SEARCH

SEARCH BY CITATION

Keywords:

  • microenvironment;
  • extracellular matrix;
  • hypoxia;
  • angiogenesis;
  • inflammation;
  • tumour-associated macrophages

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

It is now recognized that the host microenvironment undergoes extensive change during the evolution and progression of cancer. This involves the generation of cancer-associated fibroblasts (CAFs), which, through release of growth factors and cytokines, lead to enhanced angiogenesis, increased tumour growth and invasion. It has also been demonstrated that CAFs may modulate the cancer stem cell (CSC) phenotype, which has therapeutic implications. The altered fibroblast phenotype also contributes to the development of an altered extracellular matrix (ECM), with synthesis of ECM isoforms rarely found in normal tissues, including tenascin-C isoforms and the fibronectin EDA isoform. There is also emerging evidence of how the tensile strength of the tumour-associated ECM may be modified and lead to altered signalling in tumour cells. The hypoxic environment of the tumour stimulates angiogenesis and also impacts on other aspects of cell signalling, including the c-met pathway and lysyl oxidase-mediated signalling, which can directly promote tumour cell invasion. The inflammatory infiltrate associated with many solid tumours also modulates tumour function, having both anti- and pro-tumour effects. All of these components of the microenvironment provide potential targets for therapeutic attack, with a number of molecules already in clinical trials. It is also becoming evident that characterizing the tumour microenvironment can provide important prognostic and predictive information about tumours, independent of the tumour cell phenotype. Copyright © 2010 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

In recent years the tumour microenvironment has become the focus of intense research, with the understanding that the alterations that occur in the stroma around a tumour might prove useful in prognosis and generate new therapeutic targets. However, the importance of the microenvironment is not a new concept. The idea that stromal cells might promote cancer development was first recognized in 1863 when Rudolph Virchow observed leukocytes in the stroma of neoplastic tissue and hypothesized that malignancy originated at sites of chronic inflammation 1. This focused only on the inflammation and did not consider the role of other stromal elements, although when Paget put forward his ‘seed and soil’ hypothesis in 1889 this did encompass all components 2. It was not until 1982 that Bissell et al outlined a modern theory that the microenvironment in which a tumorigenic cell evolves is as critical to its evolution as the genetic mutations that it accrues 3. This review considers several of the key elements that make up the stromal environment and how changes in these compartments influence cancer development and progression (Figure 1). We also consider how an understanding of the tumour microenvironment may have predictive and prognostic importance, and provide new avenues for therapeutic attack. Whilst such changes do play a role in lymphoma and leukaemia, this review limits itself to the microenvironment of solid tumours.

thumbnail image

Figure 1. Changes to the normal microenvironment promote tumour invasion. Altered function of carcinoma-associated fibroblasts and induction of an inflammatory infiltrate lead to release of pro-angiogenic factors. Development of a desmoplastic stroma, partly in response to hypoxia, leads to tumour-specific interactions with tumour cell-surface receptors that enhance invasion

Download figure to PowerPoint

Influence of cancer-associated fibroblasts on tumour behaviour

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Fibroblasts are among the most abundant cell type in the microenvironment of solid tumours, being particularly prominent in carcinomas of breast, pancreas, colon and prostate. There is abundant evidence that cancer-associated fibroblasts (CAFs) can contribute to tumour growth and spread, mediated through their release of classical growth factors, such as EGF, TGFβ and HGF, as well as a range of chemokines shown to influence different aspects of tumour cell behaviour (Figure 2) 4.

thumbnail image

Figure 2. Origins and functions of carcinoma-associated fibroblasts (CAFs). A number of origins have been proposed for CAFs, including bone marrow-derived cells (BMDCs), response of normal fibroblasts to tumour-derived signals, genetic and epigenetic alterations in normal fibroblasts, or epithelial–mesenchymal transition (EMT) of normal or tumour epithelial cells. Their effects on tumour cells include promotion of tumour growth and invasion, stimulation of angiogenesis, generation of an altered extracellular matrix (ECM) and modulation of the cancer stem cell (CSC) phenotype

Download figure to PowerPoint

Orimo et al5 showed that fibroblasts derived from primary human invasive breast carcinomas significantly enhanced tumour growth in xenograft models compared to their normal counterparts. They demonstrated that these CAFs produced higher levels of stromal-derived factor (SDF)-1, which mediated the recruitment of endothelial progenitor cells into the tumour mass, leading to enhanced angiogenesis as well as directly promoting tumour cell growth via interaction with tumour cell CXCR4. Interestingly, they found that these unique characteristics of CAFs were maintained even in the absence of contact with tumour cells, implying that CAF function is not merely a response to tumour-derived signals 5. Other stromal-derived growth factors also contribute to the tumour-promoting effect of CAFs. Yang et al6 found that prostate-derived fibroblasts capable of promoting LNCaP prostate cancer tumorigenesis exhibited significantly higher levels of connective tissue growth factor (CTGF) than non-tumour promoting fibroblasts. Using a xenograft model system they showed that CTGF in tumour-associated stroma induced a significant increase in angiogenesis and enhanced tumour growth. CTGF is potently stimulated by TGFβ 7, and is over-expressed in the stroma of several cancer types, including breast, pancreatic and oesophageal carcinoma 8–10. Yang et al6 went on to show that TGFβ-treatment of non-tumour-promoting fibroblasts led to up-regulation of CTGF and subsequent acquisition of tumour-promoting activity.

