The Role of Hypoxia in Re-Educating Macrophages in the Tumour Environment

evidence demonstrating that TAMs initially have an M1-like phenotype in areas of chronic inflammation where tumours commonly develop. These, however, respond to secreted cytokines, chemokines, growth factors and stress signals in the (hypoxic) tumour , IL-8, adrenomedullin, CXCR4, and angiopoietin-2. Induction of these genes suggests a potent, pro-tumoural macrophage however, required for macrophage functional maturation. These findings further suggest that macrophage differentiation may be facilitated by hypoxia. Tumor microenvironment represents an extremely dynamic niche shaped by the interplay of different cell types (e.g. tumor cells, stromal cells), their soluble products (e.g.cytokines, chemokines and growth factors) and varied physico-chemical conditions (e.g low oxygen concentration or hypoxia). Recent studies have identified myelomonocytic cells as key players in regulating the tumor microenvironment and hence, tumor progression in a variety of cancers. In view of these findings, the present book attemps to provide a comprehensive account of the diversity of tumor microenvironment across different cancers and how myelomonocytic cells have taken the center-stage in regulating this niche to direct cancer progression. A better understanding of the myelomonocytic cells and the mechanisms by which they regulate cancer progression will open new vistas in cancer therapeutics.


Introduction
Monocytes and macrophages are myeloid cells which originate in the bone marrow and are essential in the primary defence against infection by bacteria, viruses and other pathogens. These cells circulate as monocytes in the bloodstream before undergoing extravasation and migration into adjacent tissues, where they differentiate into resident macrophages. Considerable monocyte extravasation occurs at the initial stages of inflammation, wound healing, tumour onset and various other diseases in response to chemotactic signals. In many instances these inflamed and/or diseased tissues have been shown to include areas of extremely low oxygen tension, termed hypoxia, by the measurement of oxygen concentrations using microelectrodes, use of hypoxic cell markers and/or expression of specific hypoxia-upregulated proteins. Such hypoxic areas are evident in the majority of malignant human cancers, including those of the breast, brain, cervix, head/neck, and soft tissue sarcomas (Raleigh et al., 2001;Vaupel et al., 1989), and are caused by an inability of the supporting vasculature to keep up with the oxygen demands of the rapidly increasing tumour mass (Shannon et al., 2003;Vaupel et al., 2005).
As with inflammation, extensive monocyte extravasation is also an early event in cancer development. Infiltrated monocytes differentiate into tumour-associated macrophages (TAMs), a process which is driven by tumour-secreted chemoattractants (Murdoch et al., 2004). Moreover, TAMs accumulate in high numbers within hypoxic areas, which drives a change in their gene expression through the modulation of such transcription factors as hypoxia-inducible factors (HIFs) 1 and 2 (Burke et al., 2003;Talks et al., 2000), activating transcription factor-4 (ATF-4), and early growth response-1 (egr-1) (Elbarghati et al., 2008). Subsequently, a wide panel of protumour genes are upregulated by hypoxic macrophages which could support tumour growth, survival and metastasis (Fang et al., 2009). This is thought to explain the correlation between high numbers of TAMs and poor patient prognosis in many types of human tumours (Fujimoto et al., 2000;Hamada et al., 2002;Hanada et al., 2000;Heidl et al., 1987;Leek et al., 1996;Lissbrant et al., 2000;Salvesen and Akslen, 1999). Fig. 1. Tumour hypoxia drives monocyte infiltration, polarization and transcription of hypoxia-regulated genes. a. Tumours and infiltrated macrophages secrete chemoattractants, resulting in the recruitment of monocytes from the blood. Hypoxic conditions commonly found within tumours enhance the polarisation of macrophages toward a protumour phenotype, which leads to the upregulation of a wide array of tumour-supporting genes (such as those shown in the figure) and the downregulation of MHC II. Almost all murine TAMs derive from a population of monocytes defined by Ly6C hi CX 3 CR1 lo expression, which continuously seed tumours. Two types of murine TAMs, MHC II hi and MHC II lo , have been shown to be located in normoxic and hypoxic areas of tumours, displaying M1 and M2 characteristics, respectively. TIE2-expressing macrophages (TEM) are recruited in response to release of Ang2 (as well as upregulation of Tie-2) by the hypoxic core. TEMs associate with blood vessels and promote tumour angiogenesis. Monocyte/macrophage-derived factors in black, tumour-derived factors in blue. b. Hypoxic conditions result in the stabilisation of HIF-1 and -2 in macrophages, which are then able to bind to a constitutively expressed common subunit, located in the nucleus. The active transcription factors then bind to HREs in a variety of genes (shown at bottom left) which regulate the immunosuppressive and protumoural functions of macrophages.

