Tumour cells do not act in isolation but exist in a complex and dynamic ecosystem with their microenvironment. Malignant cancer cells of solid tumours are associated with non-malignant host stroma, consisting of extracellular matrix (ECM) components, fibroblast cells, mesenchymal cells, blood and lymph vasculature, and tumour-infiltrating immune cells, cytokines, and chemokines [1]. The immune cells in the tumour microenvironment (TME) are of both innate and adaptive type and primarily include tumour-associated macrophages (TAMs), neutrophils, dendritic cells (DCs), myeloid-derived suppressor cells (MDSC), natural killer (NK) cells, and CD4 + and CD8 + T lymphocytes and B lymphocytes, along with several extracellular immune factors [2, 3]. All immune components of the TME constitute the tumour immune microenvironment (TIME), and the varied functions and spatial organization of immune cells in the tumour microenvironment influence tumour progression, anti-tumour immune responses, and the efficacy of immunotherapeutic interventions. Extensive research from recent decades has illustrated the crucial roles of the host immune system in controlling anti-tumour and pro-regulatory immune response through cancer immune surveillance and tumour interaction. For instance, higher populations of MDSCs and TAMs in the TIME have been observed to promote tumour progression, whereas the increased recruitment of cytotoxic T lymphocytes has been associated with improved anti-tumour immune response and better prognosis [4]. Alterations in cancer cells and the surrounding stromal tissue due to environmental stresses like hypoxia can modify the immune response to tumours [3].

The immune cells of the TIME (Fig. 1) can be broadly classified into immunosuppressive (tumour-promoting) and immune effector (tumour-antagonizing) cells [5]. The tumour-antagonizing immune cells mainly consist of effector T cells, including CD8 + cytotoxic T lymphocytes (CTLs) and effector CD4 + T cells, NK cells, dendritic cells (DCs), M1-polarized macrophages, and N1-polarized neutrophils. The CTLs are the major subset of lymphocytes for killing the cancer cells; when presented with tumour antigens from DCs, CD8 + T cells can be induced into effector CD8 + CTLs with cytotoxic capacity. NK cells are also an important subset of tumour-antagonizing immune cells with a similar function, with respect to CD8 + T cells, and are attracted to cancer tissues under the guidance of chemokines secreted by DCs. They attack tumour cells by releasing perforin and granzymes to induce apoptosis. DCs mainly function as professional antigen-presenting cells (APCs), which can present antigens and provide costimulatory signals for T-cell activation and interact with NK cells and B cells [5]. TAMs are the primary tumour-infiltrating immune cell types in the TIME [6]. They are generally categorized into classical activated M1 macrophages, which typically have anti-tumour functions, as well as alternatively activated M2 macrophages, which are immunosuppressive. The latter exhibit pro-tumour functions, including inhibition of T-cell-mediated anti-tumour immune response. Both M1 and M2 macrophages have a high degree of plasticity and can be converted into each other upon tumour microenvironment changes like hypoxic stress [7].

Tumour immune microenvironment (TIME)

Other tumour-promoting immune cells mainly consist of regulatory T cells (Tregs) and MDSCs. Tregs are a specialized subset of CD4 + T cells identified by the expression of the FOXP3 gene, and they induce tumour tolerance by the production of TGF-β and suppression of effector T cells [8]. Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid cells, which can be classified into two main subtypes: polymorphonuclear (PMN-MDSC) and monocytic (M-MDSC). Besides directly repressing DCs, NK cells, and T cells, and promoting immune tolerance, MDSCs also contribute to angiogenesis and metastases [9, 10].

During tumour development and progression, cancer and stromal cells often have restricted access to nutrients and oxygen. Most solid tumours consist of regions that are permanently or transiently subjected to hypoxia, owing to aberrant vascularisation and poor blood supply [11].

