Introduction

It has been well documented that human tumors are generally infiltrated by inflammatory cells (Whiteside, 1993; Mihm et al., 1996; Balkwill and Mantovani, 2001). Although these infiltrates of inflammatory cells can vary in size and composition from tumor to tumor, their presence has been taken as evidence that the host is not ignorant of the developing tumor, but rather attempts to interfere with tumor progression, a process referred to as immune surveillance (Zitvogel et al., 2006). In this context, inflammatory infiltrates in tumors are considered to be a host attempt at the detection of emerging tumor cells and their elimination (Zitvogel et al., 2006). Indeed, numerous reports in the literature have linked the presence of inflammatory infiltrates in human tumors with an improved prognosis or better patient survival (Kornstein et al., 1983; Baxevanis et al., 1994; Naito et al., 1998; Pages et al., 2005). More recent data based on analyses of multiple immune markers, including heat maps and microarrays, suggest that type, density and location of immune cells in the tumor have a prognostic value (Pages et al., 2005; Galon et al., 2006; Galon et al., 2007). At the same time, equally numerous reports have indicated a lack of significant correlations between lymphocytic infiltrate intensity and improved prognosis or have linked immune cell infiltration to a poor prognosis (Stewart and Tsai, 1993; Sheu et al., 1999; Nakano et al., 2001). These contradictory results emerging from reputable laboratories have remained unexplained for many years, until it became possible to explore functional properties of tumor-infiltrating lymphocytes (TIL), which often represented the major component of immune infiltrates in tumors (Whiteside, 1993; Mihm et al., 1996). In nearly all cases, TIL obtained from human tumor tissues showed inhibited proliferation in response to mitogens or antigens, compromised signaling through T cell receptor, decreased ability to mediate cytotoxicity of tumor targets or to produce Th1-type cytokines upon stimulation with tumor antigens (Kiessling et al., 1996; Reichert et al., 1998b, 2002; Kuss et al., 1999; Uzzo et al., 1999). Functional impairments in TIL were more pronounced in patients with advanced cancer than in early disease, and they seemed to differ in the frequency as well as magnitude, depending on the tumor type or its tissue of origin (reviewed in Whiteside, 1993). Importantly, the functional status of TIL was now shown to be an independent and significant correlate of improved prognosis and longer overall survival in patients with malignancy (Reichert et al., 1998a). In time, as the methods used to study functions and attributes of tissue-infiltrating immune cells improved, and the understanding of their local interactions with other cells enlarged, the role of a microenvironment in shaping cellular events in health and disease came to be appreciated. A recent comprehensive multivariate analysis of cellular interactions in the tumor microenvironment based on the nature, function, density and localization of immune cells within human colorectal cancers demonstrated that immune reactions within the tumor influence clinical outcome (Pages et al., 2005; Galon et al., 2006, 2007). The current view of the tumor microenvironment is that it exerts a key influence on tumor progression and that re-shaping of its character might offer unexpected therapeutic benefits.

Cells found in the tumor microenvironment

A tissue microenvironment of developing tumor is comprised of proliferating tumor cells, the tumor stroma, blood vessels, infiltrating inflammatory cells and a variety of associated tissue cells. It is a unique environment that emerges in the course of tumor progression as a result of its interactions with the host. It is created by and at all times shaped and dominated by the tumor, which orchestrates molecular and cellular events taking place in surrounding tissues (Figure 1).

Figure 1
figure 1

A diagram depicting the tumor microenvironment. Interactions of various cells with each other, fibroblasts (F) in the tumor stroma and blood vessels (V) are indicated by arrows. Although the tumor (TU) generates signals inducing dysfunction and death of immune cells, the latter are a source of signals promoting tumor growth. TAM, tumor-associated macrophages; DC, dendritic cells; EL, effector lymphocytes; Treg, regulatory T cells; TA, tumor-derived antigens; ROS, reactive oxygen species; PGE2, prostaglandin E2.

