Tumor hypoxia: From basic knowledge to therapeutic implications

Abstract

Diminished oxygen availability, termed hypoxia, within solid tumors is one of the most common characteristics of cancer. Hypoxia shapes the landscape of the tumor microenvironment (TME) into a pro-tumorigenic and pro-metastatic niche through arrays of pathological alterations such as abnormal vasculature, altered metabolism, immune-suppressive phenotype, etc. In addition, emerging evidence suggests that limited efficacy or the development of resistance towards antitumor therapy may be largely due to the hypoxic TME. This review will focus on summarizing the knowledge about the molecular machinery that mediates the hypoxic cellular responses and adaptations, as well as highlighting the effects and consequences of hypoxia, especially for angiogenesis regulation, cellular metabolism alteration, and immunosuppressive response within the TME. We also outline the current advances in novel therapeutic implications through targeting hypoxia in TME. A deep understanding of the basics and the role of hypoxia in the tumor will help develop better therapeutic avenues in cancer treatment.

1. Introduction

The essence of tumor hypoxia is the imbalance between the oxygen demand of cancer cells and insufficient oxygen supply due to the dysfunctional tumor vasculature. Cancer cells exploit hypoxia to defend against the attack of immune cells or therapeutic interventions, therefore contributing to immune evasion and therapy resistance. Almost all solid tumors show a phenotype of hypoxia in their intratumoral area, which is often referred to as the tumor microenvironment (TME) [1]. It is challenging to thoroughly eliminate the negative effects of tumor hypoxia, partially because hypoxia not only induces the proliferative, invasive, or resistant properties of tumor cells but also elicits the dysfunction of immune cells or agitates the stromal environment to further facilitate tumor development.

The exploration of hypoxia and delineation of the related oxygen sensing signaling has a long-term history [2], and was well recognized by the awarding of the 2019 Nobel Prize in Physiology or Medicine to Drs. William G. Kaelin Jr, Sir Peter J. Ratcliffe, and Gregg L. Semenza [3]. Along with the discovery of the oxygen sensing signaling, numerous molecular pathways or biological processes that are associated with hypoxia were later uncovered and elucidated, eliciting an era of biomedical interventions on tumor hypoxia [4]. In this review, we will focus on summarizing the basic knowledge of the oxygen-sensing pathway in cancer, the major molecular mechanisms, and signaling pathways and machinery, including the proline hydroxylases (PHDs or EglNs) and novel oxygen sensor proteins from the 2-oxoglutarate (2-OG) dependent dioxygenases, hypoxia-inducible factors (HIFs), as well as the von Hippel-Lindau (VHL) E3 ubiquitin ligase system involved in different stages of oxygen sensing and cellular adaptation. In addition, the effects of hypoxia on the tumor microenvironment (TME), especially from the perspective of how hypoxia regulates angiogenesis, metabolism, and immune response in the TME, will be summarized. We will also discuss the current preclinical or clinical applications of cancer therapy by targeting hypoxia and hypoxic biological processes or signaling pathways.

2. Oxygen sensing in the tumor microenvironment

The crucial role of oxygen (O₂) in radiation response was first acknowledged in 1953 [5]. More recently, investigations have demonstrated that hypoxia is a characteristic feature of most solid tumors. Solid tumors are dynamic heterogeneous structures and display significantly lower oxygen tensions compared to their surrounding normal tissue. Cancer cells typically outgrow their vascular supply and develop hypoxic and necrotic tumor areas located at a distance of more than 100 μm away from blood vessels [6]. Hypoxic conditions bordering necrotic areas select for cancer cells with reduced apoptotic potential [6] (Fig. 1). In addition, this region tends to be more resistant towards radiotherapy as radiotherapy normally needs oxygen to produce fresh radicals to kill cancer cells [7].

Fig. 1.

The hypoxia signaling in tumor.

Tumors contain regions of normoxic cells that are close to blood vessels, developing hypoxic and necrotic tumor areas with increased distance from a functional blood supply. (Poorly oxygenated tumors are characterized with increasing chemotherapy/radiation resistance and metastasis risk that cause therapy failure. In normoxia cells, HIF-αs are hydroxylated by proline hydroxylases (EglN1/2/3) in the presence of O₂. Hydroxylated HIF-αs are recognized by pVHL, which mediates the proteasome degradation of HIF-as. In response to hypoxia, proline hydroxylation is inhibited. VHL is no longer able to bind and target HIF-αs for proteasomal degradation, which leads to HIF-αs accumulation and translocation to the nucleus. There, HIF-1α dimerizes with ARNT, binds to hypoxia-response elements (HREs) of target genes and recruits transcriptional co-activators such as p300/CBP for full transcriptional activity.

Hypoxia in solid tumors can be caused by three primary pathogenetic mechanisms: chronic or diffusion-limited hypoxia, acute or perfusion-limited hypoxia and anemia hypoxia [5]. Chronic hypoxia is caused by an increase in diffusion distances, chronically hypoxic cells are far away from capillaries which can result in a more clinically aggressive phenotype. Acute hypoxia is caused by inadequate blood flow in tissues when aberrant blood vessels are shut down. This is often transient and causes activation of stress-response, tissue damage and an increase in free-radical concentrations. Anemic hypoxia is a “relative” anemia caused by the decreased O₂-carrying capacity of the blood, which implies that fewer hemoglobin molecules are available for binding oxygen. Anemic hypoxia could be tumor-associated or therapy-induced anemia, this type of hypoxia is particularly intensified in tumors or tumor areas exhibiting low perfusion rates.

Tumor hypoxia is a hallmark of increased malignant progression and diminished therapeutic response. A large amount of clinical and experimental evidence shows that poorly oxygenated tumors have a negative impact on the effectiveness of curative treatment [8]. Facilitation of cancer cell survival by tumor hypoxia will: 1) reduces the effectiveness of radiotherapy; 2) cause resistance to chemotherapy and chemoradiation and 3) increases metastasis risk that may facilitate patient mortality (Fig. 1). Hypoxia regions tend to harbor cancer stem cells (CSC), which correlates strongly with both treatment failure and tumor recurrence. In breast tumors, hypoxic response in CSCs is modulated by signaling pathways including HIF-1α signaling and p-AKT/WNT/β-catenin signaling essential for phenotype and self-renewal of breast CSC [9, 10]. To meet metabolic demands, hypoxic tumors continually produce angiogenic factors, such as VEGF [11, 12]. However, newly formed tumor vessels are chaotically organized, immature and fail to support adequate blood supply, which further exacerbates local hypoxia [13]. Prolonged treatment of anti-angiogenic therapies has been known to further reduce oxygen tension, resulting in subsequent treatment failures. The upregulation of VEGF also facilitates both intravasation and extravasation of circulating tumor cells and even CSC [14]. In addition to the reasons above, there are some other complicated pathogenetic mechanisms for hypoxia to cause bad clinical outcomes and poor treatment response in cancers, as we will outline later, HIF-α subunits (Hypoxia-inducible factor α, mainly including HIF-1α and HIF-2α) are the key ones.

