Hide-and-seek: the interplay between cancer stem cells and the immune system

Carcinogenesis, Feb 2017

Sultan, Mohammad, Coyle, Krysta Mila, Vidovic, Dejan, Thomas, Margaret Lois, Gujar, Shashi, Marcato, Paola

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:


Hide-and-seek: the interplay between cancer stem cells and the immune system

Carcinogenesis Hide-and-seek: the interplay between cancer stem cells and the immune system Mohammad Sultan 1 Krysta Mila Coyle 1 Dejan Vidovic 1 Margaret Lois Thomas 1 Shashi Gujar 0 1 Paola Marcato 0 1 0 Department of Microbiology and Immunology, Dalhousie University , 5850 College Street, Halifax, Nova Scotia B3H 4R2 , Canada 1 Department of Pathology The enhanced ability of cancer stem cells (CSCs) to give rise to new tumors suggests that these cells may also have an advantage in evading immune detection and elimination. This tumor-forming ability, combined with the known plasticity of the immune system, which can play both protumorigenic and antitumorigenic roles, has motivated investigations into the interaction between CSCs and the immune system. Herein, we review the interplay between host immunity and CSCs by examining the immune-related mechanisms that favor CSCs and the CSC-mediated expansion of protumorigenic immune cells. Furthermore, we discuss immune cells, such as natural killer cells, that preferentially target CSCs and the strategies used by CSCs to evade immune detection and destruction. An increased understanding of these interactions and the pathways that regulate them may allow us to harness immune system components to create new adjuvant therapies that eradicate CSCs and improve patient survival. Introduction Our defense against invading pathogens is due to innate and adaptive immune cells and the extensive network of cytokines that facilitate their interaction ensure the eradication of foreign antigens. Beyond ridding the body of invading pathogens, the defensive role of immune cells extends to eliminating abnormal cells, including mutated transformed cells, which can prevent tumorigenesis. In the early 1900s, Paul Ehrlich introduced his magic bullet theory hypothesizing that cancer would occur at high rates if it was not for the surveillance exerted by the immune system ( 1 ). However, the theory faced strong opposition. Evidence of tumor formation at sites of inflammation caused the rise of opposing opinions against the hypothesized antitumorigenesis role of the immune system ( 2 ) and stalled cancer immunology research for almost 50 years. In the late 1950s, enhanced understanding of the immune system and tumorigenesis led Frank Macfarlane Burnet and Lewis Thomas to individually reinvestigate the possible role of the immune system in cancer. Although Thomas postulated that under normal conditions the immune system prevents tumorigenesis in a similar fashion to graft rejection 3(), Macfarlane Burnet hypothesized that cancer cells contain neospecific antigens that can provoke an immune response leading to the elimination of neoplastic cells ( 4,5 ). Together, their work laid the foundation for the immune surveillance hypothesis—that the immune system eliminates nascent cancer cells and prevents tumor formation (6). Other groups disagreed with the immune surveillance hypothesis and suggested that the immune system would be unable to differentiate tumor cells from normal cells due to their similarities ( 7 ). The next wave of research provided evidence for the protumorigenic nature of innate immunity after enhanced tumor growth was observed in proinflammatory settings ( 8 ). In the 1990s, the use of various immune-deficient mouse models revealed that the immune system may have both supportive and inhibitory roles in tumorigenesis and metastasis. More recent studies into the opposing role of the immune system in cancer progression led to the development of several new concepts, which define our current understanding of Abbreviations cancer immunology. The pressure of immune surveillance, in addition to the highly adaptive and mutagenic nature of tumor cells, leads to the emergence of specific cancer phenotypes that are capable of forming tumors in the presence of a functional immune system (i.e. immunoediting) ( 9 ). As discussed in detail later in this review, immunoediting mechanisms, such as downregulation of antigen presentation to avoid detection by immune cells and recruitment of protumorigenic/anti-inflammatory immune cells to promote cancer progression, are well utilized by a subset of highly tumorigenic cancer cells, cancer stem cells (CSCs). CSCs [also commonly referred to as tumor-initiating cells (TICs)], cancer-initiating cells or stem-like cancer cells are believed to initiate cancer, mediate metastasis and contribute to therapeutic resistance and recurrence ( 10–12 ). CSCs were first defined in the 1990s by Dr John Dick’s laboratory, which isolated a population of acute myeloid leukemia (AML) cells with distinct cell surface marker expression (i.e. CD34+CD38−) that had significantly enhanced leukemogenesis properties in immunocompromised mice ( 13 ). Shortly after, the same group was able to identify CSCs, or leukemia-initiating cells, in three other forms of leukemia ( 14 ). Investigation for the presence of cell populations with tumor-initiating and self-renewal properties led to the discovery of the first CSC population in solid tumors by Dr Michael Clark’s group in breast cancer 1(0). They illustrated that breast cancer cells with CD44+CD24− surface marker expression had increased tumorigenicity and yielded tumors with heterogeneous cell surface marker expression. Subsequently, CSCs were isolated and described in other forms of solid tumors. Many of these CSC populations were identified by CD133+, such as in glioblastoma ( 15 ), liver ( 16 ), colon ( 17 ) and pancreatic cancer ( 18 ), or by CD44+CD24−, such as in pancreatic cancer ( 19 ) and prostate cancer ( 20 ). Additional CSC cell surface markers have since been discovered, such as CD90 for glioma, liver and breast cancer; CD166 for colon and prostate cancer; CD117 for lung and ovarian cancer; CD166 for colon and prostate cancer; and CD20 and CD271 for melanoma ( 21,22 ). In addition to identification by cell surface marker expression, the CSCs of many cancers can be identified by vital dye exclusion due to the increased expression of efflux pumps (23). In fact, transporter ABCG2 identifies pancreatic CSCs, and transporters ABCB5 and ABCB1 (MDR1) are used to identify melanoma CSCs ( 24–26 ). Additionally, increased aldehyde dehydrogenase (ALDH) activity as measured by the Aldefluor assay is used to identify the CSCs of a number of cancers including breast, colon, lung, melanoma, pancreatic and prostate cancer ( 10,27–29 ). It is important to note that the identification of cells that express these markers do not necessarily define them as CSCs, but rather identifies a population of tumor cells that is enriched for the cells with the CSC phenotype. The identification of tumor cells based on a combination of these markers (e.g. ALDH+CD44+CD24−) identifies tumor cells of greater tumorigenicity than those identified based on single markers (11). This suggests that the most robust CSC studies should use multiple markers and assays to confirm identification of the population. In addition to increased tumorigenicity, further investigation of CSCs revealed that they have increased resistance to conventional treatments, which could contribute to incomplete response to chemotherapy and radiation, and therefore cause relapse and decreased patient survival. Several mechanisms have been described that cause decreased sensitivity to these conventional treatments including increased expression of detoxification enzymes such as ALDHs ( 30 ), quiescence ( 31 ), hyperactivation of the DNA damage response ( 32 ) and increased drug efflux pump activity ( 33 ). The interaction between CSCs and the immune system is of much interest, as their increased tumorigenicity suggests they may have enhanced immunoediting mechanisms. In this review, we will discuss the role of major immune cells in promoting CSC expansion and therefore promoting tumorigenesis. Furthermore, we will review the evidence for CSC-specific avoidance of immune cell detection and destruction, as well as studies illustrating that CSCs can elicit protumorigenic immune cell activities. Finally, we will discuss the therapeutic potential of targeting these immune cell interactions to eliminate CSCs. Role of immune cells in regulating CSC maintenance and differentiation The role of the immune system in promoting and maintaining tumorigenesis has been largely attributed to innate immune cells [i.e. macrophages, myeloid-derived suppressor cells (MDSCs), dendritic cells (DCs) and T regulatory (Treg) cells] ( 34 ). The mechanisms for this have been largely explored outside of the context of CSCs. Below, we summarize the role of distinct innate and adaptive immune cells in tumorigenesis followed by the evidence for their role in CSC-specific expansion and maintenance (Figure 1, Table 1). Macrophages Macrophages are important innate immune effector cells. Along with neutrophils, they constitute the first line of defense against invading pathogens. Normally, macrophages bind and ingest invading pathogens, act as antigen-presenting cells (APCs) and produce proinflammatory cytokines ( 35 ). In cancer, tumorassociated macrophages (TAMs) are an essential component of the tumor microenvironment. Macrophages/TAMs can either target cancer cells and mediate their destruction or promote tumorigenesis and metastasis, depending on their predominant phenotype: M1 (proinflammatory) or M2 (anti-inflammatory), respectively ( 35–37 ). The predominant TAM phenotype is governed by the cytokine profile and the interaction with other immune cells. Adding to the complexity, the distinction between the M1 and M2 phenotypes is not always made in the cancer literature. The interaction between TAMs and CSCs has been exten sively studied. It can be direct or indirect and can have a range of consequences on CSCs including effects on chemoresistance properties, maintenance and differentiation. TAMs produce many cytokines including milk fat globule epidermal growth factor 8 (MFG-E8) and interleukin 6 (IL6), which activate signal transducer and activator of transcription 3 (STAT3) and the Hedgehog signaling pathway in CSCs, promoting their drug resistance (Figure 1A) ( 37 ). In hepatocellular carcinoma, IL6 produced by TAMs activates STAT3, which causes the expansion of CD44+ CSCs, and ultimately enhances tumorigenesis. This TAMmediated expansion of hepatocellular CSCs was halted in vitro and in vivo by inhibiting IL6 signaling with anti-IL6-R antibody tocilizumab ( 38 ). TAMs have also been found to induce transforming growth factor β1 (TGFβ1)-driven epithelial-to-mesenchymal transition (EMT) in a hepatoma cell line, promoting CSC properties; this effect was blocked by depleting TGFβ1 (Figure 1B) ( 39 ). Staining of a 96-pancreatic ductal adenocarcinoma (PDAC) patient tumor microarray for CSC markers CD4+4CD133+ and TAM marker CD204+ revealed a potential association between CSC and TAM prevalence in the tumors, which was associated with shorter overall and disease-free survival 4(0). Additionally, a recent study showed that breast CSCs interact directly with TAMs and tumor-associated monocytes via CD90/CD11b anchoring (Figure 1C) ( 41 ). This physical interaction leads to the activation of the EphA4 receptor, which activates nuclear factor-kappaB (NF-κB) as well as Src signaling pathways in CSCs. The juxtracrine signaling cascade culminates with the secretion of cytokines [IL6, IL8 and granulocyte macrophage colonystimulating factor (GM-CSF)], by the CSCs, which maintain the CSC state. CSC and TAM interactions can be mediated by other components of the extracellular matrix. Breast CSCs isolated from cell lines via cell surface markers upregulate hyaluronan synthase 2 (HAS2), which produces the extracellular matrix component hyaluronan (Figure 1D) ( 42 ). Pericellular hyaluronan mediates the interaction between CSCs and TAMs and promotes the secretion of platelet-derived growth factor BB (PDGF-BB) by TAMs. PDGF-BB activates stromal cells (e.g. fibroblasts) in the tumor microenvironment, which enhance CSC self-renewal ability by secreting fibroblast growth factor 7 (FGF7) and FGF9. Importantly, inhibiting HAS2 in CSCs with 4-methylumbelliferone halted tumor growth and resulted in fewer incidences of metastasis. This suggests that the inhibition of HAS2 and the interaction between CSCs and TAMs may be a potential novel therapeutic strategy. Finally, TAMs may indirectly drive CSC differentiation by promoting the production of interferon gamma (IFN-γ) by other immune cells such as natural killer (NK) cells ( 43,44 ). Importantly, the direct and indirect effects of TAMs and other immune cells such as NK cells on CSC maintenance and differentiation may also have the eventual consequence of altering the tumor’s ability to evade the immune system (e.g. expression of major histocompatibility complex class I; MHC-I) (44). Myeloid-derived suppressor cells MDSCs are a heterogeneous population of immature myeloid cells derived from myeloid progenitor cells, and they regulate immune responses to infection, inflammation and cancer ( 45 ). MDSCs affect immune responses by inhibiting NK cells, DCs, natural killer T (NKT) cells and T cells ( 46 ). Within the tumor microenvironment, MDSCs induce immune suppression ( 46 ), and there is some evidence of CSC-specific interactions with MDSCs. MDSCs were reported to induce microRNA-101 ( 60–67 ) ( 62,69 ) ( 62,68,69 ) ( 98–101 ) ( 70,71 ) ( 71,72 ) (74) ( 60–62 ) ( 84–86,88–91 ) ( 44,78 ) ( 62,87,92 ) (94) Immunotherapy and potential avenues for targeting cancer stem cells Immunotherapeutic approach Targeting PD1/PDL1 on T cells and CSCs, respectively Targeting STAT3 signaling DC-based vaccines (mir-101) expression in ovarian cancer cells 4( 7 ). Expression of mir-101 inhibits C-terminal binding protein-2 (CTBP2), resulting in increased expression of stemness genes and