Monocyte infiltration into areas of hypoxia
Following their infiltration into tumours, macrophages have been shown to accumulate specifically in hypoxic areas, a phenomenon which is thought to be guided by hypoxiainduced chemoattractants and maintained by the suppression of TAM motility in these areas by hypoxia (reviewed by Murdoch et al. (2004)). As would be expected, these oxygen-deprived regions of tumours have been found to have elevated levels of VEGF, produced by both tumour cells and macrophages (Brown et al., 1995;Lee et al., 1998;Lewis et al., 2000). In addition, VEGF expression in a murine model of Lewis lung carcinoma was shown by immunohistochemistry to correlate with pimonidazole stained areas, a marker for hypoxia (Kim et al., 2001). As mentioned previously, this factor is a chemoattractant for monocytes and macrophages and therefore is likely to play a major role in the accumulation of TAM at these sites . However, it is worth noting that there is not always a correlation between hypoxia and VEGF expression in human tumours (Janssen et al., 2002;Raleigh et al., 1998). Matschurat and colleagues (2003) found that another monocyte chemotactic, EMAP II, is expressed at high levels in perinecrotic areas of methylcholanthrene fibrosarcomas and B16 murine melanomas in an inactive form, pro-EMAP II. Additionally, they showed that hypoxic tumour cell supernatants in vitro demonstrated an increase in EMAP II at the protein level, which was not supported by an induction at the mRNA level. This suggests that the active protein can be induced under hypoxia without the need for transcription, possibly through cleavage of pre-EMAP II to its active form by proteases released from necrotic cells (Zhang and Schwarz, 2002), providing a rapid mechanism for EMAP II upregulation and subsequent macrophage infiltration. This effect explains why macrophages are found at sites positive for EMAP II expression in uveal melanoma (Clarijs et al., 2003).
Also known to be regulated by hypoxia are endothelins, a family of secretory vasoactive peptides involved in vasoconstriction. They also have co-mitogenic functions, enhancing the effects of other such growth factors as PDGF by initiating intracellular signalling through endothelin receptors, ET-RA and ET-RB. Studies of endothelin regulation under hypoxic conditions demonstrated a co-localisation of hypoxia and endothelin ET-2 expression in murine mammary tumours (Grimshaw et al., 2002a). This is significant since ET-2 is thought to bind to ET-RB on macrophages and act as a chemoattractant, explaining the correlation seen between ET-2 expression and ET-RB-positive macrophages in breast tumours (Grimshaw et al., 2004;Grimshaw et al., 2002b). Furthermore, ET-1 (which acts through endothelin-1 receptor A) was recently shown to enhance the invasion and migration of both tumour cells and macrophages. The contribution of these factors to metastasis was supported by the finding that tumour expression of ET-1 and activity of its receptor are required for the development of lung metastases, through a process which is dependent on macrophage infiltration of the lung (Said et al., 2011).
More recently, Wang et al. (2012) identified stromal-derived factor-1 (SDF-1/CXCL12) as a tumour-derived chemoattractant and survival factor for TAMs. In a murine glioma model they showed that SDF-1 kd tumours have a different association of TAMs with hypoxia, implying that the secretion of this factor by tumour cells is critical for the accumulation of TAM in hypoxic areas of murine glioma. This factor is known to bind to its receptor, CXC receptor 4 (CXCR4), which is upregulated through a HIF-1-dependent mechanism in www.intechopen.com The Role of Hypoxia in Re-Educating Macrophages in the Tumour Environment 93 monocytes and macrophages (as well as endothelial cells and tumour cells). Therefore, SDF-1 and its receptor, CXCR4, are very important for the chemotaxis of TAMs to hypoxic tumour sites (Schioppa et al., 2003).
