Tumor Microenvironment

1

The Microenvironment in Cancer

Nicole N. Parker and Dietmar W. Siemann

Department of Radiation Oncology and University of Florida Shands Cancer Center, University of Florida, Gainesville, Florida, USA

1.1 Introduction

In the development of a cancer, the transformation of epithelial cells into a neoplastic and progressively invasive tumor occurs though the acquisition of several procancer characteristics that can take years or decades to develop. The particular stages of transformation have been established and a general consensus exists about the properties of a successful malignancy. While many therapeutics have been developed to combat these properties, these therapies are not universally successful, and their efficacy depends on the type and site of the primary tumor, its degree of vascularization, the proliferative compartment of the tumor, and in particular, the tumor microenvironment. The latter is the key support system of a cancer, and is an important source of critical protumorigenic factors that facilitate growth, invasion, angiogenesis, and metastatic ability. The focus of this book is to examine how the reliance of tumors on their microenvironments for development and preservation of key cellular functions is now recognized not only as a major contributor to cancer aggression and treatment resistance but also as a potential target for novel therapeutic intervention strategies.

1.2 A highly selective process is required to obtain the cancer phenotype

Many studies have focused on the predetermining factors that cause hyperplasia, or the hyperproliferation of cells within their normal environment. On the path to a cancer, regions of hyperplasticity must subsequently become dysplastic, or display a highly disordered pattern of proliferation with little or no growth regulation. Predisposing genomic lesions in various genes of these dysplastic cells confer a proliferative advantage over normal cellular counterparts. Therefore, the out-proliferation exhibited by dysplastic populations requires additional genetic instability or genomic modifications, and these cells are considered neoplastic: they have an advantageous rate of proliferation with a lack of regulation, and possess other procancerous features at the time of transformation. Oncogenic lesions, coupled with inhibition of tumor suppressors, together contribute to cellular transformation. Genomic, proteomic, post-translational, and epigenetic mutations are responsible for activating oncogenes and inhibiting tumor suppressor genes.

1.3 The cancer phenotype

Although a multitude of potential cancer etiologies may occur for cancer development, several essential characteristics are present in malignancies (Hanahan and Weinberg, 2000). A key element of malignant transformation is the loss of regulatory control mechanisms present in normal somatic non-stem cells that are growth arrested and do not divide. Cancer cells not only possess heightened rates of cell proliferation and aberrant cell cycle checkpoints, but also lose contact-inhibited growth regulation. As a result, unlike detached normal cells which die by anoikis or cell detachment-induced apoptosis, cancer cells continue to grow unabated, breaking through basement membranes and invading extracellular spaces around tissues and organs.

Extracellular matrix remodeling and cellular changes in adhesion molecules are both required for a cancer cell to become more motile. Rearrangement of the actin cytoskeleton facilitates cell motility and plasticity, as does downregulation of adhesion proteins that bind tightly to the extracellular matrix (Chapter 3). A cancer cell modifies its adhesive properties and implements a program of non-adhesion through multiple modifications.

The progressive growth of a tumor ultimately results in an inability of normal tissue blood vessels to oxygenate and provide nutrients to tumor cells most distal to the blood supply. As a consequence oxygen-deficient (hypoxic) regions develop within the tumor (Chapters 2 and 9). The ability of transformed cells to survive hypoxic conditions requires a switch from aerobic to anaerobic glycolysis, a major approach by which cancer cells circumvent the cytotoxic effects of oxygen deprivation (Gatenby and Gawlinski, 2003). As a result of glycolysis, lactic acid byproducts accumulate in cells undergoing this process. Although this acidification is generally toxic to cells, cancer cells upregulate acid transporter proteins and efficiently secrete acid products into the surrounding environment. A side effect of acid secretion is an increase in local extracellular acidity, fluid retention, and subsequently, an increase in interstitial pressure (Chapter 9). Still, the ability of cancer cells to metabolically adapt by preferentially undergoing glycolysis even in the presence of oxygen not only provides a survival advantage over non-transformed cells but also ensures the persistence of only the most successful cancer cells (Gatenby and Gillies, 2004).

The outgrowth of a tumor that is beyond the diffusion limits of nearby blood vessels, which supply nutrients and oxygen, leads to another critical phenotypic advantage of cancer cells: the ability to induce angiogenesis, or the process of developing new blood vessels from existing vascular structures. Cancer cells accomplish this through the upregulation and release of proangiogenic factors that can destabilize endothelial cells and induce vascular outgrowth from normal blood vessels. Endothelial cells proliferate toward the source of the chemoattractant angiogenic factors to form a new capillary network for the tumor mass. However, unlike normal vasculature, which is extremely ordered, this newly developed tumor neovasculature is highly aberrant in structure, lacking organization and vessel integrity. Although many proangiogenic factors have now been identified, vascular endothelial growth factor (VEGF) is believed to be the major inducer of tumor angiogenesis. It has not only been implicated in many cancer types (Fukumura et al., 1998) but importantly, VEGF expression has been shown to correlate with tumor angiogenesis and aggression, poor patient outcome, and is a predictor for metastasis and high tumor grade in multiple cancer types (Brychtova et al., 2008). Interestingly, some studies have demonstrated that acidic microenvironments can induce vesicle lysing, thereby secreting VEGF into the tumor microenvironment and contributing to a feed-forward mechanism in which tumor-induced hypoxia and cellular acidification lead to the formation of neovasculature.

It is generally believed that tumors cannot grow to a size larger than a few cubic millimeters without inducing a neovasculature. Once cancer cells induce revascularization, thereby ensuring a more constant nutrient supply, this growth restriction is effectively removed. The requirement for additional tumor space necessitates the ability of cancer cells to invade into surrounding tissue. It is most advantageous for cancer cells to digest adjacent extracellular matrix and force the local reorganization of normal epithelia and surrounding stromal elements.

1.4 The extracellular matrix

The extracellular matrix is comprised of various cell types and secreted proteins that help maintain the organization of higher-order cellular structures. In addition to containing various cell types, the matrix is deposited as a mix of such proteins as collagens, fibronectin, laminins, hyoluronan, plasminogens, proteases, and numerous others, which collectively form an inflexible scaffold to which cells attach. In addition, other secreted cellular proteins such as cytokines and extracellular matrix remodeling proteins normally reside in the extracellular matrix (Chapters 3 and 4). These proteins are released when the matrix is degraded, and upon their release become activated due to proteases and other activating enzymes present in the extracellular environment, further contributing to the regulation of extracellular matrix turnover.

Many cell types are present in the extracellular matrix and the tumor milieu. Examples include fibroblasts, which are an integral inducer of matrix remodeling, as well as endothelial cells, hematopoietic-derived cells, and immune cells, which normally monitor this environment for foreign (i.e., non-host) bodies (Chapter 5). In cancer, particularly at the later stages of transformation and invasion, normal immune functions are subverted, leading to recognition of the tumor as part of the host, rather than as an invading foreign entity.