3. The regulatory machinery of the oxygen-sensing pathway

3.1. Hydroxylation and regulation of HIF-αs

HIF-αs are regulated by oxygen-dependent posttranslational modifications, which are mediated by three 2-oxoglutarate-dependent (2-OG) dioxygenases Egl nine homologs in vivo (EglNs, EglN1, EglN2, and EglN3, also called PHD2, PHD1 and PHD3 respectively). Under normoxic condition, HIF-1α is hydroxylated mainly on proline residues 402 and 564 (in HIF-2α – Pro405 and Pro531) by EglNs, which is vital for von Hippel-Lindau (encoding pVHL) E3 ubiquitin ligase complex recognition and binding. The pVHL complex adds ubiquitin to HIF-αs, which leads to the degradation of HIF-αs by proteasomes. Under hypoxic condition, due to the decreasing oxygen concentration, EglNs activity is reduced, which leads to limitation of hydroxylation of HIF-αs, subsequent ubiquitination and degradation. As a result, accumulated HIF-αs form the dimer with a constitutively expressed protein ARNT (aryl hydrocarbon receptor nuclear translocator, or HIF-1β) and associates with coactivator(s) p300/CBP to form an active transcription complex [15]. HIF complex is functional and binds to target genes containing hypoxia-responsive elements (HREs) across the genome, which promotes transcriptional regulation of targets in human cell lines and has been revealed by a cell line pan-genomic analyses of HIF [16, 17]. Proteins encoded by HIF target genes are involved in many aspects of cancer progression. For instance, VEGF, as we described above, is involved in angiogenesis and other examples including TGFα and EGFR in cell proliferation; GLUT1 in glucose uptake and metabolism; SDF1 and its receptor CXCR4 in chemotaxis [18] (Fig. 1). Both HIF-1α and HIF-2α dimerizes with ARNT but may activate transcription of a similar but distinct set of target genes depending on the cellular context. For example, HIF-2α potentates, while HIF-1α antagonizes c-Myc and β-catenin activity [19]. Mounting evidence suggests that HIF2α promotes pVHL-defective renal carcinogenesis. However, HIF1α suppresses tumor formation by VHL^−/− renal carcinoma cells, therefore HIF1α has the potential of being a kidney cancer suppressor gene [19]. Recently, Cowman S et al. found that HIF-1α but not HIF-2α is increased in ccRCC tumor cells. Compared with HIF-2α, HIF-1α is expressed highest in tumor-associated macrophages and is significantly associated with clinical benefit from antiangiogenic therapy [20]. TAD (the amino-terminal transactivation) domains of the HIF-α subunits may determine the specificity of targets of HIF-1α or HIF-2α [21]. For example, JMJD2c binds with TAD of HIF-1α but not HIF-2α and stimulates HIF-1α mediated transcription in Hela cells [22]. Both HIF-1α and HIF-2α transactivation function is negatively regulated by FIH (factor-inhibiting HIF)-dependent hydroxylation on asparagine site, which blocks the HIF-α recruitment of p300/CBP [23].

Even though most of the studies regarding HIF-1α and epigenetic regulators is about the interaction with p300/CBP, growing evidence has revealed that HIF-1α can also recruit other epigenetic factors. The H3K4 methyltransferases SET9 can bind to HIF-1α and promote HIF-1α protein stability in hypoxia [24]; the histone methyltransferase SET1B is recruited by HIF-1α to facilitate activation of the HIF response [25]; JMJD2C functions as a coactivator for HIF-1α, and HIF-1α can recruit JMJD2C into the hypoxia response elements of HIF-1 target genes [22].

3.2. The VHL-E3 ubiquitination system

VHL (encoding pVHL) is a critical tumor suppressor in clear cell renal cell carcinoma (ccRCC), which accounts for approximately 85% of all kidney cancer cases [26]. pVHL consists of two main protein binding domains, an α- and a β-domain; the α-domain of pVHL binds to elongin C, which in turn recruits elongin B and Cullin-2, whereas the latter recognizes and binds to substrates for ubiquitination and degradation by the 26S proteasome [15]. The canonical role of pVHL is its involvement in oxygen sensing and adaptive response to hypoxia. pVHL complexes recognize hydroxylated HIFαs, targeting them for proteasome degradation by ubiquitination [27, 28].

Growing evidence suggests that while HIF represents one of most important effectors for pVHL response, there exist other HIF-independent signaling pathways. For example, the inhibition of HIF-2α-mediated pathways by a specific HIF-2α antagonist PT2399 does not prevent all ccRCC tumor growth [29]; some other evidences indicate that VEGF can be induced by K-ras under hypoxia in a HIF-independent manner [30]; multiple pathways and transcription factors (TFs) such as NF-kB, AP-1, and CEBP are known to be induced by hypoxia in a HIF-independent pattern [31]. Elvidge and colleagues demonstrated that some genes regulated by hypoxia were not regulated by HIFs, suggesting that there are other HIF-independent signaling controlled by oxygen sensing pathway [32]. Now, more and more evidence identified non-HIFα substrates of pVHL. In ccRCC, pVHL binds hydroxylated ZHX2 and SFMBT1 for ubiquitinated degradation [26, 33]. In breast cancer, G9a stabilization is mediated through pVHL interaction and subsequent proteasome-mediated degradation, which links G9a as an epigenetic sensor of a hypoxic microenvironment [34]. NDRG3 has been identified as a bona fide substrate of the EglN1/pVHL system, which indicates that the EglN1/VHL system can control both HIF-dependent and HIF-independent hypoxia responses [35]. Also, several other putative pVHL substrates have been identified, including aPKC, RBP1/7, CERKL, TET2 and so on [15, 36]. By performing proteome and ubiquitome co-analysis together with the ccRCC CPTAC data, Wang and colleagues identified 57 proteins including TGFBI and NFKB2 that were ubiquitinated and downregulated by pVHL [37]. Based on these findings, a combinatorial targeting of HIF and other pVHL substrates may prove to be more effective than monotherapy in cancer therapy such as ccRCC. For example, Rouven Hoefflin and colleagues showed that the inhibitor of SPHK1 signaling regulated by pVHL-SFMBT1 axis [33] demonstrated excellent therapeutic activities in several experimental ccRCC models that are resistant to HIF-2α inhibitor PT2399 treatment [38]. These findings suggest that functional prolyl hydroxylation is beyond HIF. Although previous study suggested that other prolyl hydroxylation substrates beside HIF could not be validated through in vitro hydroxylation assays with recombinant prolyl hydroxylases [39], it exists the possibility that there may be some co-factors in vivo required to catalyze robust prolyl hydroxylation reactions for other potential substrates. Nonetheless, it is imperative to study the multi-functionality of prolyl hydroxylation in hypoxia signaling and cancer, which may yield additional therapeutic targets in cancer and other diseases. Undoubtedly, HIF-2α is a key therapeutic target for ccRCC. However, uncovering potentially additional HIFα-independent targets of VHL could provide new therapeutic opportunities and new combination treatment with HIF-2α inhibitor.

