Low levels of oxygen or hypoxia is often an obstacle in health, particularly in pathological disorders like cancer. The main family of transcription factors responsible for cell survival and adaptation under strenuous conditions of hypoxia are the “hypoxia-inducible factors” (HIFs). Together with prolyl hydroxylase domain enzymes (PHDs), HIFs regulates tumor angiogenesis, proliferation, invasion, metastasis, in addition to resistance to radiation and chemotherapy. Additionally, the entire HIF transcription cascade is involved in the “seventh” hallmark of cancer; inflammation. Studies have shown that hypoxia can influence tumor associated immune cells toward assisting in tumor proliferation, differentiation, vessel growth, distant metastasis and suppression of the immune response via cytokine expression alterations. These changes are not necessarily analogous to HIF's role in non-cancer immune responses, where hypoxia often encourages a strong inflammatory response.
Oxygen, or “fire-air” as it was first known in 1772,1 is indispensable for aerobic life forms. An obstructed supply of oxygen can be detrimental to cells, tissues and the whole organism. Multicellular organisms have therefore evolved sophisticated biological systems in order to adequately deliver molecular oxygen to all cells in the body. Hypoxia, or low levels of oxygen, is found in many different pathological disorders. Heart disease, stroke, arthritis, wounds, transplanted organs and tumors represent biological maladies which face oxygen deprivation.
The presence of hypoxic regions in a solid tumor has been known since the 1950s due to pioneering work by Gray and Thomlinson.2, 3 With the advancement of oxygen electrodes it has now become well established that oxygen concentrations in human tumors can reach very low levels. For example, median pO2 levels in pancreatic cancer are 2.7 mmHg, while normal prostate tissue pO2 levels reach 51.6 mmHg.4 Hypoxic areas in a neoplastic growth occur due to the high proliferation rate of tumor cells which surpasses that of vasculature expansion leaving many regions without sufficient access to oxygen. To top it off, many blood vessels, which grow within a tumor, are irregular, torturous and generally leaky or nonfunctional.5–7 This generates oxygen gradients depending on distance from the blood vessel, leaving some cells under acute, or even, chronic hypoxia.
Cellular hypoxia has been shown to induce mutations8 and increase selection for malignant and metastatic cells.9, 10 More critically, the presence of hypoxia in solid tumors hampers radiotherapy, where sensitivity to irradiation (DNA damage) requires oxidation in aerobic conditions.11, 12 In addition, the irregular vasculature hinders the distribution of anticancer drugs as does altered pH and proliferation rate of hypoxic cells.13–16
The hypoxic tumor additionally accumulates different inflammatory cells, which regulate and in many cases contribute to tumorigenic growth and metastatic potential. Consequently, tumor hypoxia indicates a poor prognosis in patients.11, 17, 18 This review focuses on the transcriptional response to cellular oxygen deficiency in tumor inflammatory cells, thereby exposing the possible therapeutic potential in connection with its role in other diseases.
Hypoxia Inducible Factor and Its Regulators
To survive the loss of a molecule as vital as oxygen, several systems of molecular adaptation have evolved, the most notorious of which is the hypoxia-inducible factor (HIF). Cellular adaptation to hypoxic stress can be achieved via several methods, for example by shifting metabolism toward glycolysis,19 encouraging new vessel development (angiogenesis)20 and deregulating apoptosis21 (reviewed in detail by Fang et al.22). Most of these adaptations can be piloted by HIFs.18, 23, 24
The effect of hypoxia on gene expression was first published in 1988, revealing the relationship between hypoxia and induction of erythropoietin (EPO) gene expression.25 The HIF1 transcription factor was later identified in 1995, and with it the hypoxia responsive element (HRE) found in the 5′ and 3′ enhancer region of the EPO gene.26–28 HIF is a heterodimer consisting of a HIFα and HIFβ subunit. To date, three HIFα (HIf1α, HIF2α and HIF3α) subunits and one HIFβ subunit are known. While HIFβ is nuclear and constitutively expressed, HIFα is cytoplasmic and controlled by oxygen and iron dependent enzymes (Fig. 1). These are known as the HIF prolyl hydroxylase domain enzymes (PHDs) and are able to hydroxylate HIF1α at proline residues 402 and 564 and HIF2α at proline residues 405 and 531.29 HIF3α, on the other hand, seems to act as a dominant negative regulator of HIF1 induced gene expression.30 In oxygen deficiency, PHDs are rendered non-functional, allowing HIFα and HIFβ subunits to bind together in the nucleus. The HIF heterodimers then bind to HREs found in more than 150 genes. However, in oxygen rich environments, hydroxylation of the HIFα subunit allows binding of the Von Hippel–Lindau (VHL) tumor-suppressor protein leading to subsequent ubiquitination and proteasomal degradation of the subunit. Additionally, HIF1α activity can be regulated via hydroxylation at the asparagin 803 residue by factor inhibiting HIF1 (FIH1). This blocks the recruitment of the transcription co-activators p300 and CBP.31 The HIF oxygen sensing mechanism is conserved in all analyzed metazoan species on earth.29