1.5 Motility, invasion, and metastatic ability

Successful and evolutionarily adapted cancer cells are motile, have no major attachments to extracellular substrata, and can more easily move through the extracellular and intracellular space due to decreased cell adhesion and an increase in factors which facilitate extracellular matrix degradation and remodeling. A natural effect of motility is that cancer cells invade into surrounding tissue, colonize and populate the area given a favorable microenvironment. Further, cancer cell motility facilitates the movement of the cancer cell through layers of endothelial cells surrounding blood vessels, enabled in part through cancer cell secretion of vascular destabilizing factors. As a result the vasculature is perturbated and cancer cells gain access to the circulation, which is the major mode of transport for cancer cells to reach distant organs.

In the metastatic cascade, the tumor microenvironments of both the primary tumor and the target sites colonized by cells shed from the primary tumor are of critical importance to the successful spread of neoplastic cells (Joyce and Pollard, 2009) (Chapters 6–8 and 12). The classic ‘seed and soil’ mechanism describes a situation in which only permissive target microenvironments enable the attachment and subsequent proliferation by a metastatic cell. In addition, cells in the primary site of a cancer shed factors and progrowth signals that contribute to the tumor microenvironment at the secondary sites of tumor formation (Chapter 8). In this way, metastatic tumor cells establishing new colonies continue to receive progrowth support signals while they are colonizing the secondary site and during subsequent phases of secondary tumor growth.

In addition, the evolution of the microenvironment can impact premetastatic cells in a manner that leads to an invasive, advanced, and evolutionarily favored metastatic phenotype that can survive extravasation, intravasation, and can establish new tumors at distant sites. Accumulating evidence suggests that the hypoxic conditions that select for successful tumor types also contribute to the metastatic potential of that tumor (Chapter 14). Therefore, eradication of the hypoxic regions of a cancer has short-term and long-term benefits (Chapters 16–18), in that both tumor bulk and metastatic capability are reduced.

1.6 Impact of the tumor microenvironment on the control of cancer

The tumor microenvironment is a growing target for consideration of cancer therapeutics due to its varied influence on the cells and on the physical aspects of chemotherapeutic delivery (Chapters 2 and 15). Several drawbacks to traditional chemotherapies that do not account for the microenvironment are: the tumor vasculature, which is highly disordered and leaky; tumor core hypoxia, which confers radiation resistance on tumor cells in this state; cells furthest from blood vessels become growth-arrested, preventing efficacious chemotherapeutic inhibition of proliferating cells; and the upregulation of acid transporter and other transporter proteins, which efficiently excrete chemotherapeutics from cancer cells and hinder successful cancer treatment (Chapters 2, 9, and 15). Importantly, offspring of chemotherapeutic survivors can pass this genetic property to daughter cells, making subsequent populations of tumor cells highly resistant to subsequent therapy. One further consideration regarding the tumor microenvironment is the stem cell population, a slow-growing subset of cancer cells, which is inherently resistant to therapies targeting cells that are actively cycling. Stem-or stem-like cancer cells are pluripotent, highly plastic, and dedifferentiated entities that easily and steadily repopulate tumors following therapy. Considering these scenarios, targeting the tumor microenvironment becomes an increasingly logical and attractive therapeutic option in cancer management.

1.7 Targeting the tumor microenvironment

Classical anticancer therapies including radiotherapy and chemotherapy are toxic to cancer cells but such treatments are typically also associated with inadvertent damage to critical normal tissues. Newer and more specific therapies have become more prevalent in the treatment of specific cancers as the molecular mechanisms of carcinogenesis become better characterized. The approach of uncovering molecular etiologies of cancer coupled with the development of targeted therapies that exploit essential signaling pathways (Chapters 10, 13 and 17) will undoubtedly contribute to the future arsenal of anticancer therapeutics.

Because the microenvironment of tumors not only severely impairs the treatment efficacy of conventional anticancer therapies but also differs significantly from those found in normal tissues, research is beginning to focus on the tumor microenvironment as a separate cancer-associated entity that may be targeted (Chapters 10, 13 and 17). Indeed, several strategies have already been identified that exert an anticancer effect through the specific targeting of the tumor microenvironment. Oxygen-poor cells display greater resistance to radiotherapy, and methods for reversing the radioprotective effects of hypoxia in order to enhance the treatment efficacy of radiotherapy have received considerable attention (Chapters 9 and 16). The other compartment of the tumor microenvironment that has been extensively targeted is the tumor vasculature (Chapter 18). As an essential part of tumor survival, such a strategy seeks to deprive the tumor of critical nutrients and means to spread. The use of antiangiogenic and vascular disruptive therapies provide powerful adjuncts to conventional anticancer treatments. All such potential therapeutic interventions will be critically dependent upon the establishment of novel approaches to non-invasive imaging of the tumor microenvironment (Chapter 11).

1.8 Summary

The microenvironment of tumors creates a significant hindrance to the control of cancers by conventional anticancer therapies. The physical conditions present are imposing and manifold, and include elevated interstitial pressure, localized extracellular acidity, regions of oxygen and nutrient deprivation, and contraction of the extracellular matrix. No less important are the functional consequences experienced by the tumor cells residing in such environments: adaptation to hypoxia, cell quiescence, modulation of transporters, and enhanced metastatic potential. Together these factors lead to therapeutic barriers that may render the chance of tumor elimination as minimal.

However, the aberrant nature of the tumor microenvironments also offers unique therapeutic opportunities. Reducing tumor hypoxia can improve drug delivery and enhance radiotherapy. The inhibition of fibroblasts and other cell types exploits the tumor’s reliance on the microenvironment for various factors and properties necessary for its survival. Targeting the tumor vasculature would destroy the nutritional support network of the tumor. These approaches and many others directed against the tumor microenvironment are under active investigation in the laboratory and the clinic.

Because the molecular underpinnings of cancer development are becoming increasingly well characterized, future studies will undoubtedly identify distinct molecular markings that are characteristic of the tumor microenvironment. Such advances will lead to the development of targeted therapies that will selectively impair the neoplastic cell populations residing in these environments. Ultimately, by combining such therapies with conventional anticancer treatments it may be possible to bring cancer growth, invasion, and metastasis to a halt.

References

Brychtova, S., Bezdekova, M., Brychta, T., and Tichy, M. (2008) The role of vascular endothelial growth factors and their receptors in malignant melanomas. Neoplasma, 55, 273-279.

Fukumura, D., Xavier, R., Sugiura, T. et al. (1998) Tumor induction of VEGF promoter activity in stromal cells. Cell, 94, 715–725.