However, not all proteins recognized by pVHL are labeled for proteasomal degradation. It has been shown that AKT is hydroxylated by EglN1 on proline residues 125 and 313, and pVHL interacts with hydroxylated AKT and suppresses its activity in an E3 ubiquitin ligase-independent fashion [40]. Roe and colleagues demonstrated that pVHL binds to p53 and blocks the Mdm2-mediated ubiquitination and nuclear export of p53 [41]. EglN1 hydroxylates TBK1 on proline residue 48, resulting in pVHL-TBK1 interaction that helps recruit phosphatase PPM1B, therefore, suppressing TBK1 activity [42]. These findings highlight a new mechanism through pVHL that can affect cancer progression and response to treatment.

3.3. 2-Oxoglutarate-dependent dioxygenases

As cellular oxygen sensors by regulating the hypoxia-inducible transcription factors HIF-1α and HIF-2α, EglNs proteins belong to the class of 2-oxoglutarate-dependent (2-OG) dioxygenases (2OGDDs). 2OGDDs require dioxygen, divalent iron (Fe²⁺) and 2OG as co-substrates, and produce a hydroxylated product, CO₂, and succinate [43]. There are approximately seventy known and putative 2OGDDs in the GenBank DNA database, including Proline/lysine hydroxylases, JmjC domain-containing enzymes, nucleic acid oxygenases, fatty acid and small-molecule oxygenases [44]. The details of 2OGDD enzymes have been reviewed recently in other excellent reviews, which will not be repeated here [43]. Some 2OGDDs have specific role in cancer, but the mechanism of 2OGDDs in cancer has not been fully understood (Table 1).

Table 1.

List of known 2 OG-dependent dioxygenases and their roles in cancer.

2OGDD	Cancer types	Function
Proline/lysine hydroxylases
EglN1	PG	Mutated and inactivated [201, 202]
EglN2	BRCA	Overexpressed [203]
EglN3	Glioma	CpG sites were methylated, suppressor [204]
P4HA1–3	_	_
P4HB	_	_
PLOD1–3	_	_
LEPRE1	Colo, BRCA and Lung	Overexpressed [205]
LEPREL1–2	_	_
P4HTM	_	_
BBOX2	_	_
DNA/RNA-modifying enzymes
TET1	Many cancer types	Mutated and inactivated, silenced [206–208]
TET2	Many cancer types	Mutated and inactivated [209–211]
TET3	Glioma	Epigenetically repressed [212]
ABH1,2,3,4,6	_	_
ABH5	AML	Overexpressed [213]
FTO	AML and CESC	Overexpressed [214, 215]
JmjC domain-containing enzymes
KDM2A	BRCA and LUAD	Amplified and overexpressed [216, 217]
KDM2B	PACA	Overexpressed [218]
	Aggressive brain tumors	Silenced [219]
KDM3A	_	_
KDM3B	AML	Suppressor [220]
KDM4A	Many cancer types	Amplified and overexpressed [221]
KDM4B	OV and EGC	Overexpressed [222, 223]
KDM4C	BRCA, ESCs and ESCA	Amplified [224–226]
KDM4D	ccRCC	Overexpressed [227]
KDM5A	Glioma	Silenced [228]
	Many cancer types	Amplified and overexpressed [229–231]
KDM5B	Many cancer types	Amplified and overexpressed [232]
KDM5C	Many cancer types	Mutated and inactivated [233, 234]
KDM5D	PCa	Copy number loss [235]
KDM6A	Many cancer types	Mutated and inactivated [236]
	BRCA	Overexpressed [237]
KDM6B	Haematologic cancers	Overexpressed [238]
KDM7A	_	_
KDM8	BRCA and PCa	Overexpressed [239, 240]
HR	BRCA	Suppressor [241]
JARID2	AML	Chromosomal deletion, suppressor [242]
JHDM1C	_	_
JMJD1C	_	_
JMJD4,7,8	_	_
JMJD6	Many cancer types	Overexpressed [234, 243, 244]
MINA53	Many cancer types	Overexpressed [245]
PHF2	Many cancer types	Suppressor [246]
NO66	PCa	Overexpressed [247]
PHF8	HCC	Overexpressed [248]
UTY	_	_
Other hydroxylases
ASPH	Many cancer types	Overexpressed [249]
ASPHD1–2	_	_
BBOX1	BRCA and HCC	Overexpressed [250, 251], Suppressor [252]
FIH1	_	Regulates HIF [23]
HSPBAP1	PCa	Overexpressed [253]
OGFOD1	BRCA	Overexpressed [254]
OGFOD2	_	_
PAHX-AP1	_	_
PHYH	_	_
PHYHD1	_	_

PG, Paraganglioma; BRCA, Breast invasive carcinoma; Colo, Colorectal Cancer; Lung; Lung Cancer; AML, Acute Myeloid Leukemia; CESC, cervical cancer; LUAD, lung adenocarcinoma; PACA, Prostate cancer; OV, Ovarian cancer; EGC, Gastric cancer; ESCs, Esophageal squamous cell carcinomas; ESCA, Esophageal carcinoma; PCa, Prostate cancer; HCC, Hepatocellular carcinoma.

As major cellular oxygen sensors, EglNs have O₂ Km values at/near atmospheric oxygen levels [43]. Notably, Liu et al. show that EglN2 binds with and hydroxylates H3 at proline 16 directly (H3P16oh). This work provides a further intriguing link between oxygen sensor pathway, histone modifications and gene transcription [47]. Besides EglN proteins, some other 2OGDDS such as JmjC domain-containing enzymes can also act as potential oxygen sensors [45]. Batie et al. reported various histone methylation marks were increased in HIF-independent as early as 30 min under acute hypoxia. In these histone markers, H3K4me3 level was induced by the inactivation of KDM5A [46]. This is consistent with the observation that KDM5A has the highest O₂ Km of the KDM5 family members (90 μM) [43]. KDM6A is highly oxygen-sensitive and controls cellular differentiation. Two residues (M1190 and E1335) are involved in KDM6A’s oxygen sensor function, which are not conserved in KDM6B [47]. These new studies clarify that histone modifiers can act as new oxygen sensors directly rather than through the HIF pathway and the underlying regulatory mechanisms still need to be investigated.

4. Hypoxia shapes the landscape of the tumor microenvironment

4.1. How does hypoxia affect the TME?

The fast-proliferating rate of tumor cells is often accompanied by poor vasculature formation, therefore leading to the development of intertumoral pathological hypoxia. Although HIF could stimulate the formation of new blood vessels in the tumor region to compensate for the shortage of oxygen, however, the tumor vasculature is featured with poor organization and irregular architecture which is easy to collapse under the pressure of tumor formation and stroma growth [48, 49]. This phenomenon causes an irregular oxygen supply and an overall hypoxic environment in the tumor.