In addition to promoting growth of established tumours, there is strong evidence that altered fibroblast signalling may be critical in the initiation of carcinogenesis, in regulating its phenotype and in mediating metastatic spread of tumours.

In a series of now classic experiments by Cunha et al11, the impact of tumour-associated stroma on epithelial behaviour was elegantly demonstrated. The SV40-immortalized but non-tumorigenic prostate epithelial cell line BPH-1 was transplanted into mice with either fibroblasts isolated from primary prostate cancers (i.e. CAFs) or normal prostate fibroblasts. Poorly differentiated carcinomas developed in the presence of CAFs, whilst minimal epithelial cell growth, and no tumour development, was supported in the presence of normal fibroblasts. Interestingly, when BPH-1 cells were grown with rat or mouse urogenital mesenchyme, epithelial cell growth was stimulated but there was no tumorigenic growth, suggesting that the pro-tumour function of CAFs does not simply relate to enhanced mitogenesis 11. It is important to note that CAFs did not stimulate tumour development in non-immortalized benign prostate epithelium, suggesting that some epithelial abnormality is necessary to respond to the pro-tumorigenic effects of CAFs. More recently, Bhowmick et al12 generated mice in which the TGFβII receptor was selectively ablated in fibroblasts and showed that these mice spontaneously developed neoplastic lesions, including invasive carcinoma of the fore-stomach. They demonstrated that these lesions were associated with expansion of the stromal compartment and enhanced HGF expression 12.

In 2010 Erez et al demonstrated that CAFs from dysplastic skin support tumorigenesis by mediating tumour-enhancing inflammation through activation of a pro-inflammatory gene expression signature. They were able to link this effect to NF-κB signalling in the CAF, as when IKKβ was knocked down there was reduction in recruitment of macrophages and vascularization 13. These models suggest that altered signalling from fibroblasts can influence early tumour development by disrupting normal epithelial cell–stromal interactions.

A recent study has highlighted a further, potentially important, role for CAFs in the regulation of tumour behaviour—that of modulating the cancer stem cell (CSC) phenotype. In the normal colon, Wnt signalling plays a critical role in maintaining intestinal stem cell and crypt homeostasis 14, whilst disruption of Wnt signalling is recognized as an early event in the development of colorectal cancer: mutation of APC or β-catenin results in accumulation of β-catenin in the nucleus, where it associates with the T cell factor/lymphoid enhancer factor (TCF/LEF) family of transcription factors, leading to sustained transcription of Wnt target genes 15. Using a TCF/LEF reporter directing the expression of green fluorescent protein (TOP–GFP), Vermeulen et al16 demonstrated that Wnt activity is heterogeneous in colon cancer cells, despite all cells carrying an APC mutation. Furthermore, those cells showing the highest level of Wnt activity, identified by high levels of TOP–GFP, showed up-regulation of stem cell-associated genes and enhanced clonogenic potential. They went on to show that this CSC phenotype could be regulated by tumour-associated myofibroblasts: co-culture of colon cancer cells with myofibroblasts or myofibroblast conditioned medium (CM) resulted in enhanced nuclear β-catenin, increased Wnt activity and enhanced clonogenic activity, with enhanced tumorigenicity when co-injected into mice. This important study indicates that the stem cell phenotype is plastic and is dependent on the tumour microenvironment. The authors showed some evidence to suggest that myofibroblast release of HGF may be involved in this regulation, and suggest that targeting the CSC–microenvironment interface may be the most effective approach to overcome stem cell resistance to current therapies 16.

The origin of cancer-associated fibroblasts

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Cancer-associated fibroblasts (CAFs) form a heterogeneous population, probably related to their diverse origin (Figure 2). Whilst activation of local stromal fibroblasts has traditionally been considered the major source of CAFs 4, experimental models using genetically marked bone marrow-derived cells have demonstrated that these cells can contribute to the tumour-associated stroma and develop markers and functions of CAFs 17. It has also been suggested that CAFs may be derived from normal or tumour cells that undergo epithelial–mesenchymal transition (EMT) 18, a theory that may explain the identification of identical genetic alterations in tumour epithelium and its associated stroma 19.