Gatenby, R.A. and Gawlinski, E.T. (2003) The glycolytic phenotype in carcinogenesis and tumor invasion: insights through mathematical models. Cancer Research, 63, 3847–3854.

Gatenby, R.A. and Gillies, R.J. (2004) Why do cancers have high aerobic glycolysis? Nature Reviews Cancer, 4, 891–899.

Hanahan, D. and Weinberg, R.A. (2000) The hallmarks of cancer. Cell, 100, 57–70.

Joyce, J.A. and Pollard, J.W. (2009) Microenvironmental regulation of metastasis. Nature Reviews Cancer, 9, 239–252.

2

Establishing the Tumor Microenvironment

Allison S. Betof¹ and Mark W. Dewhirst²

¹Department of Pathology, Duke University Medical Center, Durham, USA

²Department of Radiation Oncology, Duke University Medical Center, Durham, USA

2.1 Introduction

For years cancer researchers focused on a seemingly obvious target: tumor cells. While this research yielded invaluable knowledge, a comprehensive understanding of tumor behavior remains elusive. Just as children are influenced by their environment as they grow, so too are developing tumors. Tumors do not grow in a vacuum. Instead, they are surrounded by the extracellular matrix, blood vessels, immune cells, and other supporting structures that make up human organs. Together this support system comprises a tumor’s microenvironment. In recent years, cancer research has experienced a paradigm shift toward trying to understand the role of the microenvironment in tumor development, growth, metastasis, and treatment. This chapter will focus on the establishment of that environment.

There are two principal fields of microenvironmental research. One focuses on cellular components, cell–cell interactions, and the matrix that makes up the tumor parenchyma in addition to tumor cells. The other component is the physiological microenvironment. This involves the exchange of oxygen, nutrients, and waste products through tumor vasculature. While they are often discussed separately, these two components of the microenvironment do not exist in isolation. Though each component of the microenvironment will be addressed separately, this chapter will also review areas of overlap and interaction between the physical and physiological microenvironments.

2.2 From cancerous cells to a tumor

The complex, heterogeneous structures we know as tumors arise from cancer cells that proliferate within the space dedicated to another organ. The presence of malignant cells within normal tissue alters the environment and interrupts tissue homeostasis. Reciprocal interactions with the host stroma facilitate tumor proliferation, invasion, and metastasis. However, the question of which of these changes occurs first has become a central issue. Do cancer cells transform and then change the microenvironment where they are located to suit their needs? Or are there changes in the previously normal tissue that promote transformation and tumorigenesis in a particular location?

The answer, it now seems, is that both of these alterations can result in the formation of a tumor. For evidence that stromal changes can result in tumorigenesis in the absence of cancer cells, the reader is directed to one of several excellent reviews (Ariztia et al., 2006; Bhowmick, Neilson, and Moses, 2004b). However, for most tumors that are epithelial in origin, the initiating event is usually an epithelial mutation followed by key paracrine signals from stromal fibroblasts that promote neoplastic progression (Lengauer, Kinzler, and Vogelstein, 1998; Bhowmick, Neilson, and Moses, 2004b). Reciprocal interactions between differentiating epithelia and adjacent stromal cells are essential to reaching proper tissue maturation in normal tissue (Bhowmick, Neilson, and Moses, 2004b). In contrast, malignant transformation of epithelial cells is preceded by or occurs simultaneously with host stromal activation, inducing angiogenesis, fibroblast proliferation, and recruitment of inflammatory mediators (Tomakidi et al., 1999; Ronnov-Jessen, Petersen, and Bissell, 1996; Tlsty and Hein, 2001). Stromal fibroblasts respond to tumor-derived pathological signals by producing collagen and activating myofibroblast-associated proteins such as α-smooth muscle actin, vimentin, and desmin (Tlsty and Hein, 2001). In addition, activated fibroblasts proliferate and migrate more in vitro than fibroblasts from benign tissue (Schor, Schor, and Rushton, 1988; Schor et al., 1985).

Cancer-associated fibroblasts (CAFs) are stromal fibroblasts with key roles in transformation, proliferation, and invasion, three of the hallmarks of cancer. These fibroblasts secrete growth factors and chemokines that signal critical changes in the extracellular matrix (ECM) and contribute oncogenic signals that increase proliferation and invasion (Kalluri and Zeisberg, 2006). For example, CAFs are able to stimulate both growth and transformation of immortalized prostate epithelial cells in vitro and in vivo, whereas normal fibroblasts are not (Olumi et al., 1999). In addition, Parrott et al. used an ovarian cancer model to show that tumor fibroblasts may not just assist in tumorigenesis, but may actually be necessary to form a tumor from cancer cells (Parrott et al., 2001). This group then used histology to show that the ovarian cancer cells recruited adjacent stroma from the normal mouse tissue to form the stroma of the tumor. Thus, the CAFs play a crucial role in establishing the physical microenvironment of a growing tumor.

One possible mechanism for the tumor promoting effects of CAFs comes from their overexpression of fibroblast secreted protein-1 (FSP1) (Grum-Schwensen et al., 2005). When FSP1 knockout mice are injected with carcinoma cells, tumor formation is decreased compared to wild-type mice and the tumors that do form do not metastasize. Furthermore, adding FSP1+/+ fibroblasts in addition to the carcinoma cells restores the wild-type phenotype. A second possible mechanism is that stromal CAFs secrete stromal cell-derived factor 1α (SDF-1α, also known as the cytokine CXCL12), which promotes proliferation of mammary carcinoma cells when it binds to the receptor CXCR4 on the cancer cells (Orimo et al., 2005). These signaling molecules clearly contribute to the developing tumor microenvironment, but their precise effects and the contributions of other factors remain unknown.

2.3 A tumor is more than cancer cells and fibroblasts

In addition to the tumor cells and CAFs already discussed, the microenvironment consists of normal parenchymal and epithelial cells, ECM, normal stromal fibroblasts, soluble growth factors and cytokines, components of the immune system, nerves, and blood and lymphatic vessels. However, this is more than just a physical space. Implied within the term ‘microenvironment’ is a venue for these factors to interact.

2.3.1 The extracellular matrix

Within a carcinoma, epithelial cells are supported by a three-dimensional structure known as the ECM. The proteins that make up the ECM are produced by fibroblasts. The stroma is separated from the epithelium and endothelium by the basement membrane, a specialized type of ECM composed of collagen IV, laminin, and heparan sulfate proteoglycans (Kalluri, 2003).