The hypoxic TME is defined as a condition of poor oxygenation where partial O2 pressure drops below 10 mmHg [12]. Hypoxia arises from the imbalance of increased oxygen consumption by rapidly proliferating tumour cells and the available blood supply. Inadequate oxygen supply triggers new blood vessel formation or angiogenesis, but the distribution of the newly developed tumour vasculature network is irregular and characterized by diffusion limits, leakiness, and malformation. This leads to pockets of different oxygen tensions in the TME and contributes to the heterogeneity of the spatial architecture of the cells [13, 14]. At the molecular level, the response and adaption of tumour cells to the hypoxic TME are largely mediated by the hypoxia-inducible factor (HIF) family of transcription factors. HIFs are heterodimeric helix-loop-helix proteins consisting of an O2-sensitive α-subunit (HIF-1α, HIF-2α, and HIF-3α) and a constitutively expressed β-subunit (HIF-1β) [15]. HIF-1α and HIF-2α have crucial roles in the positive hypoxic response and are the best studied, whereas HIF-3α is considered a negative regulator [16, 17]. Most of the research on hypoxic regulation of the tumour microenvironment highlights the functions of HIF-1 which has a diverse range of effects. Cellular sensing of oxygen status regulates the stabilization of the HIF-1 protein under conditions of sufficient and insufficient oxygenation. In normoxic conditions, the conserved proline residues of HIF-1α undergo hydroxylation by prolyl hydroxylases (PHDs) and are bound by the Von Hippel-Lindau tumour suppressor protein (pVHL). This catalyses ubiquitination-dependent proteasomal degradation. Another factor regulating HIF-α in an oxygen-dependent manner is the factor inhibiting HIF1 (FIH1). Asparagine hydroxylation of HIF1-α (and sometimes of HIF2-α) driven by FIH1 prevents the interaction of HIF1 with its transcriptional co-activator factors, p300 and CBP, further inhibiting HIF1 transcriptional activity. However, in deprivation of oxygen or hypoxia, the oxygen-dependent PHDs and FIH cannot function, allowing HIF-1α accumulation and translocation to the nucleus, where it heterodimerizes with HIF-1β to form active HIF-1. Active HIF-1 recognizes and binds to specific promoter regions of various genes known as hypoxia-response elements (HREs) to drive the transcriptional activation of hundreds of target genes and pathways (Fig. 1) [17].

HIF-dependent signalling can promote the adaptation and selection of both cancer and stromal cells to the surrounding environment to foster changes that favour cancer progression (Fig. 2). Cancer cells show a distinct reprogrammed metabolic phenotype characterized by increased glycolysis and preferential production of lactate, as opposed to mitochondrial oxidative phosphorylation. This phenomenon is called the Warburg effect or aerobic glycolysis [18]. HIF activity switches the cell metabolism of the TME toward the glycolytic pathway, thus increasing glucose consumption and pyruvate, lactate, and H + production. While HIF-1 plays a major role in glycolytic gene regulation, HIF-2 is mainly involved in pluripotent stem cell maintenance and angiogenesis, resulting in enhanced pro-tumourigenic phenotype. HIF-1α is mainly expressed during acute hypoxia (in the first 24 h) in all tissues, while HIF-2α is stabilized during chronic hypoxia (later stages of prolonged hypoxic conditions) with tissue-specific expression. Although the expression of HIF-3α is detectable in a variety of human cancer cell lines, it is comparatively less investigated. HIF-3α lacks a transactivation domain, suggesting that this isoform possesses a suppressive effect toward the other HIF isoforms [17].

Cellular oxygen sensing and regulation of HIF-1 in normoxia vs. hypoxia

HIF-1α and HIF-2α are structurally similar except for their transactivation domain. HIF-1α generally binds HREs close to gene promoters, while HIF-2α targets transcriptional enhancers. Thus, although both have overlapping targets, several are unique target genes. The isoform specificity influencing the outcome of the transcriptional programmes has been investigated in several studies and has been found to vary depending upon the cell type and severity and duration of hypoxia [17, 19].

Despite having several nonredundant functions, there is no evidence in recent literature suggesting differential regulation of tumour immunity by HIF-1 and HIF-2 or under acute and chronic hypoxic conditions. Whether or not the different HIF isoforms have contrasting effects in mediating immunosuppression in the TME remains to be explored.

Hypoxic regions in solid tumours are infiltrated with several immunosuppressive cells, such as TAMs, MDSCs, and Treg cells, which limit access to NK cells and CD8 + T cells [20]. HIF-1 signalling plays a significant role in regulating the immunosuppressive nature of TMEs through a variety of mechanisms which are discussed here. By acting upon tumour-infiltrating immune cells, enhancing the recruitment of immune-tolerant cells, and directing the transcriptional activation of immunosuppressive factors, hypoxia has been identified to dampen the activities of effector cells like cytotoxic T lymphocytes (CTLs), NK cells, and DC cells. Hypoxia upregulates immunosuppressive regulatory T cells, MDSCs, and TAMs, promotes the secretion of immune-suppressing cytokines and chemokines, and interferes with antigen-presenting cells [8].

