Prolyl hydroxylase domain proteins: Localization, regulation, function and their role in erythropoiesis (Review)

Introduction

The present review aimed to provide a comprehensive analysis of prolyl hydroxylase domain (PHD) enzymes, detailing their subtypes, localization, regulatory mechanisms and functional roles, as well as their involvement in erythropoiesis and the development of related pharmacological agents. Under hypoxic conditions, the body initiates a cascade of adaptive biological responses, numerous of which are mediated by transcriptional complexes of the hypoxia-inducible factor (HIF) family. The precise equilibrium among HIF-1α, HIF-2α and HIF-3α is essential for orchestrating the transcriptional regulation of genes associated with erythropoiesis, angiogenesis, vascular homeostasis, metabolic pathways, and cellular proliferation and survival. Simultaneously, HIF exerts indirect inhibitory effects on the transcription of genes linked to other biological processes (1,2). Functionally, the HIF transcription factor exists as a heterodimer consisting of an oxygen-sensitive α and β subunit. The α subunit interacts with the nuclear transporter of the constitutively active aromatic hydrocarbon receptor and its activity is suppressed under normoxic conditions by the dioxygenase family, which requires oxygen, Fe, and 2-oxoglutarate (2-OG) for enzymatic function (3).

In response to hypoxia, HIF-1α levels rise rapidly, playing a pivotal role in acute hypoxic adaptation, particularly in regulating erythropoietin (EPO) synthesis, while HIF-2α predominantly governs long-term hypoxic responses (1,4). Hydroxylation of two conserved prolyl residues within the HIF-α subunit by PHD1-3 facilitates its recognition by the von Hippel-Lindau (VHL) tumor suppressor protein E3 ubiquitin ligase complex, leading to polyubiquitination and subsequent proteasomal degradation. In humans, three distinct PHD isoenzymes (PHD1-3) and an asparaginyl hydroxylase, known as factor-inhibiting HIF (FIH), have been identified. Hydroxylation of asparagine residues within the C-terminal transactivation domain (C-TAD) of HIF-α directly inhibits its interaction with the co-activator p300, thereby preventing the transcriptional activation of its target genes (5,6). Moreover, post-translational modifications such as acetylation and phosphorylation have been reported to further regulate HIF-α activity, adding additional layers of complexity to its control mechanisms (6).

PHDs exhibit substrate specificity and differential tissue distribution, particularly in their hydroxylation of HIF proline residues. By modulating intracellular metabolism, reactive oxygen species (ROS) generation, iron (Fe), homeostasis, nitric oxide (NO) signaling, and redox balance, PHDs serve as critical regulators of HIF activity, influencing a broad spectrum of physiological processes under hypoxic conditions. Although inhibitors targeting the HIF/PHD axis have been successfully integrated into clinical practice, the development of HIF/PHD activators or functional restorers remains limited due to significant technical hurdles. To date, no studies have reported the successful discovery of HIF/PHD activators, highlighting a major gap in current research and a potential avenue for future therapeutic advancements.

Therefore, the present review aimed to comprehensively summarize the subtypes, localization, regulatory mechanisms and biological functions of PHDs, along with the research and development of related pharmacological agents and their role in erythropoiesis. Activators or functional restorers targeting the HIF/PHD pathway represent a promising avenue for therapeutic intervention, particularly in addressing conditions such as polycythemia and chronic mountain sickness. Advancing research in this area could pave the way for groundbreaking developments, offering novel strategies to enhance clinical management and improve patient outcomes.

PHD subtypes and localization

PHD enzymes function as critical oxygen sensors, belonging to a family of oxygen-dependent enzymes responsible for facilitating the proteasomal degradation of HIF-1α under normoxic conditions (7). The hydroxylation of HIF-1α is highly dependent on molecular oxygen, making these enzymes essential regulators of cellular oxygen homeostasis. Currently, three PHD isoforms, PHD1, PHD2 and PHD3, have been identified, each exhibiting distinct oxygen-sensing properties (8). In vitro studies suggest a hierarchical hydroxylation efficiency among these isoforms, with PHD2 demonstrating the highest activity, followed by PHD3 and then PHD1.

PHDs exhibit both substrate specificity and tissue-specific expression in their hydroxylation of HIFs. PHD2 preferentially hydroxylates HIF-1α over HIF-2α and HIF-3α, whereas PHD1 and PHD3 display higher efficiency in hydroxylating HIF-2α. Their expression patterns also vary across tissues: PHD2 is predominantly found in adipose tissue, PHD1 is highly expressed in testicular cells, and PHD3 is primarily located in cardiac cells (8). In terms of subcellular localization, PHD2 is mainly cytoplasmic, PHD3 is predominantly nuclear, and PHD1 is distributed in both the cytoplasm and nucleus. Encoded by the EGLN1 gene, PHD2 serves as the principal oxygen sensor responsible for modulating HIF-2α activity (9,10).

Among the three PHD isoforms, PHD2 was initially identified as the most functionally significant hydroxylase in most cell types (11). Differences in their intracellular distribution have been elucidated through studies involving the overexpression of labeled proteins (12,13). Specifically, fusion protein analyses have revealed that GFP-tagged EGLN1/HPH-2/PHD2 localizes predominantly to the cytoplasm, whereas EGLN2/HPH-3/PHD1 is primarily nuclear. Conversely, EGLN3/HPH-1/PHD3 is distributed across both the nucleus and cytoplasm in a relatively balanced manner (13).

Notably, PHD expression has been implicated in inflammatory conditions such as ulcerative colitis (UC), a subtype of inflammatory bowel disease (IBD). In patients with UC, PHD1 and PHD2 are highly expressed in the lamina propria and colonic epithelium, while PHD3 is predominantly localized to the vascular endothelium. Moreover, PHD3 expression is significantly elevated in inflamed biopsy samples and is positively correlated with inflammatory cytokines such as IL-8 and TNF-α, as well as the apoptotic marker caspase-3 (14).

HIF proteins are heterodimeric transcription factors consisting of a tightly regulated α subunit (HIF-1α, HIF-2α, or HIF-3α) and a constitutively expressed β subunit. Under normoxic conditions, HIF-α subunits undergo oxygen-dependent hydroxylation by PHDs, allowing their recognition by the VHL tumor suppressor protein, a key component of an E3 ubiquitin ligase complex. This interaction marks HIF-α for ubiquitination and subsequent degradation via the proteasome pathway (15,16). However, under hypoxic conditions, hydroxylation is suppressed, preventing VHL-mediated degradation and enabling HIF-α to accumulate and exert its transcriptional effects (12).

Within mammalian cells, the functions of the PHD protein family (PHD1, PHD2 and PHD3) are diverse, with PHD2 emerging as the predominant oxygen sensor regulating HIF-1α stability and degradation. In addition to PHDs, FIH-1, an oxygen-dependent asparaginyl hydroxylase, plays a critical role in regulating the C-TAD of HIF. FIH-1 provides a direct mechanistic link between oxygen sensing and HIF-mediated transcription, further modulating cellular responses to hypoxia (17).

HIF-1α exhibits dynamic nucleocytoplasmic shuttling, whereas HIF-1β remains permanently localized within the nucleus. The VHL tumor suppressor protein (pVHL) is distributed across both the nuclear and cytoplasmic compartments. The nuclear translocation of PHD1 is mediated by classical nuclear localization signals (NLS), utilizing the importin α/β receptor pathway. By contrast, PHD2 nuclear import relies on a non-classical NLS, following an alternative import pathway, which results in PHD2-mediated hydroxylation of HIF-1α predominantly occurring within the nucleus. The nuclear export of PHD2 is facilitated by an N-terminal nuclear export signal and requires the export receptor chromosome region maintenance 1. Meanwhile, the nuclear import of PHD3 is regulated via the importin α/β receptor pathway and also depends on a non-classical NLS (18).

In the central nervous system, NG2 cells represent a fourth class of glial cells alongside astrocytes, microglia, and oligodendrocytes. HIF-2 activation in NG2 cells plays a pivotal role in promoting neurovascular expansion and remodeling independently of EPO. The regulatory control of HIF-2 activity in NG2 cells involves both PHD2 and PHD3. Under conditions where PHD2 and PHD3 are inactive, PHD1 assumes a compensatory role in regulating the HIF-2 transcriptional response, thereby facilitating vascular expansion and remodeling. However, the simultaneous inactivation of PHD1, PHD2 and PHD3 leads to robust HIF-2 activation, further enhancing neurovascular remodeling independent of EPO signaling (19).