The tumor microenvironment (TME) is the ecosystem that surrounds the tumor cells, which is comprised of the non-cellular and cellular components. The non-cellular component includes the extracellular matrix (ECM), various signaling molecules such as cytokines and chemokines, and a chemical milieu including the pH, pO₂, nitric oxide, and metabolites. The cellular component mainly consists of immune cells, stromal cells, and blood vessels [1]. The TME is characterized by several notable features such as low extracellular pH, high level of reactive oxygen species (ROS), hypoxia, and a general immunosuppressive niche in comparison with normal tissue [50]. Crosstalk and mutual causal relationship may exist between each of these characteristics, while hypoxia could stand out as the prerequisite above all in TME.

Overall, as a universal intrinsic hallmark in all solid malignant tumors, hypoxia affects TME in many forms and from different aspects [51, 52] (Fig. 2). For example, hypoxia promotes ECM remodeling by changing the composition of ECM to enhance the invasive potential of tumor cells [53], which could be partially via stimulating the paracrine secretion of soluble factors that increases the fibrotic and stiff content of the ECM [54]. The pro-inflammatory cytokines and chemokines are important signaling molecules secreted by immune cells such as the nature killer (NK) cells, a study reported that hypoxia modifies the transcriptome profiles of NK cells which shows negative effects on generating these immune-modulatory signaling molecules [55]. On the other hand, the hypoxia-driven metabolic programming in the tumor or stromal cells also changes the level and distribution of certain metabolites in TME, especially for some important regulatory metabolites such as lactate, adenosine, and reactive oxygen species that exert inhibitory function to the infiltrating immune cells [56]. Therefore, the impact on tumor progression and immune invasion by hypoxia could be direct or indirect, either by direct inducing changes in cellular metabolism or creating a metabolic crosstalk between tumor and stromal cells via secreting metabolites [57]. From another perspective, hypoxia also shapes the tumor evolutionary landscape through applying a selective pressure for certain molecular aberrations. A recent study by Bhandari et al. suggests hypoxia is associated with elevated genomic instability and unique features of driver-mutation signatures in a large portion of tumor types. Intriguingly, they show that hypoxia could also regulate microRNAs abundance and lead to their dysregulation in tumors [58].

Fig. 2.

The relationship between hypoxia and TME.

Increased distance from blood vessels leads to a reduction in the oxygen supply of solid cancer. Upregulated glycolytic metabolism in hypoxic cells contributes to the acidification of the tumor microenvironment. The ECM composition is changed and subsequently enhance the invasive potential of tumor cells under hypoxic condition. Hypoxia modifies the transcriptome profiles of immune cells and suppresses the immune response. Metabolic programming in hypoxic area changes the level and distribution of certain metabolites such as lactate, adenosine, and ROS in TME. In addition, hypoxic conditions can drive genomic instability and mutagenesis, increasing the probability of metastasis and creating chemotherapy/radiation resistant clones.

Additionally, hypoxia changes the TME not only by directly affecting the intertumoral environment, but also through modulating the processes and regulatory molecular machinery in cells. As the predominant hypoxia-responsive molecular determinant, the HIF guides the cellular adaptation to oxygen availability by regulating gene expression. However, the functions of HIF-1α and HIF-2α as oncogenes or tumor suppressors in tumor development are still under debate. This controversial concept comes from a large body of evidence that suggests HIF-1α or HIF-2α play divergent roles either in promoting or suppressing tumor growth in different cancer types under different contexts [59]. In addition, the spatiotemporal distribution of HIF subunits within tumors could also contribute to their complex role in different cancers [59].

As tumor progression is not just a simple proliferation and growth of tumor cells, instead, it is more like a dynamic process engaged by every component of the TME, including tumor cells, stromal cells and other non-cellular components such as ECM [60]. Overall, hypoxia creates a favorable environment for tumor progression by shaping the landscape of TME that promote angiogenesis, metabolic adaptation, immune escape, metastasis, and therapy resistance. In the following parts of this section, more details on how hypoxia affects the TME will be summarized.

4.2. Hypoxia induces angiogenesis in TME

The environment in the TME is highly hypoxic, acidic, and nutrient competitive, and these are the major unfavorable factors for cell survival [61]. To overcome these harsh conditions, cells within the TME coordinate a program to promote angiogenesis in order to restore oxygen level, replenish nutrient supply and remove metabolic waste [12]. The importance of angiogenesis in promoting tumorigenesis and tumor progression is well recognized and documented over the last fifty years [62]. Besides, angiogenesis is necessary for tumor metastases, as the neovascular network allows tumor cells to disseminate and migrate through the blood vessels from the local tumor mass to distant organs [63, 64].

When tumor size is small with diameter less than 1 mm at the tumor initiation stage, tumor cells acquire oxygen mainly through blood perfusion and permeation from the surrounding tissue. However, due to the expansion of the tumor mass, the inefficient tumor vasculature hampers the oxygen supply which leads to hypoxia in the intertumoral region. As the main player involved in the transcriptional program of angiogenesis, the HIF-1α is subsequently stabilized and activated to drive the gene expression of vascular endothelial growth factor (VEGF), one of the major target genes of HIF-1α and a primary regulator of angiogenesis [11, 12]. The hypoxia induced HIF-α also leads to the expression of other pro-angiogenic factors such as platelet derived growth factor (PDGF) [65], nitric oxide synthase [66], adrenomedullin [67], and interleukin 8 [68] to further promote angiogenesis in TME. Besides, Hypoxia may also influence angiogenesis through other processes, for instance, by altering the morphology of fiber in the ECM [69]. It should be noted that the hypoxia-induced and HIF-1α-mediated angiogenesis is not limited to tumor cells, but also other cell types within the TME, including immune cells and cancer-associated fibroblasts (CAFs) [70].

4.3. Hypoxia induced long noncoding RNAs play multiple roles in TME

Emerging reports show that modulation of long noncoding RNAs (lncRNAs) in response to hypoxia could play pivotal roles in biological processes related to tumorigenesis, metastasis, angiogenesis, metabolism, and immune response [71–76].