The concept of inherent dysfunction of host fibroblasts in patients with cancer remains a controversial area. A number of early reports demonstrating loss of heterozygosity (LOH) in stromal samples suggested that tumour-associated stroma is genetically unstable and may undergo genetic alterations that influence its function 20, 21. More recently, total genome LOH analysis and mutational screening for TP53 in microdissected breast tumour epithelium and stroma identified somatic TP53 mutations in the stroma and showed this to be associated with regional lymph node metastasis 22. This is one of the first studies to suggest the clinical importance of stromal changes in breast cancer. However, this study was performed on DNA from formalin-fixed, paraffin-embedded material, and a limited study on stroma derived from frozen material yielded no evidence of TP53 mutations 23. Similarly, a more extensive study of LOH in CAFs derived from breast and ovarian carcinomas suggested that fibroblast LOH was a very rare event and could not explain the CAF phenotype 24. Further studies are required to settle this controversy.

Other mechanisms by which the altered phenotype of CAFs could be maintained in the absence of a tumour cell population would be via epigenetic modulation of the DNA. Indeed, altered DNA methylation has been identified in fibroblasts from colorectal cancer 25 and breast cancer 26 compared to their normal counterparts, and methylation patterns were associated with altered mRNA levels, suggesting that epigenetic modification could contribute to the CAF phenotype.

Finally, the response of the microenvironment to a tumour may also be influenced by intrinsic host genetic variability. In keeping with this, some reports have indicated that even non-tumour fibroblasts from women with breast cancer differ from those isolated from women without breast cancer 27, 28. Such variability can be generated through single nucleotide polymorphisms (SNPs). Common functional promoter SNPs have been described in many genes, including the matrix metalloproteinase (MMP) enzymes that are involved in modifying the extracellular matrix, where they have the potential to influence levels of gene expression 29, 30. An analysis of the ability of primary fibroblasts to promote breast cancer cell invasion demonstrated that fibroblasts derived from women with a high-expressing MMP-3 genotype generated significantly more invasion than fibroblasts from women without this genotype, regardless of whether or not the fibroblasts were derived from patients with breast cancer 31. Similar variations influencing the immune system may also have an effect on tumour development 32. This suggests that genetic characteristics can influence how the stromal microenvironment responds to a tumour and may influence clinical outcome.

The extracellular matrix and its remodelling

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

As well as releasing a range of chemokines and growth factors, CAFs also generate an altered extracellular matrix (ECM) environment. Most solid tumours exhibit a very different profile of ECM proteins in the stroma compared to their normal counterparts, and many of these proteins interact directly with tumour cells, via integrins and other cell surface receptors, to influence functions such as proliferation, apoptosis, migration and differentiation 33. A number of proteins are consistently up-regulated in solid tumours, including tenascin-C (TNC), fibronectin (FN) and SPARC (secreted protein, acidic and rich in cysteine). Further diversity is introduced through the expression of alternatively spliced variants of some of these proteins, notably TNC and FN. TNC is a multifunctional protein that is expressed at low levels in normal adult tissues but up-regulated in situations associated with cell migration, such as embryogenesis, wound healing and in tumours 34. It exists as multiple alternatively spliced isoforms, which appear to be expressed in a tumour-specific manner. Whilst the full-length unspliced protein appears to be expressed in both pancreatic and prostate cancer 35, 36, a novel isoform containing domain C, which is undetectable in normal tissues and rarely expressed in other tumours, is abundant in high grade gliomas 37. In both breast 38 and ovarian 39 carcinoma, isoforms containing domains A and D are particularly abundant, and these have been shown to have functional significance, promoting tumour growth and invasion 40. It is not entirely clear how these tumour-associated isoforms influence tumour cell behaviour, but additional domains can introduce new integrin receptor sites that alter cell signalling 41 or may lead to up-regulation of proteolytic enzymes that can remodel the matrix and enhance invasion 42.

In a similar manner to TNC, alternative splicing of FN pre-mRNA, as well as post-translational modification, generates up to 20 variants of this complex protein 43. In tumours, there is frequent up-regulation of the ‘oncofetal’ forms of FN that contain extra-domain (ED)A, EDB and IIICS (type III connecting sequence) sequences 44. FN–EDA is said to be required for the transduction of TGFβ signals, and the conversion of fibroblasts to myofibroblasts, a key event in the tumour microenvironment 45, whereas FN–EDB is particularly associated with neovascular structures in many different tumour types 46, 47. Despite this, EDB- (and EDA)-null mice exhibit no defects in physiological or tumour angiogenesis, suggesting that there is functional redundancy, although they remain highly selective markers for the targeting of tumour-associated stroma and vasculature 48.

