcriptional activity of HIFs. Based on homology searches, three isoforms of HIFα have been identified: HIF1α, HIF2α and HIF3α⁶². All three isoforms contain a basic helix–loop–helix and two Per–Arnt–Sim domains (bHLH-PAS) in the N terminus and an oxygen-dependent degradation domain in the C terminus, whereas only HIF1α and HIF2α contain a transactivation domain^63–65 (Fig. 1). Among the three isoforms, HIF1α was first identified as a transcription factor binding to the promoter region of EPO⁶⁶. HIF2α and HIF3α were later discovered based on their sequence homology with HIF1α⁷. Human HIF1α and HIF2α share 53.4% amino acid homology, and human HIF1α and HIF3α share 57.1% amino acid homology. Interestingly, the three HIFα isoforms have different functional roles based on tissue and cell specificity. HIF1α is expressed in all cell types, whereas HIF2α was initially perceived to be expressed only in endothelial cells⁶⁷ (hence its original name as endothelial PAS domain-containing protein 1, EPAS1). However, recent studies point to a global role of HIF2α and identified its functions outside vascular endothelial cells^68,69. HIF1α and HIF2α synergistically enhance expression of shared target genes, such as EPO or VEGF⁷⁰. However, HIF1α and HIF2α can have opposing biological functions. For example, HIF1α specifically regulates the inducible nitric oxide synthase (iNOS) gene (NOS2), whereas HIF2α controls arginase 1 expression, resulting in opposite effects on regulation of macrophage polarization or cancer metastasis^71,72. Because HIF3α lacks the transactivation domain, it is considered an inhibitory element (hence the name inhibitory PAS domain or IPAS)^73,74. Compared with HIF1α and HIF2α, the functional role of HIF3α and its various isoforms and splice variants have yet to be further elucidated. Therefore, our further discussion will focus on the HIF1α and HIF2α subunits.

Fig. 1|. Molecular structure of hypoxia-inducible factors.

Hypoxia-inducible factors (HIFs) consist of α (HIF1α, HIF2α and HIF3α) and β (HIF1β) subunits, with the α-subunit functioning as the principal regulator of HIF transcriptional activity. HIF1α and HIF2α feature an N-terminal basic helix–loop–helix (bHLH) domain and two Per–Arnt–Sim (PAS-A and PAS-B) domains crucial for dimerization and DNA binding^63,65. The C-terminal region contains a transcription activation domain (C-TAD) and a second N-terminal activation domain (N-TAD) and N- and C-terminal oxygen-dependent degradation domains (NODD and CODD)⁸⁶. Unlike HIF1α and HIF2α, HIF3α lacks the C-TAD and is therefore considered an inhibitory HIF factor^73,74. Under normoxic conditions, HIF prolyl hydroxylase domain-containing proteins (HIF-PHDs) hydroxylate specific proline residues on HIF1α, HIF2α and HIF3α, which depends on the availability of oxygen, α-ketoglutarate and iron⁶². Following hydroxylation, HIFα is targeted for proteasomal degradation^80,81. This process involves the binding of the von Hippel–Lindau protein (pVHL) and the initiation of ubiquitylation, facilitated by the complex that includes elongin B, elongin C (EloBC), cullin 2 (CUL2) and RING box protein 1 (RBX1)^80,81. An additional E3 ubiquitin ligase also contributes to this process, ultimately leading to the proteasomal degradation of HIFα⁸¹. In addition, factor inhibiting HIF (FIH) hydroxylates Asn803 within the C-TAD of HIF1α and Asn847 within the C-TAD of HIF2α^62,76, obstructing the binding of coactivators cyclic adenosine monophosphate response element binding protein (CREB) binding protein and histone acetyltransferase p300 (CBP–p300) and inhibiting HIF function. In hypoxic conditions, the activities of PHDs and FIH are decreased, enabling the stabilization and accumulation of active HIFα–HIF1β complexes, leading to the subsequent binding to hypoxia-response elements in target genes and the concomitant induction of HIF target genes. aa, amino acids; Ub, ubiquitin.

All three HIFα subunits require interaction with their heterodimeric partner HIF1β (also known as aryl hydrocarbon receptor nuclear translocator, ARNT) to become transcriptionally active. HIF1β contains a bHLH-PAS domain in the N terminus for DNA binding and dimerization with HIFα (Fig. 1). Unlike HIFα, which is sensitive to oxygen-dependent degradation, HIF1β is constitutively expressed in all mammalian cells, independently of the availability of oxygen⁷⁵. As bHLH–PAS domains are essential for binding to HIFα and HIF1β units, previous efforts to generate small-molecule HIF inhibitors were focused on preoccupying the PAS-B domain of HIFα. HIF1α and HIF1β have similar-sized PAS-B domains. However, the internal cavity of HIF1α is substantially smaller than that of HIF2α, facilitating the drug discovery efforts that have led to HIF2α inhibitors instead of HIF1α inhibitors⁵⁸.

Normoxic regulation of HIFs

Oxygen-dependent inactivation of HIF1α or HIF2α is a multistep process governed by several key enzymes and molecules through hydroxylation and ubiquitylation. First, FIH — an oxygen-dependent asparagine hydroxylase — inhibits the HIF system by hydroxylating Asn803 within the C-terminal transactivation domain (C-TAD) domain⁷⁶, preventing the binding of the coactivators, cyclic adenosine monophosphate response element binding protein (CREB) binding protein (CBP) and histone acetyltransferase p300 (ref. 77) (Fig. 1). Interestingly, global deletion of Fih1 in mice has little impact on the classical pathways mediated by HIF, such as erythropoiesis and angiogenesis, but significantly impacts the metabolic pathways⁷⁸. Therefore, inhibition of FIH alone is unlikely to affect erythropoiesis or angiogenesis in vivo.

Central control of the HIF pathway is linked to the enzymatic activity of HIF-PHDs. There are three HIF-PHD isoforms: PHD1 (also known as EGLN2), PHD2 (also known as EGLN1) and PHD3 (also known as EGLN3). PHD1 and PHD2 promote hydroxylation of Pro402 and Pro564 on human HIF1α, whereas PHD3 hydroxylates only Pro564 (ref. 79). The hydroxylation process is highly dependent on the availability of oxygen and α-ketoglutarate (also known as 2-oxoglutarate (2-OG)) as substrate and iron as catalyst^11,63 (Fig. 1). After hydroxylation, HIFα binds to pVHL, and is polyubiquitylated by the complex formed by elongin B, elongin C, cullin 2 (CUL2), RING box protein 1 (RBX1) and an E3 ubiquitin ligase, leading to subsequent proteasomal degradation^80,81 (Fig. 1).

Human PHD1 and PHD2 are more than 400 amino acids (407 and 426, respectively) in length. By contrast, PHD3 is relatively short, with only 239 amino acids. All three HIF-PHDs share a conserved hydroxylase domain at the C terminus, whereas PHD3 has a short N terminus^23,82. HIF-PHDs have different specificity and expression patterns. Expression of human PHD1 is more abundant in the testis, whereas human PHD2 has higher expression in skeletal muscle and neutrophils, and human PHD3 has higher expression in heart muscle, skin and plasmacytoid dendritic cells (The Human protein Atlas).

HIF-dependent gene regulation

Once stabilized, HIFα dimerizes with HIF1β and subsequently translocates into the nucleus to modulate gene transcription. The bHLH-PAS domains on HIFα and HIF1β are essential for dimerization and DNA binding⁶⁷. At the same time, recruitment of cofactors p300 and CBP is crucial for transactivation: gene expression triggered by an intermediate transactivator protein⁸³ (Fig. 1). Once in the nucleus, the HIFα–HIF1β–p300–CBP complex binds to hypoxia-responsive elements (HREs) located in the promoter regions of HIF target genes with a signature sequence of 5′-(A/G)CGTG-3′ and activates transcription^63,84. HIF regulates the transcription of many target genes. The tissue, cell and physiological specificity of this regulation highly depends on which HIFα isoforms are dominant in a specific tissue or disease setting. A discussion of HIF1α- and HIF2α-mediated target gene activation is included in Box 1.

Pharmacological HIF activators

HIF stabilization under normoxic conditions can be achieved predominantly through two pharmacological strategies: iron chelators or 2-OG analogues^11,85. The first strategy targets iron’s crucial availability as the catalyst of the hydroxylation reaction. The second approach prevents 2-OG binding as a key substrate for HIF-PHDs and inhibits HIF-PHD-mediated proline hydroxylation. Several small molecules were studied as HIF activators during the discovery of the conserved HIF–VHL–PHD pathway in Caenorhabditis elegans in 2001, including iron chelator 2′2′-dipyridyl, 2-OG analogue N-oxalyl-2S-alanine and dimethyloxalylglycine (DMOG)²³. As iron chelators are relatively nonspecific in their function as HIF inhibitors, this pharmacological strategy was not attempted by clinical studies. By contrast, all modern HIF activators currently undergoing testing in clinical trials have been developed based on their structural analogy to 2-OG with modifications to enhance efficacy, optimize pharmacokinetic (PK) properties and reduce toxicity.

HIF-PHD inhibition

Inhibition of HIF-PHDs activates HIF by preventing proline hydroxylation and subsequent proteasomal degradation. In particular, HIF-PHDs hydroxylate Pro402 at the N-terminal oxygen-dependent degradation domain (NODD) and Pro564 on the C-terminal ODD (CODD) in human HIF1α⁸⁶ (Fig. 1). A recent study resolved the crystal structure of several clinically available HIF-PHDis in complex with PHD2 to investigate the mechanism of inhibition⁸⁷. These HIF-PHDis interact with the C5-carboxyl-binding pockets in 2-OG by binding to the active site metal of HIF-PHDs⁸⁷ (Fig. 2a). Several layers of specificity need to be considered in the design of HIF activators for clinical use. First, HIF-PHDs belong to the family of 2-OG oxygenases, which has more than 60 members and is essential in governing multiple biological processes, including protein biosynthesis and fatty acid metabolism⁸⁸. Thus, the potential impact of HIF-PHDis on other 2-OG oxygenases had to be considered throughout their development process. As currently available HIF-PHDis activate both HIF1α and HIF2α⁷⁹, selectivity between HIF1α and HIF2α is another consideration owing to the sometimes overlapping but considerably distinct HIF target genes. Finally, inhibitors with superior binding affinity to proline residues at NODD or CODD might facilitate the development of PHD1-, PHD2- or PHD3-specific inhibitors.

Fig. 2|. Crystal structures of PHD2 and HIF-PHDi complexes and HIF2α–PT2385–PT2977 complexes.

a, Left: overall view of 2-oxoglutarate (2-OG) C5-carboxyl-binding pockets for prolyl hydroxylase inhibitors (PHDis) within the PHD2 crystal structure. Two enlarged views display electron density maps of the PHD2–vadadustat (Protein Data Bank (PDB): 5OX6; resolution: 1.99 Å)⁸⁷ and PHD2–molidustat (PDB: 6ZBO; resolution: 1.79 Å)²⁹⁰ complexes with key residues shown. The 2-OG dioxygenase domain is depicted in magenta, the β2–β3 finger loop in light yellow and key residues in red. Right: chemical structures of selected PHDis approved for clinical use or late-stage development are ordered by ascending half-maximal inhibitory concentration (IC₅₀) values for PHD2 (ref. 87). b, Binding position for the ‘PT’ series compounds within the entire hypoxia-inducible factor 2α (HIF2α)–HIF1β crystal structure (PDB: 6E3S; resolution: 3.00 Å)²⁹¹ on the left (HIF2α, pale cyan; HIF1β, pink), with a close-up view showing the location of PT2385 (PDB: 5TBM; resolution: 1.85 Å)²⁹² and PT2977 (PDB: 7W80; resolution: 2.75 Å)¹⁴³ inside the HIF2α Per–Arnt–Sim (PAS)-B domain (light cyan), along with the surrounding residues (cyan) in the pocket. bHLH, basic helix–loop–helix.