In tumor cells, the hypoxia induced lncRNAs have been shown to participate in regulating gene expression at multiple levels including epigenetic, transcriptional, post-transcriptional regulation and protein levels [71, 77–80]. For example, the hypoxia induced WT1-AS promotes oncogenic transcription factor WT-1 expression through modulating histone H3K4 and H3K9 methylation in acute myeloid leukemia [81]. The lncRNA HOTAIR which induced by hypoxia may serve as scaffold by providing binding surfaces for histone modification complex and thus to impact the chromatin modification and transcription [82]. Besides the chromatin structure remodeling, the lncRNAs could modulate transcription by directly regulating specific transcription factors. Hypoxia inducible LncHIFCAR is a co-activator driver of HIF-1α via direct binding and subsequently promotes the recruitment of HIF-1α and p300 cofactor to the target promoters in oral carcinoma [83]. At post-transcription level, multiple hypoxia-induced lncRNAs are involved in mRNA stability regulation and miRNA-mediated gene silencing regulation. The most prominent example of lncRNAs regulating mRNA stability is HIF-1α mRNA is regulated by HIF1A-AS2 DANCR and lncRNA-LET [79, 84–87]. In addition, HIF1A-AS2 has been reported to sequester miR153–3p to enhance HIF-1α expression suggesting that lncRNAs involve in miRNA regulating [88]. In TME, lots of existing studies have indicated that lncRNA are involved in the crosstalk between tumor cells and other cell types including stromal cells, endothelial cells, and immune cells [89–93]. The exosomes that derived from tumor cells deliver lncRNAs to mediate the communication between tumor cells and other TME components and subsequently regulate tumorigenesis and progression. Although the lncRNAs have been shown to be able to serves as diagnostic and prognostic markers [94, 95], the clinical application of lncRNAs is still required to be further explored.

4.4. Hypoxic TME dampens tumor immunity

Tumor cells escape immunosurveillance by various mechanisms while tumor hypoxia is among the most important factors influencing immune responses [96]. It should be noted that hypoxia affects immunity divergently in an immunological context dependent manner [97]. In contrast to hypoxia that controls proliferation, development and effector function of immune cells in physiological immunological niches, tumor hypoxia typically causes immune cell dysregulation that eventually drives tumor progression and development [97]. Hypoxia restrains the continuous surveillance and attack of immune cells to tumor cells, generating an immunosuppressive environment to facilitate tumor progression. Specifically, hypoxia modulates the tumor immune microenvironment from at least three major aspects by: 1) impeding function or infiltration of immune cells, 2) recruiting immunosuppressive cells to block the immune response, and 3) upregulating regulatory molecules that further blocks the activation of immune effector cells. In the context of antitumor immunotherapy, hypoxia also confer resistance to immunotherapeutic drugs by inducing expression of membrane proteins that increases drug efflux or regulates specific cell signaling pathways that are critical to the response of immunotherapy. Therefore, a fully understanding of these processes is necessary to achieve successful therapeutic efficacy (Fig. 3).

Fig. 3.

Hypoxia shapes the metabolic and immune environment of TME.

Hypoxia in the TME changes metabolites profile and promotes secretion of cytokines/chemokines that regulates the function of immune cells, as well as mediates the anti-tumor response of various types of immune cells.

Tumor infiltrating immune cells are critical components of the tumor microenvironment and contain various cell types including T cells, B cells, macrophages, neutrophils, and dendritic cells [1]. In general, as previously mentioned, hypoxia impedes the function of tumor-killing immune cell populations while enhancing the pro-tumorigenic role of regulatory immune cells to achieve an overall favorable niche for tumor growth. However, the effect of hypoxia on immune cells is complicated due to the intricate biology of the immune cell differentiation programs and a variety of types of immune cell populations.

T cells are the predominant immune population for tumor immune surveillance. Hypoxia modulates T cells by first affecting the recruitment of T cells into tumor areas. For example, in some malignancies such as prostate cancer, the hypoxic tumor region lacks substantial infiltration by any type of T cells. Alleviation of tumor hypoxia restores T cell access and leads to increased T cell proliferation within the TME [98]. On the other hand, hypoxia increases T regulatory cells (Treg) in the TME to suppress CD8⁺ T cell mediated cytotoxic killing of tumor cells [99]. In addition, hypoxia also causes metabolic dysregulation in T cells. Studies have shown that the hypoxia/HIF-1α signaling could increase the expression of pyruvate dehydrogenase kinase (PDK1), thus decreases OXPHOS and increases glycolysis in the T cells [100, 101]. Intriguingly, in some other cases, hypoxia produces favorable roles in T cell function. A study by Palazon et al. found that hypoxia and HIF-1α drive CD8⁺ T cell migration and effector function, while loss of HIF-1α promotes tumor progression [102]. Another study also found similar results that CD8⁺ T cells could be activated and survive well under hypoxia condition and kill tumor cells more efficiently than normoxic T cells through increasing granzyme B packaging into granules [103]. Hypoxia affects multiple types of T cells in addition to CD8⁺ population. Recently, hypoxia has been reported to suppress the antitumor function, but not the infiltrating potential, of γδ T cells in brain tumors [104].

Tumor-associated macrophages (TAMs) are exceptionally plastic cells and a prevalent cell type in a variety of tumors [105]. Two main polarization states exist in macrophages which represent the classical M1 phenotype antitumoral and the alternative M2 immunosuppressive phenotype [106]. Mounting evidence suggests that hypoxia selectively induces specific macrophage phenotypes promoting tumor malignancy while repressing immunity [107]. This is because macrophages are very sensitive to oxygen fluctuation with the M1 polarization preferentially undertake in normoxia, while the M2 phenotype is induced in hypoxic circumstances [108]. The dendritic cells (DCs) play a major role on presenting the tumor antigens to naïve T cells to initiate the immune adaptive response. Hypoxia, although does not hamper DC maturation, inhibits antigen uptake as well as changes the DC chemokine expression profile in the in vitro setting, therefore may have a critical role in DC differentiation, adaptation, and activation [109]. This is in line with the finding that HIF-1α can promote the anti-inflammatory functions of DCs in vivo [110]. Other immune effector or regulatory cells could also be affected by hypoxia including myeloid-derived suppressor cells (MDSCs), whose immune suppressive effects are also enhanced by the hypoxia/HIF signaling [111], and the NK cells as hypoxic TME reduces the cytotoxicity and causes the elimination of NK cells by inducing mitochondrial fragmentation in NK cells [112].

In addition to directly affecting the function of immune cells, hypoxia can also induce several regulatory molecules through HIF-1α to further promote innate immune evasion. These regulatory molecules include the immune checkpoint inhibitor molecules (PD-L1 and CTLA-4) and CD47, which will bind to macrophages/effector T cells and block their activity. Under hypoxia, HIF-1α protein will accumulate and directly induces PD-L1 expression in multiple cell types such as tumor cells, macrophages, dendritic cells and MDSCs [113–115]. These cell surface PD-L1 could further bind PD-1 receptor to block T cell function. CTLA-4 receptor is induced and expressed on CD8⁺ T cells by hypoxia while binding CD80 and CD86 ligands on the surface of antigen-presenting cells (APCs), leading to effector T cell inhibition and regulatory T cell activation [116]. The CD47 “don’t eat me” signal controls immune evasion through interacting with signal regulatory protein α (SIRPα) expressed on TAMs [117] and MDSCs [118] and its expression is also controlled by HIF-1α [119]. Interestingly, HIF-induced VEGF is also an immunosuppressive cytokine in the hypoxic TME, through promoting the CD4⁺ T cells differentiation to Treg cells [120]. Overall, hypoxia in the TME regulates the function of immune cells, as well as the expression of immune inhibitory molecules, which in turn plays a role in promoting tumor development and immune evasion.