As well as the composition of the ECM, the mechanical properties of the stroma also have a profound impact on function. The tensile strength or stiffness of the ECM can regulate epithelial cell growth, differentiation and migration 49, 50, and reduction of ECM stiffness can repress the malignant behaviour of mammary epithelial cells 49. One way in which tensile strength can be modulated is via lysyl oxidase (LOX), an enzyme secreted primarily by fibroblasts that serves to cross-link collagens and elastin, increasing the insoluble matrix and contributing to tensile strength 51. A recent study using the MMTV-Neu mouse model of mammary cancer confirmed enhanced levels of LOX activity, increased collagen cross-linking and elevated ECM stiffness with progression to invasive disease 52. These stromal changes were associated with increased focal adhesion formation and increased FAK activity, and inhibition of β1-integrin in this context prevented tumour invasion, suggesting that changes in ECM stiffness regulate epithelial cells through integrin signalling 52. LOX pre-conditioning and stiffening of the mouse mammary fat pad also resulted in growth and invasion of premalignant mammary cells 52, suggesting that increased ECM stiffness can promote tumorigenesis as well as alter established tumour behaviour. This may offer insight as to how enhanced mammographic density leads to increased relative risk of breast cancer 53.

The ECM undergoes significant remodelling during tumour progressions and this is mediated largely by the extracellular proteinases, particularly the matrix metalloproteinases (MMPs), and the major source of these is from the stromal cells 54. MMPs have been implicated in the promotion of tumour invasion and metastasis for decades, although enthusiasm for them as a plausible therapeutic target waned following disappointing clinical trials 55. This has led to a re-examination of MMP function and a greater understanding of the complex roles of these enzymes. It is now evident that some MMPs act as tumour suppressors rather than tumour promoters 56 and, as well as inducing angiogenesis, may in some instances cleave proteins, leading to the generation of anti-angiogenic fragments such as angiostatin 57. This more detailed understanding of MMP function may facilitate future developments of MMP-targeted therapies.

Organ-specific microenvironment: the myoepithelial cell

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Many of the responses of the microenvironment are common to all organs and tumours, which is one of the features that make this an attractive therapeutic target. However, there are organ-specific differences that are important to understand. This is best illustrated in the breast, where the myoepithelial cell truly acts as the ‘Jekyll and Hyde’ of the microenvironment. Normal breast myoepithelial cells have been shown to exhibit anti-angiogenic 58, anti-proliferative 59 and anti-invasive 60, 61 properties.

However, it is clear that myoepithelial cells change as Ductal Carcinoma in-situ (DCIS) of the breast progresses. These cells show loss of hemidesmosome formation 62 and up-regulation of pro-invasive ECM proteins 38, and Allinen et al63 showed that myoepithelial cells exhibit more dramatic changes than any other cell component between normal and DCIS tissues, suggesting extensive abnormal paracrine interactions in DCIS. Indeed, xenograft studies suggest that dedifferentiation of host myoepithelial cells leads to the transition of in situ to invasive disease.

Hypoxia and angiogenesis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Just as normal tissues require a supply of nutrients and the removal of waste products, so do tumours 64. As tumours grow, areas of nutrient deprivation and oxygen deprivation (hypoxia) arise as a result of an insufficient blood supply 65. Although a limiting factor for tumour growth, hypoxia also represents a stimulus for invasion and metastasis, and a number of studies have shown that hypoxia is an independent predictor of poor prognosis 66–68. Hypoxia stimulates hypoxia-inducible family (HIF) proteins, which regulate diverse cellular processes, including metabolism 69, angiogenesis, cell proliferation, apoptosis 70 and tissue remodelling 71. In addition, hypoxia can down-regulate epithelial E-cadherin and so facilitate EMT 72, and low oxygen tension causes cancer cells to switch to anaerobic metabolism, which greatly increases the genetic instability of the cells (Figure 3) 73, 74. Indeed, in 2001 Jain et al developed an in vivo microscopy method that demonstrated that VEGF transcription in brain tumours is regulated by the tissue pO2. They found that under hypoxic conditions, VEGF-promoter activity increased. This was supported by their in vitro experiments 75.

thumbnail image

Figure 3. Effect of hypoxia on tumour cell behaviour: Hypoxia leads to up-regulation of HIF-1α, which leads to an angiogenic switch, enhancing expression of pro-angiogenic factors such as VEGF, PDGF and FGF, with down-regulation of anti-angiogenic thrombospondin. HIF-1α also up-regulates c-met (leading to enhanced sensitivity to HGF), promotes EMT and increases lysyl oxidase (LOX) activity, which alters the extracellular matrix and activates tumour cell FAK, all of which lead to enhanced tumour cell invasion

Download figure to PowerPoint

One of the key roles of HIF-1α in hypoxia is the induction of pro-angiogenic factors, including VEGF, angiopoietin-2, PDGF and FGF 76–79, and down-regulation of anti-angiogenic factors, such as thrombospondin 80. Angiogenesis is required if a tumour is to progress past a certain size and the microenvironment plays an important role in dictating when the ‘angiogenic switch’ will occur. This angiogenic switch separates cancer development from small (1–2 mm) lesions, which are dormant, to an exponential growth phase 77, 81, 82.