It has been suggested that the accumulation of infiltrating macrophages in tumours, primarily in hypoxic regions, is not just due to chemoattraction, but also to their retention in these areas. Downregulation of chemokine release by tumour cells, and of chemokine receptors by TAMs in hypoxia, effectively dampens TAM motility, thus causing large numbers of macrophages to be trapped in these sites. For example, the expression of CCR2 (the receptor for CCL2/MCP-1) and chemotactic responses to CCL2 in vitro were markedly higher for TAMs isolated from ovarian carcinomas than monocyte-derived macrophages in culture (Negus et al., 1998;Sica et al., 2000).
When cultured with human tumour ascites, the chemotactic response of fresh monocytes to CCL2 was greatly diminished, accompanied by a reduction in CCR2 mRNA levels. Furthermore, inhibition of TNF-restored CCR2 mRNA expression in monocytes cultured in the presence of ascitic fluid, demonstrating that defective CCR2 expression in TAM may be regulated, at least in part, by this cytokine in tumours (Sica et al., 2000). Therefore, it is possible that macrophage TNF-production within hypoxic areas of tumours (Guida and Stewart, 1998;Hempel et al., 1996;Scannell et al., 1993) may lead to a downregulation of CCR2 expression on TAMs, decreasing their responsiveness to chemotactic ligands.
An increase in TNF-expression is also believed to induce mitogen-activated protein kinase phosphatase 1 (MKP-1) (Grimshaw and Balkwill, 2001), a molecule which dephosphorylates extracellular signal-regulated kinase (ERK) 1/2, and p38 mitogen activated protein kinase (p38 MAPK) (Franklin and Kraft, 1997;Sun et al., 1993). Intracellular signalling via p38 MAPK and ERK1/2 is required for the chemotactic response of monocytes and monocytic cell lines to hypoxia-regulated chemokines (Ashida et al., 2001;Wain et al., 2002). Therefore, TNF-may be an important factor in the hypoxic tumour environment for the suppression of macrophage migration, via a downregulation of CCR2 and an upregulation of MKP-1 ( Figure 1a).

Hypoxia and its impact on macrophage function
For a long time it has been known that macrophages can be stimulated by environmental signals to exhibit a wide array of phenotypes (Nibbering et al., 1987;Ogle et al., 1994;van Furth, 1980). Two main polarization phenotypes of macrophages have been recognized. These include the classically activated (M1) and alternatively activated (M2) macrophage phenotypes. M1 macrophages are induced by interferon gamma (IFN-) and lipopolysaccharide (LPS). These macrophages upregulate pro-inflammatory cytokines (e.g. IL-12, IL-23, TNF, CXCL10), co-stimulatory molecules, produce reactive nitrogen and oxygen intermediates (RNI/ROI), and very little anti-inflammatory cytokines (e.g. IL-10). These cells promote inflammation, apoptosis, and microbicidal activity. Conversely, M2 macrophages are induced by IL-4 and IL-13, promote angiogenesis, cell proliferation and other tissue remodelling and protumoral functions. These cells are characterized in general by an IL-12 lo IL-10 hi phenotype, upregulate chemokines like CCL17, CCL18 and CCL22, various scavenging receptors and the production of Arginase I. Although the M1-M2 nomenclature is a useful one when assessing the phenotype of macrophages, it is however, an over-simplification. Not all macrophage fit into these two distinct populations, and so further sub-populations have been defined (Mantovani et al., 2004).
Recently, subsets of differentially polarized TAMs with distinct functions were described by Movahedi et al. (2010) in murine mammary tumours. Their findings showed that almost all TAMs from these tumours were derived from Ly6C hi CX 3 CR1 lo monocytes, where Ly6C is a monocyte/macrophage differentiation antigen regulated by IFN-, and CX 3 CR1 is a receptor for CX 3 CL1, a chemokine involved in the adhesion and migration of leukocytes. Notably, they found that hypoxic areas had higher numbers of M2-like TAMs, which increased as the tumour progressed (in certain tumours), and were shown to have potent proangiogenic effects in vivo. This also correlated with the expression of major histocompatibility complex II (MHC II), whereby MHC II hi macrophages resided in normoxic areas and displayed an M1-like phenotype, and MHC II lo macrophages resided in hypoxic areas and displayed a more M2-like phenotype ( Figure 1a). Expression of M1 molecules like Nos2 (iNOS), interleukin (Il)-1 , Il-6, Il-12 and Ptgs2 (or cycloxygenase 2, COX2) were reported in MHC II hi monocytes at the RNA or protein level. By comparison, MHC II lo monocytes expressed such M2-related molecules such as macrophage mannose receptor (MR), scavenger receptor 1 (SR-A), arginase-1 (ARG-1), CD163, stabilin-1 (STAB-1), and interleukin-4R (IL-4R ) (Movahedi et al., 2010). Fitting with their M2-like phenotype and localisation in areas of hypoxia, MHC II lo TAMs were found to have significantly elevated proangiogenic activity in vivo. The phenotypic similarity between MHC II hi TAMs and IKK -deficient macrophages implies that differences in these MHC II hi and MHC II lo TAM subsets may be driven by NF-κ activity (Movahedi et al., 2010).
