Hypoxia‐inducible factors not only regulate but also are myeloid‐cell treatment targets

Hypoxia describes limited oxygen availability at the cellular level. Myeloid cells are exposed to hypoxia at various bodily sites and even contribute to hypoxia by consuming large amounts of oxygen during respiratory burst. Hypoxia‐inducible factors (HIFs) are ubiquitously expressed heterodimeric transcription factors, composed of an oxygen‐dependent α and a constitutive β subunit. The stability of HIF‐1α and HIF‐2α is regulated by oxygen‐sensing prolyl‐hydroxylases (PHD). HIF‐1α and HIF‐2α modify the innate immune response and are context dependent. We provide a historic perspective of HIF discovery, discuss the molecular components of the HIF pathway, and how HIF‐dependent mechanisms modify myeloid cell functions. HIFs enable myeloid‐cell adaptation to hypoxia by up‐regulating anaerobic glycolysis. In addition to effects on metabolism, HIFs control chemotaxis, phagocytosis, degranulation, oxidative burst, and apoptosis. HIF‐1α enables efficient infection defense by myeloid cells. HIF‐2α delays inflammation resolution and decreases antitumor effects by promoting tumor‐associated myeloid‐cell hibernation. PHDs not only control HIF degradation, but also regulate the crosstalk between innate and adaptive immune cells thereby suppressing autoimmunity. HIF‐modifying pharmacologic compounds are entering clinical practice. Current indications include renal anemia and certain cancers. Beneficial and adverse effects on myeloid cells should be considered and could possibly lead to drug repurposing for inflammatory disorders.


INTRODUCTION
Myeloid cells consist of granulocytes, mostly neutrophils, and monocytes. Once released from the bone marrow, these cells circulate in the blood and are recruited to inflammatory sites where they execute functions that protect the host from infectious and noninfectious challenges. Myeloid cells perform efficiently even under hostile conditions, such as extreme temperatures, mechanical and osmotic stress, and Abbreviations: (p)VHL, von-Hippel-Lindau (gene product); ARNT, aryl hydrocarbon receptor nuclear translocator; BCL-2, B-cell lymphoma 2; bHLH, basic helix-loop-helix; CBP, CREB binding protein; CGD, chronic granulomatous disease; C-TAD, C terminal transactivation domain; egl-9, egg-laying defective nine; EGLN, egg-laying defective nine homologue; EPO, Erythropoietin; FIH-1, factor inhibiting HIF-1; GSD1b, glycogen storage disease 1b; HIF, Hypoxia-inducible factor; HRE, hypoxia-response element; Hsp, heat shock protein; MMP-9, matrix metalloproteinase-9; NE, neutrophil elastase; N-TAD, N terminal transactivation domain; ODD, oxygen-dependent degradation domain; PAS, Per-Arnt-Sim; PHD, prolyl hydroxylase domain containing enzyme; PPAR , peroxisome proliferator-activated receptor ; RACK1, receptor of activated protein kinase C; ROS, reactive oxygen species; SDHB, succinate dehydrogenase B; TAM/TAN, tumor-associated macrophage/neutrophil; T reg , regulatory T cell. gradient with >70 mm Hg in the cortex and 10 mm Hg in the medulla. 4 Importantly, hypoxia is also characteristic of inflamed tissues, 5,6 where activated myeloid cells migrate against a low oxygen supply. 7 With reduced oxygen supply, mitochondrial oxidative phosphorylation is strongly decreased, and most ATP is provided by the conversion of pyruvate into lactate. 7 Hypoxia-inducible factors (HIFs) are ubiquitous transcriptional regulators of gene expression in response to low oxygen availability. HIFs help myeloid cells to cope with low oxygen conditions by modifying several metabolic and inflammatory aspects.
Currently, drugs that either activate or inhibit HIF-mediated effects are being explored in clinical studies. We discuss the HIF system with its implications for myeloid cell functions together with the potential effects of HIF-directed treatments.