Research from recent decades shows that HIF-1α inhibits anti-tumour immunity by modifying the TIME cells and promoting the release of cross-immunosuppressive factors. Hypoxia drives the immunosuppressive function of Treg cells and contributes to immune tolerance by the direct binding of HIF-1 to the FOXP3 promoter region in CD4+ T cells. This promotes the transcription of Foxp3 in a transforming growth factor beta (TGF-β)-dependent mechanism to stimulate their differentiation into Treg cells [8, 21]. There is prior evidence of HIF-1α-induced TGF-β activation in tumours, such as breast cancer, where HIF-1α was recognized as a positive upstream regulator of the TGF-β1/SMAD3 pathway, leading to tumour progression and poor clinical outcome [22]. In gastric cancer, hypoxia promotes TGF-β1 secretion from tumour cells and subsequently enhances Foxp3 expression of T cells [23]. Crosstalk between HIF-1α and TGF-β is observed to drive tumour progression and aggressiveness through a combined synergistic effect. This is observed in a number of solid cancers like renal cell carcinoma and prostate cancer, where both engage in a positive interaction loop [24]. One of the features of TGF-beta1 is that it undergoes a functional change from suppression of cancer cell proliferation in early stages of cancer growth to inhibition of T-cell-mediated anti-cancer immunity in late-stage tumours [25, 26]. Studies have attempted to understand and pinpoint the underlying molecular mechanism behind these opposing functions. An interesting observation by Huang et al. in a recent study, suggests a correlation between the levels of HIF-1α expression with the TGF-beta ‘switch’ [27]. Their work confirmed that the regulatory role of TGF-β in non-small cell lung cancer (NSCLC) is affected by a change in oxygen tension. However, early-stage normoxic tumours with small sizes and relatively sufficient blood supply do not show HIF stabilization. With increasing tumour volume and the development of hypoxia, HIF-1α expression increases and binds with SMAD3 to form a transcription complex. This is the main trigger to dysregulate the early-stage function of TGF-β by altering the binding partners of SMAD3. Such an event causes TGF-β to lose its normal inhibitory effect on the proto-oncogene c-Myc and disrupts TGF-β-mediated regulation of p15/p21 proteins, thus showing a completely contrasting pro-tumourigenic effect, as opposed to its initial tumour inhibitory action. Besides being involved in Treg differentiation, TGF-β also plays an important role in the metabolic reprogramming of hypoxic tumours. Although it significantly inhibits glycolysis under normoxia, it facilitates the Warburg effect in hypoxia after the release of c-Myc inhibition. Huang et al. demonstrated how HIF-1α can change the regulatory effect of TGF-β on glucose metabolism at advanced tumour stages via the HIF-1α-SMAD3 complex [27]. This TME shift to glycolysis can, in turn, potentiate immunosuppressive events by further modulation of tumour immunity, such as by decreasing the lysosomal degradation of HIF-2 [28]. HIF-1α can also promote the recruitment of Treg cells to the TME, by stimulating the overexpression of immunosuppressive CC chemokine ligands 22 and 28, as seen in ovarian and hepatocellular carcinomas [29,30,31]. CCL28 binds to its receptor CCR10 to effectively recruit CCR10+ Treg cells to the tumour site, thus suppressing the functions of effector T cells. In basal-like breast cancer, Treg recruitment has been associated with hypoxia-induced CXCR4 upregulation in Tregs [8, 32].