Another monocyte population, thought to be distinct from MHC II lo TAMs, are the Tie2expressing monocytes (TEMs) (Figure 1a). Originally identified in tumour-bearing mice, these monocytes circulate in the mouse blood as Tie2 + CD11b + CD45 + cells. They comprise a small monocyte subset which migrate towards angiopoietin-2 (Ang-2), a TIE2 ligand that is primarily released by vascular endothelial cells; this is thought to be a possible mechanism by which TEMs are recruited to tumours (Venneri et al., 2007), and more specifically, to highly vascularised areas (De Palma et al., 2003) (Figure 1a). Their role in the promotion of angiogenesis was confirmed by De Palma et al. (2005), who found that selective depletion of TEMs in a murine cancer model resulted in the inhibition of tumour angiogenesis and growth. Furthermore, TEM depletion was found to increase the efficacy of vasculardisrupting agent (VDA) therapy of tumours, suggesting that the action of these cells counteracts the antitumour effects of VDAs (Welford et al., 2011). Understanding more about monocyte subsets uncovers new possibilities for targeting specific subpopulations, which could alter the overall balance of TAM phenotypes. Repolarisation of TAM from an M2-to an M1-like phenotype could restore their antitumour effects, leading to a better patient prognosis.
The implications of macrophage plasticity in cancer biology have gathered increasing interest, both in terms of the phenotypes driven by the tumour microenvironment, and more specifically, by the hypoxic tumour environment. Biswas and colleagues (2008) reviewed the experimental evidence demonstrating that TAMs initially have an M1-like phenotype in areas of chronic inflammation where tumours commonly develop. These, however, respond to secreted cytokines, chemokines, growth factors and stress signals in the (hypoxic) tumour microenvironment, to express more of an M2-like phenotype in established tumours. This suggests a "re-education" of macrophages, which are recruited by the tumour, initially expressing an M1-like phenotype (thus promoting an inflammatory response); however, their residency within tumours leads to their polarisation and differentiation into M2skewed TAMs, where their re-educated phenotype is one which promotes angiogenesis, tissue remodelling, immunosuppression and cell proliferation (Biswas et al., 2008).
One important feature of the tumour environment which brings about this phenotypic change in macrophages is hypoxia. Hypoxic regions of tumours commonly form due to the leaky and disorganised nature of tumour blood vessels, meaning that the rapid tumour cell proliferation often surpasses the ability of the poorly-formed vasculature to deliver required oxygen and nutrients (Shannon et al., 2003;Vaupel et al., 2001). Studies with human breast carcinomas (Leek et al., 1999) or animal tumours (Collingridge et al., 2001) have shown that hypoxic tumours contain higher numbers of TAMs. A positive correlation that is also seen between hypoxia and TAM numbers in secondary liver tumours that form as metastases from breast and colorectal tumours (Stessels et al., 2004). There is an inverse relationship between TAM infiltration and patient prognosis seen in many human cancers (Fujimoto et al., 2000;Hamada et al., 2002;Hanada et al., 2000;Heidl et al., 1987;Leek et al., 1996;Lissbrant et al., 2000;Salvesen and Akslen, 1999), which implies that these macrophages adopt a pro-tumoural phenotype, contrasting with their more classic role as pathogen and tumour killing cells and with their ability to initiate an immune response.