HISTORIC PERSPECTIVE OF HIF DISCOVERY
The scientific interest in hypoxia dates back to Paul Bert, who identified hypoxemic hypoxia as the cause of altitude sickness in the second half of the 19th century. 8,9 Over decades, hypoxia-mediated effects on erythropoiesis became a main research focus that more recently extended to inflammation. However, it took more than a century until molecular hypoxia mechanisms were characterized, the term HIF was introduced 10 and pharmacologic HIF modulators were developed along a timeline outlined in Fig. 1.
In 1906, Carnot reported a serum factor extracted from anemic rabbits that stimulated erythropoiesis in recipient animal bone marrow and termed this putative factor "hémopoïétine." 11,12 Yet, the nature of this factor remained elusive for several decades. In the 1950s , Jacobson suggested that erythropoietin (EPO) was secreted by the kidneys. 13 The investigators observed that bilateral nephrectomy abrogated the erythropoietic effect of CoCl 2 in rats and rabbits. 14 Later, the liver was identified as an additional extrarenal EPO source. [15][16][17][18][19][20][21][22] Most clinical observations supported the importance of the kidneys for erythropoiesis with complete erythroblastopenia in anuric renal failure patients 23 and after nephrectomy, 24 and polycythemia in patients with renal pathologies, such as renal cysts, hypernephroma, or hydronephrosis. [25][26][27][28][29][30][31][32][33] In the 1970s, EPO was isolated from the urine of anemic patients [34][35][36] followed by cloning and recombinant expression a decade later. [37][38][39][40][41][42] Allan Erslev and his research led to the discovery of EPO. 43 The hormone causes the body to make more red blood cells and is now the pivotal drug to treat anemia caused by cancer therapy, dialysis, and kidney disease. Erslev made rabbits anemic. When he injected their anemic plasma into normal rabbits, the rabbits increased production of red blood cells and as the number of red blood cells increased so hematocrit increased. In contrast, injection of normal plasma into normal rabbits did not lead to an increase in red blood cells. Erslev concluded that a hormone (EPO) was responsible for the increase in red blood cells. Yet, the molecular mechanisms of hypoxia-regulated EPO transcription and, as it later turned out, many additional genes were still unknown.
In the late 1980s, Goldberg et al. suggested a ferroprotein as the oxygen sensor. 44 The investigators used metals (i.e., manganese, nickel, [45][46][47][48] and cobalt as CoCl 2 ) that interact with protoporphyrin structures and compete with iron in heme prosthetic groups to induce EPO. 44,49 Locking heme-bearing proteins in a deoxy conformation with these metals induced EPO mRNA and protein. Hypoxia was not synergistic with this deoxy state, whereas carbon monoxide, which created an oxy state of hemoglobin, reduced the EPO-enhancing effect of hypoxia. Together, these experiments led to the reasonable assumption that the cellular oxygen sensor is a heme protein. However, as it turned out later, the free iron was bound to a nonheme protein that was yet to be discovered.
In the 1990s, reporter assays unmasked cis-acting elements responsive to hypoxia. 50,51 Transgenic mice carrying the human EPO gene produced nuclear factors selectively binding to 3 ′ flanking sequences of the human EPO transgene. 10 Consecutive mutational analysis of a 50 nt 3 ′ flanking sequence of the human EPO gene revealed a proteinaceous DNA binding that the authors termed HIF-1. 10 HIF-1 was characterized as a protein complex generated in hypoxic cells that binds to a DNA sequence crucial for hypoxic activation of EPO transcription. 52,53 More than 10 yr passed until prolyl-hydroxylase domain containing enzymes (PHD) were finally identified as the long-assumed sensor of cellular oxygen tension that regulate HIF abundance. Although the earlier suggested iron-binding domain was confirmed, the implied heme-involving mechanism was not. [54][55][56][57] In 2019, William G. Kaelin Jr., Sir Peter J. Ratcliffe, and Gregg L. Semenza received the Nobel Prize for explaining how cells sense and adapt to different oxygen concentrations.