The genetic inactivation of PHD1, PHD2 and PHD3 results in HIF activation, triggering the reprogramming of myofibroblast-transformed renal EPO-producing cells (MF REPs) to resume EPO secretion. Specifically, the loss of PHD2 in REPs restores EPO gene expression in damaged renal tissue, leading to erythrocytosis. By contrast, the simultaneous deletion of PHD1 and PHD3 prevents the suppression of EPO expression without inducing erythrocytosis (20). These findings underscore the predominant role of PHD2 in EPO regulation and red blood cell (RBC) production. Given the tissue-specific regulatory mechanisms of HIF by different PHD isoforms, their physiological roles vary significantly across organ systems. As research progresses, pharmacological agents targeting angiogenesis and EPO synthesis have been integrated into clinical practice for the treatment of related disorders. Therefore, further investigation into PHD enzyme functions holds substantial clinical significance.

The study of PHD enzyme localization primarily involves the fusion of PHD to the N-terminus of a fluorescent protein, followed by transient transfection of the fusion construct into human osteosarcoma (U2OS) cells. Subsequent visualization is performed using three-dimensional two-photon confocal fluorescence microscopy (13).

PHDs belong to the family of α-ketoglutarate (α-KG)-dependent dioxygenases and serve as key regulators of cellular oxygen sensing and metabolic control. They play a crucial role in maintaining oxygen homeostasis by mediating the hydroxylation of target proteins, thereby influencing their stability and function. A summary of the major PHD isoforms, subcellular localization, tissue distribution, substrates, and biological functions is provided in Table I.

Table I. Isoforms, subcellular localization, tissue expression, substrates, and biological functions of PHDs.

Table I.

Isoforms, subcellular localization, tissue expression, substrates, and biological functions of PHDs.

Isoforms	Subcellular localization	Tissue expression	Substrates	Biological functions	Disease associations
PHD1	Primarily localized	Highly expressed in	Substrate specificity of PHDs depends on isoform	Primarily involved in	Primarily localized in the
(EGLN2)	in the nucleus and	skeletal muscle,	and physiological context.	fundamental oxygen sensing	nucleus and mitochondria,
	mitochondria,	heart, and kidneys.	Classical substrates:	and energy metabolism	associated with energy
	associated with		– HIF-1α/2α: All PHD isoforms hydroxylate proline	regulation.	metabolism.
	energy metabolism.		residues in HIF-α subunits (e.g., Pro402/Pro564	Closely linked to
			in HIF-1α), triggering ubiquitin-mediated	mitochondrial function
			degradation.	and may modulate metabolic
			– PHD2 is the predominant hydroxylase for HIF-1α,	adaptation in muscle.
			while PHD3 plays a more prominent role in HIF-2α
PHD2	Mainly cytoplasmic	Ubiquitously	regulation.	Hypoxia adaptation: PHD2	Mainly cytoplasmic but
(EGLN1)	but can associate with	expressed, with high	Non-classical substrates:	restricts tumor angiogenesis	can associate with the
	the plasma membrane	activity in vascular	– PHD3 specifically hydroxylates CRTC2 at	via HIF-1α degradation.	plasma membrane or
	or endoplasmic	endothelial cells and	Pro129/Pro625, facilitating its nuclear translocation	PHD2 inhibitors	endoplasmic reticulum,
	reticulum, dynamically	the tumor micro-	and activation of gluconeogenic genes (particularly	(for example, roxadustat)	dynamically relocalizing
	relocalizing in response	environment.	under fasting or diabetic conditions).	are used to treat renal	in response to oxygen
	to oxygen fluctuations.		Other substrates:	anemia.	fluctuations.
PHD3	Resides in the	Highly expressed in	– Nuclear factor κB (NF-κB), RNA polymerase II,	Neuroprotection: PHD3	Resides in the cytoplasm,
(EGLN3)	cytoplasm, where it	the liver, adipose	among others, suggesting PHD involvement in	contributes to ischemic	where it interacts with
	interacts with substrates	tissue, and nervous	inflammation and transcriptional regulation.	preconditioning in neurons	substrates such as CRTC2
	such as CRTC2 and	system.	The principal hydroxylase of HIF (HIF-1α),	via HIF-2α regulation.	and HIF-1α, but can
	HIF-1α, but can		facilitating its ubiquitin-mediated degradation,	Metabolic disorders: The	translocate to the nucleus
	translocate to the		thereby regulating hypoxic responses	PHD3-CRTC2 axis is	under specific conditions.
	nucleus under specific		(e.g., angiogenesis, erythropoiesis). PHD2 knockout	aberrantly activated in type
	conditions.		is embryonically lethal, underscoring its critical role	2 diabetes, and PHD3
			in HIF regulation.	inhibition has been shown to
			Beyond its role in HIF regulation, PHD3 is involved in metabolic control (e.g., gluconeogenesis). It hydroxylates CRTC2 (CREB-regulated transcriptional coactivator), modulating the expression of hepatic gluconeogenic genes such as PEPCK and G6Pase.	lower blood glucose levels.

[i] PHD, prolyl hydroxylase domain; HIF, hypoxia-inducible factor.

Functions of PHD enzymes

The oxygen-dependent hydroxylation of HIF-α subunits serves as a key signal for ubiquitination and subsequent proteasomal degradation, playing a crucial role in regulating HIF protein abundance and maintaining oxygen homeostasis (21–25). All three PHD isoforms (PHD1-3) are capable of hydroxylating HIF-α subunits, although their functional significance varies (12,26). Among them, genetic studies have highlighted PHD2 as the most critical regulator of HIF-1α stability. Notably, PHD2 expression is upregulated under hypoxic conditions, establishing an HIF-1-mediated autoregulatory feedback loop that fine-tunes oxygen-dependent responses (11).

Beyond their role in HIF-α hydroxylation, PHDs can also modulate HIF activity through hydroxylase-independent pathways. While these enzymes primarily control the stability of HIF-α proteins, PHD2 and PHD3 themselves are subject to upregulation via HIF-dependent feedback mechanisms. Functionally, different PHD isoforms contribute to distinct physiological and pathological processes, including angiogenesis, erythropoiesis, tumorigenesis, cell proliferation, differentiation and survival. As a result, disruptions in PHD expression or function can lead to diverse biological consequences, reflecting the unique regulatory roles of each isoform (27). The selectivity of PHD isoforms toward different HIF-α subunits exhibits a degree of specificity. PHD2 is preferentially induced by HIF-1α, whereas PHD3 responds to both HIF-1α and HIF-2α (28). Each PHD isoform regulates the HIF system in a functionally distinct, non-redundant manner. Under specific experimental conditions, the relative contribution of each PHD enzyme depends significantly on its expression levels, with the hydroxylation rate being directly correlated with the abundance of active PHD proteins (29). The oxygen-dependent hydroxylation of specific proline residues in HIF-1α allows recognition by the VHL tumor suppressor complex, a key component of the E3 ubiquitin ligase system. This interaction leads to the polyubiquitination of HIF-1α, triggering its rapid degradation via the ubiquitin-proteasome pathway (26,30–32).

All three PHD isoforms contribute to HIF regulation, and their cell-type specificity and inducibility provide HIF with the flexibility to fine-tune hypoxia responses across different tissues. The relative specificity of pharmacological PHD inhibition presents an opportunity for selective modulation of the HIF system, which holds considerable clinical significance. For instance, inhibiting PHD2 broadly enhances HIF activation across multiple cell types, even under normoxic conditions. By contrast, selective inhibition of PHD3 predominantly amplifies the hypoxic response in tissues where this enzyme is highly expressed (29). This approach could be leveraged to activate the HIF system for treating ischemic or hypoxic diseases. Future research must further elucidate the in vivo specificity of PHD effects on different HIF-α subunits, as well as the distinct roles of PHD isoforms in various biological processes. A deeper understanding of these mechanisms is essential for minimizing adverse effects associated with PHD inhibitors and optimizing their therapeutic efficacy.

When PHD function is inhibited, the oxygen-dependent degradation of VHL and HIFα is disrupted. Similar to PHDs, the HIF-associated regulatory factor FIH, an asparaginyl hydroxylase, suppresses the transactivation function of HIFα by hydroxylating its TAD, thereby inhibiting the p300/CBP signaling pathway, which plays a crucial role in cell cycle regulation and apoptosis. Although the pVHL/PHD/HIF axis is one of the most extensively studied oxygen-sensing mechanisms, numerous aspects remain to be elucidated (33). The mechanistic action of the pVHL/PHD/HIF pathway, the alterations in HIF2α-ARNT dimerization under hypoxia, and the downstream targets of HIF are illustrated in Fig. 1.

Figure 1. Mechanistic action of the pVHL/PHD/HIF pathway, the alterations in HIF2α-ARNT dimerization under hypoxia, and the downstream targets of HIF. HIF, hypoxia-inducible factor; PHD, prolyl hydroxylase domain; FIH, factor-inhibiting HIF; ROS, reactive oxygen species; miR, microRNA.

Regulation of PHD-catalyzed hydroxylation reactions

PHD1 and PHD2 are large enzymes, each consisting of >400 amino acids (407 and 426, respectively, in humans), with a highly conserved hydroxylase domain located in their C-terminal region, sharing ~55% sequence identity in this region. By contrast, their N-terminal regions exhibit greater sequence diversity and comparatively lower activity (26). PHD3, the shortest of the three isoforms, contains only 239 amino acids in humans. Although it retains a hydroxylase domain, its N-terminal region is markedly distinct, comprising only a short, unique segment.

