Chronic inflammation, the tumor microenvironment and carcinogenesis

[Cell Cycle 8:13, 2005-2013; 1 July 2009]; ©2009 Landes Bioscience

Review

Chronic inflammation, the tumor microenvironment and carcinogenesis Tamas A. Gonda, Shuiping Tu and Timothy C. Wang* Division of Digestive and Liver Diseases; Department of Medicine; Irving Cancer Research Center; Columbia University; New York, NY USA

Key words: chronic inflammation, tumor microenvironment, bone marrow derived cells, gastric cancer, myofibroblasts, myeloid derived suppressor cells

Chronic inflammation often precedes or accompanies a substantial number of cancers. Indeed, anti-inflammatory therapies have shown efficacy in cancer prevention and treatment. The exact mechanisms that turn a wound healing process into a cancer precursor are topics of intense research. A pathogenic link has been identified between inflammatory mediators, inflammation related gene polymorphisms and carcinogenesis. Animal models of cancer have been instrumental in demonstrating the diversity of mechanisms through which every tumor compartment and tumor stage may be affected by the underlying inflammatory process. In this review, we focus on the interaction between chronic inflammation, tumor stem cells and the tumor microenvironment. We summarize the proposed mechanisms that lead to the recruitment of bone marrow derived cells and explore the genetic and epigenetic alterations that may occur in inflammation associated cancers.

Introduction Beginning with some of the earliest observations in cancer biology, a connection between chronic inflammation and carcinogenesis has long been suspected.1 A substantial number of epidemiologic, gene association and molecular studies have now confirmed this important association. With better characterization of infiltrating immune cells, the precise role of inflammation in cancer has begun to be elucidated, leading to a resolution of the initial contradiction that inflammation is protective in certain tumors2-4 yet detrimental in others.5 Overall, it appears that chronic inflammation more often stimulates then inhibits tumor development.6 The notion has emerged that the majority of tumors derive a benefit and indeed are dependent on immune cells. The persistence of chronic inflammation plays a critical role

*Correspondence to: Timothy C. Wang; 1130 St. Nicholas Avenue; Room 925; New York, NY 10032 USA; Tel.: 212.851.4581; Fax: 212.851.4590; Email: [email protected] Submitted: 05/04/09; Accepted: 05/11/09 Previously published online as a Cell Cycle E-publication: http://www.landesbioscience.com/journals/cc/article/8985

www.landesbioscience.com

in initiating, sustaining and advancing tumor growth,7 and thus modulating the immune response may still be an alluring goal for therapeutic intervention.8 The focus of this review is on the interaction of the cellular components of the tumor microenvironment and the role of inflammatory mediators in carcinogenesis. We also emphasize the increasingly recognized role of inflammation in the recruitment of bone marrow derived stem cells to the tumor microenvironment.

Background on Inflammation and Cancer Although a pathogenic role for chronic inflammation has been suggested in multiple tumor systems in tumor initiation, progression and metastatic potential9,10 the mechanism of this important association is still not understood completely.11,12 In a subset of cancers, an association with a recognized infection has been found (Hepatitis B and C, Helicobacter pylori, Clonorchis sinensis, Schistosoma mansoni), whereas in others the etiology of inflammation is non-infectious or idiopathic (PID, IBD, BE, chronic pancreatitis).13 In some cases, particularly with viral infections, a direct effect of the pathogen on neoplastic transformation of epithelial cells has been shown, but overwhelming evidence point to a more generalized role of the resultant inflammatory process as the primary mediator of the carcinogenic process. Not surprisingly, both eradication of infection (i.e., H. pylori in gastric MALT lymphoma and gastric cancer) and anti-inflammatory therapies (i.e., COX-2 inhibitors) have been successful in preventing progression and even curing some of these cancers. However, it is also evident that not all inflammation is tumorigenic. For example, the adaptive immune response in some tumors plays a potentially effective role in immune surveillance.14 One of the fundamental, unanswered questions in the field is the role that specific immune and inflammatory cells play in neoplastic transformation of the epithelium. Broadly, it can be suggested that either the tumor alters the immune response so that it becomes a pro-tumorigenic inflammation (whereby, for example, reactive oxygen species or cytokines further propagate tumor growth) or a sustained chronic inflammation plays an active and primary role in transforming tissue stem cells into tumor cells. Recent evidence suggests that both models may be true. In addition, chronic inflammation and