HIF subunit isoforms and dimerization with aryl hydrocarbon receptor nuclear translocator (ARNT)
HIFs are transcription factors with an N-terminal basic helix-loophelix (bHLH) followed by a Per-ARNT-Sim (PAS) domain 58

Oxygen-dependent HIF regulation by proteasomal degradation
The HIF isoforms and HIF-1 are all constitutively transcribed and translated. The cellular abundance of the former, but not the latter, is controlled by oxygen concentration. Under normoxic conditions, the HIF proteins reside in the cytoplasm where they interact with heat shock protein (Hsp)90 through the bHLH-PAS domain. [73][74][75] However, under normoxia, HIF proteins are continuously degraded in the proteasome mediated by the von-Hippel-Lindau (pVHL) protein. 76 pVHL serves as a substrate-recognizing subunit of an E3 ubiquitin ligase complex 54,77,78 and forms a ternary complex with elongins B and C 79 thereby recruiting Cul-2 and Rbx-1. [80][81][82] The resulting multimeric complex acquires E3 ligase activity and, in concert with the E1 ligase Uba and the E2 ligases Ubc5a, Ubc5b, and Ubc5c, 83 leads to oxygen-and iron-dependent ubiquitination and subsequent proteasomal degradation of the HIF subunits. 57,84,85 Mass spectrometry established that hydroxylation of proline residues within the HIF ODD was indispensable for pVHL recognition of degradation-designated HIF subunits under normoxic conditions. Subsequently, new dioxygenase isoforms were identified that were responsible for posttranslational oxygen-dependent HIF hydroxylation. 55 Thus, the cellular oxygen sensor was finally characterized as PHD. C. elegans expresses a HIF system that is homologous HLH PAS-A PAS-B ODD C-TAD N-TAD b F I G U R E 2 Schematic of hypoxia-inducible factor (HIF)-1 and HIF-2 protein structure and hydroxylation sites at proline and asparagine residues. The basic submotif and the helix-loop-helix domain (bHLH) are located close to the N terminus, followed by the Per-ARNT-Sim (PAS) domain. The PAS domain comprises repetitive amino acid sequences PAS-A and -B. The oxygen-dependent degradation domain (ODD) overlaps with the N terminal transactivation domain (N-TAD), followed by the C terminal transactivation domain (C-TAD). Hydroxylation of proline residues within the ODD and of asparagine residues within the C-TAD of HIF-1 and HIF-2 are highlighted. The nonequilibrium hydroxylation by the prolylhydroxylases (PHD) and the asparagine hydroxylase factor inhibiting HIF-1 (FIH-1) including substrates and products is depicted exemplarily for two of the three hydroxylation sites of HIF-1

F I G U R E 3 Hypoxia-inducible factor (HIF) pathway components and regulating mechanisms in myeloid cells.
HIF and HIF subunits are constitutively transcribed and translated. The HIF-1 antisense strand transcript ahif decreases HIF protein expression. In hypoxia, nonhydroxylated HIF dimerizes with HIF and binds to hypoxia-response elements (HRE) enhancing target gene transcription upon recruitment of cotransactivators, including CREB-binding protein (CBP) and p300. Under normoxia, prolyl-hydroxylase domain-containing enzymes (PHD) hydroxylate HIF subunits, thus enabling von-Hippel-Lindau gene product (pVHL) binding to HIF subunits and subsequent proteasomal degradation. Factor inhibiting HIF-1 (FIH-1)-mediated HIF hydroxylation under normoxia prevents HIF association with indispensable co-transactivators for target gene transcriptional enhancement. Phosphorylated receptor of activated protein kinase C (RACK1) induces HIF proteasomal degradation, which is inhibited by endothelin-receptor-dependent calcineurin activation. NO inhibits PHD activity. Hypoxia, infectious (gram-positive bacteria as blue circles, gram-negative bacteria in red) and noninfectious (endothelin in violet and ET receptor) stimuli activate the HIF pathway in myeloid cells to humans and was instrumental in PHD characterization. The egl-9 gene, so named because of a presumed egg-laying defect of the genedeficient worm, encodes an oxygen-dependent prolyl-hydroxylase and egl-9-deficient mutants up-regulated the human HIF homolog constitutively. 56 Subsequently, three human PHD isoforms were identified, encoded by the genes EGLN1 (egg-laying defective nine homolog 1), EGLN2, and EGLN3, respectively. 86 A conserved 2-histidine-1carboxylate motif serves as iron-binding structure. 56 The catalyzed proline-4-hydroxylation requires dioxygen, divalent iron (Fe 2+ ), and the co-substrates 2-oxoglutarate and ascorbate. 55 concert HIF-1 lysosomal degradation by cyclin-dependent kinase regulated chaperone-mediated autophagy. 95,96 The HIF subunit specific E3 ubiquitin ligases hypoxia-associated factor 97 and mammary tumor integration site 6 (Int6) 98 initiate HIF-1 , and HIF-2 proteasomal degradation irrespective of oxygen tension and pVHL, respectively.