Whereas the major focus on hypoxia has been its role in enhancing angiogenesis (for reviews, see 78, 83), recently a number of angiogenesis-independent mechanisms for hypoxia-induced tumour progression have been described 84, 85. In one key study, using a series of cell lines derived from breast, lung, cervical and ovarian cancers, among others, Pennacchietti et al showed that HIF-1α binds to the c-Met promoter, leading to over-expression of c-Met and enhanced sensitivity to HGF. This leads to an ‘invasive switch’ in the tumour cells, increasing degradation of the ECM and allowing tumour cells to move freely towards more oxygen-rich areas 85. This has therapeutic implications, since targeting angiogenesis alone may not be sufficient, and indeed may even aggravate, this invasive response to hypoxia.

HIF-1α can also activate Wnt/β-catenin signalling, as demonstrated in prostate cancer, promoting a more motile and invasive phenotype 86. Lysyl oxidase (LOX) is also regulated by hypoxia and HIF-1α 84. LOX has complex functions, being reported to have both tumour suppressor and tumour promoter properties 51, 87, 88. It is clear, however, that LOX contributes to the enhanced invasive properties of hypoxic tumour cells. This is demonstrated by the inhibition of hypoxia-induced invasion achieved by LOX shRNA expression in breast, cervical, head and neck, pancreatic, colon and lung cancer cells 89, and expression of LOX is significantly associated with reduced distant metastasis-free survival and overall survival in ER-negative breast cancers, and in head and neck cancers 90, 91. It appears that LOX enhances invasion both through matrix remodelling and by directly influencing actin polymerization and FAK activation 92. The central role of LOX in mediating tumour invasion makes it an attractive therapeutic target but, given its divergent roles, it has been suggested that therapeutic strategies should be focused on ablating the extracellular functions of LOX 84, 93.

The inflammatory microenvironment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Inflammation has recently been proposed to be the seventh hallmark of cancer 94. An inflammatory component is present in many solid tumours, and indeed inflammation predisposes to cancer at a number of sites 94, 95. Macrophages form a major inflammatory population in most tumours and are important determinants of the inflammatory milieu. Transgenic mice and human studies have demonstrated the close link between macrophages and tumour progression 96–98. An important study in which mice, which spontaneously develop mammary tumours (PyMT-MMTV), were crossed with mice that lacked the macrophage growth factor colony stimulating factor (CSF-1), demonstrated that these mice developed tumours much more slowly and had fewer metastases than the control mice 96, 99. This was associated with delayed angiogenesis, which was re-established when the macrophages were restored. This work led to the description of tumour-associated macrophages (TAMs). These are abundant in most forms of solid tumour and display a characteristic phenotype. TAMs produce many tumour-promoting factors, eg EGF and VEGF 100, 101, release cytokines and enzymes that promote invasion, angiogenesis and metastasis, eg VEGF and MMP9 102, 103 and down-regulate expression of anti-angiogenic factors, eg IL-12 (Figure 4A) 104.

thumbnail image

Figure 4. (A) Effect of inflammatory cell infiltrate on tumour progression. The inflammatory response, through M1 macrophages, B cells and cytotoxic T cells, exert tumour-suppressor functions. Tumour-derived cytokines can modify the inflammatory infiltrate, polarizing it towards pro-angiogenic and pro-tumour properties, with induction of T-reg cells and the M2 tumour-associated m(TAM) phenotype. (B) Infiltration of tumour-associated macrophages around breast ductal carcinoma in situ (DCIS); a duct containing DCIS is surrounded by an inflammatory infiltrate, the majority of which stains positively for arginase (green) and the pan-macrophage marker CD68 (red), indicating an M2 pro-tumour phenotype. Nuclear stain is DAPI

Download figure to PowerPoint

Broadly speaking, macrophages can be defined as classically (M1) or alternatively (M2) activated 88. In normal immunological responses, classical activation is involved with Th1 cytokine responses to pathogens, eg IFNγ and LPS. M1 macrophages are characterized by high levels of MHCII, IL-12 and TNFα and production of nitric oxide (NO) and reactive oxygen species (ROS). M2 macrophage differentiation is associated with Th2 cytokines (eg IL-4 and IL-13) resulting from wounding or humoral responses 105. M2 macrophages can be identified using a number of markers, since they express arginase, mannose receptor, high-level IL-10, low-level MHC II and IL-12 (Figure 4B). Other macrophage populations exist which are not so easily classified, leading to the concept of a spectrum of macrophage phenotypes with a range of functions 106. TAMs have been suggested to be biased towards an M2 phenotype but contain elements of both M1 and M2.

