Clinical & Experimental Metastasis

, Volume 26, Issue 1, pp 19–34

The tumor microenvironment and metastatic disease

Research Paper

DOI: 10.1007/s10585-008-9182-2

Cite this article as:
Lunt, S.J., Chaudary, N. & Hill, R.P. Clin Exp Metastasis (2009) 26: 19. doi:10.1007/s10585-008-9182-2


The microenvironment of solid tumors is a heterogeneous, complex milieu for tumor growth and survival that includes features such as acidic pH, low nutrient levels, elevated interstitial fluid pressure (IFP) and chronic and fluctuating levels of oxygenation that relate to the abnormal vascular network that exists in tumors. The metastatic potential of tumor cells is believed to be regulated by interactions between the tumor cells and their extracellular environment (extracellular matrix (ECM)). These interactions can be modified by the accumulation of genetic changes and by the transient alterations in gene expression induced by the local tumor microenvironment. Clinical and experimental evidence suggests that altered gene expression in response to the hypoxic microenvironment is a contributing factor to increased metastatic efficiency. A number of genes that have been implicated in the metastatic process, involving angiogenesis, intra/extravasation, survival and growth, have been found to be hypoxia-responsive. The various metastatic determinants, genetic and epigenetic, somatic and inherited may serve as precedents for the future identification of more genes that are involved in metastasis. Much research has focused on genetic and molecular properties of the tumor cells themselves. In the present review we discuss the epigenetic and physiological regulation of metastasis and emphasize the need for further studies on the interactions between the pathophysiologic tumor microenvironment and the tumor extracellular matrix.


Extracellular matrix Gene expression Heterogeneity Hypoxia Interstitial fluid pressure Metastasis Tumor microenvironment 


As a solid tumor grows, the rate of cancer cell proliferation surpasses the ability of the existing vasculature to supply growth factors, nutrients, and oxygen and to remove the catabolites produced by the cells. The result of this imbalance between supply and demand is regions of hypoxia, low glucose levels and low pH. Another feature of this pathophysiological microenvironment is elevated interstitial fluid pressure (IFP). All of these anomalies are the result, to varying degrees, of a structurally and functionally abnormal vasculature. There is a high level of heterogeneity in the pathophysiological tumor microenvironment both between different tumors and within an individual tumor. Furthermore, the tumor microenvironment has been extensively linked to a more aggressive phenotype, playing a role in tumor progression and metastatic disease.

The development of metastasis is complex, requiring multiple individual steps to successfully establish a tumor at a secondary site. The process requires a tumor cell to acquire the ability to migrate through the primary tumor mass, intravasate into, and survive in, the blood or lymphatic vascular system, extravasate from the vascular system into a secondary organ and initiate angiogenesis in order to enable proliferation at that site (see Fig. 1). As tissue structure and function is intimately connected to and controlled by a series of tumor cell–cell and cell–extracellular matrix (ECM) interactions, the metastatic process involves multiple changes at a molecular level that disrupt and modify these interactions [1, 2]. These include signalling through cell adhesion molecules (CAMs), such as integrins and cadherins, and tissue remodelling through the action of proteinases, such as plasmin or metalloproteinases (MMPs), as well as apoptotic machinery, chemokines, growth factors and signalling molecules, all of which act together to control processes such as proliferation, survival/apoptosis, migration and invasion [3]. These interactions are deregulated in the tumor microenvironment and there is significant heterogeneity in these different interactions and functions between tumor types. Furthermore, it is of particular interest that the expression of many of the genes involved in this process can be affected by exposure to conditions induced by the pathophysiologic environment of tumors, particularly hypoxia. In the current manuscript, we emphasize some of the linkages between the pathophysiologic tumor microenvironment and the tumor cell–cell/cell–ECM interactions in more malignant disease, highlighting the need for further studies on interactions between these different aspects of the tumor microenvironment.
Fig. 1

Primary tumor hypoxia and the metastatic process. The figure shows the metastatic process broken down into 5 key stages (see Table 1 for hypoxia mediated genes involved in each stage). The primary tumor demonstrates the abnormal vasculature developed by angiogenesis. A temporary occlusion is shown, possibly due to vascular compression by the tumor cell mass, which could lead to transiently hypoxic tumor cells adjacent to the vasculature (illustrated in pale blue) as well as contributing to increased vascular resistance, and consequently, elevated IFP. Diffusion limited hypoxia is also illustrated for cells at a distance from the vasculature; in this case there are cells at intermediate levels of hypoxia as the oxygen is consumed. The numbers shown are for illustration purposes only; the actual values will depend on a variety of parameters, including oxygen consumption rate and the position of the cells relative to the arterial versus venous end of the capillary. Tumor cells are shown in the lymphatic and blood vessels––the means of transport to the secondary site

Metastasis and tumor hypoxia

Clinical and experimental link between hypoxia and metastases

Hypoxia has been known to affect tumor cell radiosensitivity for more than 50 years but over the last decade it has become clear that tumor hypoxia also plays an important role in tumor progression and a more aggressive phenotype [1, 2, 4, 5]. Multiple clinical studies using the Eppendorf pO2 histograph have demonstrated a connection between tumor hypoxia (low oxygen concentration, pO2 < 10 mmHg) and disease progression in a wide variety of human tumors including carcinomas of the cervix, prostate and the head and neck and soft tissue sarcomas [6, 7, 8]. A number of these studies have linked hypoxia directly to distant metastases for both cervix carcinoma and soft tissue sarcoma [9, 10, 11]. Furthermore, studies in animal models showed directly that enhancing hypoxia in tumors by exposure of tumor bearing animals to cycling hypoxia in vivo increased metastatic dissemination to the lungs [12] or lymph nodes [13].