Full size image

Immune cells present in the tumor include those mediating adaptive immunity, T lymphocytes, dendritic cells (DC) and occasional B cells, as well as effectors of innate immunity, macrophages, polymorphonuclear leukocytes and rare natural killer (NK) cells (Whiteside, 2007). NK cells, which mediate innate immunity and are rich in perforin- or granzyme-containing granules, are conspicuously absent from most tumor infiltrates or even pre-cancerous lesions (Whiteside et al., 1998). Although NK cells represent ‘the first line’ of defense against pathogens (Lanier, 2003) and mediate potent antitumor cytotoxicity in vitro, in tumor milieu, they are infrequent, despite the fact that tumor cells frequently downregulate expression of HLA antigens and are enriched in MICA and MICB molecules (Chang et al., 2005). These features make the tumor susceptible to NK cell-mediated cytotoxicity (Lee et al., 2004), and their paucity in tumor infiltrates may be an example of the evasion mechanism preventing NK-cell recruitment to the tumor site.

Tumor-infiltrating lymphocytes, containing various proportions of CD3+CD4+ and CD3+CD8+ T cells, are usually a major component of the tumor microenvironment (Whiteside, 2007). Many of these T cells are specific for tumor-associated antigens, as indicated by clonal analyses (Miescher et al., 1987) and tetramer staining of CD8+ T cells isolated from human tumors (Albers et al., 2002). In some tumors, for example, medullary breast carcinomas, infiltrating lymphocytes form lymph node-like structures suggesting that the immune response is operating in situ (Coronella et al., 2002). Also, TIL are a source of tumor-specific lymphocytes used for adoptive transfers after expansion in IL-2-containing cultures (Zhou et al., 2004). TIL clones with the specificity to a broad variety of the tumor-associated antigens can be outgrown from human tumors, confirming that immune responses directed not only at ‘unique’ antigens expressed by the tumor, but also at a range of differentiation or tissue-specific antigens, are generated by the host (Romero et al., 2006). Although accumulations of these effector T cells in the tumor might be considered as evidence of immune surveillance by the host, they are largely ineffective in arresting tumor growth. Among CD4+ T cells present in the tumor, a subset of CD4+CD25high Foxp3+ cells is expanded (5–15% of CD3+CD4+ T cells in TIL) relative to their significantly lower frequency in the peripheral circulation of patients with cancer (Woo et al., 2001; Strauss et al., 2007a). These cells are regulatory T cells (Treg) capable of suppressing proliferation of other T cells in the microenvironment through contact-dependent mechanisms or IL-10 and TGF-β secretion (Figure 2). They come in different flavors (for example, nTreg, Tr1) and are a characteristic feature of the microenvironment in human tumors (Bergmann et al., 2007; Strauss et al., 2007a).

Figure 2
figure 2

Accumulation and expansion of Treg in the tumor microenvironment may be a result of the cross talk between the tumor and DC. The tumor coopts DC differentiation, and in the presence of tumor-derived factors, immature DC develop abnormalities in APM, have decreased expression levels of MHC molecules, upregulate B7-H1 (PD-L1) and produce excess of IL-10 and TGF-β1. Cross-presentation of TA by these DC to T cells (as in the lower part of figure) lead to emergence of Treg and their expansion. Once generated, Treg interfere with functions of anti-tumor effector cells (CTL). Even if cross-presentation of TA by DC to naive CD8+ T cells is successful (as in the upper part of figure), Treg proceed to block CTL functions. Reproduced with changes from Ferrone and Whiteside, 2007.