There are considerable differences in matrix composition between tumor stroma and the stroma associated with normal tissue. Cells from the tumor secrete a variety of proteins into the ECM that are involved in cell adhesion, motility, communication, and invasion (Mbeunkui and Johann, 2009). Notably, some of these molecules are involved in degrading the ECM. This degradation occurs near the tumor-host interface, where the tumor-derived proteases overwhelm the host’s endogenous inhibitors leading to extensive remodeling and stimulating alternative signals from the cell surface (Handsley and Edwards, 2005). Remodeling is achieved by combined efforts of secreted and membrane-anchored matrix metalloproteinases (MMPs), adamalysin-related membrane proteases, bone morphogenic protein-1-type metalloproteinases, endoglycosidases, and tissue serine proteases including tissue plasminogen activator, urokinase, thrombin, and plasmin (Carmeliet, 2003; Werb, 1997). Interestingly, most of these factors involved in remodeling the tumor-host interface originate from host cells, not from the growing carcinoma (Bowden et al., 1999; Coussens et al., 2000; Nakahara et al., 1997; Sternlicht et al., 1999).

Infiltrating inflammatory cells release MMP-9, MMP-12, and MMP-8 from intracellular stores, but they also release cytokines including IL-1β and tumor necrosis factor-α (TNF-α) that stimulate fibroblasts to produce more MMPs (Van Kempen et al., 2003). In addition to fibroblasts and inflammatory cells, paracrine signals stimulate other stromal cells to be the predominant source of microenvironmental MMPs during tumorigenesis (Van Kempen et al., 2003; Coussens and Werb, 2002).

Secreted MMPs degrade both ECM and other proteins in the microenvironment, including growth factors, cytokines, and receptors. Therefore, the effects of individual MMPs on the microenvironment are diverse, but they are critical players in establishing the environment surrounding tumor cells. One potent example is MMP-7, which is expressed by malignant breast epithelial cells. MMP-7 degrades the ECM, disrupts the basement membrane, and cleaves E-cadherin, weakening the connection between breast epithelial cells (Fingleton et al., 2001; Noe et al., 2001). MMPs are found in the microenvironment in zymogen form, and they require activation by other MMPs and related molecules. The reader is cautioned that MMP overexpression alone, which has often been reported in tumors, does not explain MMP activity in the microenvironment, since immunohistochemistry often does not distinguish between the zymogen and active forms (Coussens, Fingleton, and Matrisian, 2002). Under normal conditions MMPs are also regulated by endogenous tissue inhibitors of metalloproteinases (TIMPs), so the balance between TIMPs and MMPs critically affects the microenvironment (Mbeunkui and Johann, 2009).

To balance these remodeling and degradative enzymes, synthesis of matrix components is upregulated by both the tumor and host. Activated fibroblasts synthesize collagen type I and fibronectin in response to the binding of mast cell tryptase to protease-activated receptor-2 (Coussens and Werb, 2002; Frungieri et al., 2002). Macrophages also contribute interleukin (IL)-1β and nitric oxide synthase, both of which augment type I collagen synthesis (Van Kempen et al., 2003). The newly synthesized collagen that forms the peritumoral stroma is loosely woven and disorganized, contributing to the overall disorder surrounding a developing tumor (Ruiter et al., 2002).

2.3.2 Immune cells in the microenvironment

The immune system makes an invaluable contribution to the tumor microenvironment. Inflammatory cells secrete growth factors, cytokines, and chemokines that stimulate epithelial proliferation and generate reactive oxygen species (ROS) that damage DNA, promoting tumor initiation and progression (Coussens and Werb, 2002). They also release proteolytic enzymes, leading to matrix remodeling and angiogenesis. The first cells to respond to tumor growth factors are mast cells, which are attracted by stem cell factor (SCF), vascular endothelial growth factor (VEGF), epidermal growth factor (EGF), basic fibroblast growth factor (bFGF), and platelet-derived growth factor (PDGF) (Hiromatsu and Toda, 2003). Upon arrival, mast cells degranulate to release VEGF (Norrby, 2002), serine proteases (Yong, 1997), and MMP-9 (Fang et al., 1997).

Macrophages are typically the next inflammatory cells to infiltrate the growing microenvironment. Once activated by tumor-derived transforming growth factor beta-1 (TGF-β1), these tumor-associated macrophages (TAMs) secrete TNF-α, IL-1, IL-6, IL-8, and/or bFGF (Ono et al., 1999; Varney et al., 2002). TAMs are thought to play an integral role in signaling angiogenesis, which is necessary for continued tumor growth (Van Kempen et al., 2003). Other inflammatory cells also make invaluable contributions that are beyond the scope of this chapter, but many excellent reviews on the subject are available.

2.4 Communication between the tumor cells and stroma

It is clear that many components go into making up a heterogeneous tumor that is capable of surviving, growing, and invading other tissues. However, this complex system requires organization. Communication between the tumor cells, host stroma, and tumor stroma is mediated by soluble autocrine and paracrine signaling molecules. The most common of these factors are members of the FGF, insulin-like growth factor (IGF), EGF, hepatocyte growth factor (HGF), TGF-β, and PDGF families (Ariztia et al., 2006). The majority of these fibroblast-derived factors enhance proliferation and tumorigenesis (Bhowmick, Neilson, and Moses, 2004b). However, TGF-β exhibits different properties.

Secretion of TGF-β regulates cell proliferation and normal fibroblast transformation (Moses et al., 1981; Roberts and Wakefield, 2003). Early studies showed that TGF-β inhibits the growth of epithelial cells, and subsequent transgenic mouse experiments reinforced the hypothesis that TGF-β is a tumor suppressor (Tucker et al., 1984; Pierce et al., 1995; Bottinger et al., 1997; Gorska et al., 2003). However, TGF-β is also known to both stimulate tumor progression by inducing epithelial-tomesenchymal transition (EMT) in a variety of carcinomas and sarcomas and promote angiogenesis (Akhurst and Derynck, 2001). Therefore, TGF-β exhibits both tumor suppressive and tumor promoting behavior. At least some of the protumor effects of TGF-β appear to be mediated by HGF (Bhowmick et al., 2004a). For a more comprehensive discussion on the role of growth factors in the tumor microenvironment, the reader is directed to an excellent review (Bhowmick, Neilson, and Moses, 2004b).

In addition to these growth factors, cytokines and chemokines play an essential signaling role in the microenvironmental milieu. Analysis of the gene-expression profiles of cells from normal breast and breast carcinomas showed that expression of the cytokines CXCL14 and CXCL12 by myoepithelial cells and myofibroblasts augments epithelial cell proliferation and invasion (Allinen et al., 2004). CXCL12, in addition to promoting proliferation, enhances recruitment of endothelial progenitor cells (EPCs), thereby supporting angiogenesis (Orimo et al., 2005). The proangiogenic effects of CXCL12 may be mediated by MMP-9, since the cytokine is known to activate MMP-9 in bone marrow cells and MMP-9 knockout mice are unable to respond to CXCL12-triggered EPC recruitment (Heissig et al., 2002). CXCL12 also affects inflammatory cells in the tumor microenvironment, and these effects are likely synergistic with those of TGF-β (Littlepage, Egeblad, and Werb, 2005).