The mechanisms by which tumor hypoxia might increase metastatic potential are not fully known, but likely include the induction of mutagenic events, genetic instability and enhanced epigenetic regulation of gene expression [1, 4, 5, 14, 15, 16]. The nature of the connection is difficult to explore in the clinic, although several studies have demonstrated that there are alterations within the tumor microenvironment as a consequence of hypoxia [17, 18, 19, 20]. However, a large number of experimental studies provide support for the link between hypoxia and metastatic disease, as discussed further below. Key early experimental studies showed that transient exposure of KHT and B16F10 murine tumor cells to hypoxia in vitro, prior to transplantation in vivo, resulted in enhanced metastatic ability [21, 22], and this effect of hypoxic exposure in vitro to enhance metastatic ability was also observed with human tumor (melanoma) cell lines [23].

Hypoxia in tumors: acute and chronic

The development of tumor hypoxia is intrinsically linked to the formation of neovasculature by the process of angiogenesis, which involves the expansion of vascular endothelial cells, degradation of the local extracellular matrix, and migration of the endothelial cells towards the tumor [24, 25]. This requires a series of complex molecular events resulting in the simultaneous up-regulation of pro-angiogenic factors and down-regulation of angiogenic inhibitors [26]. The vessels developed are immature and may be dilated, tortuous, and lacking in enervation [6]. They are hyperpermeable due to an often incomplete, or absent, basement membrane, and they are prone to excessive branching, blind ends and neovascular shunts [27, 28, 29].

Diffusion-limited hypoxia, a consequence of tumor cells that are distant from the vascular supply, was the original concept of hypoxia in tumors, proposed by Thomlinson and Gray in 1955 [30]. Such cells were believed to be exposed to prolonged or chronic hypoxia and studies suggested that tumor cells could remain viable in such environments for hours to a few days [31, 32]. It was suggested later that perfusion-limited hypoxia (aka acute or fluctuating hypoxia), due to fluctuations in blood flow, might play an important role in solid tumors [33, 34]. Studies of blood flow and oxygen levels in animal tumors have now established that perfusion of blood vessels can change dynamically in time, leading to rapid but transient episodes of severe hypoxia in the tumor cells [35, 36, 37]. Thus, tumor oxygenation can fluctuate over time; some tumor cells most probably oscillate between levels of hypoxia and more oxygenated states [38] and tumor cells adjacent to vasculature may be exposed to short term hypoxia. The actual distance from blood vessels at which hypoxia occurs likely varies widely due to unstable delivery of oxygen within tumor blood vessels and the variable oxygen consumption of tumor cells. Regions of both acute and chronic hypoxia probably contribute in different degrees to the overall level of hypoxia in different tumors (see Fig. 1).

Experimental studies that clearly define the absolute oxygen level and time of hypoxic exposure are needed to further elucidate the effects of temporal fluctuations in oxygen concentrations that occur within solid tumors. Subarsky and Hill [39] have demonstrated that there are cell line specific effects of graded oxygen levels on invasive potential, and intermediate levels of oxygen (3–4%) may increase metastatic dissemination. Other studies report metastatic efficiency increases at less than 2% oxygen, [40, 41] suggesting model specific heterogeneity in response to low oxygen concentrations. Most studies of the effects of hypoxia on expression of metastasis-related genes reported to date have exposed cells to prolonged periods at a fixed level of oxygen rather than fluctuating exposures. There is also a high degree of variability in the level of hypoxia used in different studies, with oxygen concentrations ranging from anoxic (<0.1% O2) to severe hypoxia (0.1% O2) or levels approaching normoxia (3–5% O2). Furthermore, the actual levels of hypoxia induced in these cultures (which depend critically on factors such as the cell density and the surface area to volume ratio of the media) have not often been measured. Nevertheless, the duration and different level of hypoxia used and the resultant gene profile have highlighted both the heterogeneity and intricacy of gene expression changes in hypoxia. An important aspect of oxygen heterogeneity that has not yet been addressed is what happens to gene expression in cells which fluctuate over a range equivalent to 0–2% oxygen in the gas phase.