Transcription factors HIFs 1 and 2
The best understood transcription factors mediating the response of macrophages to hypoxia are the hypoxia-inducible factors (HIFs) 1 and 2 (Burke et al., 2003;Fang et al., 2009;Talks et al., 2000) (Figure 1b). Both HIFs are heterodimers consisting of an individual subunit and a common subunit which is constitutively expressed. HIF-1 and HIF-2 are tightly controlled, such that, in the presence of oxygen they are quickly degraded by the ubiquitin-proteasome pathway within the cytoplasm. However, hypoxic stress causes an increase in production and stabilisation of these subunits, which are then able to complex with the subunit within the nucleus and bind to hypoxic response elements (HREs) of certain oxygen-sensitive genes to drive transcription (Jiang et al., 1996;Semenza, 2002). The hypoxia-responsive genes regulated by HIFs are known to be involved in tumour proliferation, metabolism, angiogenesis, apoptosis and metastasis (reviewed by (Harris, 2002)). The data supporting the expression of HIFs by macrophages, especially TAMs, is currently unclear. Talks et al. (2000) showed that hypoxia predominantly upregulates HIF-2 in the pro-monocytic cell line, U937, and the HIF over-expression studies by White et al. (2004), suggested that HIF-2 might be more important for macrophage pro-angiogenic responses to hypoxia. In contrast, human macrophages exposed to tumour-specific levels of hypoxia in vitro, as well as those in hypoxic areas of several human tumours in vivo, were shown to be capable of inducing high levels of HIF-1 as well as HIF-2 (Burke et al., 2002). phenotype. Using small interfering RNA (siRNA), this gene expression was shown to be mediated via HIF-1 and 2 signalling, thus implicating these transcription factors in the generation of the hypoxia-driven, tumour-promoting macrophage phenotype.
Both HIF-1 and 2 bind to the HRE sequence contained in the promoter region of the VEGF gene and cause its upregulation (Ema et al., 1997;Flamme et al., 1997;Tian et al., 1997). Evidence that TAMs themselves upregulate VEGF in poorly vascularised tumour areas  suggests that hypoxia, at least in part, causes TAMs to align with tumour cells in their pro-angiogenic function to increase the supply of oxygen to these areas. Interestingly, the binding of HIF-1 and 2 to the promoter region of VEGF in GM-CSF-cultured macrophages is thought to have antagonistic effects on angiogenesis, whereby HIF-1 induces VEGF production and HIF-2 induces the production of the soluble VEGF receptor, sVEGFR-1, in low oxygen conditions. The secretion of sVEGFR-1 is able to neutralise VEGF biologic activity, inhibiting its angiogenic effect; this indicates that the binding of these two transcription factors may have opposing effects on the regulation of angiogenesis (Eubank et al., 2011).
Less is known about the third member of the HIF family, HIF-3 , which shows high similarity to HIFs 1 and 2 and also forms heterodimers with the same subunit. However, experiments so far show that this factor lacks the C-terminal transactivation domain (CTAD) (Gu et al., 1998), and acts as a dominant-negative regulator of the HIF pathway by antagonising the effects of HIFs 1 and 2 (Makino et al., 2001). HIF-3 was found to be constitutively expressed in monocyte-derived macrophages (MDMs), and was not responsive to hypoxic conditions in either monocytes or MDMs (Elbarghati et al., 2008). However, it was found to be hypoxia-responsive in lung epithelial cells (A549) at both the mRNA and protein level (Li et al., 2006), and so it is clear that further investigation into this transcription factor, with regards to its expression and importance, is needed.
HIFs are not the only hypoxia-responsive transcription factors. Both activating transcription factor-4 (ATF-4) and early growth response-1 (Egr-1) are upregulated in response to hypoxia in several murine and human tumour cell types (Ameri et al., 2004;Yan et al., 1999) and macrophages (Elbarghati et al., 2008), respectively. Both ATF-4 and Egr-1 proteins were found to be transiently upregulated in macrophages following a short hypoxic incubation, but there was no induction seen in monocytes. Interestingly, hypoxic treatment caused Egr-1 protein accumulation in macrophages in both the nucleus and the cytoplasm, in contrast to HIFs 1 and 2, and ATF-4 (Elbarghati et al., 2008), and is thought to play a role in monocyte differentiation into macrophages. Kharbanda et al. (1991) showed that M-CSF-stimulated monocytes demonstrate a dose-dependent increase in EGR-1 mRNA levels, and that inhibition of monocyte differentiation with dexamethasone also abolishes this EGR-1 induction. Therefore, since hypoxia increases the levels of Egr-1 protein, it is possible that hypoxia accelerates the differentiation of monocytes into TAMs (Elbarghati et al., 2008). This is also supported by the work of Oda and colleagues (2006), who demonstrated an increase in the expression of HIF-1 and HIF-1 in differentiating THP-1 cells and human monocytes from peripheral blood. RNA interference studies determined that, although HIF-1 is not essential for macrophage differentiation, it is, however, required for macrophage functional maturation. These findings further suggest that macrophage differentiation may be facilitated by hypoxia.