To date, three PHD paralogues (PHD1–3) and a related PHD isoform (PH4; PHD4) have been identified.26, 30, 32 PHD2, also known as EGLN1 (egg-laying deficiency protein nine-like protein), has been identified as the main prolyl hydroxylase domain protein, able to silence HIF1α in normoxia.33 Unlike for PHD1 and three knockout mice, loss of PHD2 is embryonic lethal. Indeed, homozygous null (PHD2−/−) embryos die on gestation day 13.5–14.5 due to heart failure and placental defects, as demonstrated by us (Rashim Pal Singh, unpublished results) and others.34, 35 Recent research has implicated PHD2 in several disorders, for example arteriogenesis,36 polycythemia,35, 37 tumorogenesis38 and cancer metastasis.39
HIF in Cancer
Adaptations to tumor hypoxia are commonly orchestrated by the HIFs.18, 24, 40 Genes regulating angiogenesis, proliferation, cancer stem cells (CSC), immune escape and metastasis can be induced, repressed or manipulated by the HIF transcription family.41 Intriguingly, several studies have found that over-expression of HIF1α in human cancers represents a poor prognosis for tumor patients, in addition to its direct negative effect on cancer therapy.23, 42
In the past two decades a large amount of research has gone into understanding how HIFs function in neoplastic cells. In addition, our group and others have shown an important and sometimes conflicting role for PHDs during tumor cell growth and metastasis.38, 39, 43–45 However, a solid tumor consists of more than just tumor cells. The tumor micro-environment (TME) consists of fibroblasts, endothelial cells, epithelial cells and immune cells46 all of which are under the same hypoxic stress as the tumor cells. Although much attention has been invested in research on hypoxia and HIF influence on tumor cells, little has been realized on the effect of HIF in the TME.
Virchow first noticed leukocytes in tumors in the 19th century.47 Since then the role of inflammation in tumorigenesis has become established as the seventh hallmark of cancer.48, 49 The link between inflammation and cancer is of two directions: extrinsic (chronic inflammation leading to neoplasia) and intrinsic (oncogene mutations leading to inflammatory cell recruitment).50, 51 One of the best examples of the induction of cancer by chronic inflammation is inflammatory bowel disease associated with colon cancer, where inflammation is partaking in all tumorigenic steps including initiation, promotion, progression and metastasis.52 On the other hand, in an intrinsic inflammatory response, some oncogenes induce an inflammatory transcriptome allowing for increased recruitment of immune cells.53 In addition, all solid tumor growth eventually outpaces the supply of oxygen and nutrients, leading to necrosis, which in turn releases pro-inflammatory mediators that recruit more inflammatory cells.54 Macrophages have been reported to accumulate in these areas of hypoxia, likely due to the effect of hypoxia regulated release of chemoattractants such as endothelial monocyte-activating polypeptide II, endothelin-2 and vascular endothelial growth factor (VEGF) by tumor and/or stromal cells.55 Hypoxia has additionally been shown to inhibit CCR256 and CCR557 macrophage expression, which aids in immobilizing macrophages once they are inside the tumor. The ability of a tumor to recruit and manipulate immune cells and induce cell renewal and proliferation have earned them the title of “wounds that do not heal.”58
Tumors are able to recruit innate immune cells (macrophages, masT-cells, myeloid derived suppressor cells (MDSCs), neutrophils, dendritic cells (DCs) and natural killer (NK) cells) as well as adaptive immune cells (T and B lymphocytes). To date all mentioned immune cell types have the ability to play a pro-tumor role.54, 59 The most famous leukocyte tumor infiltrate is the tumor associated macrophage (TAM), which can switch from an anti-tumor or classically activated (M1) profile to a pro-tumor, alternatively activated (M2) profile once recruited to the tumor. TAMs are the most prominent population in many neoplastic growths. For example, TAMs may comprise up to 50% of the cell mass of a breast carcinoma.60 Infiltrating inflammatory cells are capable of assisting a tumor in every major aspect of cancer development: tumor angiogenesis, evading immune response, encouraging tumor proliferation and metastasis (Fig. 2). In most solid tumors the amount of recruited immune cells correlates with poor prognosis.61–63
HIFs and PHDs in Tumor Inflammation
Intriguingly, immune cells have been shown to possess the ability to induce an immune eliminating response toward a tumor.64–67 However, these cells often switch to a more suppressive and protective immune response, allowing tumors to escape detection. Much work has already been done on HIF and PHD expression in tumor cells in relation to evasion of immune surveillance and release of immuno-suppressive molecules.38, 43, 68–70 This review, however, describes HIF related effects on tumor associated immune cells.