2.5 Hypoxia and angiogenesis

Among the most important components of the tumor microenvironment are the growing and developing blood vessels. Both rapidly dividing tumor cells and the influx of many microenvironmental cells increase the demand for oxygen greatly. As the oxygenation state of a tumor is determined by both oxygen supply and demand, tissue oxygenation will fall as a result of either diminished arteriolar supply or increased oxygen consumption. Interestingly, our group has shown that the amount of hypoxia in a tumor depends more heavily on oxygen consumption than either flow rate or arteriolar pO2 (Secomb et al., 1995). Thus, the growing and developing microenvironment creates conditions that require increases in oxygen supply to maintain normoxia and support continued growth. This is the physiological microenvironment that will be the focus of the remainder of this chapter.

Before delving into the details of tumor hypoxia and angiogenesis, it is important to highlight ways in which the physical and physiological microenvironments overlap and interact. Few reviews have addressed this topic. Table 2.1 summarizes what is known about the effects of hypoxia on the various components of the physical microenvironment discussed above. Carbonic anhydrase IX (CA IX) is an endogenous hypoxia marker expressed by a variety of cells. Carcinoma-associated fibroblasts are known to express CA IX, and expression by CAFs correlates with poor outcomes (Nakao et al., 2009). Furthermore, hypoxia seems to affect collagen deposition, ECM degradation, and connections between the ECM and tumor cells in various types of tumors (Erkan et al., 2009; Higgins et al., 2007; Hull and Warfel, 1986; Leask and Abraham, 2006; Wu et al., 2010). Various MMPs are affected by hypoxia, including MT1-MMP, MMP-2, MMP-7, MMP-9, and MMP-13 (Miyoshi et al., 2006; Munoz-Najar et al., 2006; Osinsky et al., 2005; Swinson et al., 2004). This will affect matrix remodeling, leading to changes in the physical environment surrounding and supporting the growing tumor.

Most of the research on interactions between the physical environment and hypoxia has focused on signaling. IL-8 is a chemokine with important functions in immune response. Expression of IL-8 is increased in hypoxia, and this is effect is inhibited by the free radical scavenger N-acetyl-l-cysteine (Galindo et al., 2001; Wysoczynski et al., 2010). This is particularly interesting because free radicals and ROS are found in high concentrations in hypoxic human tumors. In addition, the stromal-derived chemokine CXCL12 and its receptor CXCR4, discussed above for their roles in inflammation, proliferation, and migration, are both known to be upregulated in a variety of hypoxic tumor cells (Greijer et al., 2008; Kim et al., 2009; Komatani et al., 2009; Liu et al., 2006; Marchesi et al., 2004; Nomura, Yoshida, and Teramoto, 2009; Pan et al., 2006; Schioppa et al., 2003; Zagzag et al., 2006). Based on these observations, even though the physiologic and physical microenvironments are often studied in isolation, it is clear that there are important connections between these systems that require further investigation.

In their often-cited review on ‘The Hallmarks of Cancer,’ Hanahan and Weinberg highlighted six key features of tumors: including self-sufficiency in growth signals, insensitivity to antigrowth signals, tissue invasion and metastasis, limitless replicative potential, sustained angiogenesis, and evading apoptosis (Hanahan and Weinberg, 2000). While there has been some research on the effects of hypoxia on these processes, our understanding is incomplete. Figure 2.1 summarizes what is currently known about the effects of hypoxia on each of these hallmarks of cancer.

Table 2.1 Effects of hypoxia on components of the microenvironment.

Figure 2.1 The effects of hypoxia on the six ‘hallmarks of cancer’ as described by Hanahan and Weinberg (2000). While some of these effects are clearly mediated by hypoxia-inducible factor-1α (HIF-1α), the mechanisms underlying many of these observed changes are unresolved. It is clear, however, that HIF-1α plays an integral role in the effects of hypoxia on all of these tumor characteristics. Furthermore, this figure highlights the areas of overlap between the physical and physiological microenvironments that require further investigation.

2.5.1 Introduction to angiogenesis

In 1908, Goldman first hypothesized that angiogenesis is intimately related to tumor growth, and Folkman subsequently popularized the notion that angiogenesis is necessary for tumor growth, raising the possibility of modifying this process for therapeutic gain (Goldmann, 1908; Folkman, 1971). We now understand that if tumors did not form new vessels, they could not grow beyond a very small size and would be unable to metastasize. In fact, if angiogenesis is inhibited at the beginning of tumor formation using a truncated form of the soluble VEGF receptor, tumor growth is inhibited (Li et al., 2000). Thus, it appears that angiogenesis is essential to the developing microenvironment.

In the mid-nineteenth century, Virchow was the first to recognize that tumor vessels have an abnormal architecture (David, 1988). Tumor vascular networks are collections of heterogeneous vessels (Decking, 2002; Pries, Secomb, and Gaehtgens, 1995a, 1995b; Pries and Secomb, 2009). There are long and short flow pathways, and the number of component vessels is highly varied. The topological heterogeneity is somewhat tempered by maturation of the neovasculature and pruning of unnecessary and non-functional vessels. Nonetheless, what remains is a disorganized network. This abnormal structure of individual vessels and larger networks leads to poor function, causing regions of inadequate oxygen and nutrient delivery.

Hypoxia refers to a below-normal oxygenation state in a tissue. This deficiency triggers subsequent angiogenesis, which is known to be abnormal in tumors. Thus, hypoxia and angiogenesis are tightly interrelated components of the microenvironment. As such, we will consider them together and discuss the interactions between these processes.

2.5.2 The importance of hypoxia-inducible factor-1

The oxygen tension in normal tissues generally exceeds 20 mmHg (Braun et al., 2001). When the pO2 dips below 10 mmHg, the hypoxic tissue begins to produce specific proteins mediated by hypoxia-inducible factor-1 (HIF-1), which has been referred to as the master regulator of oxygen homeostasis’ (Semenza, 2003). In addition to oxygen tension, free radicals, especially oxygen containing radicals such as superoxide (O2−), modulate HIF-1 levels (Dewhirst, Cao, and Moeller, 2008). HIF-1 is a heterodimeric transcription factor that increases the expression of genes involved in angiogenesis, adaptation to hypoxia, invasiveness, and resistance to oxidative stress (Semenza and Wang, 1992; Wang and Semenza, 1993; Semenza, 2003).