CSCassociated properties (Figure 1E). Furthermore, the pancreatic cancer tumor microenvironment can induce monocyte transformation to monocytic-MDSCs by activating STAT3 signaling in the cell (Figure 1F) ( 48 ). This coincided with increased EMT of the tumor cells and frequency of ALDH+ tumor cells. This suggests that targeting STAT3 might inhibit MDSC-mediated expansion of CSCs. Dendritic cells A well-orchestrated immune response against an invading pathogen is composed of complementing innate and adaptive responses, which ensure removal of the pathogen. DCs link innate and adaptive immunity and facilitate interaction between different immune cells ( 49 ). They are the most important APC and are critical in T and B lymphocyte activation and expansion. The key role of DCs in antigen presentation made them a prime candidate in investigation and development of immunotherapies and vaccines. However, there is also evidence that DCs can have adverse effects in cancer by promoting CSC enhanced tumorigenicity. In follicular lymphoma, follicular DCs promoted CSC chemoresistance and tumorigenesis via the CXCL12/CXCR4 signaling axis (Figure 1G) ( 50 ). Importantly, tumor formation was abolished on treatment with CXCL12/CXCR4-specific inhibitor, AMD3100, suggesting potential avenues of therapeutic intervention. Moreover, tumor-infiltrating DCs can play an important role in inducing immune tolerance and halt antitumor immunity ( 51 ). DC-induced tolerization has been attributed to hampered endocytic activity ( 52 ) and low expression of costimulatory molecules leading to inefficient antigen presentation ( 53,54 ). Although the role of CSCs in promoting DC tolerization is yet to be explored, it could provide important insight on the process of tumor initiation under immune surveillance. As their name indicates, Tregs regulate the function and activation of other immune cells. Importantly, they prevent autoimmunity by inducing tolerance to self-antigens and dampening immune cell activation (e.g. anergize T cells) ( 55,56 ). Tregs also play an essential role in the ability of cancer cells to evade antitumor immunity by promoting anergy of effector immune cells, rendering these cells unable to recognize and destroy cancer cells ( 56 ). Although not much is known about the interaction between CSCs and Tregs specifically, a recent colorectal cancer study showed Foxp3+IL17+-identified Tregs promote the development TICs by secretion of hypoxia-induced IL17 (Figure 1H) ( 57 ). Immune evasion mechanisms of CSCs Antigen presentation by the cells in our bodies is a constant process by which self-tolerance is established and abnormal or stressed cells are eliminated by the immune system. Therefore, a key initial step in adaptive immune recognition and targeting of tumor cells is antigen processing and presentation of neoantigens or altered self-antigens. Antigen processing requires proteosomal degradation of cytosolic proteins to form peptides. These peptides are then transported to the endoplasmic reticulum (ER) by the transporter associated with antigen processing (TAP) heterodimer, which is composed of TAP1 and TAP2. From the ER, peptides that originated in the cytosol can be loaded onto MHC-I molecules, which are expressed on all nucleated cells. Alternatively, antigens which have been internalized by phagocytosis, macropinocytosis, endocytosis or similar mechanisms are typically presented on MHC-II molecules by professional APCs (DCs, macrophages, B cells) ( 58,59 ). Upon antigen presentation, naive CD4+ T helper cells and CD8+ cytotoxic T lymphocytes can be activated to recognize and destroy cells presenting foreign antigen (Figure 2A). Tumors cells commonly evade immune detection by downregulating several important components of the antigen processing and presentation pathways, and thereby prevent the exposure of neo-antigens. This phenomenon appears to be enhanced in the CSCs of several cancer types (Figure 2, Table 1). For example, CD44+ putative CSCs from head and neck squamous cell carcinoma have decreased MHC-I and TAP2 expression in comparison with the CD44− putative non-CSCs (Figure 2B and C) ( 60 ). Similarly, MHC and/or TAP molecules expression was lower in the putative CSCs of melanoma, glioblastoma and colorectal cancer ( 61–64 ). However, it should be noted that at least one study reported no difference in MHC-I expression between the CSCs and non-CSCs of colon cancer ( 65 ). The decreased efficiency in antigen processing and downregulation of antigen presentation makes CSCs poor targets for the cell-mediated immune response of T cells. This contributes to the immune privileged status of CSCs within the tumor microenvironment. Downregulation of antigen processing and presentation is also evident in stem cells. Embryonic stem cells and the undifferentiated NTera2 teratocarcinoma cell line (pluripotent human embryonal carcinoma cell line) have reduced levels of restricted antigen processing machinery (e.g. TAP1) and absent expression of β2-microglobulin, which limits expression of MHC-I on the cell surface of these cells ( 66 ). Differentiation of the cells increased expression of the antigen presentation genes, connecting the differentiation state of cells with expression of MHC-I. The lack or low expression of the antigen presentation machinery in undifferentiated cells is in part Figure 2. Cancer stem cells employ several mechanisms to evade detection. (A) Several signals are required to activate T cells: MHC-I/II antigen presentation to TCR of TAP1/2-processed antigen, costimulatory signals via CD80/86 and activating cytokines. (B–E) CSCs interfere with antigen presentation and costimulatory signals to prevent cytotoxic T cells from targeting them ( 60–65,68,69 ). (B) Head and neck squamous cell carcinoma CSCs can downregulate MHC-I/II molecules to prevent antigen presentation to T cells and therefore limit their activation (60). (C) CSCs from various cancers downregulate TAP1/2 resulting in lack of antigen presentation to T cells, limiting their activation ( 60–64 ). (D) CSCs downregulate CD80 and (E) upregulate PDL1 to induce T-cell anergy ( 62,69 ). due to epigenetic regulation, including repression by histone modifications and DNA methylation. Similarly, hematopoietic stem cells also demonstrate low levels of MHC molecules ( 67), suggesting that immune privilege may be a feature of stemness maintained in CSCs. Treatment with a DNA demethylation agent increased expression of antigen presentation machinery in glioblastoma CSCs, although to a lesser extent than in differentiated tumor cells, suggesting that several mechanisms contribute to the altered expression of the genes ( 62 ). Furthermore, as discussed in a later section of this review, decreased expression of MHC-I expression also makes CSCs targets of other immune cells (e.g. NK cells).Therefore, it is often the balance of multiple signals that determines the fate of CSCs and tumor cells—eradication by the immune system versus evasion of the immune system. Antigen presentation on MHC molecules is rarely sufficient