TAMs frequently infiltrate multiple types of solid tumors in humans and mice (Solinas et al. 2009). Development and growth of mammary tumors are greatly reduced in macrophage colony-stimulating factor (M-CSF)-deficient op/op mice, which are deficient in macrophages due to the lack of M-CSF (Lin et al. 2001). Bisphosphonate treatment, which selectively depletes phagocytes including macrophages in vivo, decreases the number of TAMs and induces retardation of tumor growth (Rogers and Holen 2011). These studies suggest that TAMs are essential cells for tumor growth and progression. Both tumor cells and stromal cells in tumor tissue induce the recruitment of circulating monocytes in the peripheral blood by secreting a variety of chemoattractants such as chemokine CC ligand (CCL) 2, CCL5, colony-stimulating factor (CSF)-1 and chemokine CXC ligand (CXCL) 12. Monocytes differentiate into TAMs in response to the tumor microenvironment (Solinas et al. 2009). A recent report demonstrates that the spleen and bone marrow are reservoirs for TAM precursors in tumor-bearing hosts (Cortez-Retamozo et al. 2012). TAMs are also differentiated from tumor-infiltrated monocytic MDSCs (M-MDSCs or Mo-MDSCs) as described below.

Two distinct activation states of macrophages are referred to as classically/alternatively activated or M1/M2-polarized states (Mantovani et al. 2002; Biswas and Mantovani 2010). Classically activated M1 macrophages produce pro-inflammatory cytokines such as tumor necrosis factor (TNF)-α, IL-12, IL-23, IL-6, and IL-1β, leading to induction of T helper (Th) 1-type immune response and high expression of inducible nitric oxide (NO) synthase (iNOS). In vitro experiments have shown that TNF-α and NO are involved in direct killing of certain types of tumor cells by M1 macrophages. In contrast, alternatively activated M2 macrophages are characterized by higher expression levels of IL-10, arginase-1, scavenger receptor (SR), and macrophage mannose receptor (MMR, CD206), and lower expression of pro-inflammatory cytokines such as IL-12, IL-23, IL-1β, and IL-6. M2 macrophages have poor antigen-presentation capability and immunosuppressive activity by secreting IL-10 and transforming growth factor (TGF)-β. They also produce molecules such as angiogenic factors and matrix metalloproteinases (MMPs), which are involved in tissue remodeling. M1/M2-polarization of macrophages is regulated by immune signaling (Hu et al. 2007; Lawrence and Natoli 2011). IFN-γ and IFN-β are potent stimulation factors of macrophages and induce M1-like macrophages through Janus kinase/signal transducer and activator of transcription (STAT) 1 activation (Toshchakov et al. 2002). M1 macrophages derived from granulocyte–M-CSF (GM-CSF)-treated human monocytes highly express IRF5 compared with M-CSF-induced M2 macrophages. Over-expression of IRF5 in M-CSF-induced M2 macrophages forces them into macrophages that express M1-specific cytokines, leading to both Th1 and Th17 cell development (Krausgruber et al. 2011). The Notch-recombination signal binding protein for immunoglobulin kappa J (RBPJ) pathway determines M1/M2 polarization by controlling the expression of M1-related genes in macrophages via transcription factor IRF8 (Xu et al. 2012). These results raise the possibility that TLR signaling may directly induce the expression of M1-related genes through IRF5 and IRF8 because these transcription factors participate in TLR signal-induced transcription of cytokine genes (Honda and Taniguchi 2006). In contrast, the jumonji domain-containing 3 (JmjD3)–IRF4 axis regulates M2 macrophage development (Satoh et al. 2010). JmjD3, a H3K27me demethylase, is induced in macrophages in response to the ligands of TLR2, TLR4, and TLR9, and inflammatory cytokines (De Santa et al. 2007). STAT6 and peroxisome proliferator-activated receptor (PPAR)-γ regulate M2 macrophage polarization (Charo 2007; Ishii et al. 2009). Suppressor of cytokine signaling (SOCS) 2 and SOCS3 are involved in M1 and M2 polarization, respectively, by regulating intracellular cytokine signaling (Spence et al. 2013). Furthermore, TLR signals induce chromatin remodeling to control the gene expression through IRF3 and other transcription factors (Foster et al. 2007), which may regulate the expression levels of the transcription factors involved in macrophage polarization. Most importantly, macrophage polarization is not stable (i.e., plasticity), and the manipulation of macrophage function could be achieved by regulating multiple intracellular signaling pathways (Biswas and Mantovani 2010; Lawrence and Natoli 2011; Sica and Mantovani 2012).