Cell Cycle

2005

Chronic inflammation, tumor microenvironment and carcinogenesis

tumor derived signals appear to play role in recruiting bone marrow derived cells (BMDC) where they can give rise to stroma and perhaps tumor cells that can contribute to tumor progression (Fig. 1). Tumor initiation and immune phase. The development of a tumor is associated with the growth and expansion of not only tumor cells but also stroma, vessels and infiltrating inflammatory cells, and it is the interaction between these different cell types that propagates tumor growth. Early on, it was recognized that acute inflammation can often contribute to suppression or eradication of incipient cancer.15 However, acute inflammatory responses are by definition self limiting, and in many cases evolve Figure 1. The central role of BMDCs in chronic inflammation related carcinogenesis. In the current model both a direct and indirect (via recruitment of BMDCs) interaction between mediators of into chronic inflammatory responses which chronic inflammation and the tumor cells or tumor stroma is emphasized. Several of the mediators have been associated with tumor initia- and their role in carcinogenesis are described in the text. tion. Despite some similarities to responses against pathogens, anti-tumor responses Tumor progression, metastatic potential and inflammation. in this setting are generally weak and not sufficient to eliminate tumor cells.6 It may be the case that the initial signal is not Several lines of evidence suggest that inflammation also plays recognized as an equivalent danger signal derived from exog- an important role in the ability of tumor cells to invade and enous pathogens16 or that the “host” further downregulates the metastasize. One of the most striking observations is the ability perceived immune response against self6 or that the microen- of epithelial tumor cells which metastasize to express specific vironment prevents access of the immune system to the tumor chemokine receptors.23 This process is supported by paracrine cells.17 NFκB and STAT3 pathways have emerged as key regula- secretion of pro-inflammatory cytokines (i.e., IL1-β, IL-6, TNFα) tors of the release of pro-inflammatory cytokines and important and also certain autocrine cytokine production.24 In one study mediators of both tumor proliferation and persistence of chronic knock-down of TNFα resulted in decreased cellular CXCR4 inflammation.11,18,19 The activation of these pathways creates a and CXCL12 expression, and decreased ability of the tumor vicious cycles, whereby the pro-tumorigenic inflammation results to spread to the peritoneal cavity.25 Tumor progression further depends on tumor infiltrating leukocytes, in particular, tumorin further cytokine release. Tumor escape and progression in chronic inflammation. associated macrophages or TAMs. TAMs have been subdivided Malignancy is often associated with a suppression of an effective into pro-inflammatory M1 and anti-inflammatory M2, and their T cell response and an enhanced humoral immune response. The distribution in the tumor may determine the stage and ability of undifferentiated nature of the tumor, along with the ability of tumor progression. In a breast cancer model, depletion of TAMs the tumor cells to evade immune cells and directly suppress host was associated with normal carcinogenesis but reduced metas26,27 immunity, likely contribute to the inadequacy of an anti-tumor tasis. response. Despite the presence of autologous effector T cells, these Soluble and Other Mediators of Immune Response are suppressed by an expanding population of regulatory T cells that accumulate in the tumor tissue. These regulatory T cells may in Cancer be activated or recruited by dendritic cells that accumulate in the An interaction and active interplay between the tumor cells tumor.20 The other population of cells that have been suggested to (epithelium) and the tumor microenvironment (TME) has emerged play an important role in suppression of TILs (tumor infiltrating as a necessary component of carcinogenesis. We summarize some lymphocytes) are myeloid-derived suppressor cells or MDSCs (or of the best known mediators and how they relate to inflammation their murine analogue, the CD11b+/Gr-1+ myeloid progenitor induced carcinogenesis and the recruitment of progenitor cells. cells) that are detected in increased number in the periphery Role of oxidative stress. A dominant source of endogenous reacand spleen of tumor bearing animals and in the tumor.21,22 tive oxygen and nitrogen species (O2-, NO, H2O2, OH, ONOO-, While these MDSCs were initially recognized for their immune HOCL) are innate immune cells (macrophages, leukocytes). In the suppressive ability, recent studies suggest that these cells are pro- acute phase of inflammation, the “respiratory burst” that results in inflammatory and are also able to promote angiogenesis.22 Thus, the release of these compounds plays an important role in the elimtumor-associated inflammation appears to be associated with both ination of pathogens.7 A persistence of inflammation (i.e., chronic pro-inflammatory and anti-inflammatory signals that allow tumors inflammation) results in prolonged exposure of cells to ROS/RNS that can lead to genomic alterations (DNA strand breaks and base to grow and escape immune surveillance. 2006