One of the most widely studied consequences of tumoral hypoxia is the formation of new blood vessels (angiogenesis) to sustain the oxygen-deprived cancer cells through HIF-driven activation of proangiogenic genes, especially vascular endothelial growth factor (VEGF) [33, 34]. Elevated VEGF levels are associated with poor clinical outcomes in several tumours because, in addition to angiogenic effects, VEGF has an important role in the suppression of anti-tumour immunity. VEGF inhibits the maturation of DCs and subsequent activation of CD8+ cytotoxic T lymphocytes (CTLs). This induces an immunosuppressive TME by strongly activating Treg cells, TAMs, and MDSCs. Moreover, tumour-derived VEGF, interleukin (IL)-10, and prostaglandin E3 cooperatively induce Fas ligand expression in endothelial cells, leading to exhaustion and killing of CD8+ CTLs [35]. Under TME hypoxia, VEGF is transcriptionally activated by HIF-1α, which supports the escape of tumour cells from immune surveillance. It does so by recruiting TAMs, Tregs, and MDSCs into the TME, either directly or through upregulation of VEGF [36]. The VEGF/HIF pathway is now being targeted in several cancers, and functional crosstalk among TAMs, Tregs, and MDSCs in the hypoxic TME have been strongly associated with HIF-induced VEGF production [35, 37]. Additionally, hypoxia and TGF-β are major factors that can increase VEGF production both independently and in cooperation. Hypoxic induction of TGF-β expression produces a feedback loop, which increases VEGFA production, providing another therapeutic target [24].

High concentrations of lactic acid production in the TME due to anaerobic metabolism can also enhance VEGF expression [38]. VEGF and TGF-β are two important TME factors, which are favourable for the differentiation of macrophages into immunosuppressive M2 TAMs. TAMs coexpress a mixture of both tumour-type-specific M1 and M2 markers, and TAM polarization can be strongly influenced by their spatial arrangement in tumours. In hypoxic niches, M1 TAMs can polarise into M2-like proangiogenic and immunosuppressive phenotypes. Various factors enriched in these regions, including prostaglandin E2, TGF-β, VEGF, IL-4, IL-6, and ROS, facilitate this polarisation. Additionally, hypoxic tumour cells produce lactic acid which induces M2-associated genes, to further promote the transition of M1 to M2 TAMs [39, 40].

Hypoxia also dramatically alters the function of MDSCs in the TME and redirects their differentiation toward TAMs via HIF-1α [41]. In hepatocellular carcinoma, HIF-1α is reported to promote the migration and differentiation of TAMs from immature myeloid cells via VEGF exposure [42]. TAMs can also secrete MMP7 in hypoxic tumour regions, which can cleave Fas-ligand from the neighbouring cells and protect cancer cells from Fas-ligand-mediated killing by T cells and NK cells [43].

Apart from triggering the differentiation of MDSCs into M2 TAMs, hypoxia is also involved in accumulating and maintaining the function of MDSCs [9]. HIF can induce the recruitment of CX3CR1-expressing MDSCs by activating the transcription of CCL26 in tumour cells and can increase MDSC-mediated T-cell repression by directly binding to the HRE located in the promoter of microRNA (miR)-210 [44].

Besides recruiting immunosuppressive cells, HIF-1α can also negatively regulate functions of effector cells, by directly interfering with T-cell receptor signal transduction [45]. Hypoxia enhances the synthesis of CD39 and CD73 enzymes, which are important factors in the immunosuppressive mechanism, involving adenosine production in the TME. Adenosine is produced by hydrolysis of tumour cell-derived ATP and ADP and released in the TME through membrane channels, cell death, or granular components. Interaction of free adenosine with the adenosine A2A receptors (A2AR) on T cells that are transcriptionally induced by HIF-1 and HIF-2 in hypoxic TMEs leads to the accumulation of immunosuppressive intracellular cAMP and subsequent inhibition of T-cell proliferation and cytotoxicity [17, 46]. High Treg infiltration and TGF-β expression by HIF-1α hinder NK cell functions [47]. Zhang et al. demonstrated that increased HIF-2α levels could suppress natural killer T (NK-T)-cell activation by downregulating the expression of Fas-ligands and simultaneously inducing A2AR expression [48]. Exposure of tumour cells to hypoxia inhibits specific CTL-mediated lysis by a mechanism involving nuclear translocation of HIF-1α, phosphorylation of STAT3, and VEGF secretion by tumour cells [49].

Other diverse pathways of direct and indirect action of HIF on a variety of effector immune cells are described further.

HIF-1α induces the upregulation of several immune checkpoint ligands on tumour cells and tumour-associated cells, including PD-L1 and HLA-G.