Hypoxia also plays a key role in determining the phenotype of infiltrating monocytes. TAMs are recruited to sites of hypoxia and necrosis 100, 107, 108 where, possibly due to high levels of endothelial monocyte-activating polypeptide (EMAP) II, endothelin II produced by the tumour cells 109 and necrotic debris 110, they become trapped and immobilized 111. TAMs up-regulate HIF1/2 112, 113 in response to the hypoxic environment, leading to expression of HIF-responsive genes VEGF and Tie-2114, providing TAMs with a potent pro-angiogenic phenotype. Tie-2 binds angiopoietin-2 (Ang-2), which is found at high levels in breast and other tumours 115, and this causes down-regulation of anti-angiogenic IL-12 and tumoricidal TNFα 116. Thus, in a complex network, the microenvironment controls macrophage phenotype, which then impacts on tumour cell behaviour.

Other components of the inflammatory infiltrate also modulate tumour behaviour. T-regulatory (T-reg) cells were first identified in 1971 by Gershon and Kondo 117, followed by reports of T cells suppressing the anti-tumour immune response 118, and then by identification of CD4+ T cells that suppressed autologus cytotoxic anti-tumour immune response 119. These T cells are now classified as CD4+, CD25+, FoxP3+ and are thought to protect the host from autoimmune disease by suppressing self-reactive cells and therefore also blocking anti-tumour responses.

The importance of T-regs was demonstrated when unfractionated cells from tumour-draining lymph nodes, taken 9 days after tumour challenge, gave complete rejection of established tumours. However, transfer of even a four-fold higher number of cells harvested at day 12 did not prevent tumour progression, owing to the presence of tumour-induced suppressor T cells, generated in the intervening 3 days, counteracting the anti-tumour response 120. Further, it has been shown that intra-tumoral depletion of CD4+ T cells can lead to the eradication of established tumours and the development of long-term anti-tumour memory 121. It has also been suggested, in a pancreatic model, that the cancer promotes the accrual of T-regs to suppress the anti-tumour response 122.

It is clear that the inflammatory infiltrate associated with solid tumours can exert pro- and anti-tumour functions and, further, that this cellular compartment is highly targetable. It is essential now to establish, and harness, the mechanisms promoting an anti-tumour response and ablate the inflammatory pro-tumour activity.

Clinical exploitation of the tumour microenvironment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Considerable experimental evidence points to a significant role for the microenvironment in the modulation of tumour development, growth and spread, and it is important to consider how this knowledge can be harnessed for clinical benefit. There are two broad categories in which knowledge of the tumour microenvironment can be exploited, which are discussed below.

Prognostic and predictive value

Bergamaschi et al123 were one of the first groups to classify a cancer subtype on the basis of ECM-related gene profiles. They generated four ECM-gene classes of breast cancer, with one group, characterized by high levels of protease inhibitors, being associated with significantly improved prognosis, and another group, characterized by high-level expression of integrins, metallopeptidases and SPARC, exhibiting poor prognosis and significantly reduced survival. Finak et al124 carried out Laser Capture Microdissection (LCM) of stroma from breast cancers and matched normal tissue in order to characterize stromal gene expression patterns. They identified a 26-gene stromal-derived prognostic predictor (SDPP) that could stratify patients into good and poor prognostic categories independently of classical prognostic factors, including tumour grade, size, LN status and ER, PR and Her2 status. The SDPP was also a more powerful predictor of prognosis than other prognostic signatures derived from whole-tumour samples 124, underlining the critical role of the stroma in determining disease response and outcome. The discriminating genes reflect underlying biological processes including angiogenesis, hypoxic response, tumour-associated macrophage responses, as well as regulators of Wnt signalling and development. Whilst many of these biological processes have previously been related to poor prognosis in different cancers, their individual prognostic power is relatively weak; it is the integration of these pathways that provides a powerful, independent prognostic predictor.

In a similar approach in the prostate cancer CR2-Tag mouse model, Bacac et al identified a stromal response signature, rich in lysosomal proteases, ADAM-TS and matrix metalloproteinases, reflecting high levels of stromal remodelling 125. When this signature was evaluated in datasets from human prostate cancer patients, up-regulation of human orthologues of these genes was significantly associated with reduced disease-free survival. Furthermore, the same signature could discriminate between good and poor prognosis in a series of breast cancer patients, but not in lung, gastric or renal cell carcinoma, suggesting an organ-specific pattern of response in the microenvironment.