to activate T cells. In the absence of a costimulatory signal, T cells become apoptotic or anergic—they cannot be activated. This costimulatory signal can be provided by CD80 and CD86 on the APC, which interact with CD28 on T cells. Alternatively, inhibitory molecules like PDL1 (B7-H1), which can be found on APCs and tumor cells, bind to PD1, causing T cells to become anergic or apoptotic (Figure 2D and E). Upregulation of inhibitory molecule PDL1 is a known mechanism utilized by tumor cells to inhibit T-cell activity ( 68 ). Similarly, putative CSCs from glioblastoma and head and neck squamous cell carcinoma expressed very low levels of the costimulatory molecule CD80 and high levels of the inhibitory molecule PDL1 ( 62,69 ). This suggests that CSC interactions with T cells may be especially inhibitory and promote T-cell anergy ( 62,64 ). As discussed in the final section of this review, elevated levels of PDL1 on CSCs could provide a potential avenue for CSC-targeting immunotherapy. CSCs promote the expansion of protumorigenic immune phenotypes Although certain immune cells (TAMs, MDSCs, DCs andTregs) contribute to the maintenance of CSCs, the inverse is also true—CSCs induce the expansion of protumorigenic immune cell activities and eliminate antitumorigenic cell activities (Figure  3, Table 1). This CSC-induced activity on components of the immune system has the same end result of promoting tumorigenesis. Macrophages CSCs can promote macrophage polarization toward the immunosuppressive M2 phenotype, away from the proinflammatory M1 phenotype (Figure 3A) ( 70 ). Glioblastoma CSCs produce soluble colony-stimulating factor (sCSF), TGFβ1 and macrophage inhibitory cytokine (MIC-1) promoting macrophage recruitment, polarization toward the M2 phenotype and inhibition of macrophage phagocytic activity. This helps shield the tumor from the immune system. Furthermore, incubating macrophages in glioblastoma CSC-conditioned medium promoted the secretion of immunosuppressive cytokines including IL10 and TGFβ1 and enhanced their ability to inhibit proliferation of T cells. The capacity of CSCs to induce immunosuppressive macrophages could be reversed by inhibiting STAT3 signaling in the CSCs ( 71 ). In addition, CSCs can inhibit the expansion of antitumorigenic immune cells. A xenotransplant model of AML demon strated the ability of leukemia stem cells to inhibit the activity of antitumor macrophages through SIRPα signaling via engagement with the CD47 receptor (Figure  3B). Treatment with SIRPα-Fc fusion protein disturbed the SIRPα-CD47 engagement and enhanced the ability of macrophages to phagocytose AML stem cells without affecting normal hematopoietic cells ( 72 ). Neutrophils CSC-induced expansion of protumorigenic immune cells goes beyond macrophages and includes other innate effector immune cells, such as neutrophils. Neutrophils are the most common leukocytes in the blood. They have a well-characterized role in the elimination of pathogens and a more recently discovered role in the regulatory circuits of adaptive immunity. The role of neutrophils in cancer progression and metastasis has been generally associated with their capacity as inflammatory immune cells ( 73 ). A recent study has also linked neutrophils with glioma stem-like cells (GSLCs) by examining glioma patient tumor samples (Figure  3C) ( 74 ). They identified niches where GSLCs reside in close proximity to endothelial cells. These niches were positive for stromal derived factor-1α (SDF1α; also known as CXCL12), CXCR4, osteopontin and protease cathepsin K.  SDF1α/CXCL12 binds to CXCR4 on GSLCs and retain them in the niche, whereas osteopontin promotes neutrophil and macrophage infiltration in the niche. In turn, these infiltrating leukocytes produce neutrophil elastase and matrix metalloproteinase 9 (MMP-9), which together with cathepsin K, may degrade SDF1α/CXCL12. The action of the proteases on SDF1α/CXCL12 may release GSLCs from their niche, facilitating glioma invasion. Tregs CSCs can alter the adaptive immune landscape by targeting effector T cells and inducing the expansion of immunosuppressive, cancer-promoting Tregs. Two recent studies showed that the CSCs from glioblastoma and head and neck squamous cell carcinoma impair the proliferation of effector T cells 6( 0,62 ) and induce the expansion of Tregs (Figure  3D) ( 60 ). Similarly, cytokines produced by ABCB5+ identified malignant melanomainitiating cells inhibit effector T cells and instead induce recruitment of CD4+CD25+FoxP3+-identified Tregs to the tumor site ( 61 ). NK and NKT cells may play a key role in targeting CSCs In contrast with most of the evidence discussed thus far, there are indications that certain elements of the immune system may be effective at eliminating CSCs, specifically NK and NKT cells (Figure 4, Table 1). NK cells A subset of granular innate lymphocytes, NK cells are characterized by the presence of specialized lytic organelles containing preformed cytotoxic and proteolytic enzymes such as perforin, granzymes and granulysin ( 75 ). NK cells recognize and destroy virus-infected cells, cells undergoing stress response and abnormal cells in an antigen-independent fashion (circumventing the adaptive immunity requirement of prior sensitization). Moreover, as is often in the case of tumor cells, NK cells target cells that have low expression of, or lack, MHC-I molecules ( 76 ). This targeted response can be initiated via binding of the NK cell’s activating receptor NKG2D with its ligands (stress-inducible MHC-like molecules MICA/B and ULBPs), the binding of natural cytotoxicity receptors (NCRs) such as NKp30 and NKp44 with their respective ligands, or by antibody-dependent cellular cytotoxicity (ADCC) where the CD16 Fc receptor on NK cells binds to cell-bound antibodies. Subsequently, NK cell killing is mediated by the release of cytotoxic and proteolytic enzymes and/or the binding of death receptors Fas and DR5. Importantly, NK activity is necessarily highly regulated to prevent the killing of healthy normal cells (Figure 4A). NK cells also express inhibitory receptors such as killer-cell immunoglobulin-like receptors (KIRs), which bind to inhibitory MHC-I molecules expressed on cells, and prevent NK cell lysis. A balance between the activating and inhibitory receptors, as well as activation by cytokines such as IL2, is required for NK cells to attain their full lytic potential ( 77 ). NK cells may also play a key role in regulating stem cell maintenance and differentiation, potentially killing extraneous non-differentiated stem cells (which have low or lack MHC molecules) ( 44,66,78 ) and driving differentiation of cells required for the regeneration of damaged tissue ( 79,80 ). This dual function of NK cells is dependent on the predominant phenotype of NK cells (cytotoxic toward stem cells or anergic and cytokine producing, which drives the differentiation of the cells) ( 79,80 ). NK cells play an especially important role in the elimination of nascent tumor cells due to their recognition of abnormal or stressed cells ( 