Oxygen dependence

Oxygen serves as a fundamental substrate for PHD-catalyzed hydroxylation reactions. The Km values for oxygen across the three PHD isoforms range from 230 to 250 µM, which slightly exceeds the oxygen concentration in aqueous solutions equilibrated with ambient indoor air. Since intracellular oxygen levels are typically lower, this higher Km ensures that hydroxylase activity remains strictly oxygen-dependent, as long as other essential substrates and cofactors are adequately available (34). Under hypoxic conditions (0.5–2% oxygen), HIF-1α levels increase, a process that requires functional mitochondria. Notably, inhibition of cytochrome c oxidase using respiratory chain inhibitors such as NO leads to the destabilization of HIF-1α, even in low-oxygen environments (21). This phenomenon is likely attributed to the rapid mitochondrial oxygen consumption during oxidative phosphorylation, which significantly reduces cytoplasmic O2 availability (35).

Intracellular Fe(II) concentration

PHD enzymes belong to the Fe(II)/2-OG-dependent dioxygenase family, with PHD2 displaying particularly slow reactivity with O2, a feature linked to its role in hypoxia sensing. When PHD2 forms a stable complex with Fe(II) and 2-OG, the water molecules coordinated to Fe(II) remain tightly bound at the enzyme's active site. Before O2 can bind, these water molecules must be displaced. A single amino acid substitution, replacing glutamic acid (D315E) with aspartic acid, which directly interacts with Fe(II), significantly reduces PHD2's affinity for Fe(II). However, this modification simultaneously enhances the reaction rate with O2 by 5-fold, thereby maintaining PHD2's catalytic efficiency (36).

Competitive inhibition by 2-OG analogs

2-OG plays a crucial role as an intermediate in the tricarboxylic acid (TCA) cycle and serves as an essential co-substrate for PHD activity, aiding in the coordination of Fe(II) within the catalytic center (26). The hydroxylation of HIF-α proline residues by Fe(II)- and 2-OG-dependent PHD enzymes is essential for cellular oxygen sensing. Notably, premature 2-OG binding or prolyl hydroxylation of HIF-α can drastically reduce the affinity of HIF-α for PHD2, by nearly 50-fold, thereby inhibiting its interaction with the enzyme. Consequently, when 2-OG availability is limited, PHD enzymatic activity decreases, preventing HIF-α degradation and stabilizing the hypoxic response (37).

Proline substrate specificity

The stabilization of HIF proteins under hypoxic conditions represents a key adaptive mechanism in oxygen-deprived environments. The degradation of HIF-α subunits is tightly regulated by the three PHD isoenzymes (PHD1-3), which catalyze proline hydroxylation (38). To date, no proline hydroxylase activity has been detected in non-HIF proteins or peptides under conditions relevant to HIF-α hydroxylation (39). Among PHDs, the C-terminal oxygen-dependent degradation domain (C-ODD) exhibits higher activity than the N-terminal oxygen-dependent degradation domain (N-ODD), with the latter being largely inactive in PHD3. Interestingly, PHD2 demonstrates greater efficiency in hydroxylating the N-ODD of HIF-2α than that of HIF-1α (40). However, the hydroxylation efficiency of HIF-2α N-ODD by PHD2 remains lower than that of HIF-1α N-ODD (40). It has been suggested that an antiparallel structural motif spanning residues 197 to 380 in PHD2 plays a role in interacting with the ODD domain, with hydrophobic amino acids contributing to substrate recognition (41). Furthermore, certain mutations in PHD2 are associated with altered HIF-2 stability. For instance, a heterozygous germline mutation at residue H374 leads to PHD2 destabilization and loss of enzymatic function, ultimately resulting in upregulated HIF-2α activity (9).

The hydroxylation regulation of PHDs is illustrated in Fig. 2. Once synthesized, HIF-α is rapidly degraded if sufficient non-mitochondrial oxygen is available. HIF PHD1, PHD2 and PHD3 confer oxygen-dependent proteasomal degradation to the HIF-α subunit. Identifying alternative targets of HIF hydroxylases is crucial for fully elucidating the pharmacology of prolyl hydroxylase inhibitors (PHIs) (42).

Figure 2. Hydroxylation regulation of PHDs. PHD, prolyl hydroxylase domain; FIH, factor-inhibiting HIF.

Despite significant technological advancements, screening, detecting, and validating alternative functional targets of PHD and FIH remain challenging. A major limitation is the lack of high selectivity for PHD isoform-specific inhibitors, which can lead to a range of adverse effects. The most successful clinical application to date is in the treatment of anemia associated with chronic kidney disease (CKD), where PHD inhibitors have been widely used. However, research on PHD activators or agonists remains in its early stages.

Biological efficacy of PHDs

Transition metal ions, particularly Fe, play a pivotal role in the hypoxia-regulated PHD-HIF-EPO signaling axis, which governs erythropoiesis, angiogenesis, anaerobic metabolism, adaptation, cell survival and proliferation, thereby maintaining cellular and systemic homeostasis (43). The regulation of HIF is primarily mediated by the highly conserved EGLN/PHD family, which hydroxylates HIF in an oxygen-dependent manner. This hydroxylation facilitates recognition by VHL family proteins, leading to ubiquitination and proteasomal degradation of HIF. Interestingly, PHD3 has been found to enhance HIF signaling through the hydroxylation of the glycolytic enzyme pyruvate kinase muscle subtype 2 (PKM2). Beyond its role in the HIF pathway, PHD also influences synaptic transmission by modulating α-amino-3-hydroxy-5-methyl-4-isoxazolepropanoic acid (AMPA) receptor trafficking and regulating transient receptor potential cation channel member A1 (TRPA1) activity in response to oxygen levels in sensory neurons. Moreover, PHD activation is modulated by poly(rC)-binding protein 1 (PCBP1), which functions as an Fe chaperone, as well as by the (R)-enantiomer of 2-hydroxyglutarate (2-HG), highlighting the intricate interplay between multiple regulatory pathways in PHD function (8).

Regulation of lifespan and renal adaptability

In Caenorhabditis elegans, the activity of EGL-9, a PHD homolog, is regulated by the hydrogen sulfide-sensing cysteine synthase-like protein CYSL-1, which, in turn, is modulated by the acyltransferase RHY-1. Notably, mutations in vhl-1 significantly extend lifespan through a HIF-1-dependent mechanism. The long-lived phenotype of vhl-1 mutants is suppressed by mutations in egl-9 and rhy-1, whereas RNA interference targeting rhy-1 extends the lifespan of wild-type worms while shortening that of vhl-1 mutants (44). Additionally, in renal physiology, the downregulation of PHD by high salt exposure triggers the activation of peroxisome proliferator-activated receptor α (PPARα). This PHD/HIF-1α regulatory axis introduces a novel transcriptional control mechanism, engaging downstream signaling pathways of NO synthase and heme oxygenase to promote adaptive renal responses (45).

PHD regulation of EPO production

Renin-expressing cells can be categorized into two subgroups: Classic glomerular renin-producing cells and mesenchymal renin-positive cells. The latter population functions as a reservoir of natural EPO-producing cells, displaying a rapid EPO response under acute hypoxia via the stabilization of HIF-2. Interestingly, it is the combined deficiency of PHD2 and PHD3, rather than PHD2 loss alone, that induces EPO expression in glomerular renin-positive cells. Sustained HIF-2 activation in these cells transforms them into EPO-producing cells. The strong expression of PHD3 in glomerular renin-positive cells serves to prevent this HIF-2-driven transformation, suggesting that PHD3 plays a crucial role in maintaining the functional stability of the renin-expressing cell phenotype (43).

Neuroprotection and PHD-mediated regulation of HIF

In an in vitro hypoxia-ischemia (HI) model using oxygen-glucose deprivation (OGD) in rat pheochromocytoma (PCC) (PC-12) cells differentiated with nerve growth factor, EPO treatment has been found to enhance PHD2 transcription and translation. This upregulation of PHD2 inhibits HIF-1α expression, reduces ROS formation, and decreases matrix metalloproteinase-9 (MMP-9) activity, ultimately improving cell survival following OGD-induced injury. Conversely, silencing PHD2 with small interfering RNA (siRNA) reverses the neuroprotective effects of EPO, indicating that PHD2 serves as a key mediator in EPO-induced HIF-1α suppression and neuroprotection in HI conditions (46).

PHD deficiency and metabolic consequences

The absence of PHD3 leads to increased blood lipid levels and elevated hematocrit without affecting atherosclerotic plaque size in low-density lipoprotein receptor knockout mice (47). Additionally, studies have shown that the complete loss of PHD1, PHD2 and PHD3 in the liver results in severe fatty liver disease and erythrocytosis due to excessive hepatic EPO production. Mice with hepatocyte-specific deletion of all three PHD isoforms (PHD1/2/3hKO) exhibit a 1246-fold increase in hepatic EPO expression while renal EPO levels drop to 6.7% of normal. These mice also develop hematocrit levels reaching 82.4%, accompanied by severe vascular malformations and liver steatosis (48).