Full size image

Macrophages present in tumors are known as tumor-associated macrophages or TAMs. They are re-programmed to inhibit lymphocyte functions through release of inhibitory cytokines such as IL-10, prostaglandins or reactive oxygen species (ROS) (Mantovani et al., 2003; Martinez et al., 2008). Myeloid suppressor cells (MSC) accumulating in human tumors are CD34+CD33+CD13+CD15(−) bone marrow-derived immature dendritic cells, an equivalent to CD11b+/Gr1+ cells in mice (Serafini et al., 2006). They promote tumor growth and suppress immune cell functions through copious production of an enzyme involved in L-arginine metabolism, arginase 1, which synergizes with iNOS to increase superoxide and NO production, blunting lymphocyte responses (Ochoa et al., 2007) and by induction of iNOS in surrounding cells (Tsai et al., 2007). Relatively little is known about human MSC. A recent report describes expansion of CD14+HLA-DR−/low myeloid-derived cells exerting immune suppression through TGF-β production in the peripheral circulation of patients with metastatic melanoma treated with GM-CSF-based vaccines (Filipazzi et al., 2007). The recruitment of MSC to the tumor site is orchestrated by the tumor (see Figure 3). Tumors produce many factors, including IL-10, VEGF, GM-CSF, which promote MSC accumulation and block DC maturation as well as lymphocyte functions (Serafini et al., 2006). Current data support the active role of MSC in tumor-induced immune suppression in mice and in man.

Figure 3
figure 3

Tumors recruit MSC from the bone marrow by means of tumor-derived soluble factors. Immature myeloid cells migrate to lymph nodes, where DC cross-prime T cells, and interfere with this process. They also migrate to the tumor site and become tumor-associated MSC, which are adept in blocking T cell functions through the production of arginase I and activation of iNOs.

Full size image

Polymorphonuclear leukocytes are infrequently seen in infiltrates of human tumors, with the exception of nests of eosinophils that may be present in association with tumor cells in various squamous cell tumors, for example. In contrast, granulocytes tend to be a major cellular component of many murine tumor models (Loukinova et al., 2000). This disparity may be because of a different nature of infiltrates, which in man are chronic rather than acute. Acute cellular responses may be long gone by the time human tumors are diagnosed, biopsied and examined.

Inflammatory cells present in the tumor microenvironment either contribute to tumor progression or actively interfere with its development. It is clear today that the former takes precedence, largely because the tumor generally proceeds to establish mechanisms responsible for its ‘immune evasion’ or escape from the immune intervention. The tumor not only manages to escape from the host immune system, but it effectively contrives to benefit from infiltrating cells by modifying their functions to create the microenvironment favorable to tumor progression. To this end, immune cells infiltrating the tumor together with fibroblasts and extracellular matrix forming a scaffold supporting its expansion, contribute to establish an inflammatory milieu that nourishes the tumor and promotes its growth. Tumor escape from the host is facilitated by the ability of human tumors to actively subvert anti-tumor immunity by downregulating or completely suppressing local and systemic innate as well as adaptive antitumor immunity by a variety of mechanisms as discussed below.

Inflammation and cancer

Almost 20 years ago, H.F. Dworak coined the phrase describing human tumors as ‘wounds that do not heal’ (Dworak, 1986). Indeed, changes occurring in the microenvironment of the progressing tumor resemble the process of chronic inflammation, which begins with ischemia followed by interstitial and cellular edema, appearance of immune cells and, finally, growth of blood vessels and tissue repair (Aller et al., 2004). Chronic inflammation is clearly involved in shaping the tumor microenvironment and has been referred to as ‘host reaction’ to the tumor, although it might be more appropriate to think of it as ‘tumor promoting’ reaction.

An initial goal of the inflammatory response is to destroy an invader, which in this case is the tumor. Therefore, the ‘immune phase’ of tumor-driven inflammation involves a recruitment and influx of antitumor effector cells to the tissue site. However, compared with vigorous cellular and humoral responses that are generated in tissues upon infections by exogenous pathogens, those mediated by the tumor are weak. This is probably because most tumor-associated antigens are considered ‘self,’ in contrast to infections with bacteria or viruses which are viewed by the host as ‘danger signals’ (Gallucci and Matzinger, 2001). Indeed, it is entirely possible that accumulations of regulatory T cells (Treg) in the tumor microenvironment represent an attempt of the host to downregulate response against ‘self’ with an unfortunate concomitant suppression of antitumor immunity.