81,82 ). Furthermore, the selective pressure on cancer cells to downregulate MHC-I to avoid cytotoxic T-cell destruction makes them susceptible to killing by NK cells. The innate role of NK cells in cancer surveillance makes them attractive candidates for immunotherapy; especially in the context of expanded and activated allogeneic NK cells obtained from the peripheral blood of healthy donors, since autologous NK cells harvested from cancer patients often exhibit anergy (83). As discussed below, in terms of CSC-specific targeting, the majority of evidence suggests that NK cells exhibit increased targeting of CSCs; however, there is also some of evidence of enhanced evasion of NK cell-mediated attack by CSCs ( 43,44,62,78,84–92 ). A number of studies have illustrated that although CSCs are generally resistant to killing by resting/non-activated NK cells, activated NK cells have increased anti-CSC activity (Figure  4B) ( 84–86,88–91 ). In 2009, Castriconi et  al. reported that both allogenic and autologous NK cells, upon activation by IL2 and IL5, exhibited heightened activity toward patient-derived glioblastoma cells with CSC-associated characteristics (i.e. enhanced tumorigenicity, capable of multilineage differentiation) ( 84 ). Analysis of the CSC population revealed that these cells had increased expression of NK cell-activating ligands and low MHC-I expression. Avril et  al. later reported that glioblastoma-derived CSCs were highly susceptible to allogeneic NK cells that were activated by lectin or IL2 ( 85 ). More recently, Ames et al. reported that activated allogeneic and autologous NK cells preferentially target CSCs derived from patient tumors of liposarcoma, Ewing’s sarcoma and PDAC, as well as putative CSCs identified in cell lines from pancreatic cancer, breast cancer, glioma and sarcoma. Their analyses revealed that CSC killing by NK cells was dependent on NKG2D binding to upregulated activating ligands MICA/B and death receptors FAS and DR5 ( 86 ). Similarly, IL2-activated allogeneic NK cells eliminated primary and metastatic osteosarcoma cells, including the TIC cell population ( 91 ). NK cell killing was dependent on NKG2D activation with its ligand (NKG2DL) on tumor cells and TICs. The addition of NKG2DL-inducer spironolactone enhanced NK cell killing and reduced the osteosarcoma TIC population ( 91 ). Although activated allogeneic NK cells exhib ited greater cytotoxic activity, both allogeneic and autologous activated NK cells preferentially killed the cancer-initiating cells of colorectal adenocarcinomas ( 88 ). The cancer-initiating cells had reduced expression of NK cell inhibitory MHC-I molecules but elevated expression of ligands to NK cell-activating receptors NKp30 and NKp44. Finally, a body of evidence suggests that NK cells can control the CSC population of tumors by two mechanisms; the killing of undifferentiated CSCs by cytotoxic NK cells and the differentiation of CSCs by cytokine production of anergic NK cells ( 44,78 ). Tseng et al. observed that oral squamous CSCs are more sensitive to activated allogeneic NK cell-mediated cytotoxicity compared with their differentiated counterparts and the cytotoxicity of NK cells was dependent on the differentiation state of the cells (78). The same group found that glioblastoma CSCs are highly susceptible to activated allogeneic NK cell-mediated killing by direct lysis or ADCC, in comparison with their serumdifferentiated counterparts ( 82 ). However, glioblastoma CSCs can become insensitive to NK cell-mediated killing either due to NK cells anergy or as result of their differentiation, thus, limiting the ability of NK cells to curtail glioblastoma progression. The inclusion of a secondary strategy for targeting NK-induced differentiated tumor cells may improve tumor eradication. Notably, NK cell-mediated differentiation of CSCs increased their expression of MHC-I, which may facilitate their targeting by cytotoxic T cells ( 44 ). Taken together, these findings suggest that CSCs from multiple cancer types are especially susceptible to lysis by activated cytotoxic NK cells, that NK cells affect the differentiation state of the heterogenous tumor and that if the inactivation of NK cells induced by the tumor microenvironment is overcome, the use of activated allogenic NK cells as adjuvants in immunotherapybased approaches targeting CSCs is a viable strategy. Despite the wealth of evidence suggesting that activated NK cells have enhanced CSC-targeting ability, there is some evidence suggesting that glioblastoma, AML and breast CSCs may be resistant to NK cells (Figure 4C) ( 62,87,92 ). First, Di Tomaso et al. characterized CSCs isolated from glioblastomas and revealed that the cells had decreased expression of inhibitory MHC-I, which would make them sensitive to NK cell killing; however, the CSCs also had decreased expression of activating ligands, MICA and UPLBs, which suggests a potential resistance to NK cells ( 62 ). Importantly, unlike the study by Tseng et al. that measured NK cell killing of glioblastoma CSCs ( 82 ), the effect of NK cells on the glioblastoma CSCs was not assessed in the study by Di Tomaso et al. ( 62 ). Thus, it remains uncertain if the decreased expression of activating ligands in the CSC population observed by Di Tomaso et al. is consequential. For AML, She et  al. reported that cell line KG1a bore leukemia stem cells that were resistant to lysis by allogeneic NK cells activated by IL15 and CpG oligodeoxynucleotide and had low expression of NKG2D ligands MICA/B and ULBPs ( 92 ). Wang et al. reported that CSCs isolated from breast cancer patient tumors had low levels of MICA/B and were resistant to lysis by both allogeneic and autologous IL2-activated NK cells, although in general the autologous NK cells were less cytolytic than the allogeneic NK cells ( 87 ). The downregulation of MICA/B in the breast CSCs was mediated by increased expression of the oncogenic microRNA miR20-a. Hindrance of immune surveillance by tumor-infiltrating NK cells culminated in increased lung metastasis ( 87 ). The contrasting results between these studies ( 44,62,78,82,84–88,91,92 ) could be due to differences in how studies were performed, such as if the CSCs were isolated from patient tumor tissues versus cell lines, discrepancy in the markers used to identify the putative CSCs, differences in the tumor microenvironment, which affect expression of the activating ligands, and the protocols employed to activate the NK cells. In these studies, activated allogeneic NK cells were utilized; however, strategies for activation were not consistent, which may affect the response of the immune cells to CSCs of certain cancers. Additionally, many of the studies only addressed the cytotoxic effect of NK cells on CSCs, without addressing potential effects on differentiation. Future studies are required to determine how generally applicable NK cell targeting of CSCs is across all cancers, and whether a combination of markers that better identify the discrete CSC population should be employed. NKT cells Expressing