Evidence that hypoxia induces a protumour phenotype in TAMs is not just limited to observed changes in RNA and protein expression; functional studies with hypoxic or HIFexpressing TAMs have also confirmed this phenotypic shift in macrophages. TAM-induced endothelial cell migration and tubule formation, reported by Chen et al. (2011), confirms the angiogenesis-promoting actions of TAM suggested by RNA and protein expression. More specifically, HIF-1 has been implicated in these protumour functional effects of TAMs. Doedens et al. (2010) report a dose-dependent suppression of T-cell proliferation by macrophages, demonstrating this immunosuppressive effect to be enhanced under hypoxia in a HIF-1 -dependent manner.

HIF relation with other pathways
It is clear that hypoxia drives a tumour-promoting phenotype in macrophages -it does this through the activation of hypoxia-responsive transcription factors, predominantly HIF-1 and 2, and their crosstalk with other signalling pathways. One such example is Toll-like receptor (TLR) signalling; TLR receptors are known to activate the innate immune system upon recognition of various pathogen-associated molecular patterns (PAMPs), including Lipopolysaccharide (LPS), bacterial DNA, and double-stranded RNA (Kaisho and Akira, 2006). In humans there are 10 functional members of the TLR family (TLR1-TLR10), of which, TLR4 is possibly the most involved in macrophage hypoxic response. This particular receptor recognises LPS, but has more recently been shown to be a receptor for certain endogenous molecules associated with damaged cells and tissues (Zhang and Mosser, 2008).
In their study of the relationship between hypoxic stress and TLR activity of macrophages, Kim et al. (2010) showed that hypoxia (and the hypoxia mimetic, CoCl 2 ) increased TLR4 messenger RNA and protein expression in the murine macrophage cell line, RAW264.7. This was unique to TLR4 and not seen with any of the other TLRs. Through the manipulation of macrophage HIF-1 gene expression, they demonstrated that hypoxic upregulation of TLR4 was dependent upon HIF-1 signalling, as well as showing that overexpression of HIF-1 enhanced TLR4 expression. Using chromatin immunoprecipitation (ChIP) they discovered that HIF-1 binds to the TLR4 promoter under hypoxic conditions, and the resultant induction of TLR4 in these macrophages increased the expression of interleukin-6 (IL-6), cyclooxygenase-2 (COX-2) and interferon-inducible protein-10 (IP-10) (Kim et al., 2010). Therefore, it is likely that, at sites which are challenged by hypoxic stress, macrophages upregulate TLR4 and become more sensitive to infection and inflammatory signals.
In addition to HIF-1 regulating TLR4, Sumbayev (2008) demonstrated that the inverse is also true. In human myeloid cells, TLR4 signalling (induced by the gram-negative bacterial ligand, LPS), activates crosstalk of HIF-1 and apoptosis signal-regulating kinase 1 (ASK1) pathways. Through the activation of p38 mitogen-activated protein kinase (p38 MAPK), ASK1 was found to stabilise HIF-1 , and knockdown of HIF-1 led to a reduced TLR4dependent induction of pro-inflammatory cytokines. Similarly, TLR7 and 8 (involved in the recognition of viral single-stranded RNA) were also found to induce HIF-1 , although ASK1 was not found to be involved (Nicholas and Sumbayev, 2009).
Evidence has been given for the role of HIFs 1 and 2 in these pro-tumour actions of TAM in hypoxic areas, but more recently it has emerged that HIF-1 may also be responsive under normoxic conditions. Such stimuli as LPS, cytokines (e.g. TNF ), growth factors, insulin, thrombin and vasoactive peptides cause HIF-1 stabilisation in normoxia, via nuclear factorkappa B (NF-κB) signalling. This key transcription factor was shown by Rius et al. (2008) to be upregulated following a 2-4h exposure of murine bone marrow-derived macrophages (BMDMs) to low oxygen. They also demonstrated that basal levels of NF-κB were required for the accumulation of HIF-1 protein in hypoxic cells, using macrophages from an IKKbeta knock-out (IKK -/-) mouse. This implies that IKK , an important activator of NF-κB through phosphorylation-induced degradation of IkB inhibitors, has important contributions to macrophage response to hypoxia. Since NF-κB has a crucial and well characterised role in inflammation, IKK represents a significant molecule which may link the hypoxic response to innate immunity and infection (Rius et al., 2008).