4.5. Hypoxia disturbs metabolism in TME

Tumor hypoxia develops largely because of imbalance between the high demand of oxygen consumption and its insufficient supply in the TME. Furthermore, the immature blood vessel formation directly causes a shortage of various nutrient supplies and the related metabolic activities in the TME. With this premise, hypoxia must reshape the landscape of metabolism in TME, not only by forcing the metabolic reprogramming in cells to adapt to the TME, but also by altering the level and composition of metabolites and nutrients within the hypoxic tumor region.

Unfortunately, tumor cells basically win the battle for oxygen and nutrient utilization and show higher metabolic adaptability and plasticity than the normal cells [121]. This is usually reflected on the tumor cells rewiring the aerobic respiration from the mitochondrial oxidative phosphorylation (OXPHOS) to aerobic glycolysis (or known as the Warburg effect [122]), an inefficient but rapid way to ferment glucose, even when oxygen is readily available. Consequently, tumor cells consume more glucose to generate cellular building blocks including nucleic acids and amino acids, and more “waste” products such as lactate which subsequently leads to an acidic environment to furtherly suppress the activity of infiltrating immune cells. The glycolytic shift is largely due to the HIF signaling dependent induction of the glycolysis driver genes under either hypoxia, oncogenic signaling or pseudo-hypoxia conditions due to pVHL loss in kidney cancer [123, 124]. These genes encode transporters and key enzymes involved in glucose uptake and catabolism including the glucose transporter proteins (GLUT1 and GLUT3) [125, 126], hexokinases (HK1 and HK2) [127], phosphoglycerate kinase (PGK) [128], pyruvate kinase M2 [129], and lactate dehydrogenase (LDHA) [130], etc.

The repression of mitochondrial respiration is also mediated by HIF under hypoxia through multiple mechanisms. Firstly, HIF modulates glycolytic and mitochondrial metabolism by directly regulating some key metabolic enzymes. For example, HIF-1α or −2α activates the expression of lactate dehydrogenase A (LDHA), a key metabolic enzyme of glycolysis that converts pyruvate to lactate, in the cancer cells [131]. This leads to the accumulation of lactate and regeneration of NAD⁺ which is an electron acceptor to maintain the glycolytic flux [132]. On the other hand, HIF-1 also transactivates the gene coding PDK1 which mediates the phosphorylation and inhibition of pyruvate dehydrogenase (PDH), thereby reducing cellular acetyl-CoA production and suppression of TCA activity in mitochondria [100, 101]. Secondly, HIF also could mediate an indirect way since activated HIF-1α suppresses c-MYC transcriptional activity to the mitochondrial genes [133] or transcriptionally represses PINK1 [134]. Moreover, hypoxia also represses mitochondrial biogenesis by controlling Forkhead-box protein O3a (FOXO3a) mediated mitochondrial biogenesis regulation [135]. It is worth mentioning that the hypoxia-induced mitochondrial ROS could also inhibit the turnover of HIF. Inhibition of the activity of PHDs by increased ROS leads to stabilization of HIFα subunits and HIF-mediated transcription [136].

Besides repression of OXPHOS and enhancement of glycolysis, other metabolic pathways are also modulated in cells under hypoxia. For instance, tumor cells activate the pentose phosphate pathway (PPP) and synthesis of amino acid to meet the high demands of biomass synthesis [137, 138]. Hypoxia can also regulate lipid metabolism. A previous study by Qiu et al. demonstrated that HIF-2α, but not HIF-1α, promotes lipid storage to sustain ER homeostasis by transcriptional activating the lipid droplets coat protein gene PLIN2 in ccRCC [139]. Furthermore, HIF can directly suppress fatty-acid oxidation through inhibition of medium/long chain Acyl-CoA dehydrogenase (M/LCAD) [140] or cause a lipid deposition phenotype as most commonly observed in ccRCC by repressing carnitine palmitoyltransferase 1A (CPT1A) and subsequent lipid transportation into mitochondria [141]. Cells under hypoxia can also rely on reductive carboxylation of glutamine for de novo lipogenesis [142]. Therefore, hypoxia has a broad effect on metabolic fate of the cells in TME.

Because of the metabolic reprogramming induced by hypoxia, the alteration of metabolites that present in the TME also has profound impacts on the cell fate and causes symbiotic or competitive interactions between different tumor resident cells (Fig. 3). In the 1920s, Otto Warburg first observed that cancer cells could uptake excessive amount of glucose and preferentially produce lactate [143]. The lactate generated by tumor cells induced by hypoxia is rapidly secreted into the extracellular space by monocarboxylate transporters (MCTs) to prevent intracellular acidification, however, this process could gradually lead to the accumulation of lactate and decrease the pH level of the TME [144]. Elevation of lactate level in the TME is reported to be associated with poor clinical outcome [145], which is largely due the broad effect of lactate on negatively modulating immune response [146] or act as a signaling molecule involved in various mechanisms to directly promote cancer cell proliferation, survival, and metastasis [147]. Accompanying with elevated levels of lactate, extracellular glucose level is dramatically decreased in the TME. It should be noted that not only the cancer cells are highly glycolytic, T cells also highly dependent on glucose and glycolysis to maintain cell proliferation and function. Glucose deficiencies therefore impedes T cell activation and limit antitumor responses [139, 148]. Apart from glucose, glutamine is another critical metabolite readily scavenged by proliferating cells in hypoxic environment [149–151]. Similar as glucose oxidation is repressed under hypoxia, glutamine oxidation in mitochondria is also blocked and detoured to lipogenesis instead [152, 153]. Glutamine restriction in the TME could affect tumor cells, but also has a deep impact on immune cells. As several studies reported, limitation of glutamine changes T cell differentiation from CD4⁺ T cells to FOXP3⁺ Treg cells [150, 154]. Another important immune regulatory metabolite that is activated by hypoxia is adenosine, which has been well accepted for its immunosuppressive function and as a metabolic checkpoint for immune therapy [155]. Hypoxia activates adenosine pathway not only by inducing the release of ATP, which is the precursor of extracellular adenosine synthesis, but also by upregulating ectonucleotidases including CD39, CD73, as well as adenosinergic receptors like A2BR through a HIF-1α dependent manner [156–158]. Upregulation of CD39 and CD73 facilitate the two-step conversions from ATP to AMP and AMP to adenosine, respectively. The increased extracellular adenosine could bind adenosine receptors to execute a broad immune repression to either inhibit effect T cells or induce Treg cells [155].