both NK cell markers (CD161 receptor; NK1.1) and the T-cell receptor (TCR), NKT cells link innate and adaptive immunity and are restricted to CD1d-presented lipid and glycolipid antigens rather than peptide antigens presented by MHC-I/II ( 93 ). A  recent study of colorectal CSCs (identified by CD44+CD24+CD133+ cell surface marker expression) with upregulated CD1d was highly sensitive to killing by NKT cells (Figure 4D) ( 94 ). The effect was amplified by cotreating CSCs with thymosin-α1 (an immune adjuvant and modulator of inflammation) ( 94 ). This could potentially lead to the development of NKT cell adjuvant-based vaccines that target the heterogeneity of solid tumors by effectively killing CSCs and other tumor cells. Immunotherapy and potential avenues for targeting CSCs In addition to the possibility of harnessing NK and NKT cells to target CSCs (described above), the interactions between CSCs and immune cells provide other potential avenues of immunotherapy-based intervention. Below we describe some of the more clinically advanced immunotherapy approaches being studied in the context of CSCs ( Table 1). Higher expression of PDL1 (B7H1) expressed by tumor cells and activated immune cells of the tumor microenvironment is associated with more aggressive and advanced disease and poorer prognosis ( 95,96 ). Cytokines produced in the tumor microenvironment such as IFN-γ contribute to the increased expression of PDL1 in the tumor. PDL1 binds to PD1 on CD8+ T cells, blocking their activation and leading to an immunosuppressive tumor microenvironment. This is known as immune checkpoint. Several immunotherapeutic strategies inhibit this interaction, which prevents T-cell anergy and facilitates T-cell activity against cancer cells (i.e. checkpoint blockade). For example, nivolumab, an anti-PD1 monoclonal antibody, has been approved for melanoma and non-small cell lung cancer treatment ( 97 ). Importantly, as mentioned previously, several studies have shown that the CSCs of some tumors (glioblastoma and head and neck cancer) express higher levels of PDL1 than other tumor cells ( 62,64,69 ). Similarly, Zhi et al. found that CD133+ colorectal CSCs have elevated levels of PDL1 and this increase was associated with EMT, highlighting a possible immune evasion mechanism employed by CSCs during metastasis ( 98 ). Interestingly, PDL1 binding to PD1 in gastric cancer TICs enhanced their proliferation, suggesting a potential growth-stimulating role for PDL1 in CSCs ( 99 ). Furthermore, the CSCs of head and neck squamous cell carcinoma had elevated PDL1 expression compared with non-CSCs due to constitutive activation of STAT3 signaling pathway ( 100 ). IFN-γ treatment further increased PDL1 expression in the CSCs. The increase in PDL1 correlated with decreased immunogenicity which can be reverted by STAT3 inhibition or antibody blockade of PD1 ( 100 ). Most recently, Gupta et al. reported that PDL1 knockdown in B16 murine melanoma cells significantly reduced the CSC population ( 101 ). Together these studies suggest that immu nosuppression conferred by PDL1 expression contributes to the maintenance and differentiation of CSCs within the tumor. Importantly, the elevated PDL1 levels in CSCs suggest that using checkpoint blockade agents such as nivolumab may be an effec tive way of targeting CSCs via immunotherapy. Inhibiting STAT3 signaling may also be a potential strategy for targeting CSCs and their effects on the immune system. In response to growth factors and cytokines, STAT3 is phosphorylated and translocates to the nucleus where it induces transcription of genes. This mediates cellular responses such as cell proliferation, differentiation and apoptosis, which in transformed cells contributes to tumorigenesis and metastasis ( 102,103 ). Specific to its role in the tumor microenvironment, activation of STAT3 leads to production of cytokines such as IL6, which causes an increase in immunosuppressive cells such as MDSCs and Tregs, and a decrease in the maturation of DCs. Therefore, STAT3 is a key inducer of an immunosuppressive tumor microenvironment. As mentioned throughout this review, STAT3 plays an important role in CSC/immune cell interaction, with evidence of inhibition of STAT3 as a mechanism of blocking CSC-associated tumor immune evasion activity ( 64,71,100 ). Several studies illustrated the elevated levels of STAT3 signaling in the CSC population and its role in promoting the immunosuppressive properties of CSCs ( 48,64,71 ). Furthermore, immunosuppressive immune cells such as macrophages can activate STAT3 signaling and promote the drug resistance of CSCs ( 37 ). In these studies, inhibiting STAT3 blocked the immunosuppres sive effects of CSCs. Together these finding highlight the importance of STAT3 signaling in the immunosuppressive network of CSCs. Although at this time it is unclear what the clinical impact of STAT3 inhibition is on the immunosuppressive properties of CSCs, several STAT3 inhibitors are undergoing clinical trials to assess their effectiveness on different types of cancers. For example, inhibitor OPB-51602 is showing promising effects Vaccine composition Phase DCs pulsed with the lysate of ALDEFLUOR+ cells DCs pulsed with the lysate of ALDEFLUOR+ cells DCs pulsed with the lysate of ALDEFLUOR+ cells DCs pulsed with the lysate of ALDEFLUOR+ cells DCs pulsed with the lysate of ALDEFLUOR+ cells DCs pulsed with the lysate of ALDEFLUOR+ cells DC transfected with hTERT and survivin mRNA + CSC mRNA DCs pulsed with the lysate of allogeneic glioblastoma stem-like cell line cultured under neurosphere-forming conditions DCs pulsed with the lysate of ALDEFLUOR+ cells DC transfected with CSC mRNA CD105/Yb-1/SOX2/CDH3/MDM2polyepitope plasmid DNA vaccine I/II I/II I/II I/II I/II I/II I/II I I/II I/II I Stages I and II clinical trials investigating the effect of DC- and DNA-based vaccine in targeting CSCs and eliminating tumors of different types of cancer are completed or currently recruiting (summarized from information available aCtlinicalTrials.gov). against non-small cell lung carcinoma ( 104 ). Despite this poten tial success, the ubiquitous presence and general importance of STAT3 signaling suggests that targeting this pathway may present some toxicity. DC-based vaccines are being developed with the goal of circumventing the downregulation of immune responses by tumor cells and facilitate tumor cell recognition and killing by adaptive immune cells. Bone marrow-derived DCs loaded with tumor antigens to induce antitumor immune responses can either be specific to a tumor antigen or generated from whole-cell lysates of tumor cells. For example, DCs presenting mutated RAS peptides (neo-antigens) or tumor-associated antigens (TAAs) such as mucin 1, HER2/Neu and melanoma-associated antigen 3 (MAGE-A3) are in clinical trials and show promising results 1(05). Some studies suggest that this approach will also be able to target CSCs ( 106 ). The elimination of CSCs by DC-based therapeutic vaccines requires either the identification of CSC-associated tumor antigens or the