Gene expression analysis has demonstrated that TAMs largely display the phenotype that is typical of the M2-polarized macrophages (Mantovani et al. 2008) (Fig. 3.1). However, TAMs polarized into the M1 or M2 phenotype coexist in tumors but localize in different areas of the tumor. It is reported that infiltration of M2-polarized TAMs is correlated with poor prognosis in many types of cancers, including melanoma, colon cancer, and ovarian cancer (Lewis and Pollard 2006; Biswas and Mantovani 2010). Conversely, the density of M1 macrophages, defined as CD68⁺HLA-DR⁺, in the tumor is positively associated with the survival time of non-small cell lung cancer patients, whereas CD68⁺CD163⁺ M2 macrophages are not associated with patient survival (Ma et al. 2010). Thus, the balance of M1 versus M2 population is considered to affect tumor growth.

Macrophage polarization is affected by several factors in the tumor microenvironment. It has been demonstrated that tumor cell-derived TNF-α, TGF-β, prostaglandin E2 (PGE2), and hypoxia induce M2 polarization of macrophages (Lewis and Pollard 2006; Biswas and Mantovani 2010). Th2 cytokines such as IL-4 drive the development of M2-polarized TAMs (DeNardo et al. 2009). The recently identified transcription factors that drive macrophage polarization described above may explain a mechanism that regulates the development and polarization of TAMs.

Immune and non-immune mechanisms mediated by TAMs that affect tumor growth have been reported (Mantovani et al. 2008). M2 TAMs directly facilitate tumor growth by secreting a variety of growth factors for tumor cells. M2 TAM-secreted immunosuppressive cytokines such as IL-10 down-regulate the anti-tumor activity of cytotoxic T cells and NK cells. IL-10 may also promote tumor progression by potentiating Treg activity. Furthermore, M2 TAMs produce proangiogenic factors such as vascular endothelial growth factor (VEGF), chemokines such as CCL2, and CXCL8, and growth factors such as platelet-derived growth factor and epidermal growth factor, which stimulate formation of new blood vessels to supply nutrients that are essential for tumor growth (Lin and Pollard 2007). Furthermore, TAMs influence the efficacy of anticancer therapies including chemotherapy, tumor irradiation, vascular-targeted therapies, and antibody therapies (De Palma and Lewis 2013).

Increasing evidence shows that TAMs play critical roles in multiple stages of tumor growth and metastasis in which TLR signaling pathways are involved (Qian and Pollard 2010; Sica 2010). Macrophages are recruited into pre-metastatic organs by tumor-derived factors where TLR4-activated macrophages facilitate invasion and metastasis of tumor cells by secreting proteolytic enzymes such as MMP7 and MMP9 to destroy the extracellular matrix (Hiratsuka et al. 2006, 2008). Versican, an extracellular matrix protein, secreted by tumor cells induces TNF-α production by macrophages through activation of the TLR2 signaling pathway. This promotes tumor metastasis into lungs (Kim et al. 2009). Another study has suggested that activation of TLR4 signaling pathway on M2-polarized TAMs partially induces epithelial–mesenchymal transition (EMT) in pancreatic cancer through increased IL-10 production (Liu et al. 2013).