Among the factors that determine the nature of inflammatory infiltrates found in the tumor microenvironment is the hypoxic environment. It is created early in the tumor development through activation of hypoxia-responsive genes in tumor cells (Denko et al., 2003). It favors the influx of those inflammatory cells that depend on the glycolytic pathway for survival, namely, phagocytic macrophages and granulocytes (Aller et al., 2004). These cells not only survive in the hypoxic environment but contribute to it by hyperproduction of ROS upon local activation. In the tumor milieu, where apoptosis of rapidly expanding tumor cells is common, infiltrating phagocytes receive ample activation signals and produce an abundance of ROS. Immunoinhibitory activities of ROS are mediated by the NF-κB pathway, which in turn is regulated by hypoxia and/or re-oxygenation (Lluis et al., 2007). It has been proposed that the NF-κB pathway plays a key role in activation of signaling in cancer cells as well as tumor-infiltrating leukocytes (Balkwill and Coussens, 2004; Greten et al., 2004; Pikarsky et al., 2004). NF-κB activation in these cells lead to secretion of TNF-α or other pro-inflammatory cytokines which initiate and drive regulated expression of the cytokine genes responsible for cell proliferation. Tumor cells depend on these cytokines for growth, and infiltrating leukocytes become programmed to continually release these growth factors. Responding to this NF-κB-driven pro-inflammatory cytokine cascade, tumor and stromal cells produce a variety of soluble mediators with wide-ranging biologic effects. Thus, cell proliferation and differentiation, matrix remodeling, blood vessel growth and cell migration/recruitment are all re-programmed to benefit the tumor. The role of TNF-α in driving tumor progression has long been emphasized by Balkwill and coworkers (Malik et al., 1989). It provides an example of how tumors usurp a normal process of inflammation to promote their own progression. It also suggests that blocking of TNF-α, for example, by anti-TNF-α antibodies, might be therapeutically useful (Harrison et al., 2007). Similarly, inhibition of NF-κB activation in the tumor microenvironment represents a potentially effective strategy for arresting tumor growth (Karin and Greten, 2004).

Mechanisms of tumor escape

The tumor microenvironment, once established, represents a consistently effective barrier to immune cell functions. This is because tumors are not passive targets for host immunity; instead, they actively downregulate all phases of anti-tumor immune responses using a spectrum of different strategies and mechanisms (Figure 4). To date, many mechanisms responsible for dysfunction of immune cells in the tumor microenvironment have been identified. Some are directly mediated by factors produced by tumors, whereas others result from alterations of normal tissue homeostasis occurring in the presence of cancer. Until recently, little was known about molecular alterations in tumor cells in situ as they progressed from the pre-malignant to metastatic phenotype. Genetic instability, now recognized as a principal characteristic of all tumors, may result in changes in their epitope profile. Molecular changes, already detectable during early stages of tumorigenesis, become more pronounced as the tumor progresses. The net result of these changes is increased resistance of tumor cells to immune surveillance. In addition, most human tumors appear to be able to interfere with one or more stages of immune cell development, differentiation, migration, cytotoxicity and other effector functions. Thus, all phases of an antitumor immune response are subject to adverse intervention in the tumor microenvironment as indicated in Table 1. These escape mechanisms and their consequences in terms of tumor progression have been recently reviewed (Whiteside, 2006).

Figure 4
figure 4

Mechanisms responsible for ‘immunoediting’ of tumor cells in the tumor microenvironment. The various mechanisms listed collaborate in immunoediting of the tumor cells. The tumor, stromal cells and infiltrating leukocytes all contribute to pro-inflammatory milieu. Infiltrating immune cells and stromal elements are re-programmed by the tumor to the pro-inflammatory mode favoring its survival. In this milieu, the tumor evolves a phenotype allowing it to escape and to counterattack immune cells. Tumor stem cells are resistant to anti-tumor therapies and thus represent yet another means of escape.