Accumulation of HIF-1 in a tumor mediates and integrates four major aspects of tumor biology: mutations and metabolism in addition to hypoxia and angiogenesis (Semenza, 2008). The most notable and well studied of the effects of HIF-1 are the transcriptional activation of genes involved in angiogenesis, most notably VEGF. VEGF is particularly important in the context of this chapter because it is a key regulator of vascular adaptation and angiogenesis. The HIF-1 complex binds to the hypoxia-responsive element upstream of the VEGF gene, directly activating VEGF transcription (Forsythe et al., 1996). Other HIF-1 activated genes involved in angiogenesis include angiopoietin 1 and 2, placental growth factor, and PDGF B (Pugh and Ratcliffe, 2003).

2.5.3 Causes of hypoxia

At its core, hypoxia results from an imbalance between oxygen delivery and consumption. Our laboratory recently described the eight key features of hypoxia relevant to the developing and evolving tumor microenvironment (Dewhirst, Cao, and Moeller, 2008). These features are all depicted in Figure 2.2, an image of regions of differential hemoglobin saturation in a mouse mammary tumor. First, the tumor receives an inadequate amount of oxygenated blood due to a limited arteriolar supply, which causes small tumor vessels far from the originating arterioles to contain very low levels of oxygen (Dewhirst et al., 1992, 1999; Sorg et al., 2008). Second, tumor vessels are oriented in such a way that some regions are well perfused (or even overly perfused) and others do not have enough vascularity (Secomb et al., 1993). Also related to vessel distribution and orientation, the third feature is that the center of a tumor tends to have fewer vessels than the tumor periphery.

Some tumor microvessels contain many red blood cells, whereas others contain few or none at all (Dewhirst et al., 1996). Thus, the fourth feature is that tumors show wide variations in red blood cell flux, defined as the number of red blood cells that traverse a particular microvessel per unit time. All of these cause an imbalance between oxygen supply and demand, which is the fifth aspect of tumor hypoxia (Secomb et al., 1995).

The remaining features of tumor hypoxia relate to blood flow patterns within the tumor. For example, hypoxic red blood cells shrink and stiffen, increasing blood viscosity (Kavanagh et al., 1993). The resulting decrease in flow velocity alters the distribution of erythrocytes at vascular bifurcations. In addition, vascular shunts have been observed that redirect arteriolar blood into draining veins, bypassing parts of the tumor (Eddy and Casarett, 1973). The last feature of tumor hypoxia is that it is temporally unstable. Changes in microvessel red blood cell flux cause intermittent periods of hypoxia, a phenomenon that has come to be known as cycling hypoxia because the variations exhibit periodicity in a variety of tumors (Dewhirst, Cao, and Moeller, 2008; Dewhirst, 2007).

2.5.3.1 Cycling hypoxia

Chaplin and Durand first reported that tumor oxygenation varies with the frequency of a few cycles per hour, and they later showed that this happens in large regions of the tumor over prolonged periods of time (Durand and Aquino-Parsons, 2001; Trotter et al., 1989; Bennewith and Durand, 2004; Bennewith, Raleigh, and Durand, 2002). Based on these studies, it is now understood that up to one-fifth of tumor cells may experience cycling hypoxia, and these cells are not typically adjacent to tumor microvessels. Our laboratory has used phosphorescence lifetime imaging in experimental tumors grown in dorsal skinfold window chambers to show that there are significant inter- and intratumoral differences in the spatial relationships between these fluctuations (Cardenas-Navia et al., 2008). Studies using blood oxygen level dependent functional MRI, an imaging technique that differentiates between oxygenated and deoxygenated hemoglobin, have suggested that immature vessels may be more likely to experience cycling hypoxia (Baudelet et al., 2004, 2006).

Figure 2.2 This mouse mammary tumor grown in a dorsal skin-fold window chamber was analyzed for regions of different hemoglobin saturations. This image shows many of the key features of tumor hypoxia. Areas of low hemoglobin saturation depict hypoxia and areas of high hemoglobin saturation represent normoxic regions. (A) Low arteriolar supply. This is one of the very few arterioles feeding this tumor. (B) Area of lower oxygenation because this area is poorly vascularized. (C) This region shows a steep longitudinal oxygen gradient, beginning with a moderately oxygenated arteriole leading to a hypoxic region, because the vessel does not contain enough oxygen to deliver to the tissue. This ends in a region of intravascular hypoxia. Furthermore, region C shows that the center of the tumor has fewer vessels than the periphery. (D). By comparing region C with region D, it is clear that there are mismatches in oxygen supply and demand. Some areas receive enough oxygen to remain normoxic, while others experience hypoxia because of inadequate oxygen supply. (E). Vascular shunt that redirects oxygenated arteriolar blood into the draining veins, bypassing parts of the tumor. The remaining features of tumor hypoxia, variations in flux and temporal instability, require serial imaging and therefore cannot be visualized in this figure. (This figure was reproduced with permission from Hardee, M.E., Dewhirst, M.W., Agarwal, N. and Sorg, B.S. (2009) Novel imaging provides new insights into mechanisms of oxygen transport in tumors. Curr Mol Med, 9, 435–441.) A full color version of this figure can be found in the color plate section.

Fourier transform analysis, a mathematical method used to identify frequencies in complex data sets, has shown dominant fluctuations in tumor oxygenation of less than two cycles per minute in a variety of preclinical tumor models (Sorg et al., 2008; Brurberg et al., 2004, 2005; Brurberg, Graff, and Rofstad, 2003; Cardenas-Navia et al., 2004). Measurements of flow using laser Doppler flowmetry and fiberoptic oxygen sensors have reported similar results in human tumors, but Fourier analysis has not been used in the clinical setting (Pigott et al., 1996; Brurberg et al., 2005). Other studies have demonstrated slower variations in oxygen tension within the tumor, on the magnitude of hours measured over a 24, 36, or 96-hour period (Brown, 1979; Bennewith and Durand, 2004; Skala et al., 2009; Skala et al., in press). There is no consensus on the kinetics of cycling hypoxia, and further investigation is necessary to understand these periodic fluctuations in oxygenation.

2.5.3.1.1 Mechanisms of cycling hypoxia

Based on experiments correlating microvessel red blood cell flux and interstitial pO2, it appears that cycling hypoxia results from changes in the flow of red blood cells through tumor microvessels (Kimura et al., 1996; Lanzen et al., 2006). As a natural extension of this observation, the question arises as to what causes these variations in erythrocyte flux. One possible source, though it likely does not explain all of the variations in red blood cell flow, is a change in the vasomotor activity of the tumor’s arteriolar supply (Dewhirst et al., 1996; Baudelet et al., 2006). A more likely explanation comes from understanding vascular remodeling. Changes in flow greatly affect the hemodynamics of a vascular network through processes including vascular pruning, formation of new vascular connections, and intussusception (Kiani et al., 1994; Patan, Munn, and Jain, 1996; Zakrzewicz, Secomb, and Pries, 2002). With respect to cycling hypoxia, intussusception is of particular interest. This process involves splitting of microvessels into smaller parallel vessels over a period of minutes (Patan, Munn, and Jain, 1996). Since flow resistance is inversely proportional to vessel radius to the fourth power, small changes in microvessel size lead to large redistributions in flow (Chien, Usami, and Skalak, 1984; Kiani et al., 1994; Pries et al., 1997). Thus, vascular remodeling resulting in changes in flow resistance likely account for the majority of erythrocyte flux causing cycling hypoxia.