As well as providing prognostic information, it appears that characteristics of the tumour microenvironment can predict response to treatment. A recently reported prospective gene array study on patients with colorectal cancer identified up-regulation of a TNF family ligand gene called APRIL in a subset of patients treated with 5-fluorouracil (5FU), and in an independent cohort they showed that stromal APRIL expression was associated with significantly reduced survival 126. Interestingly, APRIL expression had no relationship with prognosis in those patients not treated with 5FU, suggesting that stromal APRIL expression acts as a predictive marker for chemoresistance.

Similarly, Farmer et al127 used a bioinformatics approach to identify a gene expression signature that might predict response to neoadjuvant chemotherapy in breast cancer. They identified nine gene clusters, each represented by a single metagene, and showed that only the stromal gene cluster predicted response to chemotherapy. The signature included genes indicative of a reactive stroma, such as fibronectin, SPARC, PDGFR and MMPs. As with the APRIL gene 127, this signature was of no prognostic value in those patients not treated with chemotherapy, suggesting that it represents a distinct biological feature of the tumour.

Although it is not entirely clear how the stromal environment might confer resistance to chemotherapy, previous work showing that interaction of tumour cells with fibronectin can enhance tumour survival 128, 129, and that engagement of integrins with ECM leads to resistance to apoptosis 130, indicate potential mechanisms. Overall, these studies suggest that anti-stroma agents may have a place in overcoming resistance to chemotherapy.

Therapeutic targeting

The biological, prognostic and predictive evidence suggests that the complex tumour microenvironment offers a number of avenues for therapeutic targeting (Figure 5).

thumbnail image

Figure 5. Treatment strategies for targeting the microenvironment. The altered extracellular matrix (ECM) could be targeted using humanized monoclonal antibodies towards tumour-associated ECM isoforms such as fibronectin (FN), EDB and tenascin-C domain C (TNC). The inflammatory infiltrate may be targeted using TNFα antagonists, inhibitors of COX-2, such as Celecoxib, inhibitors of the NF-κB pathway, such as KINK-1, and broader immunomodulatory drugs, such as lenalidomide. Numerous drugs have been developed targeting angiogenesis, including VEGF inhibitors and the αvβ3 integrin antagonist cilengitide. Despite the disappointing results of early MMP inhibitors, more specific MMP inhibitors may still have therapeutic benefit

Download figure to PowerPoint

The essential requirement of the angiogenic switch for the progression of solid tumours has led to the development of inhibitors as a therapy (Table 1) 131. The majority of these target the VEGF signalling pathway 132 and can extend progression-free survival in colorectal, lung and breast cancer when used in combination with other chemotherapeutics 133–135. Angiogenesis is also regulated through integrins αvβ3 and αvβ5, which are expressed on endothelial cells. Inhibitors of these integrins have been shown to suppress tumour growth in preclinical models 136–138. The mechanism is thought to be through blocking their adhesive functions and hence preventing tumour growth by targeting the tumour cells as well as inhibiting angiogenesis 139, 140. Cilengitide, an αvβ3 inhibitor, has been effective in the treatment of glioma 141; however, this may be due to glioma tumour cells expressing high levels of αv integrins, rather than an effect on angiogenesis 142, 143. This hypothesis is further strengthened by the failure of cilengitide to produce results in other clinical settings 144–146. The reasons for this are unclear but a study by Reynolds et al147 demonstrated that at low doses (0.2–20 nM) cilengitide could enhance the growth of tumours in vivo by promoting angiogenesis. In a clinical setting, a large dose of the drug given at biweekly intervals could lead to plasma levels in patients reaching nanomolar concentrations between administrations, consequently promoting angiogenesis and tumour growth. Thus, further elucidation of the precise biology of angiogenesis is essential to optimize future therapies.

Table 1. Therapeutic targeting of the microenvironment: examples of small molecules, antibodies and siRNA molecules being used in preclinical studies and phase I–III clinical trials targeting different components of the microenvironment
Micro-environmental targetMolecular targetMoleculeEffectReferences
  1. TEC, tumour-associated endothelial cells; PSMA, prostate-specific membrane antigen; CTGF, connective tissue growth factor; SDF-1, stromal-derived factor 1; FGFR, fibroblast growth factor receptor; HGF, hepatocyte growth factor; uPA(R), urokinase plasminogen activator (receptor); VEGF, vascular endothelial growth factor.