The M1/M2 plasticity of macrophages is an important feature of immune tolerance and suppression in the TME.71 M1, or classically activated macrophages, are found in healthy or inflamed tissue. In addition to their ability to kill microorganisms, present antigens to T-cells and produce immune-stimulatory cytokines, M1 macrophages are cytotoxic toward tumor cells and induce tumor destructive reactions. On the other hand, however, TAMs recruited to the tumor tend to switch toward a more pro-tumor, non-inflammatory M2 profile. M2 macrophages have poor antigen presenting ability and suppress T-cell activity and proliferation72 and are better adapted to scavenging for debris, encouraging angiogenesis and remodeling damaged tissue. TAMs switch to an M2 profile mediated by their exposure to immune suppressive molecules from the TME (for example IL4, IL13 or IL10) as well as to tumor hypoxia. Indeed, these monocytic cells are often found to accumulate in hypoxic regions of a tumor, where they upregulate HIF-1α and −2α.73
To better understand the effect HIF1α has on macrophages in a tumor, Werno et al. looked in vitro using tumor spheroids cultured with WT (wild type) and HIF-1α (−/−) macrophages. They found both groups of macrophages infiltrate at a similar rate but HIF-1α(−/−) macrophages developed more into a M2-profile and less tumor cytotoxic population than WT macrophages. When stimulated with LPS/IFNγ, HIF-1α(−/−) macrophages showed decreased expression of pro inflammatory Interleukin 6 (IL6), Tumor Necrosis Factor-alpha (TNFα), and inducible nitric oxide synthase (iNOS) as was well as increased expression of CD206.74
The interaction between myeloid cells and T-cells in the tumor environment has been revealed to be vital in malignant progression. In the Mouse Mammary Tumor Virus Polyoma Middle T (MMTV PyMT) model mice lacking T-cells did not develop metastasis, which was attributed to the effect of IL4 expressing T-cells on TAMs.75 Soon after, Doedens et al. found HIF expression in tumor macrophages to be responsible for T-cell function.76 They show delayed tumor progression and an increase in tumor cell death in mice lacking HIF1α in myeloid cells (MMTV-PyMT-LysM:cre-HIF1) compared with control mice. Ex vivo they were able to show HIF1α can induce iNOS and Arg1 in tumor myeloid cells, both of which can induce immune-suppression in T-cells.77, 78 In this model, targeted deletion of HIF1α in myeloid cells led to T-cell activation and release from immunosuppression. Studies have shown that NO can inhibit the PHD activity during normoxia, leading to HIF1α-accumulation. Conversely, during hypoxia NO led to reduced HIF1α levels.79 Interestingly, results generated from MMTV-PyMT-iNOS−/− mice suggest a role for iNOS (which catalyzes the production of NO) in inducing a more aggressive tumor phenotype.80 In addition to NO, other reactive oxygen species (ROS) have also been reported to affect the rate of HIFα hydroxylation.81–83
In addition to HIF1α's prominent role in tumor inflammation, HIF2α has also been found to be highly expressed in TAMs.84 Although both HIF1 and 2 are structurally related and bind the same HREs, they are able to regulate common and unique genes (reviewed by Hu et al.85). For example, expression of HIF2α on tumor infiltrating macrophages has been shown as a prognostic risk of local recurrence after radiation therapy.86 To understand what role HIF2α is playing in these cells, Imtiyaz et al. looked at mice lacking HIF2α in myeloid cells (LysM:cre-HIF2α) and found them to be resistant to LPS endotoxemia in addition to a reduced ability to mount an inflammatory response.87 Using hepatocellular and colon carcinoma models in mice they demonstrated decreased infiltration of HIF2α TAMS, indicating that HIF2α is essential for TAM migration into these lesions and in turn promotes tumor progression. However, a direct comparison with the activity of HIF1α in these models was not performed.