By contrast, mice with dual liver-specific deletions of PHD2 and PHD3 (PHD2/3hKO) also show increased hepatic EPO production and reduced renal EPO expression, but the magnitude of these changes is significantly lower than in PHD1/2/3hKO mice. Unlike the triple knockout model, PHD2/3hKO mice maintain normal hematocrit levels, vascular integrity and hepatic lipid homeostasis (48). These findings indicate that overall PHD activity, rather than the function of any single isoform, plays a dominant role in regulating hepatic EPO production through the PHD-HIF2α-EPO signaling cascade in vivo (49).

Impact of PHDs on bone adaptation and metabolism

The expression levels of negative regulatory factors, including PHD2, FIH, and histone deacetylase sirtuin-6 (SIRT6), are significantly elevated in the skeletal muscle tissue of elite athletes, whereas the expression of the hypoxia response gene pyruvate dehydrogenase kinase 1 (PDK-1) is reduced. This suggests that exercise-induced training enhances HIF inhibition, thereby downregulating PDK-1 and contributing to skeletal muscle adaptation to physical activity (50). In bone cells, the disruption of PHD2, which is highly expressed in osteoblasts, results in severe osteoporosis. Notably, treatment with ascorbic acid effectively suppresses PHD2 expression without affecting PHD1 levels (51). However, when osteoblasts are treated with PHD inhibitors such as dimethyloxalylglycine (DMOG) or 3,4-dihydroxybenzoate ethyl, the ascorbic acid-mediated modulation of osteoblast differentiation markers is entirely abolished (52).

Osteoprotegerin (OPG), a direct target gene of HIF, plays a crucial role in regulating bone homeostasis. The inactivation of PHD2 and PHD3 enhances HIF activity, thereby promoting bone accumulation through the direct modulation of OPG balance between osteoblasts and osteoclasts. However, the simultaneous inactivation of all three PHD isoforms (PHD1, PHD2 and PHD3) leads to excessive angiogenesis-osteogenesis coupling, resulting in severe erythrocytosis and pathological bone overgrowth due to extreme HIF activation. Interestingly, the dual knockout of PHD2 and PHD3 is sufficient to prevent bone loss without compromising hematopoietic homeostasis in ovariectomized mice (53).

Additionally, PHD3 expression plays a distinct role in muscle cell migration. Rhodiola glycoside has been shown to promote skeletal muscle cell migration and paracrine signaling by selectively inhibiting the transcription of PHD3 without affecting PHD1 or PHD2 (54). By contrast, increased α-KG levels suppress osteoclastogenesis by inhibiting the NF-κB signaling pathway, which is activated by RANKL, in a PHD1-dependent manner (55).

Regulation of angiogenesis

PHD enzymes also play a significant role in vascular remodeling and angiogenesis. The endothelial cell-specific deletion of PHD2 results in severe pulmonary hypertension characterized by increased right ventricular systolic pressure and extensive muscular hypertrophy of the peripheral pulmonary arteries. However, this phenotype does not involve erythrocytosis. Studies on endothelial-specific PHD2 mutants with concurrent HIF-1α or HIF-2α inactivation have demonstrated that pulmonary hypertension is primarily driven by HIF-2α, rather than HIF-1α. The pathological effects of HIF-2α in this condition are largely mediated by upregulated expression of the vasoconstrictor endothelin-1 and diminished apelin receptor signaling, which normally promotes vasodilation (56). In HI models, the expression of PHD proteins exhibits dynamic changes over time. Following 24 h of HI, PHD3 protein levels, along with HIF-1α expression, are significantly elevated. However, after 72 h of HI, PHD3 protein levels decline, while PHD1 and PHD2 levels remain unchanged throughout the hypoxic period (57).

Role of PHDs in tumor progression and metabolism

PHD enzymes play a complex and context-dependent role in cancer, significantly influencing tumor growth and metabolism (58). The primary mechanism underlying this effect is the regulation of HIF-1α, which drives a metabolic shift from oxidative phosphorylation to anaerobic glycolysis, thereby promoting tumor adaptation to hypoxic microenvironments. This metabolic reprogramming alters oxidative stress responses and results in the accumulation of tumor-specific metabolites that contribute to cancer progression.

In clear cell renal cell carcinoma (ccRCC), the loss of pVHL leads to the accumulation of HIF-α isoforms and their downstream target genes. Interestingly, in contrast to most cell types where PHD3 negatively regulates HIF-1α, ccRCC tumors exhibit a strong positive correlation between PHD3 and HIF2A mRNA expression. High PHD3 expression in these tumors sustains elevated HIF-1α levels and enhances the expression of HIF target genes, potentially increasing the invasive potential of ccRCC cells (59).

In non-small cell lung cancer (NSCLC), the downregulation of PHD1 and PHD2 is associated with tumor initiation and progression. The loss of these PHD isoforms correlates with increased activity of downstream HIF pathway genes such as HIF-1α, PKM2 and PDK1. However, PHD3 does not appear to play a significant role in NSCLC tumor biology (60).

In colorectal cancer models with varying levels of drug resistance, silencing PHD1, but not PHD2 or PHD3, prevents p53 activation and impairs DNA repair mechanisms, ultimately leading to cancer cell death. PHD1 facilitates p53 activation through hydroxylation-dependent interactions with p38α kinase, enabling phosphorylation of p53 at serine 15. This modification enhances p53-mediated nucleotide excision repair by promoting interactions with the DNA helicase XPB, thereby protecting tumor cells from chemotherapy-induced apoptosis (61).

Current research suggests that PHD enzymes exhibit both tumor-promoting and tumor-suppressing functions, depending on their expression profiles in different cancer types and cellular contexts. The precise role of each PHD isoform in tumorigenesis remains an area of active investigation, with implications for the development of targeted cancer therapies (62).

Regulation in inflammation and glucose metabolism

PHD1 has been found to be overexpressed in pouchitis biopsies from patients with ileal pouch-anal anastomosis for UC, with its expression levels correlating directly with disease activity. Notably, treatment with the small-molecule PHD inhibitor DMOG has demonstrated the ability to restore intestinal epithelial barrier integrity by upregulating the tight junction proteins zona occludens-1 and claudin-1. Additionally, DMOG alleviates intestinal epithelial cell apoptosis, thereby mitigating inflammation in the pouch and improving disease outcomes (63).

In a mouse model of IBD, the expression of PHD1 and PHD2 progressively increases as the disease advances, whereas PHD3 levels remain unchanged. This suggests that inhibiting all three PHD subtypes may not be an optimal therapeutic strategy for IBD, as normal intestinal function appears to rely on the presence of PHD3 (64).

Beyond inflammation, PHD3 also plays a significant role in glucose metabolism. A sudden loss of PHD3 (also known as Egln3) in the liver has been shown to enhance insulin sensitivity by selectively stabilizing HIF-2α. This stabilization promotes the transcription of insulin receptor substrate 2 (Irs2), which in turn enhances insulin-stimulated Akt activation. The beneficial metabolic effects of PHD3 knockout on glucose tolerance and insulin signaling are entirely dependent on HIF-2α and Irs2, as their elimination negates these improvements (65).

Additionally, α-KG, a key respiratory substrate, has been implicated in insulin secretion. Cytoplasmic PHD enzymes regulate α-KG metabolism, and inhibition of PHDs using ethyl dihydroxybenzoate (EDHB) significantly suppresses glucose-stimulated insulin secretion (GSIS) in pancreatic β-cells (832/13 clone), as well as in rat and human islets. This suppression occurs due to reduced glucose metabolism, a lower ATP/ADP ratio, and diminished levels of critical TCA cycle intermediates, including pyruvate, citrate, fumarate and malate. Interestingly, silencing PHD1 and PHD3 with siRNA impairs GSIS, whereas PHD2 knockdown has no impact, suggesting that PHD1 and PHD3 are key regulators of glucose metabolism in pancreatic β-cells (66).

Regulation of erythropoiesis

A heterozygous PHD2 c.1121A → G (p.H374R) mutation has been identified in patients with familial polycythemia, though it is not directly associated with tumor development. However, individuals with PHD2 mutations linked to polycythemia have been later diagnosed with recurrent paragangliomas (PGL), suggesting a potential predisposition. The pathogenic mechanism underlying these conditions involves dysregulated hypoxia sensitivity, leading to the stabilization and accumulation of HIF-2α, which creates a pseudo-hypoxic state. This state, in turn, may contribute to the development of both polycythemia and hypoxia-associated endocrine tumors, including PCC and PGL (9).