Control of gene expression under hypoxic conditions

It is somewhat paradoxical that tumors are hypoxic due to the abnormal vasculature, produced by an angiogenic process driven by genes up-regulated as a consequence of hypoxia. In fact, angiogenesis is a requirement of tumor growth, as without an adequate supply of oxygen and nutrients a tumor mass may not expand; this concept applies to both the development of the primary tumor and establishment of a secondary mass [42]. A key pro-angiogenic gene is hypoxia-responsive vascular endothelial growth factor (VEGF) [43], whose expression is regulated by hypoxia-inducible factor 1 (HIF-1) which is stabilized under hypoxic conditions [44, 45]. HIF-1 has been referred to as the master-regulator [46] and drives expression of multiple genes that are involved in the metastatic process, although it should be noted that there are other transcription factors that respond to an hypoxic environment, namely NF-κB, p53, AP-1, C/EBPb, Egr-1 and SP-1 [5, 47, 48, 49]. More than a decade ago, the Semenza group identified HIF-1 as the key mediator of erythropoietin (EPO) expression following hypoxia [50, 51], a landmark in our understanding of oxygen physiology [52]. HIF-1 is a heterodimeric complex consisting of an α and β subunit, which belongs to a family characterized by the presence of a basic-helix-loop-helix PAS (Per/Arnt/Sim) domain. The constitutive expression of HIF-1β (also known as aryl hydrocarbon receptor nuclear translocator, ARNT), detectable in all cell types, is not controlled by oxygen levels. There are three human HIF-α genes, HIF-1α, HIF-2α, and HIF-3α, which are oxygen sensitive [53]. Of the three HIF-α-subunits, HIF-1α is the best studied and characterized to date; it is the most ubiquitously expressed and functions as a master regulator of oxygen homeostasis in many cell types [45, 49, 54, 55, 56]. In the presence of oxygen, HIF-1α is hydroxylated on conserved prolyl residues within the oxygen-dependent degradation domain by prolyl hydroxylases (PHD) and binds to von Hippel-Lindau protein, which in turn targets it for degradation through the ubiquitin–proteasome pathway. The HIF prolyl hydroxylases (PHD 1–3) are oxygen-sensitive enzymes that regulate the stability of HIFs and thereby induce cellular adaptations in response to hypoxia [57]. However, in hypoxic conditions, prolyl hydroxylase is inactive, resulting in stabilization of HIF-1α [45, 54, 58]. HIFs are frequently up-regulated in cancer and their metastases because transcription of their downstream target genes can promote growth and survival [59]. Although the correlation between oxygen availability and hydroxylase activity may serve as a general determinant of HIF induction, several levels of regulation are required to adapt HIF function in a variety of dynamic microenvironments. High levels of HIF-1α have been positively correlated with tumor progression and poor prognosis in patients with brain, non-small cell lung carcinoma, breast, ovarian, uterine, and cervical tumors [49]. Approximately 1–1.5% of the genome is transcriptionally regulated by hypoxia and many of these genes are known to be regulated by HIF-1α.

An understanding of HIF-2α (also known as endothelial PAS domain protein, EPAS-1) function has developed more recently [59, 60]. Overexpression of HIF-2α has been correlated with tumor progression and poor prognosis in patients with non-small-cell lung cancer, head and neck squamous cell, renal cell carcinoma [61, 62, 63]. HIF-2α is predominately expressed in normal endothelial cells. The most recently identified and consequently less-well studied subunit is HIF-3α, which is detectable in several human cancer cell lines [64, 65]. The significance of the full-length HIF-3α for tumor progression is unknown; however iPAS, a splice variant of HIF-3α is involved in negative regulation of HIF-1α transcriptional activity because it lacks a transcriptional activation domain and appears to repress oxygen-regulated gene expression [60, 66]. In most cell studies to date, PHD2 is the most abundant of the HIF prolyl hydroxylases, and, in oxygenated cells, it is the most important in setting steady-state levels of HIF-1-α and hence the activity of the system, whereas PHD3 was shown to have more influence on HIF-2-α [67, 68, 69]. D’Angelo et al. [70] have shown that hypoxia selectively increases gene expression of PHD2 which acts as a feedback mechanism to stop the hypoxic response in reoxygenated cells, thereby suggesting that proline hydroxylation might play a role in hypoxic preconditioning [70]. Erez et al. [71] demonstrated that expression of HIF-prolyl-hydroxylase-1 can inhibit stabilization of HIF-1α under hypoxia and inhibit tumor growth associated with the inhibition of tumor vascularization. These findings demonstrate the relevance of PHDs to tumor metastasis. Chronic intermittent hypoxia that occurs in tumors may induce ROS, which in turn can activate HIF-1, promoting persistent oxidative stress and further amplifying HIF-1 activation, with its downstream effects on gene expression [72, 73].