MDSCs are a heterogeneous cell population that consists of myeloid progenitor cells and immature myeloid cells (iMCs). MDSCs are defined as CD11b⁺Gr1⁺ cells in mice and CD14⁻CD11b⁺CD33⁺HLA-DR^neg/low in humans, and lack maturation markers of macrophages and DCs. MDSCs have potent immunosuppressive activity against both innate and adaptive immunity (Gabrilovich et al. 2012). Although CD11b⁺Gr1⁺ cells are normally present in healthy mice, they do not have immunosuppressive activity. Myeloid progenitor cells immediately differentiate into neutrophils, macrophages, or DCs in healthy mice. However, in tumor-bearing mice, the differentiation is blocked, which cause MDSC accumulation in spleen, blood, lymph nodes, and primary and metastasized solid tumors. They are also frequently detected in the peripheral blood of human cancer patients (Almand et al. 2001; Diaz-Montero et al. 2009; Ostrand-Rosenberg and Sinha 2009). In tumor tissues, MDSCs can be distinguished from TAMs by the expression of surface molecules. Gr1 is highly expressed on MDSCs but not TAMs, and F4/80 is expressed on TAMs but less on MDSCs. MDSCs are further characterized into two populations: Ly6G⁻Ly6C^high M-MDSCs and Ly6G⁺Ly6C^low granulocytic MDSCs (G-MDSCs) or polymorphonuclear MDSCs (PMN-MDSCs) (Youn et al. 2008; Movahedi et al. 2008; Peranzoni et al. 2010). Although G-MDSCs are a major subset in tumors and peripheral blood, their suppressive activity is relatively low compared to that of M-MDSCs. Recently, it was revealed that M-MDSCs contain precursors of TAMs and G-MDSC (Youn et al. 2013). MDSCs are distributed throughout the body. Tumor-derived chemokines or chemoattractants such as CCL2, Bv8 (also known as prokineticin-2), and S100 calcium-binding protein (S100) A8/A9 recruit MDSCs from peripheral organs into tumor sites (Huang et al. 2007; Shojaei et al. 2007; Sawanobori et al. 2008; Sinha et al. 2008).

Inflammation-associated factors promote the expansion of MDSCs. Prostaglandins, stem-cell factor (SCF), M-CSF, GM-CSF, TNF-α, IL-1β, IL-6, PGE2, and VEGF promote MDSC expansion (Gabrilovich et al. 2012). Some of these molecules are under the control of TLR signals. In fact, TLR4 activation leads to suppressive activity of MDSCs through a MyD88/NF-κB-dependent mechanism. STAT3, which is activated by stimulation with some of these inflammatory molecules, is one of the key signaling molecules that regulates the expansion of MDSCs. STAT3 is frequently observed to be activated in tumor-infiltrating immune cells (Yu et al. 2007). MDSC expansion is not observed in STAT3 conditional knockout mice or STAT3-specific inhibitor-treated mice under tumor-bearing conditions, which results in an increase of T cell responses (Nefedova et al. 2005; Kortylewski et al. 2005). Constitutive activation of STAT3 leads to the production of S100A8 and S100A9 in iMCs, resulting in inhibition of their differentiation. Increased reactive oxygen species (ROS) concentration driven by S100A8/A9-induced nicotinamide adenosine dinucleotide phosphate (NADPH) activation results in differentiation of iMCs into MDSCs (Cheng et al. 2008). Hsp72-containing tumor-derived exosomes trigger STAT3 activation through TLR2/MyD88-dependent IL-6 production by autocrine mechanisms, which results in the induction of MDSC expansion (Chalmin et al. 2010). Another report shows that STAT3 or STAT5 down-regulates IRF8 to maintain MDSC development (Waight et al. 2013). S100A8 and S100A9 expression is up-regulated in many tumors, including gastric, lung, bladder, mammary, and colon cancer (Srikrishna 2011). Activated neutrophils and macrophages in tumor or necrotic tumor cells release the S100A8/A9 complex, which act as a chemoattractant for MDSCs. The S100A8/A9 complex promotes and amplifies inflammatory responses via direct binding to TLR4 (Ehrchen et al. 2009). Inflammation-induced TNF signaling drives the peripheral accumulation of MDSCs through TNF receptor (TNFR)-2, but not TNFR-1. TNF-α inhibits differentiation and enhances suppressive activity of iMCs during chronic inflammation, resulting in generation of MDSCs. TNF-α-induced S100A8 and S100A9 proteins and their corresponding receptor, receptor for advanced glycan endproducts (RAGE), augment MDSC-suppressive activity (Sade-Feldman et al. 2013). Activation of complement cascades accompanied by TLR-induced inflammation also regulates tumor growth by modulating MDSC function. Complement component C5a, a cleaved product of C5, is generated by inflammation in the tumor microenvironment and recruits MDSCs and enhances their suppressive function against CD8⁺ T cell proliferation, which contributes to tumor growth. Enhanced suppression is achieved by increased ROS and reactive nitrogen species (RNS) in M-MDSCs but not G-MDSCs, which results in reduced T cell responses against tumor cells (Markiewski et al. 2008). Another report shows that the signaling balance of paired immunoglobulin-like receptor (PIR) family members PIR-A and PIR-B are expressed on MDSCs and are important for the regulation of MDSC differentiation. MDSCs isolated from PIR-B-deficient Lilrb3^−/− mice preferentially differentiate into an M1-like rather than M2-like immunosuppressive phenotype in Lewis lung cancer (LLC; also known as 3LL) tumor-bearing mice. LPS and IFN-γ stimulation enhances M1 polarization by suppressing STAT3 activation in the absence of PIR-B (Ma et al. 2011). Expansion of MDSCs are also regulated by other transcription factors (Condamine and Gabrilovich 2011; Sonda et al. 2011). Collectively, TLR signals are involved in MDSC expansion directly and indirectly.