Five HIF-PHDis have entered phase III clinical trials: roxadustat (AstraZeneca), vadadustat (Akebia Therapeutics), daprodustat (GlaxoSmithKline), molidustat (Bayer HealthCare Pharmaceuticals) and desidustat (Zydus Life Sciences). The PK and pharmacodynamic (PD) profiles of clinically advanced oral HIF-PHDis have been well characterized in humans (Table 1), and the pharmacological characteristics of each HIF-PHDi often determine the dosing strategies.

Table 1|.

Pharmacologic profiles of HIF-PHDis in phase III development

Inhibitor	Pharmacodynamics		Pharmacokinetics
	Relative activitya	IC50 for PHD2 (μM)b	Effective daily oral doses	Dosing schedule	AUC, mean (μg h ml−1)	tmax (h)	t½ (h)	Plasma EPO (IU l−1)	CL/F, mean (l h−1)	CLR, mean (l h−1)	Metabolism
Daprodustat (GSK1278863)87,227	PHD3 > PHD1 > PHD2	0.067	1–4mg (NDD-CKD), 4–12 mg (DD-CKD)	QD	3.55−5.20	1.5	1 (healthy) −7 (CKD)	24.7 (5 mg, DD-CKD), 34.4 (5 mg, NDD-CKD) and 82.4 (Japanese, 10 mg DD-CKD)	21.70–31.4	NR	CYP2C8, major; CYP3A4, minor
Desidustat (ZYAN1)228,229	NR	11.2	100 mg (NDD-CKD), 100–150mg (DD-CKD)	TIW (QD)	3.7−116.2	1.25–3.00	6.9–13 (healthy)	6.6 (10 mg, healthy) and 79.9 (300 mg, healthy)	2.1−2.9	NR	NR
Molidustat (BAY-85–3934)87,230,231	PHD3 > PHD1 > PHD2	0.007	25–150 mg (>75 mg in DD-CKD)	QD	0.08−1.11	0.25–0.75	4−10 (healthy)	39.8 (50 mg, healthy)	45.1−71.9	0.53–1.7	UGT1A1/1A9
Roxadustat (FG-4592; ASP1517; AZD9941)87,232–239	PHD2 > PHD3 > PHD1	0.027	70mg (<60kg), 100mg (60–90kg), 150mg (>90kg)	TIW	88.7	1–2	12 (healthy) −15 (moderate hepatic impairment)	113 (1 mg kg−1, twice weekly NDD-CKD), 397 (2 mg kg−1, twice weekly NDD-CKD), 130 (1.3 mg kg−1, DD-CKD)	1.18	0.028	CYP2C8, UGT1A9
Vadadustat (AKB-6548; MT-6548)87,240	PHD3 > PHD1 > PHD2	0.029	150–600 mg	QD (TIW)	397	2.0−2.5	5.8 (healthy), 7.9 (NDD-CKD), and 9.1 (DD-CKD)	32 (500 mg, stage 3 and 4 CKD)	1.2	0.007	UGT1A1/1A9

Shown are the effective daily dose ranges and most commonly used dosing regimens reported in phase II/III studies in DD-CKD and NDD-CKD. AUC, area under the concentration−time curve; CKD, chronic kidney disease; CL/F, apparent oral clearance; CLR, renal clearance; CYP, cytochrome P450; DD-CKD, dialysis-dependent CKD; EPO, erythropoietin; IC₅₀, half-maximal inhibitory concentration; NDD-CKD, non-dialysis-dependent CKD; NR, not reported or not published; PHD, prolyl hydroxylase domain-containing protein; QD, once daily; TIW, thrice weekly; t_max, time to maximum concentration after oral dose; t_½, the time it takes a drug to clear from the highest concentration to half this level; UGT, uridine 5′-diphospho-glucuronosyltransferase.

^aActivity against the three HIF-PHDs obtained with mass spectometry-based assays⁸⁷.

^bValue determined with an antibody-based hydroxylation assay⁸⁷ or in vitro assay in HepG2 cells.

Clinical indications for HIF-PHDis

Since the discovery of small-molecule HIF activators that target HIF-PHDs, the clinical focus has been on renal anaemia, mainly owing to the heavily involved regulation of EPO by HIF stabilization⁵⁰. Many clinical trials, including phase III trials, have compared the safety and efficacy of HIF-PHDis with those of traditional therapies for renal anaemia. Several compounds have shown treatment effects comparable to those of erythropoiesis-stimulating agents (ESAs), which led to the approval of five HIF-PHDis in Japan. In addition, roxadustat has also received approval in other countries or regions, including the EU (European Union), China, South Korea and Chile. However, FDA has not supported the approval of roxadustat owing to unsatisfactory benefit–risk profile. As of 2 February 2023, daprodustat received FDA approval in the USA for anaemia associated with chronic kidney disease (CKD) in adult patients in dialysis. Other HIF-PHDis, such as vadadustat (currently approved in 35 countries, including Japan, Australia and many European countries), are still in the process of seeking FDA approval. Other clinical indications for HIF-PHDis, such as ARDS, are still in an earlier stage. In the next section, we discuss the clinical indications of HIF-PHDis for treating renal anaemia and ARDS in detail. The potential use of HIF-PHDis in other disease conditions is summarized in Table 2 and Fig. 3.

Table 2|.

PHD inhibition in preclinical models in various diseases

Drug or intervention	Involved HIF isoforms	Mechanism and outcome	Refs.
Myocardial ischaemia–reperfusion injury
FG041	NR	Improved left ventricular function via dampening interstitial fibrosis after MI	241
Cobalt chloride	HIF1α	Reduction in infarct size through signalling mechanisms involving AP-1 and iNOS	242
GSK360A	HIF1α	Upregulation of HIF1α target gene Pdk1, promotion of a shift in cell metabolism to glycolysis and amelioration of myocardial injury via the decreased opening of the MPTP. GSK360A treatment also improves long-term ventricular function, remodelling and vascularity after MI	39,243
FG-2216	HIF1α, HIF2α	Improved cardiac function and attenuated of cardiac remodelling after MI	244,245
ICA	HIF1α, HIF2α	Improved cardiac function and reduction of infarct size after MI	246
FG-4497	HIF1α	Preconditioning of the recipient decreased the infiltration of macrophages and mildly improved long-term allograft survival	247
DMOG, siRNA-mediated repression of Phd2	HIF1α	Enhanced purinergic signalling pathways involving CD73 and ADORA2B adenosine receptors, activation of the iNOS-dependent pathway and attenuation of chemokine and ICAM1 expression after IRI	248,249,250,251,252
shRNA-mediated downregulation of Phd2	HIF1α	Improvement of fractional shortening and increase of microvessel density after experimental MI	253
Partial genetic downregulation (Phd2 hypomorphic) mice or cardiomyocyte-specific Phd2−/− mice	HIF1α, HIF2α	Smaller infarct size, less cardiomyocyte apoptosis and better preserved left ventricular function in an ex vivo Langendorff perfusion model and in vivo IRI and MI models	254,255,256
Phd1 −/−	HIF1α	Decreased infarct size and apoptotic cardiomyocytes, probably by enhancing HIF1α/β–catenin–eNOS–NF-κB and BCL-2 signalling pathway	257
Phd3 −/−	HIF1α	Attenuated cardiomyocyte apoptosis, increased angiogenesis and improved cardiac function after MI	258,259
Roxadustat	HIF1α	Reduced infarct size and suppressed plasma creatinine kinase activity in a murine IRI model by shifting metabolism from aerobic to anaerobic respiration	260
Liver ischaemia–reperfusion injury
EDHB	HIF1α	Attenuated mitochondrial permeability transition onset and liver IRI injury via HO-1. Enhanced liver regeneration after partial hepatectomy and portal vein ligation	261,262,263
Phd1−/−, shRNA-mediated downregulation of Phd1	HIF2α	Increased anaerobic glucose catabolism, reduced oxygen consumption and attenuated hepatocyte damage in liver IRI in mice	264
Double knockdown of Phd1 and Keap1 by shRNA	HIF2α	Attenuated hypoxia and oxidative stress-induced injury in the hepatocytes via decreased oxidative stress and apoptosis	265
Phd1 −/−	HIF2α	Enhanced liver regeneration by boosting hepatocyte proliferation in a Myc-dependent fashion after partial hepatectomy	266
Acute kidney injury
Cobalt chloride	HIF1α	Ameliorated tubulointerstitial damage, reduced macrophage infiltration and dampened ischaemic injury in an acute ischaemic tubulointerstitial injury model of rats	267
FG-4487	HIF1α, HIF2α	Ameliorated ischaemic acute renal failure	268
FG-2216	NR	Improved renal and cardiovascular outcomes and reduced obesity in a rat kidney disease model with metabolic syndrome	244
GSK1002083A	HIF2α	Reduced renal injury 3 days after IRI, improved injury scores, decreased Kim1 levels, CD45-positive leukocytes and Vcam1 and Icam1 transcripts	42
l-mimosine and DMOG	NR	Reduced renal injury after renal IRI with attenuated tubular epithelial cell apoptosis and less macrophage infiltration	43
Nanoparticle targeted delivery of Phd2 siRNA	NR	Reduction in renal IRI with lower serum creatinine and BUN levels	269
TRC160334	HIF1α, HIF2α	Reduced renal injury with improved urine output up to 24 h in a rat model of IRI-induced AKI	270
Roxadustat	HIF1α	Attenuated renal apoptosis and pro-inflammatory cytokines, improved renal function in cisplatin-induced kidney injury in mice	271
Enalodustat	HIF1α	Upregulated glycogen synthesis in tubular cells	272
FG-4497 (to transplant donor rats)	HIF1α, HIF2α	Inhibition of apoptosis in the renal tubular epithelium of transplants, improved short- and long-term outcomes	273
Inflammatory bowel disease
DMOG	HIF1α	Reduced colitis via attenuated apoptosis of intestinal epithelial cells and ameliorated intestinal fibrosis by suppressing TGFβ1-dependent ERK activation in fibroblasts in DSS colitis model	274,275,276,277
DMOG and cyclosporine	NR	Preserved barrier function in DSS murine colitis model	278
AKB-4924 or FG-4497	HIF1α	Reduced colitis due to HIF1A-mediated barrier protective function and increased wound healing in TNBS murine colitis model	35, 279,280,281
Phd1 −/−	NR	Decreased epithelial cell apoptosis and consequent intestinal epithelial barrier function enhancement in the DSS murine colitis model	274
Endothelial and haematopoietic cell-specific Phd1 deletion	NR	Ablation of Phd1 in haematopoietic cells but not in endothelial cells inhibits colitis via promoted M2-macrophage polarization in DSS murine colitis model	282
Infections with pathogens
AKB-4924	HIF1α	Reduced infection in uroepithelial cells and bladders via enhanced production of NO and antimicrobial peptides cathelicidin and β-defensin-2 in UTI murine model. Reduced Staphylococcus aureus proliferation and dampened lesion formation in a mouse skin abscess model	47,283
Mimosine	HIF1α	Reduced lesion size in a murine model of S. aureus skin infection via boosted phagocytes to kill the leading pathogen	284
DMOG	HIF1α	Attenuated breakdown of epithelial barrier function in Clostridioides difficile-induced intestinal injury	285
DMOG	HIF2α	Reduced mortality in pneumonia infection murine model	286
DMOG	HIF1α	Attenuated peri-apical inflammation and tissue destruction via downregulation of pro-inflammatory cytokines in murine periodontitis model	287
Haemorrhagic shock
Phd2 liver-specific knockout	NR	Ameliorated lactic acidosis and improved survival by activating the Cori cycle in mice administered with a lethal dose of lactate	49
EDHB	HIF1α	Attenuated microvascular inflammation via HO-1 upregulation and reduced leukocyte adherence during rat haemorrhagic shock and resuscitation via an iNOS-dependent pathway	288,289
GSK360A	NR	Decreased blood lactate levels and enhanced survival rates after a lethal dose of lipopolysaccharide injection in mice	49