Hypoxia and genes associated with cell–cell and cell–ECM interactions

The tumor cell–cell and cell–ECM interactions are vital in maintaining tissue structure and integrity. It is of particular interest for this review that the expression of many of the genes involved in these processes can be affected by exposure to conditions induced by hypoxia (see Table 1). A key step in tumor metastasis is postulated to involve the de-regulation of these interactions through epithelial–mesenchymal transition (EMT). EMT, which is a conserved process critical for embryonic development, involves the release of epithelial cells from the surrounding tissue through alterations in integrin and cadherin expression and subsequent re-organization of the ECM to permit migration [74]. In tumor tissue, this provides a mechanism for metastatic dissemination. E-cadherin is a trans-membrane protein that is integral in maintaining epithelial cell–cell adhesion through the formation of cadherin–catenin complexes [75, 76]. Its loss has been demonstrated in multiple cell lines including ovarian, renal cell, pancreatic and colon cancer lines, through the binding of hypoxia-induced repressors such as TWIST, SNAIL and SLUG [16, 77, 78, 79, 80, 81], resulting in loss of cell adhesion, increased cell motility, and consequent EMT [82].
Table 1

Hypoxia regulated genes with a role in metastatic progression

Gene symbol

Gene name

Hypoxia: ↑ or ↓ regulated

Function: stage of metastatic process

Tumor type




1, 2

Ovarian, Renal cell, Pancreatic, Colon

[16, 77, 78, 79, 80, 81]


Vascular endothelial growth factor-A

1, 2, 5

Melanoma, Glioblastoma, Hepatoma, Colon, Pancreatic, Lung

[23, 121, 122, 123, 125, 126, 183]


Vascular endothelial growth factor-C


Uterine, Breast, Ovarian

[184, 185, 186]


Vascular endothelial growth factor receptor





Basic fibroblast growth factor


Breast, Lung

[183, 187, 188, 189]




Breast, Melanoma, Ovarian

[190, 191, 192, 193]


Platelet-derived endothelial cell growth factor/Thymidine phosphorylase


Breast, Lung

[188, 194]


Urokinase plasminogen activator receptor

1, 2

Endometrial, Breast, Melanoma, Colon

[16, 87, 192, 195, 196, 197]



1, 2

Breast, Lung, Melanoma

[88, 90, 91]



1, 2

Lung, Melanoma, Breast, Colon

[16, 88, 91, 198]


Tissue inhibitor of metalloproteinase-1

1, 2





1, 3, 4, 5

Breast, Colon, Lung, Head and neck, Prostate

[114, 117, 118, 199, 200]


Chemokine receptor-4

1, 2, 3

Breast, Renal cell, Lung

[61, 102, 201, 202, 203]


Cell motility factor receptor

1, 2, 3

Cervix, Ovarian, Breast, Lung, Osteosarcoma, Pancreatic, Liver

[92, 204, 205, 206]


Hepatocyte growth factor

1, 2, 3

Pancreatic, Gastric, Lung, Ovarian, Cervix, Liver, Osteosarcoma

[92, 93, 206]


Lysyl oxidase

1, 2, 3

Breast, Prostate, Renal

[15, 207, 208]


Transformed 3T3 double minute 2


Fibrosarcoma, Brain, Colon, Osteosarcoma

[209, 210]




Ovarian, Renal cell, Pancreatic, Colon, Breast

[77, 211]




Ovarian, Renal cell, Pancreatic, Colon

[77, 78, 212]


Autocrine motility factor

1, 3

Breast, Pancreas, Colon

[16, 212, 213, 214, 215]

The table lists a selection of hypoxia regulated genes with roles in the metastatic process. It is not intended that the table is comprehensive. The gene symbol and corresponding gene name are shown for each gene, as well as whether the gene is up (↑) or down (↓) regulated in response to hypoxia. The stage(s) in the metastatic process where each of the genes is believed to be involved is indicated by a reference number pertaining to Fig. 1. The tumor types/cell lines that have demonstrated hypoxia mediated expression of the listed gene are shown, along with the relevant references

Other important adhesion molecules mediated by hypoxia include the beta-1 integrins which have been shown to be up-regulated in several pancreatic cancer cell lines [83]. The level of constitutive activity of different beta-1 integrins has been found to correlate with invasive capacity, and the use of monoclonal antibodies to inhibit their expression can block invasion in vitro [84]. Chang et al. [83] demonstrated that HIF-1 increased expression of beta-1 integrin in pancreatic cancer cell lines, and further that the use of anti-sense HIF-1 inhibited its expression and reduced metastases presentation in vivo. Survivin, which is an anti-apoptotic protein, was also found to be regulated by HIF-1 expression in this study, and the authors postulated that the combined reduction of survivin and beta-1 integrin were responsible for the reduced metastases observed [83]. Since HIF-1 can regulate the expression of many genes involved in metastatic disease, it is possible that other proteins aside from those described may be involved. Similarly, many HIF-1 regulated genes have alternate methods of activation, a point emphasised in a recent study by Zhang and Hill [41]. The ability of hypoxia to enhance the metastatic potential of a human fibrosarcoma cell line was found to be diminished following exposure to a Ras inhibitor (farnesyl thiosalicylic acid) that had previously been shown to down-regulate HIF-1α expression. However, although VEGF-A expression was shown to be inhibited following treatment, Carbonic Anhydrase-IX (CA-IX) and Lysyl oxidase (LOX) were not. Furthermore, HIF-1α expression was not inhibited. These data demonstrate that different cell types/lines can respond differently to similar stimulae.