Full size image
Table 1 Mechanisms orchestrated by the tumor that contribute to its escape from the host immune systema
Full size table

Of the various escape mechanisms listed in Table 1, two have received special attention in recent years, possibly because they appear to be ubiquitous and are clearly associated with disease progression. Accumulations in tumors of Treg (CD4+CD25bright Foxp3+ T cells) and myeloid-derived cells (CD34+CD33+CD13+CD11b+CD15−) are common features of human tumors, and the frequency as well as suppressor activity subsets of these cells mediate locally and systemically have been linked to poor prognosis in patients with cancer (Almand et al., 2000).

Under normal physiologic conditions, Treg have a beneficial role in preventing autoimmunity (Shevach, 2000). However, in cancer, they expand, migrate to tumor sites, downregulate autologous effector T-cell proliferation and suppress anti-tumor responses of both CD4+CD25− and CD8+CD25− T cells using distinct molecular pathways (Roncarolo et al., 2006; Bacchetta et al., 2007). They are a heterogenous population of regulatory CD3+CD4+ T cells, comprising natural Treg, antigen-specific Tr1 cells and other less well defined subsets of suppressor cells (Roncarolo et al., 2006). Tr1 cells are induced in the tumor microenvironment, which is rich in IL-10, TGF-β, and prostaglandin E2 (PGE2), all of which have been shown to promote Tr1 generation (Bergmann et al., 2007). Today, the nature of human Treg is only partially defined. The phenotype, functions (antigenic specificity, stability, trafficking or survival), lineage, differentiation and the relationship between the various Treg subsets are under intense investigation. No single specific marker is sufficient for distinguishing Treg subpopulations. Given the expansion of these populations in the circulation and tumor tissues of cancer patients (Woo et al., 2001; Liyanage et al., 2002; Curiel et al., 2004; Shevach, 2004; Strauss et al., 2007b), it is important to perform Treg phenotypic and functional evaluations to be able to define their role in the regulation of tumor-specific responses. From a practical point of view, it is important to distinguish Treg from activated CD4+CD25+ T cells which mediate helper functions and are sensitive to activation induced cell death (AICD). In contrast, Treg appear to be resistant to apoptosis (Strauss et al., 2007c). Recent reports suggest that oncologic therapies, surgery, radiation, chemotherapy, expand Treg and enhance their suppressor functions (Banerjee et al., 2006; Zhou et al., 2006; Strauss et al., 2007b). These intriguing data support the need for serial follow-up studies of Treg in cancer patients treated with oncologic therapies.

MSC also suppress T-cell responses in the tumor microenvironment. Most human tumors secrete TGF-β or induce TGF-β secretion from MSC that accumulate in the tumor microenvironment (Gallina et al., 2006; Serafini et al., 2006). Pak et al. first reported accumulations of CD34+ cell-derived myeloid cells with immunosuppressive ability the peripheral blood of HNC patients (Pak et al., 1995). These cells correspond to CD11b+/Gr-1+ myeloid progenitor cells in mice (Serafini et al., 2006). In tumor-bearing mice, MSC accumulate in the spleen and peripheral circulation, reaching very high proportions and exerting potent immunosuppression, thus favoring tumor growth. MSC also control the availability of essential amino acids such as L-arginine and produce high levels of ROS. MSC present in tumors constitutively express iNOS and arginase 1, an enzyme involved in metabolism of L-arginine, which also synergizes with iNOS to increase superoxide and NO production, blunting lymphocyte responses (Bronte et al., 2003; Ochoa et al., 2007; Tsai et al., 2007). Another enzyme that might be produced by MSC is indoleamine-2,3-dioxygenase (IDO) involved in the catabolism of tryptophan, an essential amino acid for T-cell proliferation and differentiation (Munn and Mellor, 2007). The frequency of MSC with high levels of suppressive functions were found to be increased in the peripheral blood of patients with various cancers (Almand et al., 2001). Further, maturation defects in DC of patients with cancer have been described (Almand et al., 2000) and are attributable, in part, to vascular endothelial growth factor (VEGF) production by human tumors (Gabrilovich et al., 1999; Fricke and Gabrilovich, 2006). GM-CSF, which is also a frequently secreted product of tumor cells, recruits MSC and induces dose-dependent in vivo immune suppression and tumor promotion (Serafini et al., 2006). At the same time, GM-CSF is widely used as immune adjuvant in antitumor vaccines (Dranoff et al., 1993). In fact, GM-CSF was observed to significantly enlarge a subset of TGF-β-producing MSC phenotypically defined as CD14+HLA-DR−/low in the circulation of patients with metastatic melanoma (Filipazzi et al., 2007). This dual role of GM-CSF (stimulatory and suppressive) suggests that GM-CSF and MSC are involved in maintaining immune homeostasis under normal physiologic conditions but in the tumor presence are subverted to promote its escape.