AKI, acute kidney injury; AP-1, activator protein 1; BUN, blood urea nitrogen; DMOG, dimethyloxalylglycine; DSS, dextran sulfate sodium; EDHB, ethyl 3,4-dihydroxybenzoate; eNOS, endothelial nitric oxide synthase; ERK, extracellular signal-regulated kinase; HIF, hypoxia-inducible factor; HO-1, haem oxygenase-1; ICA, 2-(1-chloro-4-hydroxyisoquinoline-3-carboxamido) acetate; ICAM1, intercellular adhesion molecule 1; iNOS, inducible nitric oxide synthase; IRI, ischaemia–reperfusion injury; KEAP1, Kelch-like ECH-associated protein 1; KIM1, kidney injury molecule 1; MI, myocardial infarction; MPTP, mitochondrial permeability transition pore; NF-κB, nuclear factor κB; NGAL, neutrophil gelatinase-associated lipocalin; NO, nitric oxide; NR, not reported or not published; PHD, prolyl hydroxylase domain-containing protein; PDK1, 3-phosphoinositide-dependent protein kinase 1; shRNA, short hairpin RNA; siRNA, small interfering RNA; TGFβ1, transforming growth factor β1; TNBS, 2,4,6-trinitrobenzene sulfonic acid; UTI, urinary tract infection; VCAM1, vascular cell adhesion molecule 1.

Fig. 3|. Potential therapeutic applications of HIF-PHDis in various diseases.

In normoxia (central panel) hypoxia-inducible factor prolyl hydroxylase inhibitors (HIF-PHDis) enhance the stabilization of HIFα. This is achieved by inhibiting proline hydroxylation of HIFα by HIF-PHDs, thereby preventing the recognition of von Hippel–Lindau protein (pVHL) and subsequent proteasomal degradation of HIFα. Once stabilized, HIFα forms a complex with HIF1β, histone acetyltransferase p300 and cyclic adenosine monophosphate response element binding protein (CREB) binding protein (CBP), translocates to the nucleus, and binds to hypoxia-responsive elements (HREs) in the promoter regions of HIF targets to activate their transcription²⁸. Pharmacologically enhanced HIF stabilization by HIF-PHDis has shown benefits in diseases such as ischaemia–reperfusion injury (heart and liver), acute kidney injury, inflammatory bowel disease, central nervous system injury, renal anaemia and acute respiratory distress syndrome (ARDS)²²⁶. In both the kidneys and liver, HIF-PHDis stabilize HIF2α, thereby enhancing the production of endogenous erythropoietin (EPO)^94,98. This process is particularly beneficial in conditions such as renal anaemia, in which EPO production is typically insufficient⁹⁴. EPO promotes the survival and differentiation of bone marrow erythroid progenitors, specifically colony-forming unit erythroid (CFU-e) cells and proerythroblasts, resulting in an increase in red blood cell production⁸⁹. In the liver, the stabilization of HIF2α leads to increased production of transferrin (TF), the plasma protein that carries iron, and a decrease in hepcidin production, which serves as a negative regulator of ferroportin (FPN)^100,293. FPN is responsible for the export of iron from duodenal enterocytes²⁹³. Additionally, HIF-PHDis can improve iron absorption and metabolism by inducing divalent metal transporter 1 (DMT1), ferrireductase duodenal cytochrome B (DCYTB) and FPN1 in the enterocytes, further promoting iron homeostasis²⁹³. With increased iron availability and delivery by TF to marrow erythroblasts, there is an increase in both the size of red blood cells and their haemoglobin content. HIF-PHDis may alleviate ARDS by stabilizing HIFs. HIF1α stabilization in alveolar type II epithelial cells (ATII cells) during ARDS optimizes carbohydrate metabolism³¹. HIF1α also activates extracellular adenosine signalling — particularly through increasing extracellular adenosine levels via enhancement of CD73 and repression of equilibrative nucleoside transporter ENT1 or ENT2 and adenosine kinase^294–296. Subsequently, extracellular adenosine binds to HIF1α-dependent ADORA2B to reduce lung inflammation¹⁰⁹. Furthermore, HIF1α attenuates severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) replication through inhibition of angiotensin-converting enzyme 2 and transmembrane protease, serine 2, two receptors that are important to viral entry, and repression of viral RNA replication¹²¹. On the other hand, HIF2α maintains endothelial cell barrier integrity by inducing vascular endothelial protein tyrosine phosphatase (VE-PTP), and the resultant VE-cadherin dephosphorylation-mediated assembly of adherens junctions (AJs), thereby preventing vascular permeability and lung oedema¹¹³. HIFs also regulate immune cell functions, with HIF-mediated glycolysis being essential for macrophage-mediated pathogen clearance^117,118. HIF-dependent netrin-1 induction in myeloid cells inhibits natural killer cell (NK cell) invasion during endotoxin-induced lung injury¹¹⁹.

Renal anaemia.

Red blood cells (RBCs) are essential to carry oxygen in the blood to meet oxygen demand, and the production of RBCs depends on EPO binding to high-affinity receptors (EpoRs), which are primarily expressed on colony-forming unit erythroid cells and proerythroblasts in the bone marrow⁸⁹. EPO is produced mainly by the fetal liver during embryonic development⁹⁰. However, after birth, the production of EPO is switched from the liver to the kidneys in fibroblast-like interstitial cells in the cortex and outer medulla, especially during hypoxic stress^91–93. Therefore, CKD is commonly associated with anaemia resulting from insufficient EPO production.

Conventional ESAs, including recombinant human EPO and long-acting EPO analogues, such as darbepoetin alfa, are the most common treatment approaches for CKD-associated anaemia. Adverse effects associated with ESAs include increased risk of thrombotic events, higher incidence of cardiovascular events and the development of antibodies against EPO, which neutralize both ESAs and endogenous EPO and cause antibody-mediated pure red cell aplasia^94,95. In addition, recombinant EPO is given as an injectable drug, which can be associated with reduced patient compliance and discomfort caused by the injections⁹⁴. As a result, small molecules that could stimulate EPO production are of great interest in clinical development as a treatment for CKD-associated anaemia with the hope of reducing the risk of antibody development and the incidence of adverse cardiovascular events and having an orally available treatment option for CKD.

Molecular studies suggested that HIF2α, not HIF1α, is crucial for EPO production in the kidneys⁹⁶, and experimental studies showed increased EPO production with HIF activator treatment⁹⁷. Beyond the kidneys, HIF2α also contributes to EPO production in hepatocytes during hypoxia⁹⁸, and PHD3 is crucial for the selective activation of HIF2α in hepatocytes⁹⁹. In addition to EPO production, HIF-PHDis have the potential to improve iron absorption and metabolism by inhibiting hepcidin¹⁰⁰ and inducing divalent metal transporter 1 (DMT1)¹⁰¹ and ferroportin 1 (FPN1)¹⁰², thereby further promoting iron homeostasis as an additional contribution to an effective renal anaemia treatment (Fig. 3).

HIF-PHDis have been and are being assessed in clinical trials for various disease conditions (Table 3). Roxadustat was the first to complete phase III trials, when it showed non-inferior efficacy to epoetin alfa in patients with CKD undergoing dialysis⁵⁴ and increased mean haemoglobin level, a readout for RBC mass, compared with placebo control in patients with CKD not receiving dialysis⁵³. In the later study, on average, patients in this trial who received roxadustat showed an increase in haemoglobin of 1.9 ± 1.2 g dl⁻¹ compared with 0.4 ± 0.8 g dl⁻¹ decline in the placebo group⁵³. Similarly, phase III clinical trials using vadadustat indicated non-inferiority to darbepoetin alfa regarding its effect on erythropoeisis⁵². Furthermore, recent studies also demonstrated that daprodustat was non-inferior to darbepoetin alfa regarding maintenance of haemoglobin levels and cardiovascular outcomes in patients with CKD not receiving dialysis¹⁰³. Compared with injectable ESA, daprodustat was non-inferior in patients with CKD undergoing dialysis¹⁰⁴. Currently, HIF-PHDis are being studied in clinical trials as a treatment for chemotherapy-related anaemia and in paediatric patients (NCT05301517, NCT04925011). In addition, post-marketing surveillance of roxadustat is being carried out in Japan (NCT04408820) to track long-term outcomes.

Table 3|.

Ongoing clinical trials with HIF-PHDis

Drug	NCT number	Conditions	Interventions	Phase	Study design	Primary outcome measures
Anaemia
Daprodustat	NCT05682326	Anaemia	Daprodustat	III	Interventional, single group, OL	Number of participants with AEs and SAEs or with AESIs (56 weeks); number of participants with AEs leading to study intervention discontinuation (52 weeks)
Desidustat	NCT05515367	CKD; CKD-related anaemia	Desidustat	IV	Interventional, single group, OL	Proportion of patients with treatment-emergent AEs (52 weeks) or SAEs (52 weeks)
Molidustat	NCT04899661	Renal anaemia	Molidustat	NA	Observational model: cohort prospective	Incidence of safety events (24 months)
Roxadustat	NCT04502537	CKD-related anaemia	Roxadustat EPO	NA	Observational model: cohort prospective	Mean value of Hb levels over time (52 weeks) Achievement rate for target Hb level (52 weeks)
	NCT04408820	Renal anaemia	Roxadustat	NA	Observational model: cohort prospective	Proportion of participants with ADR, serious ADR, thromboembolism, hypertension, hepatic function disorder, malignant tumours, retinal haemorrhage, seizures, serious infection, central hypothyroidism, myopathy events or renal function disorder (140 weeks); proportion of participants with ADR within 4 weeks after switching to roxadustat; change from baseline in Hb levels Mean value of Hb levels over time; achievement rate for target Hb level; mean Hb level at 4 weeks after switching to roxadustat
	NCT05301517	Chemotherapy-associated anaemia	SEPO, roxadustat	III	R, PA, OL, interventional	Change in Hb level (weeks 9–13)
	NCT04925011	CKD-related anaemia	Roxadustat	III	Interventional, single group, OL	Proportion of patients with mean Hb ≥ 11.0 g dl−1 (weeks 16–24)
	NCT03303066	Primary MDS classified as very low, low or intermediate risk with <5% blasts	Roxadustat; Placebo	II III	Interventional, R, PA, quadruple masking	Proportion of patients with a Hb response to roxadustat without transfusion (26 weeks)
	NCT03263091	Primary MDS (very low, low or intermediate IPSS-R with <5% blasts)	Roxadustat Placebo	III	Interventional, R, PA, quadruple masking	Percentage of participants who achieve transfusion independence ≥56 consecutive days in the first 28 weeks of treatment
	NCT04925661	CKD Renal anaemia	HEC53856, roxadustat Placebo	I	Interventional, R, PA, quadruple masking	Incidence of AEs (12 weeks)
Vadadustat	NCT04707768	CKD-related anaemia	Vadadustat Mircera	III	Interventional, R, PA, OL	Mean change in Hb between baseline (average pretreatment Hb) and the primary evaluation period (average Hb weeks 20–26, inclusive) Number of participants with treatment-emergent AEs and treatment-emergent SAEs (week 56)
Heart failure
Roxadustat	NCT05691257	Heart failure with CKD and anaemia	Roxadustat EPO and/or iron agents	IV	Interventional, NR, PA, OL	Change in Hb from baseline (8 weeks) Change in NT-proBNP, left ventricular systolic or diastolic function, left atrial volume (24 weeks)
	NCT05053893	Cardio-renal syndrome	Roxadustat; sacubitril valsartan sodium tablets; EPO; ACEI/ARB	NA	Interventional, R, PA, OL	Changes in Hb level and EF before and after treatment (90 days) Incidence of acute HF, acute MI, severe hyperkalaemia and severe anaemia (90 days)
Myocardial infarction
Roxadustat	NCT04803864	ST elevation MI	Roxadustat	II	Interventional, R, PA, single masking	Infarct size (30 days), MACE (1 year), left ventricular function and cardiac enzyme (0–3 days)
Acute kidney injury
Roxadustat	NCT05010460	Coronary artery bypass	Roxadustat Placebo	II	Interventional, R, PA, quadruple masking	Acute kidney injury (0–48 h after surgery)