VHL-associated diseases also result in the stabilization of HIFs, frequently observed in endocrine tumors such as PCC and PGL. The aberrant upregulation of HIF not only directly promotes tumor growth but also reduces apoptosis in endocrine tumor cells, further exacerbating disease progression. Based on the classification, localization and functions of PHDs, it is evident that excessive erythrocytosis is primarily driven by dysregulation of PHD2, which leads to unchecked HIF-2α accumulation. This, in turn, results in the overexpression of downstream target genes, particularly through the hyperactivation of the EPO pathway.

Chuvash polycythemia, a well-characterized genetic disorder, is associated with a homozygous 598C>T germline mutation in the VHL gene. This mutation leads to aberrant HIF-1α upregulation, even under normoxic conditions, driving the excessive production of EPO and several other hypoxia-responsive genes. The C>T missense mutation in VHL causes an arginine-to-tryptophan substitution at residue 200 (Arg200Trp), weakening the interaction between VHL and HIF-1α. This defective interaction reduces HIF-1α degradation, thereby leading to the persistent overexpression of EPO, SLC2A1 (GLUT1, encoding solute carrier family 2), TF (encoding transferrin), TFRC (encoding transferrin receptors CD71/p90), and VEGF (encoding vascular endothelial growth factor) (67,68).

Overall, the HIF/PHD axis plays a complex and multifaceted role in diverse physiological and pathological processes (Fig. 3). Significant progress has been made in the development of PHD inhibitors for clinical applications (69). However, different PHD subtypes exert distinct effects across various biological functions, including angiogenesis, erythropoiesis, cancer progression, cellular growth, differentiation and survival. Therefore, further investigations are required to elucidate the precise impact of PHD enzymes on HIF-α subtype specificity in in vivo models, as well as their individual contributions to distinct biological pathways (27).

Figure 3. Role of HIF/PHD axis in diverse physiological and pathological processes including erythropoiesis and iron metabolism, metabolism/redox, growth and apoptosis, angiogenesis and vascular regulation, migration/motility, transcriptional regulation and extracellular matrix metabolism. FIH, factor-inhibiting HIF; PHD, prolyl hydroxylase domain; HIF, hypoxia-inducible factor.

Pharmaceutical development and therapeutic applications of PHD inhibitors

The hydroxylation of HIF-α proline residues is catalyzed by Fe- and 2-OG-dependent dioxygenase enzymes known as PHDs, whose activity is strictly oxygen-dependent. As a result, inhibiting PHD function leads to the stabilization of HIF-α, thereby activating hypoxia-associated signaling pathways even under normoxic conditions. This mechanism has been leveraged for therapeutic purposes, with small-molecule PHD inhibitors developed as clinical treatments for renal anemia. Beyond their application in anemia management, recent research has highlighted the broader medical potential of PHD inhibitors in various disease contexts. Emerging evidence suggests that modulating PHD activity may have therapeutic implications for ischemic disorders, chronic inflammatory diseases, metabolic syndromes and even neurodegenerative conditions. These findings underscore the expanding role of PHD-targeted therapies and warrant further investigation into their diverse clinical applications (70).

PHD inhibitors have also been reported in the treatment of various other diseases (70). There is increasing evidence supporting the potential of targeting the HIF pathway in acute myeloid leukemia (AML) therapy. Studies suggest that the selective PHD2 inhibitor IOX5 exerts its effects by stabilizing HIF-1α, thereby triggering a cascade of events detrimental to AML cells. By stabilizing HIF-1α, IOX5 disrupts pro-leukemogenic signaling pathways, shifts metabolic processes to a state less favorable for cancer cells and induces apoptosis through BNIP3 upregulation, highlighting its potential as a non-toxic therapeutic strategy for AML. Furthermore, when combined with BCL-2 inhibitor venetoclax, which is already used in clinical settings, IOX5 enhances anti-AML efficacy. Notably, IOX5 selectively targets PHD without interfering with other enzymes such as FIH-1, a key feature for minimizing off-target effects (71).

Roxadustat, another PHD inhibitor, has shown promise in alleviating anemia in patients with lower-risk myelodysplastic syndrome, reducing their dependence on RBC transfusions. Additionally, roxadustat has been identified as a novel therapeutic candidate for Fe-refractory Fe-deficiency anemia (IRIDA). In mouse models of IRIDA, it activates the HIF-2α-ferroportin (FPN) axis, demonstrating significant therapeutic efficacy and clinical translation potential. Mechanistically, roxadustat stabilizes HIF-2α in the duodenum, leading to FPN transcriptional activation and increased intestinal Fe absorption, independent of hepcidin levels, ultimately ameliorating hepcidin-activated anemia (72).

Moreover, in ischemia/reperfusion injury models, roxadustat preconditioning has been shown to induce ischemic tolerance by shifting aerobic respiration to anaerobic metabolism, thereby maintaining ATP production under hypoxic conditions and reducing myocardial ischemic damage (73). Additionally, roxadustat activates the HIF-1α/VEGF/VEGFR2 pathway, promoting angiogenesis, and has demonstrated therapeutic potential in diabetic wound healing in rat models (74). Furthermore, roxadustat induces the accumulation of HIF-1α and WNT7a in the tibialis anterior muscle, facilitating muscle regeneration following cyclophosphamide-induced muscle injury and promoting the formation of significantly larger muscle fibers (75). These findings suggest that roxadustat could be a promising drug for diabetic wound healing and the prevention of age-related skeletal muscle atrophy, though clinical validation is still required.

Given the broad biological functions of HIF, roxadustat's ability to stabilize HIF has opened new avenues for treating hypoxia-related diseases, including neuroprotection in nerve injury (76), oxygen-induced retinopathy (77), pulmonary fibrosis (78), acute lung injury (79), fracture healing (80) and radiation-induced apoptosis (81).

The therapeutic development of HIF-PHD modulators has primarily focused on inhibitors designed to elevate hemoglobin (Hb) levels, particularly for the treatment of renal anemia. To date, no studies have reported on optimal recovery doses or the development of PHD activators.

A comprehensive meta-analysis encompassing 30 studies and a total of 13,146 patients has assessed the long-term efficacy and safety of HIF-PHD inhibitors, including roxadustat, daprodustat, vadadustat, molidustat, desidustat and enarodustat, in the management of anemia associated with CKD. The findings demonstrate that HIF-PHD inhibitors significantly enhance Hb levels, total Fe-binding capacity, and transferrin concentrations compared with placebo or erythropoiesis-stimulating agent (ESA) treatments. Additionally, these inhibitors contribute to a reduction in cholesterol levels, indicating potential metabolic benefits.

Regarding safety, patients receiving HIF-PHD inhibitors exhibit a higher incidence of serious adverse events compared with those in the placebo group. However, the overall risk profile is comparable to that observed with ESA therapy. Adverse effects commonly associated with HIF-PHD inhibitors include diarrhea, nausea, peripheral edema, hyperkalemia and hypertension, with a higher likelihood of vomiting, headaches and thrombotic events when compared with ESA treatment. Despite these risks, HIF-PHD inhibitors remain effective in elevating Hb levels, optimizing Fe metabolism, and demonstrating favorable long-term tolerability in CKD-associated anemia. To mitigate adverse effects related to excessive Fe utilization, it is recommended that HIF-PHD inhibitors be administered in conjunction with Fe supplementation for extended treatment regimens (82).

The research studies on HIF-PHD inhibitors for renal anemia are as follows [Table II;(83–94)]: Beyond clinical trials, investigations have explored the pharmacokinetic interactions of HIF-PHD inhibitors both in vitro and in healthy volunteers. In vitro studies have demonstrated that the formation of chelates between vadadustat and Fe-containing compounds varies depending on the specific type of Fe reagent in aqueous and simulated intestinal fluid environments. Notably, when vadadustat is co-administered with oral Fe supplements, its area under the plasma concentration-time curve (AUC0-∞) and maximum plasma concentration (Cmax) are reduced due to gastrointestinal chelation. This interaction suggests that Fe supplements may impair vadadustat absorption, necessitating a dosing interval to optimize its therapeutic efficacy (95).

Table II. Clinical studies of hypoxia-inducible factor-PHD inhibitors for renal anemia.

Table II.

Clinical studies of hypoxia-inducible factor-PHD inhibitors for renal anemia.