Cell Cycle

2009; Vol. 8 Issue 13


modifications), lipid and protein peroxidation and activation of signal transduction by post translational modification (i.e., RNS is a mediator of MAPK and AP-1 signal transduction; NO can induce post translation modification of p53 via ATM or ATR) or specific induction of anti-apoptosis genes (AIF).28-30 These mechanisms link ROS/RNS production, and hence the innate immune system, to tumor proliferation, invasion, angiogenesis and metastasis. Additionally, oxidative stress has also been associated with impaired DNA methyltransferase activity (DNMT-1) or MBP binding and hence provides a link to epigenetic alterations.31 A provocative recent study has demonstrated that ROS-induced stress can result in recruitment of histone deacetylase-1 and DNMT1 and result in promoter hypermethylation and gene silencing.32 Role of specific cytokines. As discussed earlier, both genetic studies and animal experiments have confirmed that cytokines play a key role in linking chronic inflammation and cancer. They are activated by both inflammatory and tumor cells and their effect is similarly important for both sustaining chronic inflammation, promoting progression of tumor cells and microenvironment proliferation and inhibiting immune mediated tumor surveillance. In general, cytokines can be divided into those that are pro-inflammatory (i.e., IL-1, 6, 8, 11, 12, 18, 23, IFNγ, TNFα and MIF) and anti-inflammatory (i.e., IL-4, 10, IFNα and β, TGFβ); however, it has become clear that many of these molecules can have dual roles. TNFα and IL-1β are essential in the initiation of chronic inflammation, and their role via the activation of NFκB pathways has provided a strong link to cancer. Recent work from our laboratory has shown that IL-1β overexpression alone, in the absence of Helicobacter infection, is sufficient to cause gastric cancer.33 TGFβ is another example of a pleiotropic cytokine that may play a role of tumor suppression as well as tumor progression via stimulation of the cancer associated inflammation and remodeling of the microenvironment. Complete deletion of TGFβ has been associated with increased tumor incidence; however most tumors progress and metastasize in the presence of high levels of TGFβ. TGFβ has also been associated with the recruitment of MDSCs into the TME, decreased CD8+ T cell and NK cell accumulation, and thus an inflammatory milieu that is more permissive of tumor growth. Along with IL-6, TGFβ plays a role in T cell differentiation towards a pro-inflammatory phenotype. IL-6 alone has been associated with promotion of both chronic inflammation and carcinogenesis in the colon and stomach via, primarily, the excessive activation of STAT 1 and 3. Recently, IL-6 has been conclusively linked to inflammation mediated tumor initiation and proliferation in colon cancer.34 Two groups have demonstrated that IL-6 deficient mice were not able to develop colitis-associated carcinoma in a DSS/AOM mouse model and that the IL-6 effect was mediated by STAT-3 in intestinal epithelial cells.35,36 Importantly, the source of IL-6 in this model was shown to be bone marrow derived myeloid cells. The effect of IL-6 in tumors is not limited to intestinal epithelial cells (IECs), as it may also lead to activation of dendritic cells and T helper cells. This may contribute to the persistence of a cytokine production and further propagation of both the chronic inflammatory state and a pro-tumorigenic cytokine milieu.34 www.landesbioscience.com

Taken together these models suggest that cytokines (IL-1β) play a crucial role in the recruitment of bone marrow derived myeloid cells, which serve as a source for IL-6 production, and IL-6 induction of STAT3 in epithelial cells can lead to tumor initiation. Inhibitors of these cytokines and their signal transducers (STAT-3/JAK) are starting to emerge as promising targets of cancer therapy. Matrix remodeling proteases. Matrix metalloproteinases (MMPs) are especially important in establishing cross-talk and co-regulation between the various cellular components of the tumor microenvironment. Overall, MMP expression is often associated with a worse prognosis, and most MMPs are expressed by tumor-associated stromal and immune cells (although there are exceptions, such as MMP7, which is predominantly expressed by epithelial cells). Work in a variety of systems has revealed that extracellular proteases act not just on the remodeling of ECM components, but also through modulation of a range of cytokines, chemokines, and growth factors. MMP9 has emerged as especially important in the regulation of epithelial proliferation and through the processing pro-growth factors (i.e., VEGF) in angiogenesis. MMP 9 expression has been associated with a more invasive and aggressive phenotype in breast and skin tumors.37 Recent evidence also suggests that MMP9 may be necessary for the recruitment of bone marrow derived progenitor cells, such as endothelial progenitor cells (EPCs), and intra-tumoral expression of MMP9 by MDSCs is sufficient to restore angiogenesis in irradiated mice.38 MMP7 is another member of this class that has been shown play a role in metastasis, through its ability to process RANKL, which plays a role in osteolysis. In gastric cancer, MMP7 has been found to be directly induced in epithelial cells by Helicobacter infection, and its expression correlated with both epithelial and stromal proliferation.39 Despite correlations between MMP9 or MMP2 expression or the ratio of MMP9 and its inhibitor TIMP2, therapies targeted at these (and other) MMPs have not yet shown the expected efficacy. This lack of expected efficacy is maybe due to either their parallel function in normal tissue homeostasis and cancer progression or to their varied expression between tumor types and tumor stages.