5. The therapeutic implications of targeting hypoxia in TME

Tumor hypoxia has emerged as an attractive therapeutic area due to its essential role for cancer. Upon discovery of a myriad of mechanisms, hypoxia shapes the TME and promotes tumor initiation, progression, metastasis, immune evasion, and drug resistance. A large opportunity of biomedical interventions could be and/or have been made to tackle the cancer through targeting tumor hypoxia. In general, some various strategies and drugs have been developed at the stage of preclinical or clinical application. Based on the therapeutic mechanism of action, these hypoxia targeting strategies could be classified into several categories, such as 1) modifying tumor oxygenation in TME, 2) targeting HIF and HIF-related hypoxia signaling, 3) developing hypoxia-activated prodrugs, and 4) targeting other hypoxia-associated biological processes and pathways. Here we provide the rationale of each of the strategies, as well as outline some of the representative therapeutic applications that overcome the hypoxic barrier in the preclinical or clinical setting (Table 2).

Table 2.

Summary of the representative therapeutic interventions for tumor hypoxia.

Treatment/Drug	Action/Effects	Clinical trial	Refs
1. Modify tumor oxygenation
Hemoglobin transfusion	Deliver oxygen	/	[159]
Erythropoietin (EPO)	Stimulate hemoglobin production	/	[161]
Perfluorocarbons	Deliver oxygen	/	[162]
Hyperbaric oxygen	Deliver oxygen	/	[163]
Carbogen breathing	Promotes oxygen delivery and increases blood flow	/	[164]
Hyperthermia	Increases tumor blood flow and perfusion	/	[165, 166]
Atovaquone	Alleviate tumor hypoxia	/	[167, 168]
Papaverine	Alleviate tumor hypoxia	/	[169]
Metformin	Decreases the rate of tumoral mitochondrial oxygen consumption	NCT04275713, Phase II; NCT02394652, Phase II; NCT03510390, Phase N/A	[170]
2. Target HIF and HIF-related hypoxia signaling
PT2385	HIF-2α antagonist	/	[172]
Belzutifan (PT2977)	HIF-2α antagonist	/	[173]
HIF-related pathways inhibitors	Inhibit HIF mediated hypoxia regulatory pathway	/	[174–178]
2-OG oxygenase inhibitors	Inhibit 2-OG oxygenase	/	[179]
USP37	HIF-2 deubiquitinase	/	[181]
3. Hypoxia-activated prodrugs
Evofosfamide (TH-302)	Hypoxia specific cytotoxins	NCT03098160, Phase II	[185]
E09 (Apaziquone)	Hypoxia specific cytotoxins	NCT00461591, Phase III; NCT00598806, Phase III	[186]
Others	Hypoxia specific cytotoxins	NCT02096354, Phase II; NCT02454842, Phase II; NCT02449681, Phase II; NCT00394628, Phase I/II; NCT01037556, Phase I/II	[187, 188]
4. Target other hypoxia-associated biological processes and pathways
Girentuximab	CAIX inhibitor	NCT00087022, Phase III	[192]
SLC-0111	CAIX inhibitor	NCT02215850, Phase I	[193]
AZD3965	MCT1 inhibitor	NCT01791595, Phase I	[194]
AZ93	MCT4 inhibitor	/	[195]
CB-839	Glutaminase inhibitor	/	[198]
Aminooxyacetate	Aminotransferase inhibitor	/	[199]
Adenosine pathway inhibitors	Inhibitors of the adenosine pathway	/	[200]

5.1. Modify tumor oxygenation in TME

The most straightforward approach to tackle the tumor hypoxia is to increase the oxygen level in the hypoxic tumor region and this could be achieved from various perspectives. Under physiological conditions, the oxygen supply is positively correlated with concentration of the oxygen carrier hemoglobin. Attempts have been made by performing hemoglobin transfusion to improve radiotherapy outcomes in the hemoglobin low-level patients [159], however, the trials failed probably due to a transient effect of the oxygen elevation and adaptation in the tumors [160]. Other similar approaches were also examined including using erythropoietin (EPO) to stimulate hemoglobin production [161], or using artificial blood substances like perfluorocarbons to deliver oxygen [162]. Increasing oxygen delivery is an actionable strategy to improve tumor hypoxia, which leads to the clinical trials to treat patients with hyperbaric oxygen and achieved significant local tumor control [163]. Besides, carbogen breathing (95% oxygen + 5% carbon dioxide) was used as an alternative of hyperbaric oxygen treatment, as it promotes oxygen delivery via vasodilation and increases blood flow stimulated by carbon dioxide [164]. Several studies suggest that the mild Hyperthermia, a physical therapy by applying heat (40 to 45 °C) to treat tumors, could increase tumor blood flow and perfusion and thus increase oxygen delivery [165, 166]. Due to the technical difficulties to achieve cytotoxic temperatures or overcome tissue thermotolerance, Hyperthermia is mainly used as adjuvant therapy in combination with chemotherapy or radiotherapy. Nevertheless, none of these approaches above has yet reached a controllable clinical outcome so far. Apart from increasing oxygenation, suppressing oxygen consumption is the other side of the coin to eliminate tumor hypoxia. Numerous inhibitors targeting the mitochondrial electron transport chain (ETC) complexes have been developed, including the FDA-approved drugs Atovaquone and Papaverine, which recently have been found to alleviate tumor hypoxia in preclinical models [167–169]. Metformin is another widely used drug to decrease the rate of tumoral mitochondrial oxygen consumption, although its original usage is to reduce insulin availability for the treatment of type 2 diabetes [170]. There are several ongoing clinical trials using metformin as a tumor oxygenating agent such as two trials for treating cervical cancer (NCT04275713 and NCT02394652, Phase II) and one for treating head and neck squamous cell carcinoma (NCT03510390).

5.2. Target HIF and HIF-related hypoxia signalling

In the hypoxia signaling cascades, HIF is the central hub that connects the upstream oxygen sensing and downstream cellular adaptation. In addition, expression of HIF is also significantly upregulated after radiation or chemotherapy, suggesting HIF pathways are also actively involved in the treatment resistance [171]. Therefore, HIF is served as a pivotal target for intervention. However, due to the nature of HIF-αs as transcription factors, it has been challenging to develop agents to target this transcriptional factor complex. Great progress has been achieved during that last decade. One of the most successful cases is the development of HIF-2α specific antagonists such as PT2385 [172] and Belzutifan (PT2977) [173], while the latter one has been approved by FDA in 2021 for adult patients with von Hippel-Lindau disease including advanced-stage ccRCC. Beside HIF-2α inhibitors, there are also many HIF-1 inhibitors under development and several novel agents have undergone clinical evaluation [2]. Other approaches targeting HIF signaling, including a variety of inhibitors to interfere with mRNA expression, protein synthesis, DNA binding, or transcriptional activity of HIF, have been summarized previously and will not be discussed here [174].