ACEI, angiotensin-converting enzyme inhibitor; ADR, adverse drug reaction; AE, adverse event; AESI, adverse event of special interest; ARB, angiotensin II receptor blocker; CKD, chronic kidney disease; EF, ejection fraction; EPO, erythropoietin; Hb, haemoglobin; HF, heart failure; IPSS-R, International prognostic scoring system – revised; MACE, major adverse cardiovascular event; MDS, myelodysplastic syndrome; MI, myocardial infarction; NA, not applicable; NR, non-randomized; NT-proBNP, N-terminal pro-brain natriuretic peptide; OL, open label; PA, parallel assignment; R, randomized; SAE, serious adverse event.

Acute respiratory distress syndrome.

Besides renal anaemia, ARDS is another clinical indication with ongoing clinical safety and efficacy trials using HIF activators for treatment. Consistent with previous studies in ischaemic–reperfusion injury^16,105, HIFs are also stabilized during many inflammatory conditions^106,107, including ARDS^31,108,109. HIF stabilization during ARDS is an endogenous protective pathway that can be further exploited pharmacologically for early ARDS treatment or prevention¹¹⁰. For example, during ARDS, HIF1α is stabilized in alveolar epithelial cells and functions to optimize their carbohydrate metabolism³¹ and extracellular adenosine signalling¹⁰⁹. Additionally, HIF1α deficiency resulted in increased influenza A virus replication in A549 cells via reduced transcription of glycolysis genes and subsequent promotion of autophagy¹¹¹. In addition, isoform 3 of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (PFKFB3) is a novel target gene of HIF1α, crucial for metabolic adaptation in various experimental models of ARDS or samples from patients with ARDS¹¹⁰. Finally, during the repair stage after acute lung injury, HIF1α promotes the proliferation of alveolar type II epithelial cells (ATII cells) via induction of stromal cell-derived factor 1 and C-X-C chemokine receptor type 4 (ref. 112).

HIF2α also contributes to lung protection during ARDS, for example, by maintaining the barrier integrity of endothelial cells. Mechanistically, HIF2α induces vascular endothelial protein tyrosine phosphatase and enhances endothelial barrier integrity¹¹³ (Fig. 3). Mice with endothelium-specific deletion of Hif2a show increased vascular permeability and lung oedema during endotoxin-induced lung injury¹¹³. Moreover, endothelial cell-specific deletion of Hif2a in mice resulted in increased tracheal endothelial cell death and impaired barrier function through angiopoietin 1–TIE2 and Notch signalling¹¹⁴.

HIF is also important for regulating immune cell functions, mainly in myeloid cells^115,116. For example, HIF-mediated glycolysis is crucial for the polarization of macrophages towards M1 or classically activated phenotype macrophages¹¹⁷, which are essential for clearance of certain pathogens¹¹⁸ (Fig. 3). Similarly, hypoxia exposure of tissue-resident alveolar macrophages significantly increases glycolysis and enhances cell survival during lipopolysaccharide (LPS) treatment in vitro, while improving outcomes during murine influenza infection⁴⁴. Besides controlling metabolism, HIF1α is also essential in modulating neuronal guidance cues in macrophages via induction of netrin 1 to inhibit natural killer cells, conferring lung protection during LPS-induced acute lung injury¹¹⁹.

In ventilator-induced lung injury mouse models, treatment with the HIF activator DMOG was associated with attenuated albumin leakage into the lungs and improved survival time³¹. Furthermore, pretreatment of rats with DMOG reduces lung pathology and inflammation during acute lung injury induced by a combination of hypoxia and LPS¹²⁰. During influenza infection, treatment with roxadustat on the day of infection improved mice recovery and survival, accompanied by reduced inflammatory cytokine production⁴⁴. Similarly, roxadustat treatment reduces angiotensin-converting enzyme 2 (ACE2) expression and restricts replication of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in alveolar epithelial cells in vitro¹²¹ (Fig. 3). Furthermore, in hamster models of COVID-19, treatment with roxadustat 1 day after infection significantly reduced the clinical score and reduced respiratory viral load¹²². These studies led to the assessment of HIF-PHDi vadadustat as a preventive or treatment for ARDS associated with COVID-19 (NCT04478071) in a phase II clinical trial in which patients with COVID-19 with concomitant hypoxia (oxygen saturation less than 94%) were randomly assigned to receive oral vadadustat or placebo for up to 14 days. The trial has been completed, and we anticipate the findings to be informative about the clinical safety of HIF-PHDis in critically ill patients and the potential benefit in ARDS prevention or early treatment. Further clinical studies are required to assess HIF-PHDis in patients with other forms of pneumonia or different ARDS causes (such as ARDS caused by aspiration pneumonia or post-operative ARDS).

Potential side effects of HIF-PHDis.

HIF-PHDis have shown an excellent short-term safety record in large phase III clinical trials and were not associated with serious side effects^123,124. These findings encourage further clinical studies of HIF-PHDis for acute treatment indications, such as prevention of peri-operative organ injuries, including acute kidney injury, myocardial injury or ARDS^125–127. There was an isolated event of acute lethal fulminant hepatitis by the HIF-PHDi FG-2216 in 2007 during a phase II clinical trial, in which a patient developed fatal hepatic necrosis temporally related to the treatment, and other patients developed abnormal liver enzyme tests¹²⁸. Therefore, the FDA suspended this clinical trial, and no other clinical studies with FG-2216 have been carried out¹²⁸. An extensive investigation of this incident suggests that the liver injury is likely directly associated with the toxicity of FG-2216 instead of being a consequence of HIF activation. In recent phase III clinical trials with newer generations of HIF-PHDis in patients with CKD, liver injury was not seen^{36,52,54,104,129}. Although the short-term use of HIF-PHDis did not reveal any serious side effects — other than elevation of haemoglobin in patients with renal anaemia — further studies have to confirm longer-term safety and optimal dosing strategy of HIF-PHDis in different patient populations, such as critically ill patients.

Long-term use (taken regularly for more than 3 months) of HIF-PHDis could be associated with unwanted side effects, including on-target effects caused by chronic HIF activation and off-target effects associated with the specificity of HIF-PHDis for individual hydroxylases. First and foremost, as HIF activation has been associated with cancer progression, especially of clear cell renal cell carcinoma (ccRCC), long-term HIF activation might lead to cancer development or progression, for example, via the stimulation of VEGF¹³⁰. However, evidence from patients with gain-of-function mutations of HIF genes suggested no increased risk of developing cancer, although only a few families with fewer than 100 patients were studied^131–133. Moreover, in rodent models, oral treatment with the HIF-PHDi roxadustat over 104 weeks did not affect survival or development of neoplastic lesions at up to 60 mg kg⁻¹ in mice and up to 10 mg kg⁻¹ in rats¹³⁴. Thus, HIF activation alone might not be sufficient to lead directly to cancer development without a secondary mutation of key oncogenic pathways. In phase III randomized trials that tracked patients for up to a year, cancer incidence was similar between HIF-PHDi and control groups^52,104. However, the long-term outcomes of patients receiving HIF-PHDis still need further investigation.

Another potential on-target side effect of HIF-PHDis involves excessive erythropoiesis and thromboembolic complications, similar to those associated with ESA use, even when targeting a normal haemoglobin level as a therapeutic goal. Indeed, treatment with roxadustat was associated with a higher incidence of thromboembolic events than treatment with darbepoetin alfa in patients with CKD undergoing haemodialysis⁹⁴. However, such rare events are likely to be clinically manageable, including monitoring and consideration of anticoagulatory approaches.

Increased risk of major adverse cardiovascular events (MACEs) is also a potential concern associated with the long-term use of HIF-PHDis because patients with CKD have an increased susceptibility to major cardiac events. Analysis from several pooled phase III studies of roxadustat in patients with CKD undergoing haemodialysis showed no significant differences in the risk for MACE compared with epoetin alfa, including a decreased risk of MACE in patients with heart failure requiring hospitalization¹³⁵. Vadadustat was not associated with increased MACE risk in dialysis-dependent patients⁵¹. By contrast, the most recent phase III trial of vadadustat in patients with CKD non-dependent on dialysis identified an increased risk of MACE, particularly among study patients outside the USA⁵². Additional study is needed to determine whether the increased risk of MACE is due to treatment with the HIF-PHDis or might be caused by differences in haemoglobin targets⁵². HIF-PHDis have potential off-target effects by inhibition of other 2-OG-dependent hydroxylases. Although most currently available HIF-PHDis showed reasonable specificity, several small-molecule HIF-PHDis inhibit collagen prolyl-4-hydroxylases (CP4Hs) and, in turn, suppress C1q secretion by macrophages¹³⁶. Chronic suppression of C1q might increase the risk of infection and development of systemic lupus erythematosus¹³⁷. As a result, further studies are needed to investigate the impact of HIF-PHDis on other off-target enzymes in vivo. However, it is reassuring that none of these side effects caused a safety signal in the large phase IIIa clinical trials of renal anaemia with HIF-PHDis^51,54,104.

HIF inhibitors

HIF stabilization can be detrimental in many disease conditions, such as certain cancers and pulmonary hypertension. Therefore, pharmacological strategies have been developed to inhibit HIFs and have undergone advanced clinical testing. Early pharmacological interventions focus on inhibition of HIF expression or promotion of HIF degradation. Recent small-molecule inhibitors, such as PT2977, inhibit the formation of the transcriptionally active heterodimer of HIF2α and HIF1β. In this section, we discuss HIF inhibitors based on their targeting approach for different HIFα isoforms, with a first section focusing on nonspecific targeting of both HIF1α and HIF2α, followed by pharmacological interventions that target specifically HIF1α or HIF2α.