Oxygen-dependent HIF regulation by transcriptional inhibition
In addition to directing HIF degradation, oxygen controls the transactivation efficacy of HIF heterodimers by factor inhibiting HIF-1 (FIH-1). 99 FIH-1 is an asparaginyl-hydroxylase belonging to the same oxygen-and 2-oxoglutarate-dependent dioxygenase superfamily as the PHDs. 100 However, FIH-1 activity persists even under hypoxic conditions of 1% oxygen when PHD2 activity is abolished. 101 Under normoxic oxygen tensions, FIH-1 hydroxylates Asn803 of HIF-1 , and Asn851 of HIF-2 , respectively. 102 These hydroxylation sites are located within the C-TAD. Their hydroxylation prevents indispensable co-transactivator recruitment that initiate target gene transcription.

HIF-regulated target genes and signaling pathways
CREB-binding protein (CBP) and p300 bind to HIF / heterodimers with nonhydroxylated asparagine residues in the C-TAD. 101,103,104 All HIF isoforms recognize the same HRE 5 ′ -TACGTG-3 ′ motif, but result in unique differential target gene expression 105 with the N-TAD determining target gene selectivity. 106,107 Nonetheless, comparison of cell-specific target gene regulation highlighted the importance of the cell type for the HIF-controlled transcriptome. 106,107 HIF-1 primarily controls metabolic pathways, including adaptation to anaerobic energy supply by up-regulating glycolysis and the hexose monophosphate pathway. [108][109][110] These effects facilitate cell survival in low-oxygen conditions. In addition, HIF-1 regulates apoptosisrelated genes, for example, members of the B-cell lymphoma 2 (BCL-2) family, 111 and proinflammatory genes including IL-1 , 112 IL-6, 113 and IL-8. 114 HIF-1 also prevents excessive cellular reactions to hypoxia by up-regulating PHD transcription. 115,116 In contrast, HIF-2 fine-tunes embryonic development and cellular differentiation. 105 Despite the initial discovery of HIF-1 in the context of EPO expression, we and others have shown predominant EPO transcriptional control by HIF-2. 117,118

HIF-CONTROLLED MYELOID CELL FUNCTIONS IN HUMANS
Human neutrophils express all PHD isoforms 119 and do not express HIF-1 protein under normoxia. 120 However, neutrophil HIF-1 protein is induced at low oxygen tensions. 120 Some, 120 but not all studies 121 described HIF-2 expression in human granulocytes under normoxia with preserved response to hypoxia. Possibly, differences are explained by the use of different culture media as for example the presence of the NO donor GEA3162 120 inhibits PHD activity. 122,123 In human monocytes, hypoxic induction of the HIF-1 natural antisense transcript ahif contributes to a negative feedback mechanism on HIF-1 activity. 124 Evolutionary adaptation of high-altitude populations, as well as monogenetic mutations affecting the HIF pathway, provide insight in HIF-controlled mechanisms that help myeloid cells to cope with low oxygen concentrations and to maintain their functions. Additional information comes from human individuals or isolated myeloid cells exposed to hypoxia.