During metastatic dissemination, degradation of the ECM and basement membrane can facilitate migration of the tumor cells through the stroma and into/out of the vasculature. The urokinase-type plasminogen activator (uPA) and its receptor (uPAR) regulate the expression of the serine protease, plasmin. uPA is a pro-enzyme that is converted into its active form on binding to its receptor, enabling the conversion of plasminogen to plasmin which in turn allows the localized degradation of the ECM [85]. In addition, plasmin can activate latent forms of MMP-2 and MMP-9, further facilitating degradation of the ECM [86]. It has been shown that hypoxia induces the expression of uPA, uPAR, MMP-2 and MMP-9 in a variety of cell lines [87, 88]. Furthermore, several studies have shown that this hypoxia-mediated expression translates to an increase in metastatic invasion both in vitro [87] and in vivo [89]. Moreover, Rofstad et al. [89] demonstrated that the use of a MAb against uPAR resulted in an almost complete absence of metastatic disease in mice bearing certain human melanoma xenografts.

The balance between proteinases and their inhibitors is perturbed during tumor cell invasion. Hypoxic exposure of a human breast carcinoma line in vitro resulted in the down-regulation of TIMP and concomitant up-regulation of MMP-9. These changes were found to increase the in vitro invasive capacity of this cell line, an effect that could be blocked using an inhibitor of MMPs [90]. As with most metastases-promoting molecules identified to date, the importance of specific MMPs in hypoxia mediated invasion appears to be cell line specific; the aforementioned study found no change in MMP-2 expression despite changes in MMP-9 levels, and Subarsky et al. [39] showed no changes in MMP-2, -9 or -14 levels in two cell lines (human fibrosarcoma and human breast cancer) with divergent invasive capacities under low oxygen conditions [39]. In contrast to these data, other studies have demonstrated hypoxia-mediated up-regulation of MMP-2 activity and a positive correlation with metastatic ability in lung and melanoma tumor models [88, 91].

Hepatocyte growth factor (HGF) is a cytokine with roles in both normal and tumor tissue cell growth which signals through its receptor, the protooncogene c-MET. Hypoxia has been shown to increase transcription of c-MET thereby sensitising the cells to HGF, significantly increasing the invasive capacity of tumor cells [92]. Strong evidence for the importance of HGF/c-MET signalling in metastatic disease has also been provided in a recent study by Corso et al. [93] using a novel RNAi lentiviral inducible system that enabled the specific inhibition of c-MET in transfected tumor cells. Using this system, it was shown that inhibition of c-MET in a human gastric carcinoma cell line not only prevented the development of experimental metastases in vivo, but also facilitated the regression of established metastatic lesions, clearly demonstrating an important role in the maintenance, as well as the development, of metastatic disease.

The stromal derived factor-1 (SDF-1)/chemokine C-X-C receptor type 4 (CXCR4) signalling complex plays a key role in tumor cell motility and homing. SDF-1 (also known as CXCL12) is the ligand to CXCR4 [94, 95]. In normal tissue SDF-1 signals through CXCR4 to regulate the homing of hematopoetic stem cells to the bone marrow [96]. An important study in 2001 [97] showed high levels of SDF-1 expression in the normal human tissue of common breast cancer metastatic sites, namely the bone marrow, lung, liver and lymph nodes. In addition, its receptor CXCR4 was found to be expressed in human breast tumors, but not normal breast tissue, as well as in a range of human breast cancer tumor lines. In vivo studies using a human breast cancer line showed that the formation of both spontaneous and experimental lung metastases could be significantly reduced using a monoclonal antibody against CXCR4, demonstrating its importance in the metastatic process [97], a role that has been supported in additional studies [98, 99, 100, 101]. CXCR4 has been shown to be up-regulated by tumor hypoxia, facilitating the development of metastatic disease [61, 102]. It has been suggested that the presence of SDF-1 at secondary sites concomitant with the expression of its receptor on circulating tumor cells enables cell adhesion and extravasation at the secondary site [95].