Nonspecific HIF inhibitors

Many nonspecific HIF inhibitors were initially discovered as small molecules that inhibited tumour growth and it was found subsequently that their tumour-growth-inhibiting properties were mediated by HIF inhibition. For example, 2-methoxyoestradiol is an endogenous oestrogen metabolite that inhibits solid tumour growth in preclinical and clinical studies¹³⁸. 2-Methoxyoestradiol inhibits both HIF1α and HIF2α at the mRNA and protein levels, thereby implicating the HIF pathway as a key mechanism of tumour growth inhibition¹³⁹. Digoxin, a commonly used cardiac glycoside for cardiovascular disorders¹⁴⁰, emerged from an in vitro screen as a potent inhibitor of HIF1α and HIF2α that induces cancer cell growth arrest¹⁴¹. Daily treatment with digoxin (2 mg kg⁻¹) limits growth of P493-Myc cell line-derived tumour xenografts in vivo in murine models¹⁴¹. Although the mechanism of HIF inhibition by digoxin still remains somewhat unclear, studies suggest that it might involve inhibition of the translation of HIF1A mRNA to protein¹⁴¹. Of note, the clinically used dose of digoxin for the treatment of cardiac indications is much lower (in the 5–15 μg kg⁻¹ level) than what is needed for HIF inhibition (in the 1–10 mg kg⁻¹ level). With the narrow therapeutic window of digoxin¹⁴², achieving HIF inhibition would likely be associated with a high toxicity rate.

Acriflavine represents another nonspecific HIF inhibitor that destabilizes the HIFα–HIF1β heterodimer by directly interacting with amino acid residues Arg266 and Val305 of HIF1β⁶⁵. Acriflavine also binds distinctly to residues Gln306, Trp318, Leu344, Ser345 and Glu348 of the HIF2α PAS-B domain⁶⁵. By contrast, belzutifan exhibit a different binding mode, primarily associating with residues Met252, Phe254 and His293 in the HIF2α PAS domain¹⁴³. Interestingly, the proflavine binding pocket uses nearly identical amino acids within HIF2α–HIF1β and HIF1α–HIF1β heterodimers, potentially explaining the nonspecific HIF inhibition. Unfortunately, acriflavine failed to advance in clinical development owing to its potential side effects and short half-life. Thus, based on its similarity in gene target induction to acriflavine, 32–134D was recently identified as a nonspecific HIF inhibitor with a chemical structure unrelated to that of acriflavine⁵⁹. 32–134D is well tolerated and promotes tumour eradication in HCC Hepa1–6 cell line-derived tumour murine models⁵⁹.

Most nonspecific HIF activators failed in a clinical setting owing to concerns about their toxicity²⁸. By contrast, more recently generated nonspecific HIF inhibitors, such as 32–134D⁵⁹, could be promising candidates for clinical trials, as they are likely better tolerated and have acceptable toxicity profiles in humans.

Strategies to inhibit HIF1α

Since the discovery of HIF1α in the late 1990s, there has been an ongoing effort to screen small-molecule inhibitors using library screens in cell lines that report HIF binding activity to HREs in vitro. Topotecan (NSC-609699), which is closely related to topoisomerase (Topo)-I inhibitors, significantly reduces hypoxia-induced VEGF expression while inhibiting HIF1α protein accumulation and preventing binding to DNA in U251 cells¹⁴⁴. Mechanistically, topotecan inhibits translation of HIF1α in a Topo-I-dependent manner, instead of promoting protein degradation, heterodimer formation or reduction of mRNA expression¹⁴⁵. PX-478 (S-2-amino-3-[4′-N,N,-bis(chloroethyl)amino]phenyl propionic acid N-oxide dihydrochloride) is another chemical agent that inhibits HIF1α and decreases tumour growth in murine models^146,147. PX-478 inhibits HIF1α in a pVHL-independent manner, mainly via reducing HIF1A mRNA levels and inhibiting its translation¹⁴⁸. Along with indirect inhibitors, several direct inhibitors of HIF1α have been identified and investigated by preclinical studies. For example, cyclo-CLLFVY, a cyclic peptide inhibitor of HIF1α, prevents the dimerization of HIF1α with HIF1β by occupying the PAS-B domain while not impacting on HIF2α in cell lines¹⁴⁹. Furthermore, considerable effort has been made to identify inhibitors that disrupt the binding of HIF1α with p300, as summarized in a recent review article¹⁵⁰. Further investigation is needed to address the safety and efficacy of these small molecules in humans. Besides small-molecule inhibitors, EZN-2968/RO7070179 is a locked nucleic acid antisense oligonucleotide (ASO) that binds to HIF1A mRNA to inhibit protein expression and downregulate HIF target genes in vitro with no impact on HIF2α¹⁵¹. Furthermore, EZN-2968 treatment reduces HIF1α at both mRNA and protein levels, accompanied by a decrease in Vegf mRNA level in vivo in a dose-dependent manner in the liver up to 5 days after one single intraperitoneal injection¹⁵¹.

Strategies to inhibit HIF2α

Like HIF1α, early efforts in small-molecule library screens had identified several compounds that inhibit HIF2α in cell lines using HRE reporters. For example, C76 (methyl 3-{2-[cyano(methylsulfonyl)methylene] hydrazino}thiophene-2-carboxylate) emerged as a small molecule that inhibits HIF2α independently of mechanistic target of rapamycin (mTOR) signalling¹⁵². Mechanistically, C76 dampens HIF2α translation by facilitating the binding of iron regulatory protein 1 to the iron-responsive element of the EPAS1 mRNA¹⁵². However, owing to the wide range of genes controlled by iron regulatory proteins¹⁵³, the specificity of C76 and its potential off-target effects are still unclear. Thus far, C76 has shown benefit in preclinical models of pulmonary hypertension in mice and rats with decreased right ventricular systolic pressure and hypertrophy, as well as reduced fibrotic changes and smooth muscle remodelling in the lung¹⁵⁴. However, it has not yet advanced to clinical trials.

HIF2α was considered undruggable until the discovery of a series of compounds capable of occupying the PAS-B domain of HIF2α, thereby blocking the interaction between HIF2α and HIF1β (Fig. 2b). A group of investigators initially discovered this series of compounds at Peloton Therapeutics and hence these compounds are referred to as ‘PT’ compounds. Importantly, this discovery was made possible by seminal work that revealed the internal cavity within the PAS-B domain of HIF2α, identified binding of small molecules from NMR-based screening and confirmed disruption of PAS domain interactions in vitro¹⁵⁵. Subsequent studies enabled in vitro high-throughput screens to discover and optimize compounds to efficiently disrupt HIF2α function in cells¹⁵⁶. PT2385 is the first-in-class HIF2α inhibitor as a clinical candidate for ccRCC¹⁵⁷. The identification of PT2385 involved a multistep optimization process that originated from a ‘parent molecule’ with structures predicted in the literature to bind to the PAS-B pocket of HIF2α. Using an in vitro drug screen approach, several lead compounds were selected based on half-maximal effective concentration (EC₅₀) for in vivo PK/PD profiling studies in rats and larger animals¹⁵⁷ (Table 4). PT2385 was determined to have an optimal in vivo PK/PD profile and potent antitumour capabilities in preclinical xenograft models and thus advanced to clinical trials¹⁵⁷ (Table 4). Finally, retrospective crystallography analysis indicated that PT2385 binds to two key amino acids, Tyr281 and His293, on HIF2α to block its interaction with HIF1β. Phase I clinical studies of PT2385 demonstrated good efficacy and safety profile^158,159. In patients with ccRCC who had received previous therapy, administration of PT2385 800 mg twice daily had a median time of maximal concentration (t_max) of 2 h and terminal half-life (t_1/2) of 17 h with a mean blood concentration of 3.1 μg ml⁻¹ (ref. 159). However, the PKs of PT2385 show high variability between patients, dampening its potential to move forwards in the clinic^158,159 (Table 4). As a result, a successor molecule, PT2977, was developed with improvements in its PK profile and higher efficacy in blocking HIF2α activity in vitro and in preclinical animal models via reducing the glucuronidation by switching the geminal difluoro group of PT2385 to a vicinal difluoro group¹⁶⁰ (Table 4). Oral treatment with 120 mg of belzutifan once daily significantly reduced plasma EPO levels, as a direct indication of on-target HIF2α inhibition, with a t_max of 1–2 h and t_1/2 of 11.2–21.5 h in patients with ccRCC¹⁶¹. In 2021, the FDA approved PT2977 (belzutifan, MK-6482) for the treatment of ccRCC¹⁶² in patients with VHL disease, marking a phenomenal accomplishment based on more than 20 years of research on HIF2α biology and therapeutic targeting. Besides PT2385 and belzutizan, several other small-molecule HIF2α inhibitors from Nikang (NKT2152), Novartis (DFF332) and Arcus (AB521) are in various stages of clinical development (Table 5).

Table 4|.

Pharmacological profiles of HIF inhibitors in phase II development in ccRCC

Property	PT2385 (refs. 157,159)	PT2977/belzutifan/MK-6482 (refs. 160,161)
Pharmacodynamics
Change in EPO (%)	Nearly 80 at day 8	Nearly 80 at day 8
Pharmacokinetics (in vitro)
Luciferase EC50 (nM)	27	11
VEGFA EC50 (nM)	46	17
Free fraction adjusted EC50 (ng ml−1)	95	13
Free fraction adjusted EC85 (ng ml−1)	540	75
Pharmacokinetics (in patients)
Dosing schedule	800 mg, BID, PO	120 mg, QD, PO
tmax (h)	2.3 (1.7)a, 1.7 (0.7)b	1.60 (0.5–23.7)c, 1.50 (0.5–6.1)d
t½, (h)	16.1 (9.6)a, 19.0 (11.7)b	16.84 (8.6)c, 15.95 (7.2)d
Cmax (μg ml−1)	1.2 (0.7)a, 2.2 (1.6)b	1.35 (0.5)c, 1.8 (0.7)d
AUClast (μg h ml−1)	11.7 (8.1)a, 26.2 (23.7)b	15.1 (5.8)c, 20.7 (10.5)d
AUC0–∞ (μg h ml−1)	NR	25.1 (16.6)c, 36.8 (29.17)d
CL/F (l h−1)	NR	10.2 (6.0)c, 7.3 (4.1)d
Vz/F (l)	NR	217.4 (121.3)c, 137.9 (37.6)d

Data are mean (s.d.) except for t_max for PT2977, which is median (range). AUC_last, area under the curve from 0 to last measurable concentration; AUC_0–∞, area under the curve extrapolated to infinity; BID, twice daily; ccRCC, clear cell renal cell carcinoma; CL/F, apparent clearance; C_max, maximum concentration; EC₅₀, half-maximal effective concentration; EPO, erythropoietin; NR, not reported or not published; QD, once daily; PO, orally; t_1/2, terminal half-life; t_max, time of maximal concentration; VEGFA, vascular endothelial growth factor A; Vz/F: apparent volume of distribution.

^aMean PT2385 plasma concentration on day 1.

^bMean PT2385 plasma concentration on day 15.

^cMean PT2977 plasma concentration at week 1.

^dMean PT2977 plasma concentration at week 3.

Table 5|.