Adaptive genetic variations in HIF pathway components provide an opportunity to study consequences for myeloid cell functions
Despite living above 4000 m, Tibetan communities, in contrast to communities residing at similar altitudes in the Andes, have mostly normal red-blood-cell and hemoglobin values. A missense mutation in the EGLN1 gene results in a PHD2 variant with a lower K m and higher V max value for oxygen. 125,126 Consequently, HIF hydroxylation is facilitated even under hypoxia. Other studies in Tibetans correlated SNPs in EPAS1 (HIF-2 ) with hemoglobin levels [127][128][129] and SNPs in the EGLN3 (PHD3) and PPP1R2P1 (protein phosphatase 1 regulatory inhibitor subunit 2) genes with altitude polycythemia. 130 However, these Tibetan adaptations of the HIF pathway provide an interesting opportunity to study myeloid cell functions, immunity, and inflammatory disorders.
Chuvash polycythemia, named after the Chuvash republic in Russia, is another endemic genetic variation of the HIF pathway accompanied by elevated EPO and VEGF plasma levels. 131 The C598T base exchange in the third VHL exon causes a missense mutation (R200W) 132 that stabilizes predominantly HIF-2 over HIF- 1 . 133 Th1 (IL-2, IL-12, IFN , TNF , GM-CSF) and Th2 cytokine (IL-4, IL-5, IL-10, IL-13) plasma levels in affected individuals were found to be elevated together with decreased CD4 + T-cell frequency and reduced CD4/CD8 ratio. 132 Transcriptome analysis in PBMCs from Chuvash polycythemia patients showed up-regulated HIF target genes involved in the inflammatory response (TNF , IL-1 , TLR4) as well as in myeloid cell differentiation, phagocytosis, and bacterial defense (FCGR2A, HCK, GAB2, ITGB). 131 Pro-apoptotic genes (CASP8, CASP2) and TCR elements were down-regulated. 131 The reasons for the apparent discrepancy between TCR down-regulation found in this 131

Genetic diseases highlight the interplay of metabolism and HIF pathway components
Patients with VHL syndrome harbor heterozygous germline VHL mutations predisposing to hemangiomas, paragangliomas, and renal carcinomas. Neutrophils from these patients showed decreased spontaneous apoptosis as well as increased phagocytic activity against bacteria. 134X Hypoxia further enhanced these functions in both VHL neutrophils and cells from healthy controls. 134 136 In some of the GSD1b patients, constitutive neutrophil HIF-1 stabilization, attributed to the Hsp90 and ROS increase, was observed. 137 Nevertheless, the metabolic impairment in GSD1b neutrophils led to accelerated constitutive apoptosis, reduced respiratory burst, phagocytosis, and chemotaxis despite stabilized HIF-1 . 135,136 As expected, HIF stabilization improves cellular energy supply that is indispensable for neutrophil survival and functioning. In fact, HIF-1 target genes, including peroxisome proliferator-activated receptor (PPAR ), were up-regulated in GSD1b neutrophils. PPAR up-regulation contributed to neutrophil dysfunction because PPAR inhibition improved chemotaxis and the respiratory burst. 137 Accordingly, neutrophils isolated from healthy controls mimicked GSD1b-associated neutrophil dysfunction upon pharmacologic HIF stabilization and PPAR activation by rosiglitazone. 137 These data suggest that HIFs control myeloid cell functions not only by providing cellular energy supply but also through PPAR activation.
Succinate is a powerful PHD inhibitor inasmuch as it is an end product of the hydroxylation reactions mediated by PHDs. 90
Together, these experiments indicate that hypoxia exposure of humans, which leads to HIF stabilization, enhances inflammatory myeloid cell functions. However, these observations cannot establish that HIFs play a causal role in this process. Another caveat is that although the blood donors were exposed to hypoxia, isolated myeloid cells were studied under normoxia. Conceivably, normoxia led to rapid HIF degradation of in vivo stabilized HIFs, whereas HIF-induced effects on transcription and metabolism may have persisted.