A hypoxia-induced extracellular matrix protein, LOX, has recently been identified as a key member of the hypoxia regulated genes involved in metastatic disease [15, 103]. LOX was first identified through microarray analysis of tumor cell lines exposed to hypoxia [104]. Further studies demonstrated that HIF-1 was responsible for the observed hypoxic response. Clinically, LOX was found to be positively correlated with tumor hypoxia in breast cancer patients, and there was a significant relationship between LOX expression and distant metastases [15]. When a breast cancer line with repressed LOX expression was generated and grown orthotopically in the mammary fat pad, no liver metastases and significantly fewer lung metastases were formed, consistent with a requirement for LOX in the metastatic process in this model. Similar results were observed using a monoclonal antibody against LOX. A series of in vitro studies demonstrated an apparent role for LOX in invasion and migration through regulation of focal adhesion kinase activity (FAK), suggesting multiple roles for hypoxia induced LOX in the formation of metastatic disease [15]. In addition to its response to LOX, FAK, which is expressed at focal adhesions and activated by integrin binding, may be linked to hypoxia-mediated increases in tumor cell migration through alternative mechanisms. FAK is up-regulated in response to VEGF [105], and furthermore, a reduction in VEGF expression through inhibition of HIF has been shown to inhibit phosphorylation of FAK [106, 107]. This bears further study, as FAK has been implicated in tumor cell motility, invasion and survival, and thus plays an important role in the metastatic process [108]. Another important ECM protein that is hypoxia regulated is the secreted glycophosphoprotein osteopontin (OPN) [109], which is expressed by multiple different cell types (osteoclasts, osteoblasts, epithelial cells, endothelial cells) [110]. OPN has roles in cell adhesion, angiogenesis, prevention of apoptosis and the anchorage independent proliferation of tumor cells [110], and its expression has been found to correlate with increased metastatic potential in breast [111, 112, 113], prostate [114], colon [115, 116], head and neck cancer [117, 118] and soft tissue sarcoma [119]. Clinically, OPN levels were found to be elevated in the serum of head and neck patients with hypoxic tumors, and this was correlated with an adverse outcome [117]. The fact that OPN levels are elevated in the plasma or serum, and that this correlates with outcome, provides the possibility for a simple diagnostic test that could enable stratification of at risk patients. This potential has been demonstrated retrospectively in head and neck patients, where those with elevated plasma levels of OPN demonstrated an improved response to radiation when administered in combination with the hypoxic sensitizer, nimorazole [118].

VEGF-A, originally identified as Vascular Permeability Factor [120], is a potent inducer of tumor angiogenesis that is up-regulated in response to hypoxia [121] by both HIF-1 dependent [122] and independent [123] mechanisms. Its receptors, VEGFR1 and-2, are also induced under hypoxic conditions [124]. The role of this protein in metastatic disease has been extensively examined, with some studies suggesting a link between VEGF-A expression and metastatic disease [23, 125, 126], and others not [127, 128], consistent with the concept that different cell types may demonstrate tumor specific effects. However, in driving the development of neo-vasculature VEGF-A may provide a mechanism of transport for the tumor cells as well as possibly enhancing the intravasation and extravasation stages of the metastatic process due to increased vascular permeability. VEGF-A also plays a role in macrophage migration, and tumor associated macrophages have been reported to localise in hypoxic regions [129] aiding in tumor cell migration [130], although again this is not a universal finding (Kalliomaki et al. 2008 in press, BMC Cancer).

Perhaps because of the importance of angiogenesis in tumor survival, there are several other hypoxia inducible pro-angiogenic proteins that provide a potential alternative mechanism for hypoxia and angiogenesis mediated secondary disease. These include interleukin-8 (IL-8) [131, 132], platelet-derived endothelial cell growth factor (PD-ECGF) [133, 134] and basic fibroblast growth factor (bFGF) [135]. A study by Rofstad et al. [136] examined the effect of neutralizing antibodies against VEGF, IL-8, PD-ECGF and bFGF in different human melanoma lines, demonstrating the effects of these proteins on tumor metastasis as well as the tumor specific effect apparent in the differing response of the tumor lines to the antibodies. It is clear from the above discussion that the multiple and complex processes required for the development of metastatic disease involves the expression of many hypoxia responsive genes (see Table 1).

Tumor metastasis and pH

Another feature of the pathophysiological tumor microenvironment that arises, at least partially, as a consequence of the abnormal tumor vasculature, is low extracellular pH. This feature of the microenvironment is not as well studied as that of tumor hypoxia, and there are many regions of overlap between the two conditions, which can confound interpretation of the available data. Tumor acidity is a complex and multifactorial process; a key causal factor is the poor supply of oxygen from the inadequate vasculature, which triggers glycolysis resulting in the accumulation of lactic acid [137, 138, 139]. Although it should be noted that this process is not solely driven by tumor hypoxia (Warburg described high rates of glycolysis in oxic tumors in 1956), there is increasing evidence of complex links between these two parameters [140, 141]. In addition to the physiological effect of tumor hypoxia, changes in gene expression mediated by hypoxia also play a role. Up-regulation of HIF-1 is known to regulate the expression of various genes involved in the regulation of pH, including CA-IX and the glucose transporters Glut-1 and -3, as well as key enzymes in the glycolytic pathway [139, 142].