Ongoing clinical trials with HIF2α inhibitors

NCT number	Conditions	Interventions	Phase	Study design	Primary end point
PT2977/belzutifan/MK-6482
NCT04627064	ccRCC	Abemaciclib, belzutifan	I	Interventional, NR, SA, OL	ORR, MTD (21 months)
NCT05239728	ccRCC	Belzutifan, pembrolizumab, placebo	III	Interventional, R, PA, triple masking	DFS (54 months)
NCT04846920	ccRCC	Belzutifan	I	Interventional, NR, SA, OL	Participants with at least one AE or DLT or who discontinue treatment owing to an AE
NCT04586231	RCC	Belzutifan, lenvatinib, cabozantinib	III	Interventional, R, PA, OL	PFS (34 months), OS (44 months)
NCT04736706	ccRCC	Pembrolizumab, belzutifan, pembrolizumab–quavonlimab, lenvatinib	III	Interventional, R, PA, OL	PFS (46 months), OS (66 months)
NCT05468697	RCC	Belzutifan, palbociclib	I II	Interventional, R, PA, OL	Participants with at least one AE or DLT oaho discontinue treatment owing to an AE; ORR (4.5 years)
NCT04626518	RCC	Pembrolizumab, MK-4830, belzutifan, lenvatinib, Pembrolizumab–quavonlimab, favezelimab–pembrolizumab	I II	Interventional, R, PA, OL	Participants with at least one AE or DLT or who discontinue treatment owing to an AE; ORR (37 months)
NCT04626479	RCC	Pembrolizumab, favezelimab–pembrolizumab, belzutifan, lenvatinib, pembrolizumab–quavonlimab, vibostolimab–pembrolizumab	I II	Interventional, R, PA, OL	Number of patients with one or more DLTs or AEs or who discontinue study treatment owing to an AE; ORR (43 months)
NCT04195750	RCC	Belzutifan, everolimus	III	Interventional, R, PA, OL	PFS (39 months)
NCT04924075	PPGL, pNET, VHL disease	Belzutifan	II	Interventional, single group, OL	ORR (5.5 years), OS (5.5 years)
NCT04976634	HCC, CRC, PDAC, BTC, EC, ESCC	Pembrolizumab, belzutifan, lenvatinib	II	Interventional, R, PA, OL	Number of participants with at least one DLT or AE or discontinue study treatment for AE; ORR (60 months)
AB521
NCT05536141	ccRCC, solid tumours	AB521	I	Interventional, NR, SA, OL	Number of participants with DLTs (4 months) or AEs (4 months)
NCT05999513	Healthy participants	AB521	I	Interventional, R, crossover assignment, OL	AUC, Cmax, tmax, t1/2, CL/F and Vz/F (predose, up to 168 h after dose)
DFF332
NCT04895748	RCC	DFF332, RAD001, PDR001, NIR178	I	Interventional, NR, SA, OL	Incidence and severity of AEs and SAEs; patients with dose interruptions and dose reductions; DFF332: dose intensity for escalation and expansion (3 years); incidence of DLTs in cycle 1 (28 days) as a single agent and in combinations
NKT2152
NCT05119335	ccRCC	NKT2152	I II	Interventional, R, SA, OL	Number of participants with DLT events in the dose escalation phase (21 days); recommended phase II dose determined in the dose escalation phase (2 years); ORR (1 year)
NCT05935748	ccRCC	NKT2152 Palbociclib Sasanlimab	II	Interventional, R, PA, OL	Number of participants with DLT events (28 days); ORR (1 year)
ARO-HIF2
NCT04169711	ccRCC	ARO-HIF2	I	Interventional, SA, OL	Number of participants with AEs (2 years)

AE, adverse event; AUC, area under the concentration–time curve; BTC, biliary tract cancer; ccRCC, clear cell renal cell carcinoma; CL/F, apparent total plasma clearance; C_max, maximum concentration; CRC, colorectal carcinoma; DFS, disease-free survival; DLT, dose-limiting toxicity; EC, endometrial cancer; ESCC, esophageal squamous cell carcinoma; HCC, hepatocellular carcinoma; MTD, maximum tolerated dose; NR, non-randomized; OL, open label; ORR, objective response rate; OS, overall survival; PA, parallel assignment; PDAC, pancreatic ductal adenocarcinoma; PFS, progression-free survival; PPGL, pheochromocytoma and paraganglioma; pNET, pancreatic neuroendocrine tumour; R, randomized; SA, sequential assignment; t_1/2, terminal half-life; t_max, time of maximal concentration; VHL, von Hippel–Lindau; Vz/F, apparent volume of distribution during the terminal elimination phase.

RNA therapy has also been developed to specifically target HIF2α to reduce potential tumour drug resistance from spontaneous mutations in the binding site of modern HIF2α inhibitors. In vitro screening in ccRCC cell lines combined with in vivo xenograft models identified a tumour-specific RNA interference (RNAi) strategy by delivering HIF2α-specific small interfering RNA (siRNA) to ccRCC cells with high expression of integrins αvβ3 and αvβ5 (ref. 163). Weekly intravenous injection of HIF2α-specific siRNA during murine orthotopic xenograft in vivo significantly inhibited tumour growth over 49 days. Furthermore, a subsequent study using the HIF2α RNAi strategy (A2HIF2 or ARO-HIF2; Arrowhead Pharmaceuticals) showed a similar effect in tumour xenografts derived from patients with ccRCC in the siHIF2 phase I clinical trial (NCT04169711). One of the patients from the siHIF2 phase I clinical trial who received two-weekly injections of ARO-HIF2, had significant downregulation of HIF2α protein in lymph node biopsy and EPO level in the blood with a concomitant reduction in lymph node lesion¹⁶⁴. Together, targeted delivery of HIF2 siRNA could serve as a promising therapy to treat many disease conditions that would benefit from HIF inhibition (Fig. 4).

Fig. 4|. Potential therapeutic applications of HIF inhibition in various diseases.

PT2385 and PT2977 (also known as MK-6482 or belzutifan) are small-molecule inhibitors that selectively target hypoxia-inducible factor 2α (HIF2α), blocking its transcriptional activity by preventing heterodimerization with HIF1β^158,161. These inhibitors offer potential therapeutic options for HIF-driven diseases. For instance, clear cell renal cell carcinoma (ccRCC) is a common form of RCC associated with VHL gene mutations¹⁷². Such mutations can impair the ability of the von Hippel–Lindau protein (pVHL) to recognize and bind to hydroxylated HIF2α, leading to HIF stabilization and target gene transcription, promoting tumour growth¹⁷². PT2399, a HIF2α inhibitor, has shown promising results in reducing tumour growth in preclinical studies¹⁸¹. PT2385 and PT2977 have undergone clinical trials for patients with ccRCC, with PT2977 (belzutifan) obtaining FDA approval for VHL disease-related tumours owing to its encouraging outcomes⁵⁵. Pacak–Zhuang syndrome is a rare form of multiple paragangliomas associated with polycythaemia and caused by a gain-of-function mutation in the endothelial Per–Arnt–Sim (PAS) domain protein 1 (EPAS1) gene^188–190. Under normoxic conditions, HIF2α is hydroxylated by prolyl hydroxylase domain-containing proteins (PHD) and undergoes rapid degradation. In Pacak–Zhuang syndrome, the gain-of-function mutations in the EPAS1 gene can lead to a version of HIF2α that is resistant to hydroxylation by PHDs, preventing its degradation^188–190. As a result, HIF2α accumulates and activates its target genes even under normoxic conditions, which can contribute to the pathophysiological manifestations of the syndrome, such as paragangliomas and polycythaemia^188–190. In a single-patient trial, belzutifan demonstrated promising results in treating the syndrome, with sustained improvements observed⁵⁵. Other conditions that may benefit from targeting HIF2α stabilization include pulmonary hypertension (PH) and retinal neovascularization. ARG1, arginase 1; bHLH, basic helix–loop–helix; CDH5, cadherin 5; CXCL12, C-X-C motif chemokine ligand 12; EGFR, epidermal growth factor receptor; ET1, endothelin-1; GLUT1, glucose transporter type 1; HRE, hypoxia-responsive element; ICAM1, intercellular adhesion molecule 1; OIR, oxygen-induced retinopathy; PAI-1, plasminogen activator inhibitor 1; PDGFB, platelet-derived growth factor subunit B; RBX1, RING box protein 1; RNAi, RNA interference; siRNA, small interfering RNA; SNAI1, snail family transcriptional repressor 1; SNAI2, snail family transcriptional repressor 2; TGFα, transforming growth factor-α; VEGF, vascular endothelial growth factor.

Clinical indications for HIF inhibition

Cancer

Stabilization of HIFs is commonly observed in most human cancers, particularly in solid tumours. Intratumour hypoxia and inflammation is the main mechanism for HIF stabilization, along with genetic mutations commonly seen in VHL disease-related cancers, such as ccRCC. Many cancer-activated pathways could account for HIF activation, including somatic mutation or epigenetic regulation of tumor protein p53 (TP53), PTEN, epidermal growth factor receptor (EGFR) and human epidermal growth factor receptor 2 (HER2)²⁸. Cancer cells use the HIF pathway to drive cancer progression through various mechanisms, such as promoting vascularization, metabolic reprogramming, immune evasion, cell viability and renewal, and treatment resistance. As the role of HIFs in cancer has been covered comprehensively by previous reviews^28,165,166, we focus here on the therapeutic targeting of HIF1α in cancer and HIF2α in ccRCC and multiple paragangliomas, which have a more defined pathogenesis and are at the forefront of pharmacological targeting of HIFs.

HIF1α inhibition in cancer.

HIF1α is stabilized in solid tumours, and a high level of HIF1α is associated with poor prognosis. Genetic and pharmacological studies indicate that HIF1α drives angiogenesis, optimizes glucose metabolism, and promotes cell survival and proliferation. Considerable effort and progress have been made in therapeutic targeting of HIF1α in cancer. For example, daily treatment with topotecan inhibits tumour growth in vivo in murine xenografts derived from the U251-HRE glioblastoma cell line, accompanied by reduced expression of the HIF1α target genes VEGF and phosphoglycerate kinase 1 (PGK1)¹⁶⁷. Topotecan is used in the clinic as an antineoplastic for metastatic cancer¹⁶⁸. However, its primary effect still likely resides in its general cytotoxicity effect instead of HIF1α inhibition. PX-478 has been tested in a phase I, dose escalation study in cancer patients (NCT00522652) and showed efficient HIF1α inhibition and reasonable safety profile¹⁶⁹. Further studies are needed to investigate the safety and efficacy of PX-478 in cancer patients. The early effort to study HIF1A RNAi by ASO EZN-2968 in patients with refractory solid tumours was terminated prematurely owing to the loss of interest from the sponsoring agent¹⁷⁰. However, EZN-2968/RO7070179 was recently investigated in patients with advanced HCC¹⁷¹; overall, EZN-2968/ RO7070179 is well tolerated in cancer patients.

HIF2α inhibition in RCC.

ccRCC is one of the most common forms of RCC. There are two main aetiologies of ccRCC, sporadic ccRCC and ccRCC in VHL hereditary cancer syndromes. Both forms of ccRCC have mutations in VHL, either germline or somatic, which result in sustained HIF stabilization (primarily of HIF2α) during normoxic conditions owing to the insufficient binding of pVHL to hydroxylated HIFα⁵⁸. Consistent with a two-hit tumorigenesis model, ccRCC usually harbours a secondary mutation, such as a loss of tumour suppressor, besides the VHL mutation¹⁷². Current therapy for ccRCC targets angiogenesis by using VEGF inhibitors, such as the multiple receptor tyrosine kinase inhibitor, sunitinib, and MET and VEGF receptor 2 (VEGFR2) inhibitor cabozantinib, as first-line treatments⁵⁸. However, because tumours often develop strategies to activate alternative angiogenic pathways, complete responses to VEGF inhibitors are rare¹⁷³. Also, long-term treatment with VEGF inhibitors is associated with the development of cardiotoxicity, thromboembolism and hypertension^174,175. The mTOR pathway is overactivated during ccRCC to drive aberrant cell survival and proliferation¹⁷⁶. Thus, mTOR complex 1 (mTORC1) inhibitors, such as temsirolimus and everolimus, are approved for ccRCC¹⁷⁷. However, inhibiting mTORC1 signalling could result in the compensatory activation of mTORC2, which drives HIF2α to facilitate cell survival and proliferation, potentially dampening its long-term therapeutic effect¹⁷⁸. Recently, immune checkpoint inhibitors such as the PD1 inhibitor nivolumab have been approved as second-line therapy for ccRCC following a positive phase III study that showed beneficial effects¹⁷⁹, serving as a candidate for combination therapy with other related treatments.