First author/s, year	Stage trial	Study design	Object	Medicine	Results	Conclusion	(Refs.)
Kansagra et al,	Phase I	Randomized,	100 healthy	ZYAN1	Maximum concentration (Cmax) ranged from	Single (10–300 mg) and	(83)
2018	clinical	double-blind,	volunteers		566.47±163.03 to 17,858.33±2,899.19 ng/ml.	multiple (100–300 mg) doses
	trial	placebo-			After a single incremental dose of 10–300 mg	of Zyany1 were safe and
		controlled			The median time (t/MAX) to Cmax with a	well tolerated in healthy
		(evaluation			300 mg oral dose of Zyan1 was ~2.5 h.	volunteers.
		of safety,			The mean Cmax and area (AUC T) area values	The mean C Max and
		tolerability,			under the concentration-time curve from time	AUCt increased almost
		and pharma-			0 to time t showed a dose-proportional increase	proportionally with the
		cokinetics			regardless of single or multiple dosing.	dose of Zyan 1.
		after oral			The average elimination half-life (t½) is	The average concentration
		administration)			6.9–13 h and the cumulative dose was	of serum EPO showed a
					negligible.	dose-response trend.
					After a single dose of Zyany1, the maximum	Zyan 1 once every 2 days
					mean serum EPO showed a dose-response	was recommended for
					(i.e., 10 and 300 mg Zyany1 doses were	Phase II study based on
					6.6 and 79.9 Miu/l, respectively), and the	the data of T 1FI 2,
					average maximum serum EPO	pharmacodynamic activity
					concentration ranged from 10 to 72 h.	and drug accumulation.
Singh et al, 2021	Phase 3	Randomized,	2964 patients	Daprodustat	The mean (±SD) baseline Hb level was	In CKD patients with	(84)
	clinical	open-label	with dialysis		10.4±1.0 g per deciliter overall. The mean	dialysis, the daprodustat
	trial	(1,487 cases of	CKD patients		(±SE) change in the Hb level from baseline	group is not inferior to the
		daprodustat oral	of Hb levels		to weeks 28 through 52 was 0.28±0.02 g	ESAs group in changes
		administration;	ranging from		per deciliter in the daprodustat group and	in Hb levels since baseline
		1,477 cases of	80 to 115 g/l		0.10±0.02 g per deciliter in the ESA group	and outcomes of cardio-
		ESA)			[difference, 0.18 g per deciliter; 95%	vascular adverse events.
					confidence interval (CI), 0.12 to 0.24],
					which met the prespecified noninferiority
					margin of −0.75 g per deciliter.
					During a median follow-up of 2.5 years, a
					major adverse cardiovascular event occurred
					in 374 of 1487 patients (25.2%) in the
					daprodustat group and in 394 of 1477 (26.7%)
					in the ESA group (hazard ratio, 0.93; 95% CI,
					0.81 to 1.07), which also met the prespecified
					noninferiority margin for daprodustat.
					The percentages of patients with other adverse
					events were similar in the two groups.
Singh et al,	Phase 3	Randomized,	3872 patients	Daprodustat	The mean (±SE) change in the Hb level from	In non-dialysis CKD	(85)
2021	clinical	open labe	with non-		baseline to weeks 28 through 52 was	patients with anemia, the
	trial	(daprodustat:	dialysis CKD		0.74±0.02 g per deciliter in the daprodustat	daprodustat group was not
		darbepoetin	stage 3–5		group and 0.66±0.02 g per deciliter in the	inferior to the dapepoetin
		alfa 1:1)			darbepoetin alfa group [difference, 0.08 g per	alpha group in terms of
					deciliter; 95% confidence interval (CI), 0.03 to	Hb levels changes since
					0.13], which met the prespecified noninferiority	baseline and outcomes of
					margin of −0.75 g per deciliter.	cardiovascular adverse
					During a median follow-up of 1.9 years, the	events.
					first MACE occurred in 378 of 1937 patients
					(19.5%) in the daprodustat group and in 371
					of 1935 patients (19.2%) in the darbepoetin
					alfa group (hazard ratio, 1.03; 95% CI, 0.89 to
					1.19), which met the prespecified noninferiority
					margin of 1.25.
					The percentages of patients with adverse events
					were similar in the two groups.
Fishbane et al,	Phase 3	Randomized,	2133 patients	Roxadustat	Mean (95% confidence interval) Hb change	Roxadustat effectively	(86)
2022	clinical	open label	with dialysis-		from baseline was 0.77 (0.69 to 0.85) g/dl with	increased Hb in patients
	trial	(roxadustat:	dependent		roxadustat and 0.68 (0.60 to 0.76) g/dl with	with DD-CKD, with an
		epoetin alfa 1:1)	(DD) CKD		epoetin alfa, demonstrating noninferiority	AE profile comparable to
					[least squares mean difference (95% CI),	epoetin alfa.
					0.09 (0.01 to 0.18); P<0.001].
					The proportion of patients experiencing ≥1 AE
					and ≥1 serious AE was 85.0 and 57.6% with
					roxadustat and 84.5 and 57.5% with epoetin
					alfa, respectively.
Fishbane et al,	Phase 3	Double-blind	CKD stages	Roxadustat	The mean change in Hb from baseline was	Roxadustat effectively	(87)
2021	clinical	randomization	3–5 of Hb		1.75 g/dl [95% confidence interval (95% CI),	increased Hb in patients with
	trial	(roxadustat:	<10.0 g/dl		1.68 to 1.81] with roxadustat vs. 0.40 g/dl	non-dialysis-dependent CKD
		placebo 1:1)			(95% CI, 0.33 to 0.47) with placebo, (P<0.001).	and reduced the need for
					Among 411 patients with baseline elevated	RBC transfusion, with
					high-sensitivity C-reactive protein, the mean	an adverse event profile
					change in Hb from baseline was 1.75 g/dl	comparable to that of
					(95% CI, 1.58 to 1.92) with roxadustat vs.	placebo.
					0.62 g/dl (95% CI, 0.44 to 0.80) with placebo,
					(P<0.001). Roxadustat reduced the risk of
					RBC transfusion by 63% (hazard ratio, 0.37;
					95% CI, 0.30 to 0.44). The most common
					adverse events with roxadustat and placebo,
					respectively, were ESKD (21.0% vs. 20.5%),
					urinary tract infection (12.8% vs. 8.0%),
					pneumonia (11.9% vs. 9.4%), and hypertension
					(11.5% vs. 9.1%).
Chen et al,	Phase 3	Double-blind	154 patients	Roxadustat	During the primary analysis period, the mean	The roxadustat group had	(88)
2019	clinical	randomization	with non-		(±SD) change from baseline in the Hb level	a higher mean Hb level
	trial	(Roxadustat:	dialysis CKD		was an increase of 1.9±1.2 g per deciliter in the	than those in the placebo
		placebo= 2:1)	in 29 regions		roxadustat group and a decrease of 0.4±0.8 g	group after 8 weeks.
			of China		per deciliter in the placebo group (P<0.001).	Hyperkalemia and metabolic
					The mean reduction from baseline in the	acidosis occurred more
					hepcidin level (associated with greater Fe	frequently in the roxadustat
					availability) was 56.14±63.40 ng per milliliter	group than in the placebo
					in the roxadustat group and 15.10±48.06 ng per	group. The efficacy of
					milliliter in the placebo group. The reduction	roxadustat in Hb correction
					from baseline in the total cholesterol level was	and maintenance was
					40.6 mg per deciliter in the roxadustat group	maintained during the
					and 7.7 mg per deciliter in the placebo group.	18-week open-label period.
Perkovic et al,	The	Randomized	3872 cases	Daprodustat	The median baseline Hb was 9.9 g/dl, blood	Hb efficacy,cardiovascular	(89,90)
2022;	American	Open-label trial	of non-		pressure was 135/74 mmHg, and the estimated	(CV) safety and secondary
Testi et al,	CKD	(daprodustat:	dialysis CKD		glomerular filtration rate was 18 ml/min/1.73 m2.	efficacy outcomes,
2023	anemia	darbepoetin	anemia in		Among randomized patients, 53% were ESA	daprodustat is not inferior
	study	alfa 1: 1)	38 countries		non-users, 57% had diabetes, and 37% had a	to the control substance
					history of CV disease. At baseline, 61% of	darboetin alpha.
					participants were using renin-angiotensin
					system blockers, 55% were taking statins,
					and 49% were taking oral iron.
					Baseline demographics were similar to those
					in other large non-dialysis anemia trials.
Kurata et al,	Phase III		Non-dialysis-	Roxadustat	Roxadustat effectively increases and maintains	Roxadustat effectively	(91)
2022	clinical		dependent		Hb levels in both non-dialysis-dependent and	increases and maintains
	trial		and dialysis-		dialysis-dependent CKD patients. Roxadustat	Hb levels. Roxadustat is an
	expert		dependent		also improved Fe metabolism and reduced	attractive alternative
	opinion		CKD patients		intravenous (IV) Fe requirements. However,	treatment especially for
					pooled analyses of phase 3 studies have	patients with ESA
					revealed frequent thromboembolic events in	hyporesponsive due to
					the roxadustat group, which might be attributed	impaired Fe utilization.
					to rapid changes in Hb and inadequate Fe	So, the appropriate selection
					supplementation. Roxadustat is an attractive	of target patients and its
					alternative treatment especially for patients	proper use are crucially
					with ESA hyporesponsive due to impaired Fe	important.
					utilization.
Eckardt et al,	Phase 3	Random, open-	3,923 patients	Vadadustat	In the pooled analysis, the first MACE occurred	vadadustat was noninferior	(92)
2021	clinical	label, non-	with		in 355 patients (18.2%) in the vadadustat group	to darbepoetin alfa with
	trial	inferiority	occasional		and in 377 patients (19.3%) in the darbepoetin	respect to cardiovascular
		(vadadustat:	or dialysis-		alfa group [hazard ratio, 0.96; 95% confidence	safety and correction and
		darbepoetin	dependent		interval (CI), 0.83 to 1.11]. The mean	maintenance of Hb
		alfa)	(DD-CKD)		differences between the groups in the change	concentrations.
					in Hb concentration were −0.31 g per deciliter
					(95% CI, −0.53 to −0.10) at weeks 24 to 36 and
					−0.07 g per deciliter (95% CI, −0.34 to 0.19)
					at weeks 40 to 52 in the incident DD-CKD
					trial and −0.17 g per deciliter (95% CI, −0.23
					to −0.10) and −0.18 g per deciliter (95% CI,
					−0.25 to −0.12), respectively, in the prevalent
					DD-CKD trial. The incidence of serious adverse
					events in the vadadustat group was 49.7% in
					the incident DD-CKD trial and 55.0% in the
					prevalent DD-CKD trial, and the incidences
					in the darbepoetin alfa group were 56.5% and
					58.3%, respectively.
Provenzano et al,	Phase 1b	Double-blind	17 patients	Roxadustat	Maximum plasma concentration and area	Peak median endogenous	(93)
2020	clinical	placebo	with hemo-	(FG-4592)	under the plasma concentration-time curve	EPO levels were 96 mIU/ml
	trial	randomized	dialysis		for patients receiving roxadustat were slightly	and 268 mIU/ml for the 1-
		controlled study	end-stage		more than dose proportional and elimination	and 2-mg/kg doses,
		(roxadustat:	renal disease		half-life ranged from 14.7–19.4 h. Roxadustat	respectively, within
		Placebo = 3:1)			was highly protein bound (99%) in plasma, and	physiologic range of
					dialysis contributed a small fraction of the total	endogenous EPO responses
					clearance: only 4.56 and 3.04% of roxadustat	to hypoxia at high altitude or
					recovered from the 1 and 2 mg/kg dose groups,	after blood loss. No serious
					respectively. Roxadustat induced transient	adverse events were reported,
					elevations of endogenous EPO that peaked	and there were no treatment-
					between 7 and 14 h after dosing and returned to	or dose-related trends in
					baseline by 48 h after dosing.	adverse event incidence.
Akizawa et al,	Phase 3	Open-label,	Non-dialysis	roxadustat	Either the roxadustat or DA comparative group	The roxadustat dose required	(94)
2021	clinical	partially	dependent		received treatment (roxadustat, n=131; DA,	to maintain target Hb in NDD
	trial	random	(NDD)		n=131). Higher mean (standard deviation)	patients in Japan with anemia
		(roxadustat and	patients		doses of both roxadustat [63.15 (24.84) mg]	of CKD relative to DA dose
		darbepoetin	with CKD		and DA [47.33 (29.79) µg] were required in the	may not be impacted by
		alfa 1: 1)	anemia		highest ESA resistance index (≥6.8) quartile	low-grade inflammation.
					(P=0.003 and P<0.001, respectively). Patients	Roxadustat may be beneficial
					with adequate Fe repletion had the lowest doses	for ESA-hyporesponsive
					for both roxadustat [45.54 (18.01) mg] and DA	NDD CKD patients.
					[28.13 (20.98) µg]. High-sensitivity C-reactive
					protein ≥28.57 nmol/l and the estimated
					glomerular filtration rate <15 ml/min/1.73 m2
					were associated with requiring higher DA but
					not roxadustat doses.