Hypoxia, macrophage function and tumour progression
By contributing to angiogenesis, metastasis, invasion, immunosuppression, chemo-and radio-resistance, and altering metabolism, macrophages are known to greatly influence the survival and progression of cancer (see Biswas et al. (2008)). This phenotype of TAMs is also known to be influenced by hypoxia, which induces a distinct protumour phenotype. Expression of various growth factors, including fibroblast growth factor 2 (FGF2), plateletderived growth factor (PDGF), placental growth factor (PGF), and hepatocyte growth factor (HGF), have been found to be upregulated in vitro by macrophages under hypoxia (White et al., 2004). These factors, in addition to VEGF, function as tumour cell mitogens and support tumour growth in hypoxic regions (Fang et al., 2009;Lewis et al., 2000).
Another key process in tumour progression is angiogenesis. The expression of VEGF (a potent mitogen and well characterised pro-angiogenic factor) by hypoxic macrophages has been discussed previously. However, other key proteins reported by White and colleagues (2004) include CXCL8, angiopoietin, cyclooxygenase-2 (COX-2) and inducible nitric oxide synthase (iNOS), all of which were identified in cDNA arrays as genes that are transcriptionally upregulated in primary macrophages under hypoxia. Induction of these genes by macrophages is likely to be crucial for tumour angiogenesis.
Additionally, hypoxic TAMs also release tissue factor (CD142) (Compeau et al., 1994) and macrophage inhibitory factor (MIF) (Schmeisser et al., 2005), which are thought to be involved in invasion and metastasis. Tissue factor expression by hypoxic TAMs (as well as tumour cells, endothelial cells and fibroblasts), induces the production of thrombin, which in turn promotes tumour cell metastasis (Versteeg et al., 2004). Furthermore, MIF has been www.intechopen.com shown to promote tumour cell motility in a murine colon cancer cell line in vitro and in vivo (Sun et al., 2005). These factors may act indirectly through matrix metalloproteases (MMPs), such as MMP-9, which is stimulated by MIF and degrades the basement membrane and extracellular matrix (ECM) . This weakens the attachment of tumour cells to these structural supports and enables their subsequent invasion and metastasis. Further evidence of MMP induction comes from a co-culture of macrophages with tumour cells in hypoxia, which revealed an upregulation of macrophage MMP-7 production in low oxygen conditions in vitro and human tumours (Burke et al., 2003).
Finally, the hypoxic phenotype of TAMs also includes various immunosuppressive functions which are achieved through several mechanisms; these include the expression of immunosuppressive factors prostaglandin E 2 (PGE2) and IL-10 (Ertel et al., 1993;Murata et al., 2002), whose presence within the tumour microenvironment can downregulate the tumouricidal abilities of TAMs. In addition, PGE2 and IL-10 inhibit the functions of T cells and other effector cells of the immune system (Elgert et al., 1998), which combined with hypoxic inhibition of macrophage phagocytosis and presentation of antigens (Leeper-Woodford and Mills, 1992;Murata et al., 2002), suppresses the triggering of an adaptive immune response directed toward the tumour. Doedens et al. (2010) recently reported hypoxia-and HIF-1 -dependent suppression of T-cell proliferation by macrophages. Hypoxia, therefore, drives the macrophage towards a protumour phenotype which regulates tumour growth, angiogenesis, invasion, metastasis and immunosuppression.

Targeting tumour hypoxia for therapy
Under the stresses associated with hypoxia in areas of tumour ischemia (predominantly low oxygen and low glucose concentrations), tumour cells are forced to respire anaerobically and reduce their proliferation. This challenges many conventional cancer therapies such as chemotherapy, since their mechanism of action relies on the rapidly replication of tumour cells. In addition to this, the poorly developed tumour vasculature, which contributes to the development of hypoxia in the first place, also impedes the delivery of drugs to these areas of the tumour.