isolation of CSC populations from patient tumors for the generation of whole-cell lysates. Potential CSCassociated antigens being investigated for DC-based vaccines include CD133, CD44 and octamer-binding transcription factor 4 (OCT-4), survivin and telomerase reverse transcriptase (TERT) ( 107 ). Animal models and clinical trials have shown high efficacy in establishing antitumor immunity and selectively targeting CSCs by inducing anti-CSC antibodies and cytotoxic T cells ( 65,108 ). Together, these findings led to the development of sev eral clinical trials to assess the effect of DC vaccines loaded with CSC antigens on several types of tumors (Table 2). However, due to the potential immunosuppressive nature of CSCs, combination treatment strategies where checkpoint blockade therapies are administered concomitantly with DC-based vaccines may be required to evoke a potent anti-CSC immune response. Conclusion The immune system can play opposing roles in the progression of cancer, having both protumorigenic and antitumorigenic consequences. Innate and adaptive immune cells contribute to tumor establishment, growth and metastasis by creating an anti-inflammatory, protumorigenic environment and inducing tumor cell migration. In contrast, immune cells also inhibit tumor development and promote destruction of established tumors by direct lysis of tumor cells by cytotoxic immune cells. The immune surveillance hypothesis suggests that when this balance is shifted toward the cancer cell by immunoediting mechanisms tumors develop. Studies reviewed here suggest that CSCs play a pivotal role in shifting this balance, having profound effects on regulating immune responses. Compounded by the impact that CSCs have on tumor development, progression and chemotherapy resistance, understanding the specific interaction between CSCs and the immune system has become increasingly important. However, it is apparent that delineating CSC-specific immune interactions is complex and only partially deciphered. Reports are at times conflicting, with evidence of both CSC targeting and evasion with the same immune cell (e.g. NK cells). Regardless, these finding have triggered interest into targeting CSCs via immunotherapy-based approaches. Delineating the molecular components of CSC-specific immunity will aid in developing immunotherapy approaches that eliminate the entire tumor without leaving behind chemotherapy-resistant CSCs. In ongoing and future immunotherapybased clinical trials, it will be important not only to determine the effect on tumor regression and patient survival but also to monitor the effect on CSCs and the immune system. Funding Canadian Institutes of Health Research (CIHR; MOP-130304 to P.M.); studentship or trainee awards from the Beatrice Hunter Cancer Research Institute (to M.S., K.M.C. and M.L.T.); Canadian Breast Cancer Foundation (to M.S.  and M.L.T.); CIBC Graduate Scholarship in Medical Research (to M.S.); CGS-D award from CIHR (to K.M.C.); Nova Scotia Health Research Foundation (to K.M.C., M.S. and M.L.T.); NS graduate scholarship (to M.S., M.L.T. and D.V.); Killam Laureate scholarship (to K.M.C., M.L.T.  and D.V.). Conflict of Interest Statement: None declared. 1. Ehrlich , P. ( 1909 ) Über den jetzigen stand der chemotherapie . Ber. Dtsch. Chem . Ges., 42 , 17 - 47 . 2. Aitken , W. ( 1863 ) The Science and Practice of Medicine . Vol. 2 . Lindsay and Blakiston, Philadelphia, PA. 3. Thomas , L. ( 1959 ) Discussion . In Lawrence, H.S. (ed). Cellular and Humoral Aspects of the Hypersensitive States. Hoeber-Harper , New York, NY, pp. 529 - 533 . 4. Burnet , M. ( 1957 ) Cancer; a biological approach. I. The processes of control . Br. Med . J., 1 , 779 - 786 . 5. Burnet , M. ( 1964 ) Immunological factors in the process of carcinogenesis . Br. Med . Bull., 20 , 154 - 158 . 6. Swann , J.B. et al. ( 2007 ) Immune surveillance of tumors . J. Clin. Invest ., 117 , 1137 - 1146 . 7. Gold , P. et  al. ( 1965 ) Demonstration of tumor-specific antigens in human colonic carcinomata by immunological tolerance and absorption techniques . J. Exp. Med ., 121 , 439 - 462 . 8. Scribner , J.D. et al. ( 1972 ) Inflammation and tumor promotion: selective protein induction in mouse skin by tumor promoters . Eur. J. Cancer , 8 , 617 - 621 . 9. Dunn , G.P. et al. ( 2002 ) Cancer immunoediting: from immunosurveillance to tumor escape . Nat. Immunol ., 3 , 991 - 998 . 10. Al-Hajj , M. et al. ( 2003 ) Prospective identification of tumorigenic breast cancer cells . Proc. Natl Acad. Sci. USA , 100 , 3983 - 3988 . 11. Ginestier , C. et al. ( 2007 ) ALDH1 is a marker of normal and malignant human mammary stem cells and a predictor of poor clinical outcome . Cell Stem Cell , 1 , 555 - 567 . 12. Dalerba , P. et al. ( 2007 ) Cancer stem cells: models and concepts . Annu. Rev. Med ., 58 , 267 - 284 . 13. Lapidot , T. et al. ( 1994 ) A cell initiating human acute myeloid leukaemia after transplantation into SCID mice . Nature , 367 , 645 - 648 . 14. Bonnet , D. et al. ( 1997 ) Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell . Nat. Med ., 3 , 730 - 737 . 15. Singh , S.K. et al. ( 2004 ) Identification of human brain tumour initiating cells . Nature , 432 , 396 - 401 . 16. Ma , S. et al. ( 2007 ) Identification and characterization of tumorigenic liver cancer stem/progenitor cells . Gastroenterology , 132 , 2542 - 2556 . 17. O 'Brien , C.A. et al. ( 2007 ) A human colon cancer cell capable of initiating tumour growth in immunodeficient mice . Nature , 445 , 106 - 110 . 18. Hermann, P.C. et  al. ( 2007 ) Distinct populations of cancer stem cells determine tumor growth and metastatic activity in human pancreatic cancer . Cell Stem Cell , 1 , 313 - 323 . 19. Li , C. et al. ( 2007 ) Identification of pancreatic cancer stem cells . Cancer Res. , 67 , 1030 - 1037 . 20. Liu , C. et  al. ( 2011 ) The microRNA miR-34a inhibits prostate cancer stem cells and metastasis by directly repressing CD44 . Nat. Med ., 17 , 211 - 215 . 21. Medema , J.P. ( 2013 ) Cancer stem cells: the challenges ahead . Nat. Cell Biol ., 15 , 338 - 344 . 22. Bruttel , V.S. et al. ( 2014 ) Cancer stem cell immunology: key to understanding tumorigenesis and tumor immune escape? Front . Immunol., 5 , 360 . 23. Komuro , H. et  al. ( 2007 ) Identification of side population cells (stemlike cell population) in pediatric solid tumor cell lines . J. Pediatr. Surg. , 42 , 2040 - 2045 . 24. Schatton , T. et al. ( 2008 ) Identification of cells initiating human melanomas . Nature , 451 , 345 - 349 . 25. Wouters , J. et al. ( 2013 ) The human melanoma side population displays molecular and functional characteristics of enriched chemoresistance and tumorigenesis . PLoS One , 8 , e76550 . 26. Monzani , E. et al. ( 2007 ) Melanoma contains CD133 and ABCG2 positive cells with enhanced tumourigenic potential . Eur. J. Cancer , 43 , 935 - 946 . 27. Awad , O. et  al. ( 2010 ) High ALDH activity identifies chemotherapyresistant Ewing's sarcoma stem cells that retain sensitivity to EWSFLI1 inhibition . PLoS One , 5 , e13943 . 