Interestingly, work from 2003 and 2005 on non-tumor models demonstrated that HIF1α mediated the inflammatory response in macrophages88 and bacterial killing.89 This places HIF1α in a pro-cytotoxic light, whereas most in vivo tumor inflammation work with TAMs (as stated above) portrays HIF1α to be an inducer of an M2 pro-tumor profile. This strongly suggests that HIF1α's role in macrophages depends on their immediate environment, in addition to the specific cell types surrounding them. HIF1α's dual role could prove interesting for further investigation in the hopes of leading inflammatory cells toward a cytotoxic and aggressive M1 macrophage in the tumor.
T-cells in tumors have been shown to play an important role in immune-surveillance in mice90 and humans.91 Their inability to induce an immune response despite their high presence in solid tumors is an indicator of tumor cell proficiency in silencing and suppressing the immune response.92, 93 An interesting method of tumor immune protection has been suggested to be HIF1α related. HIF1α expression in T-cells can be induced by hypoxia and hypoxia-independent mechanisms via the T-cell receptor (TCR) and PI3K mediated pathways.94, 95 In 2001, Lukaschev et al. showed that T-cell activation leads to the up-regulation of a short HIF1α (I.1) isoform.96 The I.1 encodes for the HIF1α protein without the first 12 N-terminal amino acids but retains DNA binding and transcriptional activity.97 Successive work using mice with knock out of the short I.1 isoform and T-cell targeted knock down regulation of HIF1α (using Lck:cre mice) indicated that HIF1α may play an inhibitory role in T-cell function, where absence of HIF1α in T-cells leads to up-regulation of T-cell function.98 These results suggest that HIF1α is not only involved in oxygen homeostasis but also works as a negative regulator of T-cells. The authors propose that this pathway, normally used to protect healthy tissues from collateral immune damage, is hijacked by tumors for protection from the immune system.99 Recent studies have also indicated a role for T-cell HIF1α in vascular inflammation and remodeling,100 however this has yet to be shown in tumors.
Anti-inflammatory CD4-FoxP3+ regulatory T-cells (Treg) play an important immune suppressive role in tumors whereby they are able to overcome the anti-tumor activity of CD8 cytotoxic cells, NKs and DCs.101 Facciabene et al. identified HIF1α's role (tested in 17 human ovarian cancers) in promoting the recruitment of Tregvia over-expression of CCL28 by tumor cells.102 Interestingly, the role of HIF1α on differentiation of naive CD4 T-cells into T17 or Treg has recently been reported. Dang et al. indicated that HIF1α expression in T-cells tips the balance in favor of pro-inflammatory T17 by inducing transcriptional activation of RORγt.103 Once again, HIF seems to play a dual role inducing pro-inflammatory T17 cells from naive T-cells, or inhibiting T-cell function altogether.
Myeloid derived suppressor cells
MDSCs are a heterogeneous population of premature myeloid cells found to perform an important inhibitory role in tumors in addition to other pathological diseases.104, 105 MDSCs are part of an immune-suppressive network shown to be responsible for defects in T-cell activation in cancer. In humans, MDSCs are defined as a population which lack mature markers of myeloid and lymphoid cells but express CD11b and CD33. In mice, MDSCs carry markers for both CD11b and Gr1, where they make ≤ 3% of splenocytes in a tumor-free mouse and expand to more than 20% in a tumor-bearing mouse. In a cancerous growth, these cells primarily differentiate into TAMs.106–108 Work done by Corzo et al. identified HIF1α to be primarily responsible for MDSC differentiation and function in the TME. Comparing MDSC populations in a tumor with the same population in a spleen, they found that non-tumor MDSCs could suppress antigen specific CD8 T-cells but failed to inhibit nonspecific T-cell function. Tumor MDSCs were able to suppress both and differentiate more readily into macrophages. The use of DFO (desferrioxamine, a hypoxia mimetic) as well as bone marrow transplantation with HIF1α knock-out (KO) bone marrow led to a suppressive activity and significant induction of arg1 and iNOS expression. Therefore, HIF1α seems to be responsible for their functional nonspecific suppressive capacity and differentiation in a tumor.109
Other inflammatory cells
HIF1α has been exposed to promote innate immune functions of DCs110, 111 and mast cells,112, 113 albeit in non-tumor immune models. Once again, HIF1α seems to play an important role in DC activation in inflammatory situations unlike their smoldering role in tumors.66, 114 In addition, HIF1α has been shown to be responsible for regulating neutrophil survival under hypoxia via NFκB.115 Cummins et al. have associated hypoxia's activation of NFκB to the presence of IKKβ's evolutionarily conserved LxxLAP consensus motif in its activation loop for hydroxylation by PHD1.116 Notwithstanding the above, little is known about the role of these immune cells under hypoxic conditions in a tumor.