Interaction of Bone Marrow Derived Cells and the Tumor Microenvironment Cytokines and other tumor derived soluble factors not only recruit to tumors well-defined inflammatory cells but also, as increasingly recognized, bone marrow-derived progenitor cell populations. Bone marrow derived cells that are recruited to the site of chronic inflammation have demonstrated remarkable plasticity and have been shown to differentiate into cells of diverse lineages. In vitro, this differentiation is governed by secreted factors. Similarly, in vivo their differentiation seems to be governed by the tissue (tumor) microenvironment. Indeed, this stem cell-like property has led to speculation that rather than tissue stem cells, BMDCs may represent an alternative source of tumor initiating cells. The concept of cancer stem cells (CSC) arose from the observation that only a fraction of tumor cells possess the capacity for self-renewal, and has been supported by observations that these are the cells

Cell Cycle

2007


that can initiate tumors in immunodeficient mice.40 While the cell surface markers that define cancer stem cells have been identified for a number of tumor types, the precise cellular origin of most cancer stem cells has remained elusive. In animal models of cancer, and in models of hepatitis, colitis, pancreatitis and gastric inflammation that have included transplantation of tagged bone marrow, we and others have demonstrated that BMDCs cells can differentiate into stromal myofibroblasts, epithelial cells, and contribute to angiogenesis. Importantly, the degree of engraftment of BMDC to tissue seems to be highly related to Figure 2. Increase in ASMA positive cells in gastric cancer. (A) Normal gastric epithelium. the degree of chronic inflammation. The degree Majority of ASMA positive cells are found in the muscularis mucosae and in the lamina propria. of transdifferentiation remains controversial, (B) Intestinal type gastric cancer. Proliferation of ASMA positive cells (CAFs) is seen. (C) Diffuse type gastric cancer. Marked proliferation of CAFs is observed. and how it occurs is an area of very active research, with some recent evidence suggesting the possibility of cell fusion.41 It may be proposed that chronic which presumably removes the source of tumor derived factors, inflammation not only alters the tumor microenvironment by completely normalizes MDSCs abundance and reestablishes a soluble mediators but also through recruitment of cells that differ- normal myelopoiesis in tumor-bearing hosts. While the soluble factors that play a role in recruiting MDSCs entiate into the tumor and its microenvironment. Bone marrow derived cells comprise a heterogeneous popula- have not been completely defined, it is likely that IL1-β is one tion that includes hematopoetic stem cells and mesenchymal stem of these factors. Studies have shown that xenograft tumors with cells (MSC). The latter is defined as the population of stem cells IL-1β overexpression show greater accumulation of MDSCs and that adhere to plastic when BM is placed in culture. MSCs can more rapid tumor progression,48 while 4T1 mammary carcinoma differentiate into adipocytes, chondrocytes, osteocytes and marrow tumors implanted into IL-1R–deficient mice exhibit delayed mesenchyma. Myeloid cells are derived from hematopoietic accumulation of MDSCs and slower growing tumors.49 Once stem cells, and the types that are seen in human tumors include mobilized and recruited to the tumor, MDSCs can promote tumor macrophages, TIE2 expressing monocytes, hemangiocytes and progression.50 MDSCs from mammary carcinomas can promote DCs, as well as neutrophils, eosinohils, mast cells and myeloid tumor invasion and metastasis though a process involving metalderived suppressor cells (MDSC). While myeloid cells are clearly loproteinase activity and aberrant TGFβ signaling. A recent shift in the understanding of this recruitment a diverse group of cells, it is important to consider them as one group as they may be able to differentiate or shift phenotype in process was provided by our group: it now appears that at least response to tumor derived signals. MDSCs are of particular recent one population of myeloid cells (MDSCs) can be recruited by interest as they circulate in the periphery at increased numbers in IL-1β alone in very young H/K-ATPase-IL-1β transgenic mice, tumor bearing animals, and are recruited by specific chemokine/ prior to the development of gastric dysplasia. This may be an cytokine signals to tumors. These tumor derived factors comprise a important observation as it provides evidence that perhaps the prevariety of biologically active compounds, including growth factors, neoplastic inflammation alone can play a role in the recruitment cytokines and chemokines and can profoundly affect myelopoiesis, of MDSC. Interestingly, while these cells have been thought of in the past as primarily immunosuppressive, our work has shown mobilization and activation of myeloid cells.42 MDSC in mice are characterized by the coexpression of Gr-1 and that accumulation of MDSCs is also associated with production CD11b and have been shown to inhibit T-cell activation in various pro-inflammatory factors such as IL-6, TNFα and SDF1. These tumor models, including inflammation-induced carcinomas.8-11 In pro-inflammatory factors in turn further mobilize and recruit naïve wild-type mice, a small number of CD11b+Gr-1+ cells (