MDSCs suppress anti-tumor T cell responses by several mechanisms (Fig. 3.2). The immunosuppressive activities of MDSCs are divided into four categories. First, inhibition of T cell proliferation is mediated by depleting nutrients in the microenvironment. MDSCs highly express arginase-1, which rapidly decreases the concentration of l-arginine in the microenvironment. Reduced level of l-arginine concentration causes a profound inhibition of T cell proliferation by the inability to up-regulate cyclin D3 and cyclin-dependent kinase 4 upon antigen stimulation, cytokine production, and expression of the CD3ζ chain of the T cell receptor (TCR) (Zea et al. 2005). MDSCs also inhibit T cell proliferation by sequestering cystine and limiting the availability of cysteine (Srivastava et al. 2010). Second, ROS and RNS produced by MDSCs modulate immune responses. MDSC-derived peroxynitrate inhibits T cell responses by inducing nitration of tyrosine residues in TCRs, resulting in an altered TCR/major histocompatibility complex (MHC) peptide recognition (Nagaraj et al. 2007). In parallel, peroxynitrite induces tumor cell resistance to CTLs by modifying MHC class I–antigen complex (Lu et al. 2011). Among two subsets, G-MDSCs suppress antigen-specific CD8⁺ T cells predominantly by producing ROS (Youn et al. 2008). Third, MDSCs modulate lymphocyte trafficking. CD62L (l-selectin) expression on naive CD4⁺ and CD8⁺ T cells are decreased by a disintegrin and metalloprotease 17 (ADAM17) on MDSCs, leading to the inhibition of recruitment to lymph nodes (Hanson et al. 2009). Fourth, MDSCs indirectly affect T cell activation by inducing immunomodulatory cells such as Foxp3⁺ Tregs and M2 TAMs. MDSC-derived TGF-β and IL-10 are required for Treg induction (Huang et al. 2006; Serafini et al. 2008). In addition to the differentiation of MDSCs into M2 macrophages in tumors, MDSC-derived IL-10 and cell–cell interaction promote M2 polarization of macrophages, as well as impair cytokine production and antigen presentation by DCs.