HIF1α and HIF2α may have opposing roles in ccRCC. In vitro studies suggest that HIF1α serves as a tumour suppressor by promoting pro-apoptotic BNIP3 whereas HIF2α acts as an oncogene¹⁸⁰ by promoting target genes such as VEGF, enhancing cell cycle progression through cyclin D1, modulating transforming growth factor-α (TGFα) signalling pathways, and facilitating glycolysis as well as immunomodulation⁵⁸. PT2399, a small molecule related to the HIF2α inhibitors PT2385 and belzutifan, significantly reduced tumour growth in human ccRCC cell lines in vitro and in murine models¹⁸¹. In a phase II, open-label clinical trial, belzutifan provided clinical benefit in patients with ccRCC associated with VHL disease with an objective response rate of 49%⁵⁶. Notably, belzutifan treatment also showed efficacy in non-renal tumours, including pancreatic lesions, central nervous system haemangioblastoma and retinal haemangioblastomas, indicating a systemic antitumour response by HIF2α inhibition in VHL disease. Furthermore, LITESPARK-005, a phase III double-blinded clinical trial that compared belzutifan with everolimus indicated substantial benefit for belzutifan in patients with advanced-stage RCC who had been previously treated with a VEGF tyrosine kinase inhibitor or a PD1/PDL1 inhibitor¹⁸². These exciting studies solidified the use of belzutifan in the treatment of RCC.

Drug resistance could potentially hinder the long-term efficacy of belzutifan treatment. For example, a phase I clinical trial showed that PT2385 treatment is associated with several mutations that drive resistance upon chronic treatment¹⁵⁸. Mechanisms of resistance include G323E amino acid change in HIF2α, which blocks binding of PT2385, as well as mutations in TP53 (R273H, a well-recognized oncogenic mutation) as an alternative pathway¹⁵⁸. However, the TP53 mutation R273H, along with other mutations identified in ccRCC cell lines (such as R248W and P278A) are likely bystander mutations independent of HIF2α inhibition, as sensitivity to HIF2α inhibitor PT2399 is sustained in ccRCC cell lines with these TP53 mutations¹⁸³. Nevertheless, future drug development should focus on targeting mutant HIF2α and potential combination therapies¹⁸⁴. There are two basic strategies for developing potential combination therapy in ccRCC — combining either drugs that target similar pathways or those that target independent pathways. Combining drugs that target a single pathway might result in a synergistic effect to increase medication efficacy. However, drugs that target one pathway often share the same on-target toxicity, resulting in lower tolerated doses and potentially more side effects. For example, combining sunitinib and bevacizumab, a VEGF-neutralizing antibody, resulted in severe cardiovascular adverse events¹⁸⁵. Moreover, the antitumour effects of combining drugs for synergistic action could be theoretically overcome by a single mutation of the target pathway genes, quickly rendering drug resistance. Conversely, combining drugs with different mechanisms of action could be a more rational option as it potentially minimizes the overlapping on-target toxicity while requiring additional mutations in the cancer cells to develop drug resistance. Many ongoing clinical trials seek to investigate whether combining belzutifan with other FDA-approved ccRCC treatment options could improve long-term outcomes in patients with this disease (Table 5). Finally, developing multiple HIF2α small-molecule inhibitors and RNA-silencing drugs targeted to tumour cells (Table 5), possibly through different inhibitory mechanisms, will broaden the treatment options in patients and combat drug resistance in ccRCC.

HIF2α inhibition in polycythaemia and multiple paragangliomas.

Multiple paragangliomas are rare catecholamine-producing neuroendocrine tumours in the adrenal medulla commonly associated with polycythaemia. Around 35–40% of multiple paragangliomas are linked to gene mutations, including mutations in the HIF pathway, such as those encoding VHL and PHDs^186,187. Pacak–Zhuang syndrome is a disease manifestation with early onset polycythaemia and multiple paragangliomas or somatostatinomas caused by gain-of-function mutation in the EPAS1 gene, resulting in disrupted prolyl hydroxylation of HIF2α and aberrant accumulation^188–190. To mimic Pacak–Zhuang syndrome in patients, a murine model was developed by introducing a gain-of-function Epas1^A529V mutation¹⁹¹. Mice with this mutation experienced polycythaemia, increased levels of EPO in plasma and elevated somatostatin in the duodenum. PT2385 treatment only rescued the EPO phenotype but did not affect polycythaemia, indicating a HIF2α-independent pathway for the pathogenesis of polycythaemia in this model¹⁹¹. Interestingly, no paraganglioma or somatostatinoma tumours were detected in the mutant mice, suggesting the existence of additional gene mutations in these tumours and partially explaining the relatively late onset of paragangliomas in patients with Pacak–Zhuang syndrome. As the EPAS1^A529V mutation is the main driver of Pacak–Zhuang syndrome, belzutifan has been studied as a treatment in a single patient who suffered severe polycythaemia, hypertension and multiple paragangliomas. EPAS1^A529V was determined by somatic panel testing and whole-exome sequencing, and belzutifan was given as 120 mg daily treatment for 24 months⁵⁷. Treatment led to a rapid response of decreasing normetanephrine levels in plasma 9 days after treatment initiation and to the alleviation of polycythaemia, marked by reduced haemoglobin levels with a concomitant drop in EPO 17 days after treatment⁵⁷. The adrenal mass also significantly decreased 17 days after the initiation of treatment, and the effects were still evident at 696 days of ongoing treatment⁵⁷. This study indicated that belzutifan could be a promising therapeutic option for patients with Pacak–Zhuang syndrome. Moreover, owing to the significant extent to which the HIF pathway is affected in patients with multiple paragangliomas, future studies are needed to investigate the safety and efficacy of HIF inhibitors in these populations.

Pulmonary hypertension

Pulmonary hypertension is defined by a mean pulmonary artery pressure higher than 20 mmHg¹⁹² and is characterized by increased pulmonary vascular resistance caused by aberrant proliferation of endothelial cells and pulmonary artery smooth muscle cells, extracellular matrix deposition and recruitment of inflammatory cells. Several distinct forms of pulmonary hypertension have been described, based on the underlying aetiology, including pulmonary arterial hypertension, pulmonary hypertension associated with chronic hypoxia exposure and pulmonary hypertension associated with chronic lung diseases. Both HIF1α and HIF2α are stabilized during pulmonary arterial hypertension, with higher constitutive activity of HIF1α in many different cell types and higher constitutive activity of HIF2α more restricted to pulmonary artery endothelial cells⁶⁰. HIF1α reduces the proliferation of endothelial colony-forming cells via induction of cyclin-dependent kinase inhibitor 1B¹⁹³ and promotes pulmonary arterial smooth muscle cell proliferation by driving the expression of miR-9–1, miR-9–3 (ref. 194), which suppresses myocardin expression, and miR-322, which downregulates BMPR1a and SMAD5 (ref. 195). HIF1α also induced the expression of CD146 in pulmonary arterial smooth muscle cells, thereby leading to vascular remodelling¹⁹⁶. In vivo, global or endothelial cell-specific Hif1a deletion did not affect the development of hypoxia-induced pulmonary hypertension^197,198. By contrast, smooth muscle cell-specific deletion of Hif1a was associated with an exaggerated phenotype of pulmonary artery hypertension¹⁹⁹. However, myeloid or monocyte-specific Hif1a deletion in mice protected from the development of hypoxia-induced pulmonary hypertension, suggesting cell type-specific roles for HIF1α^200,201. Moreover, HIF2α exacerbates the pathogenesis of pulmonary hypertension by facilitating endothelial—mesenchymal transition by inducing snail family transcriptional repressor 1 or 2 (SNAI1/2)¹⁹⁷, dysregulating normal vascular NO homeostasis by targeting arginase 1 (ref. 202), promoting pulmonary arterial smooth muscle cell migration and contractility by targeting thrombospondin 1 (ref. 203) and enhancing inflammatory cell adhesion via induction of intercellular adhesion molecule 1 (ref. 204). Global haploinsufficiency of Hif2a results in an improvement in hypoxia-induced pulmonary hypertension in mice, mainly contributed by HIF2α in endothelial cells¹⁹⁸, as mice with endothelial cell-specific deletion of Hif2a are protected from hypoxia-induced pulmonary hypertension¹⁹⁷. Finally, gain of function of HIF2α in mice via G536W alteration resulted in spontaneous pulmonary hypertension with right ventricle hypertrophy²⁰⁵.

Along with genetic evidence, pharmacological targeting of HIF1α or HIF2α can be of benefit in several distinct rodent models of pulmonary hypertension⁶⁰. For example, daily topotecan treatment attenuates hypoxia-induced pulmonary hypertension in rats, marked by reduced pulmonary arteriolar remodelling²⁰⁶. Small-molecule HIF2α inhibitors, such as belzutifan, attenuate pulmonary hypertension in Vhl^R200W mice, which have a loss-of-function mutation in Vhl²⁰⁷ (Fig. 4). However, clinical trials using HIF inhibitors are still needed to assess their efficacy as a treatment for human pulmonary hypertension.

Retinal neovascularization

Retinal neovascularization underlies the pathogenesis of many retinal diseases involving retinal ischaemia, including diabetic retinopathy and retinal vein occlusion. During ocular neovascularization, the newly generated vessels lack tight junctions, which causes plasma leakage, vitreous haemorrhage and traction retinal detachment, resulting in severe retinal dysfunction and vision loss²⁰⁸. Preclinical studies and studies in patients with ischaemic retinal disease demonstrated that HIF1α and HIF2α are stabilized during retinal neovascularization^61,209. Moreover, functional studies identified HIF1α as a driver ofretinal angiogenesis when an adenovirus carrying a constant active form of HIF1α was injected into the eye, leading to the induction of several HIF target genes, including those encoding VEGF, angiopoietin 1 (ANGPT1) and ANGPT2 (ref. 210). In a murine model of age-related macular degeneration, Hif1a deletion attenuated pathological vessel growth and prevented vision loss²¹¹. Besides HIF1α, a study using Hif2a-haploinsufficient mice suggested that HIF2α also drives retinal neovascularization via inducing pro-angiogenic factors in retinal vascular endothelial cells²¹². Interestingly, murine retinal explants experience a rapid stabilization of HIF1α and delayed but sustained stabilization of HIF2α⁶¹. Thus, although siRNA-mediated inhibition of either HIF1α or HIF2α could prevent retinal neovascularization in oxygen-induced retinopathy mouse models, targeting both HIFα isoforms is possibly necessary for effective treatment in patients. Mechanistically, plasminogen activator inhibitor 1 is a HIF2α target gene in vascular cells that drives retinal neovascularization in patients with diabetes²¹³. Pharmacological HIF inhibitor digoxin inhibited both HIF1α and HIF2α in the retina⁶¹, and daily intraperitoneal injection reduced retinal neovascularization in a murine model of retinopathy^61,213. In addition, treatment with HIF inhibitor acriflavine reduced retinal neovascularization and improved vision in mice during oxygen-induced ischaemic retinopathy²¹⁴. Furthermore, intra-ocular injection of HIF inhibitor 32–134D improved retinal neovascularization in a murine model of diabetic eye diseases. Similarly, CBP–p300-interacting transactivator 2 (CITED2) is a peptide inhibitor of HIF1α, and intravitreal injection of CITED2 reduced retinal neovascularization in mice during oxygen-induced retinopathy²¹⁵, although the specificity of CITED2 for HIF2α was unexplored.