The association with tumor hypoxia also complicates the interpretation of the role of tumor pH in tumor progression. Similar to tumor hypoxia, there have been several studies demonstrating a correlation between a low extracellular pH and increased invasion/metastatic spread. Studies have shown that culturing different tumor cell lines, both murine (KHT fibrosarcoma and B16F1 melanoma) and human (A-07, D-12 and T-22 melanomas), under acidic conditions in vitro can enhance the formation of experimental lung metastases when injected in vivo [128, 143, 144]. Rofstad et al. [144] demonstrated that acidity induced up-regulation of the proteolysis enzymes MMP-2 and MMP-9 and the angiogenic factors VEGF and IL8 in vitro, all of which are known to be involved in the metastatic process (see previous section). More recently, it has been suggested that exposure to acidic conditions selects for clonal populations with enhanced invasive capacity, as human melanoma cells exposed to acidic conditions were found to demonstrate an increase in vitro invasion only following re-acclimatization to normal pH [145]. However, an in vivo study using the murine KHT fibrosarcoma and B16F1 melanoma tumors, where increased tumor acidity was experimentally induced, failed to enhance the development of spontaneous lung metastases [146].

The relationship between pH- and hypoxia-mediated invasion remains unclear. The experimental metastases studies [128, 143, 144] show very similar results to those demonstrating enhanced metastases following in vitro exposure to hypoxia [21, 23]. Furthermore, the genes found to be up-regulated in response to an acidic microenvironment are all genes that have been shown to respond to a hypoxic environment, increasing invasive capacity [144]. Walenta et al. [147] showed a correlation between hypoxia and lactate levels in an experimental rat tumor model, suggesting that these two parameters may act in conjunction with one another. Furthermore, Brizel and colleagues reported that high lactate levels in head and neck cancers were associated with increased metastasis [17]. However, another study showed that both hypoxia and an acidic extracellular pH were able to enhance expression of the pro-angiogenic factor VEGF in brain tumor xenografts independently of one another [148]. To date, the relationship between these features of the microenvironment remains unclear, and offers scope for further studies.

Metastasis and IFP

Another interesting feature of the pathophysiologic tumor microenvironment is elevated interstitial fluid pressure (IFP) with values ranging from 10 to 100 mmHg [149, 150, 151, 152, 153, 154, 155]. There are multiple factors that act together to elevate tumor IFP values. The highly irregular tumor neovasculature generated by angiogenesis is characterized by increased permeability and microvascular resistance, which results in the movement of fluid into the ECM. As tumors are either lacking in lymphatic vessels, or the intra-tumoral vessels are non-functional [156, 157], excess fluid accumulates in the interstitium, extending the elastic ECM and elevating IFP. It has been suggested that the tumor vasculature is the driving force in increasing tumor IFP [7, 158, 159]; however, the tumor stroma is also thought to play an active role due to anomalies within the ECM composition as well as infiltration of macrophages and other inflammatory cells, leading to the “reactive” stroma [7, 159, 160]. Heldin et al. [160] proposed a model for the regulation of IFP in normal tissue, that they extend to tumor tissue, in which fibroblasts actively regulate the tension applied to the ECM through collagen binding integrins which enable them to exert or modify tension on the collagen fibre network, thereby moderating the elasticity of the ECM in response to hyaluronan and proteoglycan expansion [160]. Consistent with a role for the collagen fibre network in modulating tumor IFP, a reduction in fibromodulin (a proteoglycan involved in collagen assembly and maintenance) in an experimental tumor model resulted in a reduced tumor IFP. This outcome was attributed to the thinner, less abundant, collagen fibres apparent in these tumors as there were no apparent changes in lymphatic or blood vessel density or structure [161]. Furthermore, there is evidence that the process of collagen fibre contractions is actively regulated in response to specific cytokines such as PDGF [162], the receptors of which are expressed in the stroma of multiple tumors [163]. Several studies have demonstrated that inhibition of PDGF signalling can reduce tumor IFP values, consistent with a dynamic role in increasing tumor IFP [164, 165, 166].

Elevated tumor IFP can act as a barrier to delivery of therapeutic agents, thereby reducing their efficacy and multiple studies have demonstrated improved uptake of chemotherapeutic drugs following a reduction in tumor IFP [164, 167, 168, 169, 170]. However, data on the relationship between IFP per se and disease progression is limited. A clinical study by Milosevic et al. [154] demonstrated a significant relationship between a high tumor IFP value and risk of local recurrence and distant metastases in cervical carcinoma patients following treatment with radiotherapy alone, although there was no correlation between tumor IFP and metastatic disease at the time of diagnosis. Experimentally, Rofstad et al. [171] demonstrated a positive relationship between IFP in oxic melanoma tumors and the development of lymph node metastases but there was substantial overlap in metastases presentation from the low and high IFP groups. In contrast, Lunt et al. [158] found no relationship between IFP and metastases presentation in a cervical xenograft model, murine fibrosarcoma model or spontaneous mammary tumor model.