Myeloid-cell exposure to hypoxia in vitro prolongs survival and increases activation responses
Several studies analyzed neutrophils that were isolated from normoxic donors and exposed to hypoxia in vitro. McGovern and coworkers found that hypoxic culture of human neutrophils did not affect the secretion of IL-6, IL-8, TNF , or IL-10, 151 whereas ROS-dependent bactericidal activity was reduced. Limited molecular oxygen leading to reduced NADPH oxidase-dependent respiratory burst was the possible explanation for the latter observation. In contrast, ROSindependent killing by hypoxic neutrophils was increased. 151 Hypoxia augmented the release of granule proteins from activated neutrophils as shown for NE, 151 myeloperoxidase, lactoferrin, and matrix metalloproteinase-9 (MMP-9). 152 Consequently, supernatants from activated hypoxic neutrophils caused more epithelial cell damage compared to normoxic neutrophils. 152 The hypoxic degranulation increase was reduced by a selective PI3K inhibitor that abrogated the hypoxic degranulation augmentation. 152 The fact that pharmacologic HIF stabilization by PHD inhibitors did not mimic augmented degranulation and increased epithelial cell injury seen with hypoxic neutrophils, questions a causal role for HIF but does not exclude involvement of components upstream from HIF mediated by PHDs or FIH-1. Hypoxia increased efferocytosis, the phagocytosis of apoptotic neutrophils by monocytes or macrophages. This effect was, at least in part, mediated by HIF-1 -dependent induction of the class B scavenger receptor CD36 and its ligand thrombospondin-1 conveying apoptotic material. 166 HIF-mediated CD36 induction is supported by the observation that CD36 and HIF-1 expressing macrophages correlated in biopsies from patients with inflammatory bowel disease. 166 Altogether, these studies support the notion that hypoxia exposure in vitro prolongs myeloid cell survival and promotes proinflammatory responses that are important for host defense. The exact role of HIFs in these adaptive processes remains unclear and needs to be addressed in animal studies that allow HIF manipulations.

Hypoxic modulation of myeloid cells controls inflammation in animals
Various animal models were employed to study oxygen-dependent modifications of myeloid cell-driven inflammation. In rats, hypoxic preconditioning protected the animals from gastrointestinal ischemiareperfusion injury, including bacterial translocation. 167 Neutropenic rats lacked the protective effect suggesting a neutrophil-dependent mechanism. 167 In agreement with this suggestion, neutrophils consumed oxygen during respiratory burst thereby creating a hypoxic environment for epithelial cells that promoted HIF stabilization and induction of HIF target genes. As a result, the epithelial barrier was increased. 168 Neutrophils from NADPH oxidase gene-deficient chronic granulomatous disease (CGD) mice are unable to produce superoxide anions and therefore did not create a hypoxic environment. Consequently, CGD neutrophils did not increase the epithelial barrier and CGD mice displayed a more severe phenotype of chemical colitis. 168 Further evidence supporting HIF driven epithelial barrier stabilization is provided by murine colitis models showing beneficial effects of PHD inhibitors. [168][169][170][171][172]

Inflammatory and mechanical challenges imitate oxygen-dependent HIF stabilization
Myeloid cell studies in animals elucidated HIF stabilizing mechanisms above and beyond hypoxia. Bacterial antigens from gram-negative and gram-positive species 173,174 as well as TLR4 stimulation by LPS 158 stabilized HIFs in myeloid cells. Mechanistically, combined PHD downregulation and up-regulation of HIF transcription were suggested to increase HIF proteins. 162,173,174 More recently, HIF induction by physical forces was reported in murine bone marrow-derived monocytes.
Cyclical hydrostatic pressure, for example, due to in-and expiration, activated the monocytic ion channel PIEZO1 leading to paracrine endothelin-1 secretion. Subsequently, endothelin receptor stimulation activated calcineurin that dephosphorylated receptor of activated protein kinase C (RACK1). 175 Phosphorylated RACK1 competes with cytoplasmic Hsp90 for binding HIF subunits and promotes their proteasomal degradation. 176,177 Finally, mechanotransduced RACK1 dephosphorylation contributed to HIF-1 protein accumulation. 175