Apart from targeting HIF directly, intervention of the HIF-related hypoxia signaling is also strongly actionable. Indeed, almost every step in the HIF mediated hypoxia regulatory pathway is targetable. These HIF-related pathways include but not be limited to PI3K/mTOR signaling [175], topoisomerases [176], heat shock proteins [177], and histone deacetylases [178]. Numerous inhibitors have been developed and tested preclinically or clinically for targeting these pathways, with some of them have been reviewed previously [2, 179]. On the other hand, the identification of novel HIF regulators could provide an extra layer to intervene hypoxia. For example, some of the 2-OG oxygenase family such as the well-known HIF hydroxylase EglNs (or PHDs) or perhaps other new members, which regulate HIF post-translational modifications and protein stability, are ideal druggable targets to disturb the hypoxia/HIF signaling. Some PHD inhibitors such as Roxadustat were used for treating patients with chemotherapy-induced anemia [180]. It is worth mentioning that the safety of these PHD inhibitors should be carefully evaluated in patients with cancer since they act as HIF activators and may complicate the treatment outcome. Nevertheless, various 2-OG oxygenase inhibitors in the disease contexts have been highlighted previously [179], and new small compounds are under investigation. Recent studies indicate USP37 is a HIF-2α deubiquitinase mediating its protein level through reverses the degradation process in ccRCC [181], and ZHX2 as a novel HIF-1 regulator controlling tumor progression in triple-negative breast cancer [182]. While how to translate these novel upstream targets in combating the hypoxia within TME is unclear and warrants further investigation.

5.3. Hypoxia-activated prodrugs

It remains attractive to specifically target tumor hypoxia while sparing the normoxic surrounding tissue, which motivates the development of a class of hypoxia specific cytotoxins so called the hypoxia-activated prodrugs (HAPs). These compounds were elegantly designed and believed to precisely target the hypoxic region of tumors, owing to their property of selective reduction by endogenous oxidoreductases to yield active cytotoxic agents only under hypoxic conditions [183]. Several HAPs have been developed and are under extensive preclinical or clinical investigation either alone or in combination with conventional radiotherapy, chemotherapy, and immunotherapy [184]. One example is evofosfamide (TH-302) which has been shown to be non-lymphotoxic and thus well applied in combination with immunotherapy such as the blockade of PD-1 and CTLA-4 [185]. The effectiveness of TH-302 has been documented in preclinical models that can induce CD8⁺ T cells into hypoxic tumor region, meanwhile decreasing MDSCs population and reducing suppressive myeloid stroma [185]. The therapeutic efficacy of TH-302 in combination with ipilimumab against multiple cancers is under a Phase II clinical evaluation (NCT03098160). The mitomycin C derivative prodrug, E09 (Apaziquone), is another representative HAP which has shown efficacy in bladder cancer with two Phase III clinical trials (NCT00461591 and NCT00598806) ongoing [186]. Other representative HAPs that are under clinical evaluation including RRx-001 (NCT02096354, Phase II), TH-4000 (Tarloxotinib) (NCT02454842/NCT02449681, Phase II), AQ4N (NCT00394628, Phase I/II), and PR-104 (NCT01037556, Phase I/II). HAPs derive from various chemical classes and have diverse mechanisms of action, but the fundamental notion is that the reduction of these agents is typically initiated by donating a single election (1e) which leads to generation of a free radical intermediate as an oxygen sensor. A common challenge in development of HAPs is that many chemical compounds such as quinones, nitro, and N-oxides are also subject to concerted two-electron (2e) reduction thus bypassing the oxygen-sensing radical intermediate and leading to an issue of poor selectivity. It has been observed hypoxia-independent 2e reduction events of HAPs, such as PR-104A by aldo-keto reductase 1C3 and Apaziquone by NADPH: quinone oxidoreductase 1 [187, 188], which may potentially hinder their clinical application.

5.4. Target other hypoxia-associated biological processes and pathways

Given the profound impact of hypoxia that gives rise to tumor cells and TME, a broad range of biological processes and signaling pathways are regarded as the direct or indirect consequence of hypoxia. These elicit further potential therapeutic opportunities for treating hypoxic tumors including targeting hypoxia-associated biological processes like acidosis and metabolic dysregulation in the TME, or inhibiting hypoxic related pathways such as the unfolded protein response (UPR), and the DNA damage response (DDR). Due to the space limitation, the UPR and DDR pathways will not be summarized here, detailed therapeutic intervention can be found in previous review [189].

Acidosis in the TME highly parallels with hypoxia. It is mainly induced by intertumoral high rates of glucose anaerobic catabolism, lactate, H⁺ and carbon dioxide production and secretion, leading to a hostile acidic environment. While tumor cells are more tolerable to acidosis than non-malignant cells due to high expression of pH-regulating proteins and thus provide selective therapeutic targets. Extracellular- facing carbonic anhydrase 9 (CAIX), which encodes by CA9, is one of the representative pH-regulating proteins and attractive therapeutic targets. Not only does high CA9 expression predict worse patient prognosis [190], but also it is strongly upregulated by HIF-1 [191]. Many CAIX targeting drugs have been developed and some of them were at the clinical stage, including Girentuximab, a monoclonal antibody for treating kidney cancer (NCT00087022, Phase III) [192] and SLC-0111, a small-molecule compound for treating advanced solid tumors (NCT02215850, Phase I) [193]. Other drugs targeting tumor acidosis include the lactate transporter inhibitors, such as MCT1 inhibitor AZD3965 (NCT01791595, Phase I) [194], and MCT4 inhibitor AZ93 [195].

As described previously, metabolic reprogramming is a vital consequence of hypoxia within the TME. Therefore, targeting related metabolic enzymes or intermediates in dysregulated pathways could be a remarkable choice to overcome the hypoxic effect. Glycolytic inhibitors such as 2-deoxy-glucose have been efficacy in tumor cells, while the caveat is that they also abrogate the function of immune cells which will limit their further application [196]. Targeting the MCT1/2 lactate transporter by AZD3965 is an alternative way to neutralize the waste metabolite production of glycolysis, and there are phase I clinical trials undergoing for several cancers [197]. In addition, many cancer cells are highly glutamine-dependent, inhibiting glutamine metabolism by targeting the glutaminase (specific inhibitor such as CB-839) or the aminotransferase (specific inhibitor aminooxyacetate) reveals therapeutic efficacy in tumors with hypoxic traits [198, 199]. Finally, inhibition of the hypoxia induced extracellular adenosine signaling has made remarkable progress and is considered as a ‘next-generation’ therapeutics in immuno-oncology [155]. Several inhibitors of the adenosine pathway have been developed and entered clinical trials over the past few years. The drugs that either target CD73, CD39, or adenosine receptors have been systematically summarized elsewhere [200].

6. Conclusion and perspective

In conclusion, hypoxia related research remains one of most dynamic and exciting directions in cancer research. The 2019 Nobel Prize in Medicine or Physiology on hypoxia and oxygen sensing signaling marks the beginning of new era in this line of research. Targeting hypoxia signaling in cancer will be further developed from several different perspectives: 1) targeting upstream oxygen sensing signaling, including canonical oxygen sensors such as prolyl hydroxylases and other 2-OG dependent dioxygenases; 2) targeting hypoxia/VHL loss induced therapeutic vulnerabilities, including HIF signaling, synthetic lethality targets, hypoxia prodrugs and 3) targeting hypoxia signaling pathways in combination with immunotherapies in tumor microenvironment.