2.5.3.1.2 Interactions between chronic and cycling hypoxia

The overall oxygenation state of a tumor region has profound effects on whether that region will experience cycling hypoxia, so chronic hypoxia and cycling hypoxia are inextricably linked phenomena. For example, poorly perfused areas that are far from microvessels are unlikely to be affected by changes in red cell flux, and the pO2 in this region is unlikely to change much because it is already quite low. Alternatively, regions that are perfused by an abundance of vessels are equally unlikely to experience large changes in oxygen tension due to changes in flow through only one of the feeding microvessels. Our laboratory has likened this situation to the effects of tides and waves on an island in the ocean, as depicted in Figure 2.3 (Dewhirst, 2009). When tides are high, the island may be completely awash and individual waves do not affect the amount of beach that is exposed. However, at low tide, when the water no longer covers the entire island, the height of individual waves determines how much of the beach is exposed at a given time. We liken the tides to overall oxygenation state, waves to episodes of cycling hypoxia, and the amount of beach exposed to the level of hypoxia at a particular time. If the overall oxygenation of a tumor region is high (high tide), the individual waves do not cause significant hypoxia. However, if the overall oxygenation is lower, individual fluctuations (waves) cause large changes in tissue oxygenation.

Figure 2.3 The height of tides and waves on an island can be used as an analogy to understand cycling hypoxia. Tides refer to the overall oxygenation state of a tumor, and waves depict temporal variations in red blood cell flux. The exposed island is a representation of the amount of hypoxia, which is also depicted by the background color, with red indicating well-oxygenated areas and blue indicating hypoxia. (a) A region of chronic hypoxia. In this region, the low overall oxygenation state cannot be overcome by the height of individual waves. In (b), individual waves determine whether the island is oxygenated at a particular time. Therefore, this panel represents a region of cycling hypoxia. (c) an area that is so well-oxygenated that individual waves do not cause hypoxia. Therefore, this panel depicts a tumor region that is normoxic.

While this analogy provides a framework for understanding these phenomena, it clearly does not explain the whole story. The temporal variations between different reports of cycling hypoxia make the two processes difficult to differentiate. For example, while variations on the order of seconds or minutes seem clearly to be cycling hypoxia, reports of variations on the order of hours to days blur the distinction (Skala et al., 2009; Skala et al., in press).

It is well known that chronic hypoxia has important physiological consequences for tumors that lead to treatment resistance, but cycling hypoxia appears to have effects that are distinct from those of chronic hypoxia. Richard Hill’s group has shown that metastatic frequency is dependent on the degree of tumor hypoxia and that cycling hypoxia increases metastatic frequency over that of chronic hypoxia (Cairns and Hill, 2004; Cairns, Kalliomaki, and Hill, 2001; De Jaeger, Kavanagh, and Hill, 2001; Rofstad et al., 2007). Tumor-bearing mice subjected to cycling hypoxia also showed increased expression of genes associated with metastasis, including CXCR4, uPAR, VEGF, and osteopontin (Chaudary and Hill, 2009).

The effects of cycling hypoxia are not limited to metastasis. Cycling hypoxia appears to influence HIF-1α protein levels and transcriptional activity more than chronic hypoxia (Martinive et al., 2006; Ning et al., 2007; Peng et al., 2006). Cycling hypoxia may increase ROS due to repeated hypoxia-reoxygenation cycles, and a recent study involving a transgenic model of breast cancer showed evidence of significant oxidative damage to DNA and lipids caused by cycling hypoxia (Kalliomaki et al., 2008, 2009). Furthermore, it is clear that the microenvironment greatly affects mammalian target of rapamycin (mTOR) activity, and cycling hypoxia seems to have opposite effects from chronic hypoxia on mTOR function and interaction with HIF-1 (Dunlop and Tee, 2009; Hudson et al., 2002; Yuan et al., 2008). This is significant because mTOR is an integral part of the mTOR complex that is involved in modifying response to changes in nutritional and energy status and oxidative stress. Hypoxia and oxidative stress both induce the unfolded protein response (UPR), which alters protein expression, metabolism, and cell death in response to stress (Wouters and Koritzinsky, 2008). It seems likely that cycling hypoxia will affect the UPR, since genes controlled by HIF-1 are often contained in the stress granules formed by the UPR and cycling hypoxia increases oxidative stress (Moeller et al., 2004). However, further investigation into these changes is needed to better understand the pathophysiological responses to cycling hypoxia.

In accordance with the clear effects on the environment of the tumor, hypoxia, and cycling hypoxia have important implications for tumor treatment (Brown and Wilson, 2004). In the setting of radiotherapy, oxygen helps to stabilize treatmentinduced DNA damage such that the DNA cannot be effectively repaired. Therefore, chronic hypoxia decreases the cytotoxic effects of radiation. Furthermore, the delivery and activity of chemotherapeutic agents is often decreased under hypoxic conditions. In addition to the chemo- and radioresistance conferred on tumor cells by chronic hypoxia, cycling hypoxia is known to cause resistance to radiation therapy (Brown, 1979; Yamaura and Matsuzawa, 1979). Therefore, these microenvironmental effects are critical to understand in order to design better targeted therapeutic strategies.

2.5.4 Interactions between hypoxia and angiogenesis

The term ‘‘angiogenic switch’’ refers to the balance of pro- and antiangiogenic factors, leading to the initiation of angiogenesis. Low tissue oxygenation is generally considered to be the predominant weight that tips the scales in favor of angiogenesis, since hypoxia enhances HIF-1 protein levels and activity, directly upregulating VEGF (Diaz-Gonzalez et al., 2005; Laderoute et al., 2000; Mazure et al., 1996). Other microenvironmental components, such as oncogenes and growth factors, can act via the PI3K-AKT pathway to increase expression of HIF-1 to the point that it overcomes oxygen-mediated degradation (Feldkamp et al., 1999; Jiang and Liu, 2008).

For primary tumors to grow beyond a few millimeters in diameter, angiogenesis is generally required. In early malignant breast tumors, HIF-1 expression correlates with VEGF levels and angiogenesis (Bos et al., 2001). This is consistent with a model of HIF-1 causing VEGF production and angiogenesis, but it does not answer the question of whether hypoxia-mediated stabilization of HIF-1 is the principal underlying cause of neovascularization. Using a rat glioma model, Holash et al. showed that vessel cooption occurs in the first week after tumor transplantation, followed by evidence of vessel regression in week 2 and angiogenesis by week 4 (Holash et al., 1999). In this experiment, increased expression of angiopoietin 2 was hypothesized to cause vessel regression, leading to the ‘hypoxic crisis’ that triggered the onset of angiogenesis.