HIF-1 improves myeloid cell functions in infectious and noninfectious inflammation models
Myeloid-specific HIF-1 gene deletion severely reduced intracellular ATP concentrations in murine macrophages and neutrophils leading to reduced intracellular bactericidal activity, 173 adhesion, and motility of monocytes. 7 HIF-1 gene-deficient murine neutrophils displayed decreased NE and cathepsin G activities that were restored by VHL deletion, hence, by constitutive HIF-1 stabilization. 173 Likewise, pharmacologic HIF stabilization by PHD inhibition enhanced monocyte bactericidal properties in murine skin abscesses by inducing monocytic cathelicidin and IL-8 production. 178 113 We demonstrated HIF-1 and HIF-2 up-regulation in human psoriatic skin lesions. 179 Compared to control mice, myeloid-specific HIF-1 gene deficiency caused ameliorated leukocyte skin infiltration in bacterial 173  However, myeloid HIF pathway activation as well as myeloid HIF deficiency did not affect the rheumatoid arthritis-associated uveitis phenotype in an intravitreal LPS-induced mouse model. 187 Conceivably, this discrepancy is due to the divergent inflammatory stimuli underscoring the importance of the inflammatory context.

HIF-2 mitigates myeloid cell destructive capacity, but prolongs inflammation
In contrast to HIF-1 deletion, myeloid-specific HIF-2 gene deficiency in mice did not affect ATP generation in macrophages. 188 However, macrophage motility and tissue infiltration were significantly diminished accompanied by down-regulation of chemokine receptor CXCR4 and fibronectin-1. 188 Secretion of proinflammatory cytokines IL-1 , IL-6, IL-12, TNF , and CXCL2 following stimulation with IFN or LPS was significantly decreased, whereas these mice up-regulated antiinflammatory IL-10 upon LPS injection. 188 Studies in a murine LPS-induced acute lung injury model revealed differential effects of the HIF subunits during neutrophil-mediated inflammation. A HIF-2 gain-of-function mutation did not affect neutrophil effector functions such as oxidative burst and phagocytosis but decreased constitutive apoptosis similar to what was aforementioned for HIF-1 . 120 Acute pulmonary inflammation predominantly induced neutrophil HIF-1 at an early stage, whereas HIF-2 induction was prominent during the resolution phase. Myeloid HIF-2 gene deficiency shortened and alleviated pulmonary inflammation, particularly in later stages of acute lung injury, presumably by increased neutrophil apoptosis. 120 Inflammation models in zebrafish with HIF-2 gain-of-function mutation and myeloid-specific HIF-2 gene-deficient mice further underscored the fact that HIF-2 prolongs inflammation. 120,182,188 We previously demonstrated that HIF-2 controls EPO transcription. 117 In addition to its role in erythropoiesis, EPO was proposed to have anti-inflammatory effects. In a murine peritonitis model, hypoxia induced EPO as well as EPO receptors (EPOR) on infiltrating macrophages. 189 Macrophage EPOR signaling led to PPAR activation thereby inducing anti-inflammatory cytokines, down-regulating proinflammatory cytokines, enhancing macrophage efferocytosis and phagocytosis. CGD mice that were unable to mount a respiratory burst and therefore did not consume oxygen failed to develop peritoneal hypoxia. Consequently, endogenous EPO was not induced. 189 Exogenous EPO therapy restricted peritoneal inflammation. The authors discuss this observation as a consequence of HIF-1 , despite the fact that EPO is rather a HIF-2 target gene.
Tumor-associated macrophages and neutrophils (TAM and TAN) contribute to the progression of solid tumors. We showed previously that these cells also promoted chronic lymphatic leukemia in a murine disease model and that selective depletion of myeloid subpopulations retarded leukemia progression. 190 TAM express HIF-2 160 and several observations suggest that HIF-2 restrains their anticancer effects. Thus, HIF-2 flox/flox : LysM-Cre mice developed fewer chemically induced colon carcinomas. 188 In contrast, the number and size of chemically induced hepatocellular carcinomas were not reduced in HIF-2 -deficient mice, but both tumor entities demonstrated lower grading, delayed tumor progression, and decreased mitotic indices compared to wild-type (WT) mice. 188