Targeting of the tumor microenvironment for therapy

The failure of immune surveillance in tumor-bearing hosts has been one of the major incentives for the development of cancer immunotherapy, including anti-tumor vaccines, adoptive transfers of T cells and exogenous cytokine delivery. Numerous animal tumor models have provided strong evidence that in the presence of effective anti-tumor immunity, tumors fail to progress and established tumors regress (Ostrand-Rosenberg, 2004). Hence, recovery of immune surveillance and protection of immune cells from tumor-induced suppression are well-rationalized objectives of current anti-tumor therapies. As the molecular mechanisms responsible for tumor escape or involved in tumor-induced immune suppression are defined, therapeutic options for blocking tumor-induced suppression are becoming more realistic. With an improved understanding of mechanisms underlying tumor-induced immune suppression, future therapeutic strategies will likely focus on combined approaches designed to restore antitumor immune responses, eliminate tumor escape and correct tumor-induced immune deviation to enable the host immune system to more effectively control tumor growth. Table 2 lists some of the therapeutic approaches aimed at the modification of the tumor microenvironment and targeting Treg. On the basis of the currently available data, it seems reasonable to hypothesize that in situ interactions of the tumor with the host tissues, including infiltrating leukocytes, are a critical factor for tumor promotion. Therefore, disrupting or otherwise altering these interactions in favor of the host might result in therapeutic benefits. For example, the elimination of Treg before delivering antitumor vaccines might be considered to enhance their anti-tumor potential in patients with advanced malignancies. Table 2 lists some of the potential therapeutic strategies targeting Treg that are already in the clinic or are being experimented with animal models of cancer. Similar strategies for elimination or blocking of MSC activities might also be considered.

Table 2 Examples of potential strategies aimed at the recovery of immune competence through downregulation of Treg numbers or function in tumor-bearing hostsa
Full size table

Conclusions

Current evidence suggests that chronic inflammation is associated with tumor development and progression. The NF-κB pathway may form a link between inflammation and cancer. Its activation in cells present in the tumor results in sustained production of pro-inflammatory cytokines, which promote tumor survival. The tumor co-opts functions of leukocytes in the microenvironment to support its growth using a variety of molecular mechanisms, which are beginning to be elucidated. At the same time, the tumor manages to hide from the immune attack, and either mounts a ‘counterattack’ or develops resistance to immune cells. The nature and intensity of inflammatory infiltrates may vary as the tumor progresses, depending on the local milieu that is created and shaped by the tumor. Consequently, mechanisms evolved by tumors for disarming host defenses and escape from the immune control vary in different cancers, and the unique signature of each tumor is reflected by its microenvironment. Therefore, understanding of cellular and molecular interactions operative in the tumor microenvironment is of crucial importance. Changing of chronic to acute inflammation at the tumor site might be therapeutically beneficial. Molecular tools are now available for devising novel and more effective anticancer therapies targeting not only the tumor but also its microenvironment.