[i] HIF, hypoxia-inducible factor; PHD, prolyl hydroxylase domain; CKD, chronic kidney disease.

Additionally, pharmacokinetic evaluations in healthy volunteers have assessed the interaction between roxadustat and warfarin. The combination has been found to be well-tolerated, with only mild treatment-related adverse events. Importantly, co-administration does not necessitate warfarin dose adjustments, indicating that roxadustat does not significantly alter warfarin metabolism or anticoagulant activity (96).

The potential impact of HIF-PHIs on tumor development and progression has been a subject of significant concern, particularly given their close association with VHL disease. VHL disease is a syndrome characterized by detectable genetic abnormalities in the VHL gene, frequently associated with tumor formation (97). HIF regulates the transcription of numerous target genes, including VEGF, a key factor that promotes tumor growth. Additionally, two Phase 3 clinical trials conducted in China have reported a higher incidence of hyperkalemia in the roxadustat group. However, further analysis of potassium levels measured by central laboratories does not confirm an increased risk of hyperkalemia in patients treated with Roxadustat (88,98).

In the TREAT trial, patients in the darbepoetin alfa group (target Hb 130 g/l) exhibit a higher risk of fatal or non-fatal stroke, venous thromboembolism, and arterial thromboembolism compared with the placebo group (where darbepoetin alfa is only administered when Hb levels fell below 90 g/l) (99). Furthermore, in 2019, the first reported case of roxadustat-induced pulmonary arterial hypertension is documented (100).

Beyond its role as an HIF prolyl hydroxylase inhibitor, roxadustat has been suggested to exert HIF-independent effects, indicating the presence of potential off-target activities. Notably, it may increase the risk of pulmonary arterial hypertension and vascular calcification, as well as exacerbate inflammation and infection (101). Although no significant off-target effects have been observed in clinical trials or real-world applications thus far, HIF regulates a vast array of genes and has pleiotropic effects. Therefore, strict control of HIF activation levels and duration is essential, and careful consideration of roxadustat's dosage and treatment regimen is warranted to minimize unintended side effects resulting from excessive HIF activation.

PHDs and erythrocytosis

Human survival depends on a continuous and adequate oxygen supply to meet cellular metabolic demands, primarily through oxidative phosphorylation, which generates ATP. HIFs play a central role in regulating gene transcription to maintain oxygen homeostasis by balancing oxygen supply and demand. The activity of HIFs is tightly regulated through oxygen-dependent hydroxylation, primarily mediated by PHD proteins and VHL proteins. Mutations in VHL, HIF-2α and PHD2 genes can lead to hereditary polycythemia, a condition characterized by excessive RBC production and PHA due to aberrantly elevated HIF activity. Furthermore, genetic adaptations involving variations in PHD2 and HIF-2 contribute to high-altitude acclimatization by reducing erythropoiesis and pulmonary vascular reactivity to hypoxia, enabling improved survival in low-oxygen environments (102).

Population genomic studies have identified EPAS1 (HIF-2α) and EGLN1 (PHD2) as the primary genes responsible for high-altitude adaptation. These genes exhibit strong associations with lower Hb levels in the Tibetan population, suggesting an evolutionary advantage in hypoxic environments (103). PHD2, encoded by EGLN1, functions as a critical oxygen sensor, facilitating HIF-α hydroxylation and subsequent degradation under normoxic conditions. In addition to its catalytic domain, PHD2 contains a highly conserved zinc finger domain, which independently interacts with HIF-α in vitro. A C36S/C42S EGLN1 knockout mutation, which disrupts the zinc finger function, leads to increased EPO gene expression, excessive erythropoiesis, and an enhanced hypoxic ventilatory response in vivo. These physiological effects are attributed to loss-of-function mutations in EGLN1, resulting in prolonged HIF stabilization and activation (104).

A gain-of-function mutation in EGLN1 (PHD2 D4E: C127S), along with polymorphisms in EPAS1 (HIF-2α), contributes to reduced Hb levels among Tibetan highlanders. While the influence of EGLN1 haplotypes on Hb concentration varies with age in low-altitude populations, Tibetan highlanders consistently exhibit lower Hb levels due to adaptation to the EPAS1 rs142764723 C/C allele. The high-altitude-adapted EGLN1 haplotypes, c.12C>G and c.380G>C, are unique genetic variations found in Tibetans, conferring resistance to polycythemia while enabling survival in chronic hypoxia. Moreover, EPAS1 haplotypes have been incorporated into the Denisovan genome, suggesting that these genetic modifications play a critical role in shaping the lower Hb concentration observed in Tibetans living on the plateau (105).

The EPAS1 variant prevalent in the Tibetan population has been shown to reduce gene expression in human umbilical endothelial cells and the placenta. Additionally, mice with heterozygous EPAS1 knockout exhibit a delayed physiological response to prolonged hypoxia. This Tibetan EPAS1 variant is associated not only with lower Hb levels but also with a diminished pulmonary vasoconstriction response. These findings suggest that EPAS1 downregulation serves as the molecular basis for Tibetan adaptation to high-altitude hypoxia, minimizing the risk of chronic mountain sickness while optimizing oxygen utilization in a low-oxygen environment (106).