Besides nonspecificHIF inhibitors, treatmentwithHIF1α inhibitors— such as topotecan or PX-478 — improved visual function and reduced retinal neovascularization area, respectively, in a murine model of oxygen-induced retinopathy^216,217. HIF2α inhibitors have also been investigated in murine models of retinal neovascularization. For example, daily oral PT2385 treatment starting at day 12 during oxygen-induced retinopathy significantly attenuated the area of neovascularization in the retina of mice in conjunction with decreased expression of VEGF^61,213 (Fig. 4). Together, these studies demonstrated a potential beneficial effect of pharmacological inhibition of HIF1α and HIF2α in retinal neovascularization in various types of retinopathy, indicating the necessity of future clinical studies along this line.

Potential side effects of HIF inhibition

Inhibition of erythropoiesis and the development of anaemia are the most common on-target side effects associated with chronic small-molecule HIF inhibitor use, particularly when using HIF2α-specific inhibitors. Clinical trials of PT2385 and belzutifan for cancer treatment reported anaemia in 45% (23 of 51) or 90% (55 of 61) of patients, respectively^56,159. Anaemia associated with chronic HIF2α inhibitor treatment is considered an on-target event, as EPO production was significantly reduced in these patients^56,159. Although the severity of anaemia is generally mild or asymptomatic and was not associated with transfusion requirements in early phase I studies^159,161, 4 of 90 patients who received belzutifan for 21.8 months during the phase II clinical trials required blood transfusions to correct the anaemia, and 12 of 90 patients received ESA treatment⁵⁶. Thus, routine monitoring of the development of anaemia is necessary for long-term belzutifan treatment. Hypoxia is another common side effect of chronic HIF2α inhibitor treatment, which occurred in 18% (9 of 51) or 31% (17 of 55) of patients participating in phase I studies of PT2385 or belzutifan, respectively^159,161. The molecular and physiological mechanism responsible for the occurrence of hypoxia as a side effect of HIF2α inhibition is currently unclear. However, it is speculated that this phenomenon could be related to an alteration in cardiopulmonary responses to hypoxia²¹⁸, as HIF2α is crucial for increasing minute ventilation, the amount of air entering the lungs in one minute, in response to hypoxia via its functional role in the carotid body²¹⁹. Of note, in the phase II trial of belzutifan, only 1 of 90 patients had hypoxia, which was considerably less common than in previous trials and was attributed to the better baseline health of the trial participants⁵⁶. Because the efficacy of HIF2α inhibitors as treatment for RCC is mainly attributable to its anti-angiogenic effects, it will be interesting to compare the side effects between HIF2α inhibitors and classic angiogenesis inhibitors such as those that target VEGF (bevacizumab). Although VEGF inhibitors have not been reported to cause anaemia or hypoxia, they are known to cause cardiovascular toxicity, thromboembolism or hypertension. Safety reports of the HIF2α inhibitor belzutifan do not include severe cardiovascular side effects resulting in treatment cessation. As anaemia can be effectively monitored and treated relatively easily, the absence of cardiovascular side effects argues for a safety advantage of HIF2α inhibitors over direct VEGF blockers. It is speculative that the distinct safety profile of HIF2α inhibitors could stem from an incomplete blockade of VEGF^181,220. However, the detailed mechanism still needs further investigation. Besides small-molecule HIF2α inhibitors, ASO-based HIF1α inhibition by RO7070179 was associated with dose-dependent liver toxicity in patients with advanced-stage HCC, marked by a rapid increase in alanine aminotransferase (ALT) and aspartate transaminase (AST) levels¹⁵¹. However, this liver toxicity was resolved after a reduction in treatment dosage and could be associated with impaired liver function in patients with HCC¹⁵¹. Side effects associated with long-term HIF1α ASO treatment have not been described yet.

Future challenges and conclusions

It has been a remarkable journey, from the first discovery of HIFs in the early 1990s to the recent FDA approval of both HIF activators and HIF inhibitors to treat human diseases. This success stems from extensive studies on the HIF regulatory pathway, an in-depth understanding of the mechanisms of the functional roles of HIFs and PHDs, and countless trials and errors to identify the optimal molecules for clinical use. As both activators and inhibitors of HIF were recently approved for clinical use, much future work will be required to determine how to optimally use these compounds for additional disease conditions or other potential pharmacological strategies for more specific targeting.

A future challenge relates to the fact that HIF activators activate HIF1α and HIF2α simultaneously. Isoform-selective PHDi²²¹ would potentially allow specific HIF activation in those tissues or cell types with the most abundant HIF-PHD isoforms, minimizing the impact in other unrelated cell or tissue compartment⁸⁶. Similarly, activators of specific HIF isoforms would harness the unique mechanisms of action of HIF1α or HIF2α^41,105,222 without interfering with the function of the other HIFα isoforms.

Development of specific delivery mechanisms is another important direction for the future development of HIF activators or inhibitors. For example, intravenous formulation of HIF-PHDis will enable effective medication delivery in critically ill patients (such as patients with haemorrhagic shock)^49,223, who would be unsuitable for oral administration because they frequently require tracheal intubation. Inhalable HIF-PHDis will allow administration of medication directly into the pulmonary system, thereby increasing their delivery to target cells (such as alveolar epithelial cells) while simultaneously reducing the systemic impact of HIF-PHDis and associated side effects such as thromboembolic events for ARDS treatment. Moreover, non-absorbable HIF-PHDis would allow direct medication delivery to the intestinal epithelial cells, thereby enhancing the therapeutic activation of the HIF pathways for inflammatory bowel diseases. Furthermore, combining HIF inhibitors with other treatment options, such as immune checkpoint blockade inhibitors, might provide novel cancer treatment options.

A general disadvantage of targeting HIFs with HIF-PHDis or HIF inhibitors is their delayed onset of action. HIF activators are effective through the transcriptional induction of HIF target genes, and the time lag to ‘drug onset’ depends on the timing of the target protein maturation. Similarly, HIF inhibitors have a delayed onset of action as they require the prevention of transcription of the targets, and the half-life of the target proteins will dictate the onset of action. Thus, the direct activation or inhibition of HIF target genes may be more desirable, particularly in acute disease conditions such as myocardial infarction. For example, in the context of cardioprotection during ischaemia and reperfusion, the protective effects of HIF activators converge on the ADORA2B receptor^5,224,225. An immediate drug effect is highly desirable in patients with acute myocardial infarction. Therefore, treatment with an ADORA2B agonist could be favourable over treatment with a HIF activator, as the effect of the ADORA2B agonist is immediate. As such, using the knowledge gained from studying adaptive HIF responses, there are opportunities to discover additional therapeutic targets under the control of HIFs that can be targeted directly for a more acute or specific intervention than general HIF activators or inhibitors.

With the current clinical availability of safe and FDA-approved pharmacological compounds to inhibit or promote the stabilization of HIFs, it will be crucial to pursue additional clinical trials in patients beyond the original indications of these drugs, such as renal anaemia treatment for HIF activators or cancer treatments for HIF inhibitors. These results will be important to further guide patient treatments with those compounds, for example, with HIF inhibitors for the treatment of pulmonary hypertension or neovascular diseases of the eye, or HIF activators for the prevention or treatment of peri-operative organ injuries^125,226, such as AKI, ARDS, myocardial infarction or cardiac surgery.

Glossary

ADORA2B receptor

A G-protein-coupled receptor that is activated by the signalling molecule adenosine. It has a role in regulating various physiological processes, including inflammation, angiogenesis and immune cell function

Angiopoietin 1 (ANGPT1) and ANGPT2

Two growth factors that have crucial roles in the regulation of angiogenesis, blood vessel maturation and vascular stability. Whereas ANGPT1 generally promotes vessel stabilization and maturation, ANGPT2 is often associated with vascular destabilization and remodelling, facilitating the actions of other angiogenic factors such as vascular endothelial growth factor

C-terminal transactivation domain

(C-TAD). A region found at the C terminus of certain transcription factors, including hypoxia-inducible factors, which is responsible for activating target gene transcription by recruiting coactivators and other components of the transcription machinery to the target gene promoter

Erythropoiesis-stimulating agents

(ESAs). A group of medications that mimic the effects of erythropoietin, promoting the production of red blood cells in the bone marrow and used to treat conditions such as anaemia

Erythropoietin

(EPO). A glycoprotein cytokine primarily produced by the kidneys in response to cellular hypoxia. Its main biological function is to stimulate erythropoiesis in the bone marrow

Factor inhibiting HIF

(FIH). Also known as asparaginyl hydroxylase; hydroxylates a specific asparagine residue in the C-terminal transactivation domain of hypoxia-inducible factors (HIFs). This modification inhibits the interaction between HIFs and the transcriptional coactivators histone acetyltransferase p300 and cyclic adenosine monophosphate response element binding protein (CREB) binding protein (CBP), thereby reducing HIF-mediated gene transcription.

Hepcidin

Hepcidin is a liver-produced peptide hormone that regulates iron metabolism by inhibiting iron release from cells and decreasing dietary iron absorption, thereby maintaining iron homeostasis in the body

Hypoxia-responsive elements

(HREs). Short DNA sequences found in the regulatory regions of hypoxia-inducible genes, which allow for their transcriptional activation by hypoxia-inducible factors when cellular oxygen levels are low, thereby promoting cellular adaptation to hypoxic conditions

Lipopolysaccharide (LPS) treatment

The experimental administration of LPSs, molecules from the outer membrane of Gram-negative bacteria, to induce an inflammatory response in the lungs, simulating conditions such as acute respiratory distress syndrome for research purposes

Locked nucleic acid

A modified RNA nucleotide in which the ribose ring is ‘locked’ by a methylene bridge connecting the 2′-O atom with the 4′-C atom, which enhances its thermal stability and its affinity for complementary RNA and DNA strands, making it a valuable tool in molecular biology and medical research, particularly in the fields of genomics and therapeutics

Pacak–Zhuang syndrome

A rare condition characterized by the presence of paragangliomas, often in the head and neck region, and associated with somatic mutations in the HIF2A gene, which encodes hypoxia-inducible factor 2α. This syndrome may also be accompanied by polycythaemia and, in some cases, duodenal somatostatinomas

Polycythaemia

A condition characterized by an increased number of red blood cells in the bloodstream, which can thicken the blood, slow its flow and increase the risk of clotting. It can be a primary disease owing to mutations in the bone marrow cells, referred to as polycythaemia vera, or secondary, often as a response to hypoxia or certain tumours

Respiratory bursts

Rapid increases in cellular oxygen consumption, primarily in immune cells such as neutrophils and macrophages, that lead to the production of reactive oxygen species to combat pathogens during an immune response

Von Hippel–Lindau protein

(pVHL). A tumour suppressor protein involved in regulation of cellular responses to oxygen levels by targeting hypoxia-inducible factors for degradation under normal oxygen conditions