While these differences are perhaps not surprising in view of the fact that metastatic disease is highly multi-factorial and often cell-type specific, it is possible that a high IFP may act as a marker of alternative factors that predispose to metastatic disease. For example, if IFP is primarily, or largely, determined by the vasculature, a high IFP may be a marker of vascular persistence. One possibility is that tumors with a higher IFP are characterised by increased vascular leakiness, potentially due to higher levels of VEGF-A expression, and decreased pericyte coverage of the blood vessels, both of which may facilitate the intravasation process. Indeed, it has been shown using a thyroid carcinoma cell line that inhibition of VEGF-A signalling reduced the extracellular fluid volume and plasma protein leakage from the blood vessels, both of which could be interpreted as evidence for reducing vascular permeability [172]. Since VEGF-A is up-regulated by tumor hypoxia, the strong role of the tumor vasculature in the development of both tumor hypoxia and IFP might imply a relationship between these two parameters of the tumor microenvironment. However, the literature to date has shown no evidence for such a correlation, either clinically [154] or experimentally [158, 173, 174]. These data do not necessarily signify that no association exists however. None of the methods used to assess tumor hypoxia were able to distinguish between regions of chronic or transient hypoxia. It is possible that, as the microvascular pressure and the interstitial pressure reach equilibrium, temporary drops in vascular pressure may result in transient perturbations in flow and subsequent occlusion of the vessels. Furthermore, the packing density of the tumor cells may impact IFP and this may be related to increased vascular compression by the tumor mass (see Fig. 1). Both of these situations could result in regions of transiently hypoxic tumor cells, which experimental data suggest may be key in the development of hypoxia-mediated metastases [12, 13].

As discussed, hypoxia is known to mediate the expression of many genes involved in the metastatic process and many of these genes are involved in ECM-tumor cell interactions. There are no known IFP mediated changes in gene expression; however, there are genes that may either elevate tumor IFP or act in conjunction with elevated IFP levels to increase metastatic potential. VEGF-A as a potent pro-angiogenic factor, as mentioned above, and VEGF-C and VEGF-D, both of which signal through VEGFR3, are lymphangiogenic factors that increase the number of lymphatic vessels at the periphery of the tumor as well as enlarging their diameter [175]. Expression of these genes could provide an increased avenue for tumor cell migration into the lymphatic system. Elevated tumor IFP may aid in this process, irrespective of gene expression. It has been shown that when an equilibrium exists between the capillary pressure and the interstitial pressure, the movement of fluid into the interstitial space is equal to the movement of fluid into the surrounding normal tissue [176, 177], potentially forcing the tumor cells into the surrounding normal tissue where they can invade into the lymphatic system. Further research on the role of elevated IFP in metastasis in the context of the interactions between the tumor cells and tumor ECM and vascular system is required to provide mechanistic insight into metastatic disease as well as potential avenues for therapeutic exploitation.

Summary and future studies

The tumor microenvironment is complex involving pathophysiological features such as hypoxia, low pH, and increased IFP due to poorly structured vascular supply. In particular, cells can experience a wide range of oxygen levels from near arterial pO2 levels (approximately equivalent to 100 mmHg) to average tissue pO2 levels equivalent to 5% O2 (~40 mmHg) to levels of hypoxia equivalent to 2–0% oxygen depending on their distance from functional blood vessels. Many of the cells will experience fluctuating levels of hypoxia due to variations in the red cell flux in the blood vessels. These pathophysiological features of tumors, particularly hypoxia, can cause changes in gene expression that interact with the cell–cell and cell–ECM signalling interactions in the tumor (cell–matrix interactive microenvironment) which may themselves be aberrant relative to normal tissues. However, little is currently known about the effects of fluctuating hypoxia on gene expression particularly in the range 2% and below; further studies are required in this area.

Mechanisms of metastasis involve a complex array of genetic and epigenetic changes many of which appear to be specific both for different types of tumors and for different sites of metastasis. It is apparent that hypoxia can increase the metastatic potential of a tumor both through enhancing the metastatic potential of a tumor cell due to the genetic and epigenetic alterations acquired in order to adapt to the hypoxic microenvironment, and also through the expression of molecules that disrupt or modify the ability of the tumor cells to interact with the ECM facilitating key stages of the metastatic process. The number of hypoxia-responsive genes linked to the development of metastatic progression, that are involved in tumor cell–ECM interactions, indicates a need for further studies on the tumor as a whole, not just the tumor cells.

Finally, recent studies of stem cells in tumors [178, 179, 180] have suggested that tumors may contain only a small fraction of cells capable of regenerating a tumor and it is presumably these cells which have metastatic ability. There is recent evidence that signalling pathways in these cells which maintain their stem cell properties may be preferentially affected by hypoxia [181, 182]. If this is the case, studies focusing on this particular subpopulation of cells in tumors would be particularly relevant to the metastatic process as it is affected by the tumor microenvironment.

Copyright information

© Springer Science+Business Media B.V. 2008

Authors and Affiliations

  • Sarah Jane Lunt
    • 1
    • 2
  • Naz Chaudary
    • 1
    • 2
  • Richard P. Hill
    • 1
    • 2
  1. 1.Applied Molecular Oncology, Ontario Cancer Institute, Princess Margaret HospitalUniversity Health NetworkTorontoCanada
  2. 2.Department of Medical BiophysicsUniversity of TorontoTorontoCanada

Personalised recommendations