In light of this, antiangiogenic "vessel normalizing" strategies are being developed which aim to improve tumour vasculature for better anticancer treatment and reduced metastasis. Rolny and colleagues (2011) demonstrated that histidine-rich glycoprotein (HRG), a hostderived factor, is able to significantly reduce hypoxia and to polarise TAMs away from a protumour phenotype.
The concept of delivering a prodrug systemically -for subsequent activation in specific areas of the body -has been applied to cancer biology, but is most often limited by the level of expression of the activating enzyme at the target site. Rather than hypoxia inhibiting cancer therapy, some recent therapeutic strategies have focussed on utilising cellular responses to these harsh conditions, twinned with the prodrug therapeutic design, to creatively activate cytotoxic agents within the hypoxic tumour microenvironment. Griffiths et al. (2000) made use of macrophage accumulation in areas of low oxygen to deliver gene therapy to pathological hypoxia. They genetically modified macrophages to express the enzyme cytochrome p450 under hypoxic conditions, which when expressed, converts the systemically administered pro-drug cyclophosphamide into its active form exclusively in these areas and causes tumour cell death. This system was then adapted further to deliver an oncolytic adenovirus to these areas by co-transducing macrophages with a hypoxiaregulated E1A/B construct, as well as an E1A-dependent virus which can only proliferate within a prostate-tumour (using a prostate-specific promoter) . E1A/B proteins were only synthesised once the host cell (the macrophage) had infiltrated into areas of extreme hypoxia in tumours. This then subsequently activated the proliferation of the oncolytic adenovirus and its release. The virus then infected and killed surrounding prostate tumour cells -in both hypoxic and non-hypoxic areas of tumours. This three-step process (the homing of macrophages to hypoxic sites, hypoxia-responsive proliferation of the adenovirus, and the limiting of viral replication to within prostate tumour cells), makes this system very specific for killing tumour cells within the hypoxic areas prostate tumours . This hypoxia-based therapy was seen to eradicate both primary and secondary tumours in mice.
It has been suggested that the best use of this therapy would be in combination with conventional therapies, since this could potentially eliminate both the slower proliferating (hypoxic) and the highly proliferating areas of the tumour. Using a mathematical model, Owen et al. (2011) predict that the use of a macrophage-based, hypoxia-responsive therapy immediately before or during conventional chemotherapy would produce significant antitumour effects.

Concluding remarks
Macrophage accumulation in tumours is known to correlate with poor patient prognosis in the majority of cancer types (Fujimoto et al., 2000;Hamada et al., 2002;Hanada et al., 2000;Heidl et al., 1987;Leek et al., 1996;Lissbrant et al., 2000;Salvesen and Akslen, 1999). This can be explained by the tumour-promoting phenotype of TAMs, induced by the tumour microenvironment. Here we have reviewed how hypoxia, a key component of many malignant tumours (Raleigh et al., 2001;Vaupel et al., 1989), is centrally involved in the polarisation of TAMs. The shift in TAM phenotype under hypoxia has been shown not just at the expression level, but also at a functional level as well. In many of these reports, the HIF-1 transcription factor was found to be crucially important. Studies with HIF-1 knockout mice have revealed its role, not just in regulating responses to pathological hypoxia, but also to physiological low oxygen conditions as part of normal oxygen homeostasis. It is possible that HIF-1 signalling (including its activation of other pathways), is a major factor in determining the polarisation and function of macrophages in different environments (Dehne and Brune, 2009).
With considerable assistance from hypoxic, M2-like TAMs, tumour cells in hypoxic areas have the necessary support and drive to migrate and invade into adjacent tissues, evade the immune system and prevent a targeted adaptive immune response, and travel through the vasculature to form metastases at secondary sites. This has large implications in cancer therapy, especially since hypoxic tumour cells are less affected by most conventional therapeutic strategies. Inefficient targeting of such tumour cells is likely to contribute to the well-documented relapse in many chemotherapy-and radiotherapy-treated cancer patients; therefore, more creative and innovative therapeutic methodologies need to be developed based on our continually growing understanding of tumour hypoxia, to enhance patient long-term survival.
Tumor-associated macrophages promote angiogenesis and melanoma growth via adrenomedullin in a paracrine and autocrine manner. Endocr Rev 18, 4-25.