Angiogenesisαvβ3CilengitideReduce angiogenesis and increase apoptosis of TEC in vitro. In the clinic they have shown some effect on reducing tumour growth and sensitizing tumours to other treatments141, 167
 αvβ5Abegrin  
 α5β1VolociximabInhibits angiogenesis168, 169
 HSP-7017AAGSensitizes endothelial cells to radiotherapy170
 VEGFBevacizumabNeutralize VEGF133, 171–175
 PSMAAntibody (MLN2704)PSMA expressed on TEC of many solid tumours; delays progression176–178
 CTGFFG-3019Blocking antibody decreases tumour growth and metastasis179
  DN-9693Inhibits VEGF-mediated stabilization of CTGF mRNA180
Signalling inhibitorsFGFRIn vitroInhibition of cell proliferation, cell cycle and enhances cell death181, 182
 HGFNK4Antagonist of HGF183
 PKB/Akt and mTORIn vitro siRNAInhibits FN-induced proliferation184
Cytokine inhibitorsSDF-1/CXCR4Bryostatin-5Antagonizes CXCR4-mediated migration and metastasis/inhibits neovascularization185, 186
 CD105 (endoglin)SN6j (Ab)Inhibits CD105 antagonistic effect on TGFβ inhibitory functions and induces apoptosis in hypoxia187
ECM degradation inhibitorsFAPα (Seprase)SibrotuzumabReduced growth and invasion188, 189
 Tenascin-C81C6 (131I-labelled Ab)Delays tumour growth, prolonged survival190
  TTA1 (aptamer)Facilitates delivery of radioactive isotopes191
 uPA/uPARA6Reduces tumour growth, metastasis and angiogenesis192
 MMPsMarimastat (BB-2516)In vitro and xenograft models show anti-invasive, metastatic and angiogenic effects193, 194
  Rebimastat (BMS-275291)No conclusive clinical benefit found195
HypoxiaCAIXWX-G250Induces antibody-dependent cellular toxicity196, 197

Other aspects of the tumour microenvironment are also being targeted. The expression of ‘tumour-specific’ ECM proteins has been exploited to target delivery of bioactive molecules to tumours: these ECM components are highly abundant in tumours and are often more stable than antigens located on the cell surface of tumour cells. Radiolabelled antibodies specific to TNC domains A1 and D have been used successfully in the clinic to treat glioma and lymphoma 148. Using antibody phage technology, a human monoclonal antibody to the C domain of TNC has been generated, and this scFv protein shows a highly selective uptake in gliomas, making it a promising tool for the future 149. The aim is not necessarily to inhibit the function of the target but to use the target to concentrate delivery of bioactive molecules. This has been particularly successful targeting the FN–EDB 48. EDB antibodies show specific localization to a range of tumours, including brain, lung and colorectal cancers 150. The EDB-targeting antibody L19 has been used as a vehicle for TNFα and has been shown to induce necrosis in tumours 151, 152. Recently EDB has been targeted in lymphoma patients, using a radiolabelled antibody 131I-L19SIP. Two patients treated with this antibody showed a sustained partial remission, indicating that a therapeutic dose of radioactivity can be delivered to tumours using this approach 153.

Components of the immune system have also been harnessed for therapeutic gain. One of the approaches taken has been to re-establish an anti-tumour inflammatory milieu. TNFα antagonists have been shown to induce stabilization of disease and partial responses in breast and advanced cancer 154, 155, and multiple myeloma is treated very successfully with combinations of drugs, including lenalidomide 156, 157, which suppresses the production of several inflammatory cytokines 158. Work by Hagemann et al159, aimed at reprogramming the macrophage phenotype, has demonstrated that by inhibiting IKKβ macrophages change from a TAM phenotype to a tumour suppressive phenotype. A novel small molecule of IKKβ has been developed (KINK-1), which demonstrates the ability to sensitize tumours in mice to doxorubicin and reduce tumour mass and metastases 160. It remains to be determined whether there is any effect on macrophage phenotype.

A number of drugs in clinical trials for other diseases that target the immune system, such as non-steroidal anti-inflammatory drugs (NSAIDs; eg COX inhibitors such as celecoxib) for treating arthritis 161, 162, have been applied to some tumours, eg malignant melanoma and pancreatic cancer 163, 164. COX-2 inhibitors have also been shown to prevent the recurrence of sporadic 165 and genetically predisposed 166 adenomas.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

It has taken decades for the concept of the microenvironment as an important determinant of tumour behaviour to gain acceptance. Elucidating the nature of the interactions between the tumour and the multiple facets of the microenvironment will allow us to harness this relationship for clinical benefit.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Influence of cancer-associated fibroblasts on tumour behaviour
  5. The origin of cancer-associated fibroblasts
  6. The extracellular matrix and its remodelling
  7. Organ-specific microenvironment: the myoepithelial cell
  8. Hypoxia and angiogenesis
  9. The inflammatory microenvironment
  10. Clinical exploitation of the tumour microenvironment
  11. Conclusions
  12. Author contributions
  13. Teaching Materials
  14. References
  15. Supporting Information
FilenameFormatSizeDescription
path_2803_suppinfo.ppt1682KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.