Hypoxic and necrotic areas of a tumor tend to accumulate high amounts of myeloid cells. The ability of these immune cells to induce angiogenesis (growth of new blood vessels), and lymph vessels as well as metastasis has been established in the last decades. In such environments TAMs up-regulate pro-angiogenic (M2) genes such as VEGF, IL8, fibroblast growth factor 2 (FGF2), angiopoietin, in addition to Matrix Metalloproteinases (MMP-2, MMP-7, MMP-9, MMP-12) and cyclooxygenase-2.117, 118
The presence of TAMs has been correlated with the induction of tumor vasculature, known as the angiogenic switch, and poor patient outcome.50, 119, 120 Mice with a null mutation of Csf1 (colony stimulating factor 1) have exhibited reduced macrophage presence in the primary mammary tumor in addition to a delayed angiogenic switch. Over-expressing the CSF1 in wild type mice induced an early influx of macrophages to the neoplastic lesion and an early onset of the angiogenic switch.121 Targeted silencing of the vegf gene in macrophages led to the inhibition of the angiogenic switch, both in the MMTV-PyMT model as well as in the Lewis Lung Carcinoma (LLC) tumor model. However, due to the normalization of the vasculature the tumors were less hypoxic. They therefore exhibited larger volume and were more aggressive than tumors inoculated into WT animals.122 Although it is well established that vegf is a HIF1α regulated gene, breast carcinoma in mice lacking HIF1α in myeloid cells (MMTV-PyMT-LysM:cre-HIF1) did not exhibit differences in VEGF-A levels nor did their vasculature change. The tumors did, however, exhibit decreased progression76 attributed to their escape from immune suppression. Looking in vitro using myeloid HIF1α KO cells and tumor spheroids, Werno et al. have shown a decrease in the ability of macrophages to stimulate differentiation of stem cells to CD31+ cells in addition to a decrease in their cytotoxic ability.74 Roda et al. used LysMcre-HIF1 or HIF2 mice together with a B16F10 melanoma model to investigate the use of Granulocyte-macrophage colony-stimulating factor (GM-CSF) for treatment of melanoma. GM-CSF treatment of LysMcre-HIF1 mice induced sVEGFR1 (soluble form of VEGF receptor 1) which inhibits the activity of VEGF, leading to reduced tumor growth, angiogenesis and metastasis. In LysMcre-HIF2 mice, GM-CSF treatment induced VEGF but not sVEGFR-1 expression and did not decrease tumor growth, angiogenesis or lung metastasis.123
HIF2α has been shown to be more up-regulated in TAMS than in normal tissue macrophages.84 Looking at several human invasive breast carcinomas Leek et al presented a correlation between expression of HIF2α in TAMs and tumor vascularity in addition to tumor grade.124 An inverse correlation was found between TAM HIF2α and tumor thymidine phosphorylase expression. Experiments with over-expression of HIF2α but not HIF1α in normoxic macrophages led to increase in transcription of pro-angiogenic genes VEGF, IL8, platelet-derived growth factor β (PDGF β) and angiopoietin-like 4 (ANGPTL4).118
TAMs also express and secrete proteolytic enzymes such as MMP1125 and MMP7.126 These enzymes facilitate endothelial migration by digesting the extracellular matrix and thereby encourage angiogenesis. In line with this, MMP9−/− macrophages exhibited slower invasion to human tumor xenografts, which was associated with reduced angiogenesis.127
In addition, HIF1α recruits TEMs (Tie2 expressing monocytes) to a tumor.128 TEMs are a sub-population of circulating monocytes identified by their ability to express Tie2 which binds angiopoietin2 (Ang2) on tumor cells, which in turn represses the anti-angiogenic cytokine IL12.129 Facciabene et al. looked at 17 human ovarian cancer cell lines and found the chemokine ligand CCL28 highly up-regulated under hypoxic conditions. They demonstrated that HIF1α promotes angiogenesis and tolerance via recruitment of CCL28-induced CXCR10-expressing T-cells.102 The work of this group highlights the connection between immune tolerance displayed in the tumor environment and angiogenesis.