In our laboratory, using fluorescent labels for blood vessels and hypoxic regions in a dorsal skin fold window chamber tumor model, we observed that angiogenesis precedes hypoxia (Cao et al., 2005). There was no evidence of vascular stasis before angiogenesis, though angiogenesis was enhanced by HIF-1α upregulation. This led us to put forth the ‘acceleration model,’ suggesting that HIF-1α is not necessary for angiogenic initiation but instead accelerates neovascularization. This would mimic the role of angiogenesis in wound healing (Haroon et al., 2000). However, more evidence is needed to better understand the temporal and functional relationships between vessel cooption and angiogenesis.

It is not only the onset of angiogenesis that is of interest. Ongoing angiogenesis affects delivery of oxygen, nutrients, and therapeutic agents to the tumor. Using a fluorescent reporter for HIF-1 expression, it is clear that as tumors grow, some well-oxygenated microvessels are found in regions with high expression of HIF-1 (Sorg et al., 2005). It is possible that ROS and/or NO may stabilize HIF-1 in these regions, leading to VEGF expression in the absence of hypoxia and causing vascular remodeling and angiogenesis.

2.5.5 Communication in the vasculature

The development and remodeling of vessels is a complex process that requires coordination not only between the tumor and the vessels but also among and within the vessels themselves. A variety of microenvironmental signaling molecules, including NO, modulate vascular tone, and sustained changes in vascular tone affect the structure of tumor vessels (Cosby et al., 2003; Bakker et al., 2002; Martinez-Lemus, Hill, and Meininger, 2009). We are beginning to understand the mechanisms by which these vessels communicate to enact these structural modifications.

Communication is essential not only between vessels, but also within an individual vessel. Gap junctions are plasma membrane channels that connect cells, and they play a key role in intercellular communication in vessel walls. These channels enable small molecules (less than 100 Da), including ions, amino acids, and second messengers, to pass between adjacent cells. Transmembrane proteins known as connexins (Cxs) are essential components of these junctions. Radially aligned connexins form hexameric complexes known as connexons. Channels are made of two connexons, one from each cell. Connexin proteins are composed of four hydrophobic transmembrane domains, two extra-cellular domains involved in connecting to an adjacent connexon, and three cytoplasmic domains (Willecke et al., 2002).

Four connexin isoforms have been identified in the vascular wall: Cx37, Cx40, Cx43, and Cx45 (Brisset, Isakson, and Kwak, 2009). These isoforms form connections between endothelial cells or between smooth muscle cells, and to a lesser extent, allow communication between endothelial and smooth muscle cells. While knockout experiments indicate that both Cx40 and Cx43 are involved in vascular tone, Cx43 is the isoform most likely involved in the myoendothelial junction (Isakson et al., 2006). Therefore, most research on vascular connexins has focussed on Cx43.

The permeability of a gap junction is determined by the connexin isoform it is composed of, and channel opening and closing is controlled by environmental signals (Lampe and Lau, 2004). The short half-life of Cx43 in cultured cells and tissues, ranging from 1 to 3 hours, suggests tight post-translational regulation. Serine phosphorylation is the predominant modification to Cx43, and these phosphorylation events appear to correlate with changes in the formation and degradation of gap junctions (Musil et al., 1990; Cooper et al., 2000; Lampe and Lau, 2004). Additionally, phosphorylation of tyrosine residues on Cx43 is correlated with significant decreases in intercellular communication (Filson et al., 1990). Phosphorylation of Y247 by pp60v-src leads to closure of the gap junction (Lin et al., 2001). It seems likely that phosphorylation of Y265 stabilizes the interaction between Cx43 and v-Src, thereby enabling the phosphorylation of Y247 that leads to closure of the channel (Lampe and Lau, 2004). Therefore, phosphorylation of Cx43 could greatly affect communication between endothelial cells.

A variety of microenvironmental factors cause differential phosphorylation of Cx43. Binding of the growth factors EGF and PDGF to their receptors leads to serine phosphorylation (Lampe and Lau, 2004). Treatment of endothelial cells with VEGF is associated with Cx43 phosphorylation and reduction in endothelial cell–cell communication as mediated by both the c-Src tyrosine kinase and MAP kinase pathways (Suarez and Ballmer-Hofer, 2001).

2.5.5.1.1 The role of connexin-43 in hypoxia and angiogenesis

While it is known that there is a considerable amount of heterogeneity in tumor microvascular networks, the mechanisms underlying the structural and functional abnormalities are poorly understood. Recent computer modeling of tumor vessels revealed that reduced communication along vessels is a leading cause of aberrant tumor vessel structure and function (Pries et al., 2009). Similar models revealed that information must be transferred along the length of vessels in order for functional networks to form, and impaired communication leads to shunting of blood flow, one of the eight causes of tumor hypoxia addressed above (Pries, Reglin, and Secomb, 2001; Pries, Secomb, and Gaehtgens, 1998). Furthermore, red blood cell flux is disturbed by a lack of cell–cell communication, which could explain the changes in erythrocyte flow that result in cycling hypoxia.

While normal vessels coordinate blood flow through endothelial cell and smooth muscle cell gap junctions, little is known about the expression or function of connexins in tumor vasculature (Segal, 1994). Tumor vessels lack the regular organization of endothelial cells, basement membrane, and smooth muscle (Carmeliet and Jain, 2000). They are often composed of just endothelial cells or even of tumor cells themselves (Di Tomaso et al., 2005). Since tumor vessel composition is abnormal, it follows that communication between these cells is likely disrupted. Figure 2.4 is a schematic diagram of an abnormal gap junction in tumor vasculature.

Figure 2.4 Gap junction connecting two endothelial cells in a tumor vessel. The junction is composed of a hexameric complex of connexin 43 (Cx43), which forms a channel to allow the passage of small signaling molecules between adjacent cells. The panel on the left shows a welloxygenated region in which the Cx43 is not phosphorylated. This leads to a functional junction and enables intercellular communication. The blue panel on the right depicts a hypoxic region containing reactive oxygen species (ROS) and vascular endothelial growth factor (VEGF). These microenvironmental factors lead to phosphorylation of Cx43, disrupting the gap junction. As a result, the passage of materials between adjacent tumor endothelial cells is ineffective.

Of the vascular connexins, Cx43 has been implicated in dysfunctional tumor vasculature, because its expression and function are altered by the microenvironment (Errede et al., 2002). Some groups have reported that decreases in Cx43 expression in breast tumor models are associated with a more aggressive tumor phenotype (Shao et al., 2005; McLachlan et al., 2006; Pollmann et