Tumour cells and other antigen-presenting cells, like tumour-infiltrating myeloid cells, express programmed death ligand (PD-L1 or PD-L2) in response to environmental cues like cytokines, hypoxia, or growth factors. The interaction of the PD ligands with the programmed death-1 (PD-1) receptor induces T-cell apoptosis, T-cell exhaustion, and overall suppression of T-cell-mediated anticancer immunity. Inhibiting the PD‐1/PD ligand (PD‐L1, PD‐L2) binding thus reinvigorates immune rejection of tumour cells, which is a process known as immune checkpoint inhibition and can be achieved clinically using monoclonal antibodies [50]. PD-L1 expression in the TIME exploits the immune tolerance system to facilitate tumour survival and immune escape. This can be controlled by several signalling pathways. Hypoxia can upregulate the expression of PD-L1 in malignant cells and MDSCs via HIF-1α and HIF-2α. Noman M. Z. et al. reported a significant increase in PD-L1 expression in tumour-infiltrating MDSCs induced by the binding of HIF-1 to the HRE4-binding site in the PD-L1 proximal promoter. Furthermore, the blockade of PD-L1 under hypoxia enhanced MDSC-mediated T-cell activation [51]. In glioma cells, PD-L1 has been identified as a direct transcriptional target of HIF-1α, by direct binding to the PD-L1 promoter region, leading to high PD-L1 expression in hypoxic regions. Based on these results, Ding et al. hypothesized the potential success of blocking HIF-1 signalling, together with PD-L1 blockade using checkpoint inhibition therapy. Combining checkpoint inhibition with HIF blockade has indeed shown improved immunotherapeutic results in hypoxic tumours. For instance, targeting the HIF-1α/PD-L1 axis in hypoxic murine breast cancer cells has been observed to restore the activity of tumour-infiltrating lymphocytes [50].

Another indirect pathway of HIF1-mediated PD-L1 signalling has been recently studied in cutaneous melanoma. Single-cell RNA-seq analysis of human cutaneous melanoma datasets revealed a high correlation of HIF-1 and PD-L1 signalling. Although no direct HIF-1-mediated transcriptional control of PD-L1 was observed upon further investigation, HIF-1 was seen to enhance IFNγ-induced PD-L1 mRNA expression in an indirect hypoxic regulation. The authors also state from their studies that HIF-1 alone may be insufficient to induce PD-L1 expression and may cooperate with other factors to trigger its upregulation [52]. High expression levels of PD-L2 in response to hypoxic signals have also been implicated in HIF-1-mediated regulation, as seen in malignant phaeochromocytomas and paragangliomas. This interaction has been associated with TME inflammation and immunosuppression [53, 54]. However, the exact molecular mechanism behind the HIF-1/PD-L2 crosstalk remains to be investigated [54].

The immune checkpoint HLA-G is a non-classical MHC-I molecule, which normally regulates physiological immune tolerance at the foetal-maternal interface, to prevent immunological rejection of the foetus. It is also constitutively expressed in immune-privileged sites such as thymus and cornea [55]. HLA-G is abnormally expressed in tumour tissues, where its immunosuppressive roles are exploited to facilitate tumour immune escape and induction of immune cell tolerance and exhaustion [55], although its functional importance in dictating clinical prognosis has been contested in a recent study on HLA-G expression in carcinomas [56]. Regardless, HLA-G and its receptors have been identified as important immunotherapeutic targets, due to its broad range of immunosuppressive effects. This acts in combination with other crucial immune checkpoints like PD-1, to disarm anti-tumour immunity, by affecting both innate and adaptive immune responses [57]. HIF-1α activates HLA-G expression through HREs, located on the HLA-G promoter region and at exon 2; Yaghi, Layale et al. identified for the first time an HLA-G transcriptional target site of HIF-1 in their 2016 study. They detected HIF-1 to directly activate HLA-G gene expression through an HRE located in coding exon 2 by inducing acute hypoxic stress in glioma cells placed in hypoxia-mimicking conditions [58]. A comprehensive review by Garziera, Marica et al. examined several studies reporting HIF-1-induced HLA-G expression in different human cancer cell lines. They reported that opposite HLA-G transcriptional activities are observed when different tumour types are exposed to hypoxic stress, whereby HIF-1 acts as a negative or positive regulator of HLA-G, depending on the type of cell line (HLA-G− or HLA-G+). In hypoxic conditions, HLA-G− cell lines transcribed HLA-G without any observed translation of the protein. However, HLA-G-positive cell lines showed reduced HLA-G transcriptional activity and protein level, which may be linked to epigenetic regulation through methylation/demethylation of the HLA-G promoter and post-transcriptional regulations [