How MDSCs are implicated in the regulation of NK cell function is controversial. MDSCs inhibit NK cell cytotoxicity against tumor cells and IFN-γ production through direct cell–cell interaction. Membrane-bound TGF-β1 expressed on MDSCs inhibits NK cell cytotoxicity and IFN-γ production, and induces anergy of NK cells in a liver transplant model (Li et al. 2009). MDSCs also inhibit NK cell activation by blocking the expression of NK group 2D (NKG2D). However, another report suggests that the F4/80⁺ population of MDSCs express retinoic acid early inducible 1 (RAE-1), the ligand for NKG2D, activate NK cell cytotoxicity (Nausch et al. 2008).

These immunoregulatory functions of MDSCs are regulated by TLR and TLR-induced cytokine production. IFN-γ stimulation induces suppression of antigen-specific T cell responses by M-MDSCs, which requires STAT1 activation (Movahedi et al. 2008). STAT3 signaling is potentially activated by pro-inflammatory cytokines such as IL-6 and is implicated in arginase-1 expression in CD14⁺HLADR^-/low MDSCs from head and neck cancer patients (Vasquez-Dunddel et al. 2013). Hypoxia-inducible factor (HIF)-1α activation enhances immunosuppressive function and differentiation of MDSCs in the tumor microenvironment (Corzo et al. 2010). LPS-induced TLR4 signal activates HIF-1α (Frede et al. 2006), suggesting that TLR4-triggered MDSC accumulation and development may be induced via HIF-1α-mediated transcriptional regulation. Activation of TLR as well as IL-1 receptor and receptor tyrosine kinases activate Gr1⁺CD11b⁺ MDSC-like cells to promote tumor inflammation and progression through phosphoinositide 3-kinase γ (Schmid et al. 2011).

Stromal cells in the tumor microenvironment are considered to be promising targets for cancer treatment (Quail and Joyce 2013). Recent reports suggest that regulation of immunosuppressive activity of tumor-associated myeloid cells could be useful for improving the efficacy of cancer immunotherapy. Therefore, strategies for elimination of tumor-associated myeloid cells or modulation of their function in tumor-bearing hosts are currently being investigated (Ugel et al. 2009; Talmadge and Gabrilovich 2013). Molecules that are responsible for the accumulation and immunosuppressive activity of MDSCs and TAMs could become therapeutic targets. There are several classes of inhibitors or reagents that can control the population or modulate the function of tumor-associated myeloid cells as described in other review (Ugel et al. 2009).

TAMs and MDSCs are proposed to be a target of cancer immunotherapy because they frequently accumulate in solid tumors and have critical roles in tumor growth and progression. Furthermore, they display a high degree of plasticity in their phenotype. Recent results have highlighted that TLR signaling pathways have important roles for switching between immunosuppressive phenotype and anti-tumor phenotype of TAMs and MDSCs.

Adjuvant immunotherapy using purified TLR ligands or TLR agonists seems to be a promising treatment for cancer by inducing DC-mediated anti-tumor responses. Recent reports suggest that TAMs and MDSCs play important roles in adjuvant therapy. However, it is true that TLR signals such as TLR2 and TLR4 contribute to promote tumor growth by inducing immunosuppressive activity of TAMs and MDSCs. Timing of administration and selection of TLR ligands may determine the outcome of adjuvant immunotherapy for cancer because the tumor microenvironment continuously changes during the course of cancer progression where the population and the function of TAMs and MDSCs are varied. Recent studies show that poly I:C is capable of inducing not only DC-mediated anticancer immune responses that lead to the activation of NK cells and CTLs, but also anti-tumor activity of TAMs and MDSCs. Therefore, poly I:C-induced TLR3/TICAM-1 and MDA5/MAVS pathways are promising targets of cancer treatment. Further basic studies to clarify the mechanisms of anti-tumor and pro-tumor effects induced by innate immune signaling on tumor-associated myeloid cells are still required to establish adjuvant therapy for cancer.