Silencing of PHD2 in several colorectal tumors exhibited decreased recruitment of bone marrow derived cells (BMDCs) and displayed an increase in amount of CD31 vessel staining, independent of the HIF activity. The influence of PHD2 was linked to angiogenesis, mediated through NFκB-induced angiogenin and IL8. These shPHD2 tumors showed dramatic increase in volume.43 In contrast to their findings, we silenced PHD2 in several different mouse cell lines but found no differences in amount of inflammatory cells recruited to the tumors in vivo. Surprisingly, these cell lines, when inoculated into mice, exhibited a large decrease in tumor volume compared to WT tumors. Interestingly, this was accompanied by induced angiogenesis, which we found to be directly dependent on the MMP-activated anti-proliferative Transforming growth factor beta (TGFβ) pathway.38, 130
BMDCs found in a tumor, especially TAMS, have been shown to play an important role in tumor malignancy. In addition to their effect on tumor immune suppression and angiogenesis, they are involved in invasion, migration and intravasion of tumor cells.120 Studies have implicated early presence of BMDCs at metastatic sites with encouraging subsequent colonization by tumor cells.131 For example, in a breast cancer model, using primary MMTV-PyMT-cells inoculated into Csf1 knockout mice, TAMs have been shown to mediate extravasation, establishment and growth of metastasis.132 Much of the regulation of gene expression stated in the angiogenesis-section is likely to have an influence on metastasis.
Research on cancer metastasis has also indicated a link with hypoxia.133–135 Recently, the metastatic niche has become a well-established mechanism, where specialized inflammatory cells are found early-on in the metastatic organ. Wong et al. found HIF1α to be an important regulator of the breast cancer metastatic niche. They show tumor cells express LOX (lysyl oxidase), which is induced by HIF1α and is required to prepare the lung for arrival of BMDCs. Indeed, HIF1α induction of LOX and LOX ligand was important for ECM modeling which, in turn, facilitated the ability of BMDCs to invade ex vivo.136 Knock down of HIF1α led to a decrease in collagen crosslinking, recruitment of CD11b cells from bone marrow and lung metastasis in the MDA-231 breast cancer cell line.
Human macrophages exposed to hypoxia have elevated levels of MMP7 in vitro and in avascular areas of human tumors.137 Another group has, however, found MMP7 to be regulated by hypoxia in hepatoma cells in an HIF1α-independent manner.138 Grimshaw et al co-cultured macrophages and tumor cells together in the presence of endothelin-1 and −2, which are both up-regulated in hypoxic regions of a tumor and have receptors on TAMS and tumor cells. This induced MMP2 secretion by tumor cells which stimulated the invasive ability of tumor cells.139 Both MMP2 (regulated by HIF1α140) and MMP9 (regulated by HIF2α141) are HIF responsive.
HIFs have a remarkable ability to induce both pro- and anti-inflammatory effects in the main immune cells (T-cells and macrophages). So far, we know of HIF's ability to push an anti-immune, pro-tumor scenario in a tumor, but much of the evidence gathered in inflammatory models indicates its capability of doing the opposite.142 Further understanding of this dual mechanism could lead to manipulating this process to the advantage of tumor therapy via immune cells which are, advantageously, constantly homing to the neoplastic growth.
The immune cell consortium within a tumor is not exclusively pro- or anti-tumor.143–145 A tug of war seems to be constantly taking place within the tumor, specifically during initiation, between the tumor associated immune cells themselves and their microenvironment. Tipping the balance toward one scenario and not the other might lie in the hands of highly manipulative transcriptional factors such as HIF or its regulators.
Authors thank Kristin Franke for helpful discussions. We apologize to any colleagues whose work was not cited because of space limitations.