Genetic PHD deletion controls myeloid cell metabolism survival, and myeloid cell-mediated inflammation
Myeloid-specific gene-deletion of PHD enzymes facilitated the investigation of the HIF pathway in innate immunity in vivo. PHD2 is the most critical regulator of HIFs. 192 197 These observations suggest that a postnatal inducible model, but not constitutive myeloid-specific PHD2 gene-deletion resulted in a spontaneous autoimmune phenotype that was linked to HIF-2 controlling interactions of innate and adaptive immune cells. PHD1 −/− mice were also protected from chemical colitis. 198 In human ulcerative colitis tissue intestinal PHD1 expression correlated with the degree of inflammation, 198,199 which is also consistent with a protective role of HIF. Together, these studies indicate that PHD2 links innate and adaptive immunity. PHD2-dependent HIF-regulation is indispensable for limiting inflammation and autoimmunity. PHD1 and PHD3 reduce inflammation by preserving mucosal barriers with some of these effects being possibly HIF independent. PHD3 keeps specifically monocytes in check. Reasons for differential PHD effects remain illdefined but could be related to cell type, HIF subunit, and HIFindependent effects on additional pathways.

PHARMACOLOGIC HIF MODIFIERS
Strategies to either stabilize or reduce HIFs are of major clinical interest and are currently explored in clinical studies. Given the profound HIF effects on myeloid cells, it will be important to carefully observe the effect of these pharmacologic substances on inflammation and immunity. AKB-4924, 171 and JNJ1935. 206 Most of these compounds are currently investigated in phase 2 and phase 3 clinical trial programs for renal anemia treatment. Of note, roxadustat treatment was associated with an increased rate of upper respiratory infections compared to standard therapy with recombinant human EPO in phase 3 study in dialysis-dependent patients with kidney disease (18.1% vs. 11.0%). 89 Beyond the correction of renal anemia, preclinical evidence suggests that PHD inhibition offers novel opportunities for organ protection, an area of unmet clinical need. We showed potent PHD inhibition by 2-(1-chloro-4-hydroxyisoquinoline-3-carboxamido)acetate (ICA) with beneficial effects in murine models of kidney ischemia-reperfusion injury, allotransplantation, and chronic kidney disease. [207][208][209][210] Tissue and organ protective effects of PHD inhibition have also been demonstrated in models of myocardial injury, 211 brain injury, 212 lung injury, 213 and-as mentioned earlier-inflammatory bowel disease. 169,172 AKB-4924 is evaluated for the treatment of inflammatory bowel disease (NCT02914262).

HIF inhibition
Cancer research incentivized the development of HIF inhibitors.
Agents that inhibit HIF heterodimerization, DNA binding, or transactivation are classified as direct HIF inhibitors, whereas indirect HIF inhibitors reduce HIF de novo synthesis or increase proteasomal degradation. 214,215 Various compounds were reported in the literature (as reviewed in Bhattarai et al. and Ban et al. 214,215 ), but only a few substances are currently available for clinical applications. PT-2385 is a direct HIF-2 inhibitor interfering with the HIF-2 -ARNT heterodimerization that is currently under investigation for treatment of renal cell carcinoma (NCT02293980, NCT03108066) and glioblastoma (NCT03216499). 216 Furthermore, in vitro testing of the HIF inhibitor PX-478 in prostate carcinoma cells 217