PHD2 serves as a key regulator of HIFs, mediating their degradation and thereby controlling physiological responses to hypoxia, including RBC production. The PHD2 p.[Asp4Glu; Cys127Ser] variant exhibits lower oxygen Km values, which enhances the degradation of HIFs even under low-oxygen conditions. In wild-type erythroid progenitor cells, hypoxia typically stimulates proliferation; however, in patients carrying the EGLN1 c.[12C>G; 380G>C] mutation, progenitor cell proliferation is significantly suppressed under hypoxic culture conditions. This mutation, originating from a common haplotype ~8,000 years ago, has been linked to high-altitude adaptation. Specifically, the EGLN1 c.[12C>G; 380G>C] variant prevents the hypoxia-induced, HIF-mediated increase in erythropoiesis, thereby acting as a molecular defense mechanism against high-altitude erythrocytosis in Tibetans. This genetic adaptation plays a crucial role in maintaining normal erythrocyte levels in high-altitude populations, allowing them to thrive in oxygen-deprived environments (107).

Whole-genome sequencing of 10 Tibetan patients with high-altitude polycythemia (HAPC) has identified two significant genetic loci associated with erythrocytosis: the egln3/phd3 homolog (14q13.1, rs1346902) and the PPP1R2P1 (protein phosphatase 1 regulatory inhibitor subunit 2) gene (6p21.32, rs521539). The egln3/phd3b gene regulates HIF-α stability, while PPP1R2P1 is involved in ROS homeostasis and oxidative stress regulation (108).

Furthermore, the zinc finger domain of PHD2 binds specifically to a Pro-Xaa-Leu-Glu (PXLE) motif found in the ribosomal chaperone nascent polypeptide complex-α (NACA). This interaction facilitates the recruitment of PHD2 to the translation machinery, enabling co-translational hydroxylation of HIF-α. Notably, this co-translational modification is enhanced by the presence of a translational pause sequence within HIF-α, which increases PHD2's efficiency in targeting the protein. Mice carrying a mutation in Naca that eliminates the PXLE motif exhibit erythrocytosis, highlighting the functional importance of this co-translational regulatory mechanism in RBC production (109).

Based on the aforementioned theoretical framework, HIF stabilization may be detrimental in various disease conditions, including certain cancers and PHA. Consequently, pharmacological strategies to inhibit HIFs have been proposed. Early pharmacological interventions have focused on suppressing HIF expression or promoting HIF degradation. More recently, small-molecule inhibitors such as PT2977 have been developed to block the formation of transcriptionally active HIF-2α/HIF-1β heterodimers. Depending on the specific HIFα isoform being targeted, pharmacological interventions can be classified into non-specific inhibitors targeting both HIF-1α and HIF-2α or isoform-selective inhibitors specifically targeting either HIF-1α or HIF-2α.

Non-specific HIF inhibitors include 2-methoxyestradiol, digoxin, and berberine acetate. However, these compounds have faced significant limitations in clinical development. 2-Methoxyestradiol, an endogenous estrogen metabolite, has a limited clinical window; digoxin requires HIF inhibition at toxic concentrations due to its narrow therapeutic index; and berberine acetate has potential side effects and a short half-life, preventing further clinical progress (110).

Selective HIF-1α inhibitors include topotecan (NSC-609699), PX-478 (S-2-amino-3-[4′-N,N-bis(chloroethyl)amino] phenylpropionic acid N-oxide dihydrochloride), cyclic peptide inhibitors (cyc-CLLFVY), and locked nucleic acid antisense oligonucleotides such as EZN-2968/RO7070179. While these compounds demonstrate potential therapeutic benefits, further research is required to establish their safety and efficacy in humans (110).

Selective HIF-2α inhibitors include C76 (methyl 3-{2-[cyano(methylsulfonyl)methylene]hydrazinyl}thiophene-2-carboxylate), and PT2385, the first clinical HIF-2α inhibitor. Later, the FDA approved PT2977 (belzutifan, MK-6482) for the treatment of ccRCC in patients with VHL disease. Additionally, several HIF-2α inhibitors from Nikang (NKT2152), Novartis (DFF332), and Arcus (AB521) are in various stages of clinical development (110).

Given that the EPAS1A529V mutation is a key driver of Pacak-Zhuang syndrome, belzutifan has been investigated as a potential therapeutic option for a single patient with severe polycythemia, hypertension and multiple PGL (110). Targeting HIF-2α may offer therapeutic benefits for PHA, retinal neovascularization and polycythemia.

While significant advancements have been made in HIF inhibition, research on enhancing PHD enzyme activity to counteract HIF accumulation remains scarce. This area presents challenges similar to small-molecule p53 reactivators, which are further complicated by the diverse isoforms and subcellular localization of PHDs. One promising future direction involves RNA-based therapies, including siRNA and gene editing technologies, alongside the development of targeted delivery mechanisms. These approaches could enable precise modulation of HIF activity while minimizing off-target effects on unrelated tissues or cell types, providing a novel avenue for HIF activation or inhibition in disease therapy.

Prospects of PHD-targeted therapies for high-altitude polycythemia

The link between polycythemia and high-altitude adaptation was first reported over a century ago, with subsequent discoveries highlighting the central role of HIF-1 in oxygen homeostasis. As a heterodimeric transcription factor composed of an α and β subunit, HIF-1 is primarily regulated by oxygen-dependent post-translational hydroxylation of its α subunit. This regulatory mechanism is mediated by non-heme Fe hydroxylases, which specifically hydroxylate the proline and asparagine residues of HIF-1 through dioxygenation, utilizing 2-OG as a co-substrate. Three PHD enzymes and FIH-1 function as essential cellular oxygen sensors, fine-tuning HIF activity in response to fluctuating oxygen levels (111).

Genetic studies of high-altitude Tibetans have identified EPAS1 (HIF-2α), EGLN1 (PHD2), and PPARA as key genes shaped by natural selection for high-altitude adaptation. Among these, HIF signaling serves as the primary regulator of erythropoiesis and other critical physiological functions (112). Hypoxia-induced responses, spanning cardiovascular, pulmonary and metabolic pathways, provide crucial insights into disease mechanisms. Understanding these adaptive pathways may pave the way for targeted therapies for hypoxia-related conditions, including cardiopulmonary diseases, neurodegenerative disorders, metabolic syndromes such as diabetes and obesity, and high-altitude illnesses. To date, no PHD activators or functional restorers have been developed to enhance HIF-α degradation, thereby attenuating downstream gene expression and mitigating associated pathological effects. In Tibetans, genetic variants of EPAS1 downregulate EPAS1 transcription, serving as a protective mechanism against erythrocytosis. This natural adaptation helps maintain relatively low Hb levels and is associated with reduced pulmonary vasoconstriction responses (106). Building upon these genetic insights, strategies aimed at downregulating HIF-α transcription or enhancing PHD enzymatic activity could provide therapeutic avenues for HAPC and hypoxia-induced PHA. Such an approach may help maintain optimal RBC counts, improve oxygen-carrying capacity, balance erythropoiesis and apoptosis, and reduce complications arising from increased blood viscosity, ultimately alleviating patient symptoms and improving prognosis.

Despite the advancements in developing targeted inhibitors, the development of PHD activators or functional restorers remains a significant challenge (113). Although small-molecule p53 reactivators have been explored in clinical trials for restoring the function of certain p53 mutations, similar approaches have yet to be translated into clinical applications for hypoxia-related diseases (114).

Theoretical models suggest that HIF-α inhibitors or PHD functional restorers could be valuable in treating HAPC by counteracting excessive erythropoiesis. While extensive research has been conducted on HIF inhibitors in oncology, no studies have specifically investigated HIF-α suppression as a therapeutic strategy for HAPC. Furthermore, efforts to enhance or restore PHD enzyme function face formidable technological challenges, as identifying compounds with strong biological efficacy suitable for clinical development remains difficult.

Additionally, both HIFs and PHDs exhibit subtype-specific effects in hypoxic and inflammatory environments, adding another layer of complexity to drug development. However, promising advancements in targeting hypoxia-driven inflammatory pathways suggest that novel therapeutic strategies are on the cusp of entering clinical trials (69).

Acknowledgements

Not applicable.

Funding

The present study search was supported by the Basic Research Project of Qinghai Provincial Department of Science and Technology (grant no. 2025-ZJ-730) and the Clinical Research Center for Blood System Diseases of Qinghai (grant no. 2021-SF-136). Additional funding was provided by the National Natural Science Foundation of China (grant no. U24A20749) and the 2022 ‘Kunlun Yingcai’ Advanced Innovation and Entrepreneurship Top-notch Talents Project in Qinghai.

Availability of data and materials

The data generated in the present study may be requested from the corresponding author.

Authors' contributions

YH, ZHZ and YBX drafted the manuscript. WQL, GXH and KS conducted to the literature search, data organization and edited content. YBX was responsible for the structure, concept and design of the article, as well as proofreading the version to be published. Data authentication is not applicable. All authors read and approved the final version of the manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References