28. Clay , M.R. et  al. ( 2010 ) Single-marker identification of head and neck squamous cell carcinoma cancer stem cells with aldehyde dehydrogenase . Head Neck , 32 , 1195 - 1201 . 29. Silva , I.A. et  al. ( 2011 ) Aldehyde dehydrogenase in combination with CD133 defines angiogenic ovarian cancer stem cells that portend poor patient survival . Cancer Res. , 71 , 3991 - 4001 . 30. Sládek , N.E. et  al. ( 2002 ) Cellular levels of aldehyde dehydrogenases (ALDH1A1 and ALDH3A1) as predictors of therapeutic responses to cyclophosphamide-based chemotherapy of breast cancer: a retrospective study. Rational individualization of oxazaphosphorine-based cancer chemotherapeutic regimens . Cancer Chemother . Pharmacol., 49 , 309 - 321 . 31. Liu , Q. et al. ( 2009 ) Molecular properties of CD133+ glioblastoma stem cells derived from treatment-refractory recurrent brain tumors . J. Ne-u rooncol. , 94 , 1 - 19 . 32. Bao , S. et al. ( 2006 ) Glioma stem cells promote radioresistance by preferential activation of the DNA damage response . Nature , 444 , 756 - 760 . 33. Loebinger , M.R. et al. ( 2008 ) Squamous cell cancers contain a side population of stem-like cells that are made chemosensitive by ABC transporter blockade . Br. J. Cancer , 98 , 380 - 387 . 34. Coussens , L.M. et al. ( 2002 ) Inflammation and cancer . Nature , 420 , 860 - 867 . 35. Condeelis , J. et al. ( 2006 ) Macrophages: obligate partners for tumor cell migration, invasion, and metastasis . Cell , 124 , 263 - 266 . 36. Mantovani , A. et  al. ( 2005 ) Macrophage polarization comes of age . Immunity , 23 , 344 - 346 . 37. Jinushi , M. et al. ( 2011 ) Tumor-associated macrophages regulate tumorigenicity and anticancer drug responses of cancer stem/initiating cells . Proc. Natl Acad. Sci. USA , 108 , 12425 - 12430 . 38. Wan , S. et al. ( 2014 ) Tumor-associated macrophages produce interleukin 6 and signal via STAT3 to promote expansion of human hepatocellular carcinoma stem cells . Gastroenterology , 147 , 1393 - 1404 . 39. Fan , Q.M. et al. ( 2014 ) Tumor-associated macrophages promote cancer stem cell-like properties via transforming growth factor-beta1-induced epithelial-mesenchymal transition in hepatocellular carcinoma . Cancer Lett. , 352 , 160 - 168 . 40. Hou , Y.C. et  al. ( 2014 ) Coexpression of CD44-positive/CD133-positive cancer stem cells and CD204-positive tumor-associated macrophages is a predictor of survival in pancreatic ductal adenocarcinoma . Cancer , 120 , 2766 - 2777 . 41. Lu , H. et  al. ( 2014 ) A breast cancer stem cell niche supported by juxtacrine signalling from monocytes and macrophages . Nat. Cell Biol ., 16 , 1105 - 1117 . 42. Okuda , H. et al. ( 2012 ) Hyaluronan synthase HAS2 promotes tumor progression in bone by stimulating the interaction of breast cancer stemlike cells with macrophages and stromal cells . Cancer Res. , 72 , 537 - 547 . 43. Tseng , H.C. et al. ( 2015 ) Differential cytotoxicity but augmented IFN- γ secretion by NK cells after interaction with monocytes from humans, and those from wild type and myeloid-specific COX-2 knockout mice . Front. Immunol. , 6 , 259 . 44. Tseng , H.C. et  al. ( 2014 ) Induction of split anergy conditions natural killer cells to promote differentiation of stem cells through cell-cell contact and secreted factors . Front. Immunol. , 5 , 269 . 45. Gabrilovich , D.I. et al. ( 2009 ) Myeloid-derived suppressor cells as regulators of the immune system . Nat. Rev. Immunol. , 9 , 162 - 174 . 46. Lindau , D. et al. ( 2013 ) The immunosuppressive tumour network: myeloid-derived suppressor cells, regulatory T cells and natural killer T cells . Immunology , 138 , 105 - 115 . 47. Cui , T.X. et  al. ( 2013 ) Myeloid-derived suppressor cells enhance stemness of cancer cells by inducing microRNA101 and suppressing the corepressor CtBP2 . Immunity, 39 , 611 - 621 . 48. Panni , R.Z. et al. ( 2014 ) Tumor-induced STAT3 activation in monocytic myeloid-derived suppressor cells enhances stemness and mesenchymal properties in human pancreatic cancer . Cancer Immunol . Immunother., 63 , 513 - 528 . 49. Banchereau , J. et al. ( 1998 ) Dendritic cells and the control of immunity . Nature , 392 , 245 - 252 . 50. Lee , C.G. et  al. ( 2012 ) A rare fraction of drug-resistant follicular lymphoma cancer stem cells interacts with follicular dendritic cells to maintain tumourigenic potential . Br. J. Haematol ., 158 , 79 - 90 . 51. Ma , Y. et  al. ( 2013 ) Dendritic cells in the cancer microenvironment . J. Cancer , 4 , 36 - 44 . 52. Tourkova , I.L. et al. ( 2007 ) Small rho GTPases mediate tumor-induced inhibition of endocytic activity of dendritic cells . J. Immunol. , 178 , 7787 - 7793 . 53. Tourkova , I.L. et al. ( 2005 ) Restoration by IL-15 of MHC class I antigenprocessing machinery in human dendritic cells inhibited by tumorderived gangliosides . J. Immunol. , 175 , 3045 - 3052 . 54. Shurin , M.R. et al. ( 2002 ) Inhibition of CD40 expression and CD40-mediated dendritic cell function by tumor-derived IL-10 . Int. J. Cancer , 101 , 61 - 68 . 55. Curiel , T.J. et al. ( 2004 ) Specific recruitment of regulatory T cells in ova-r ian carcinoma fosters immune privilege and predicts reduced survival . Nat. Med ., 10 , 942 - 949 . 56. Wolf , A.M. et al. ( 2003 ) Increase of regulatory T cells in the peripheral blood of cancer patients . Clin. Cancer Res. , 9 , 606 - 612 . 57. Yang , S. et al. ( 2011 ) Foxp3+IL-17+ T cells promote development of cancer-initiating cells in colorectal cancer . J. Leukoc. Biol ., 89 , 85 - 91 . 58. Guermonprez , P. et al. ( 2002 ) Antigen presentation and T cell stimulation by dendritic cells . Annu. Rev. Immunol. , 20 , 621 - 667 . 59. Ziegler , K. et al. ( 1981 ) Identification of a macrophage antigen-processing event required for I-region-restricted antigen presentation to T lymphocytes . J. Immunol., 127 , 1869 - 1875 . 60. Chikamatsu , K. et  al. ( 2011 ) Immunoregulatory properties of CD44+ cancer stem-like cells in squamous cell carcinoma of the head and neck . Head Neck, 33 , 208 - 215 . 61. Schatton , T. et al. ( 2010 ) Modulation of T-cell activation by malignant melanoma initiating cells . Cancer Res. , 70 , 697 - 708 . 62. Di Tomaso , T. et al. ( 2010 ) Immunobiological characterization of cancer stem cells isolated from glioblastoma patients . Clin. Cancer Res. , 16 , 800 - 813 . 63. Volonté , A. et  al. ( 2014 ) Cancer-initiating cells from colorectal cancer patients escape from T cell-mediated immunosurveillance in vitro through membrane-bound IL-4 . J. Immunol ., 192 , 523 - 532 . 64. Wei , J. et  al. ( 2010 ) Glioblastoma cancer-initiating cells inhibit T-cell proliferation and effector responses by the signal transducers and act-i vators of transcription 3 pathway . Mol. Cancer Ther., 9 , 67 - 78 . 65. Inoda , S. et  al. ( 2011 ) Cytotoxic T lymphocytes efficiently recognize human colon cancer stem-like cells . Am. J. Pathol ., 178 , 1805 - 1813 . 66. Suárez-Alvarez , B. et  al. ( 2010 ) Epigenetic mechanisms regulate MHC and antigen processing molecules in human embryonic and induced pluripotent stem cells . PLoS One , 5 , e10192 . 67. Le Blanc , K. et al. ( 2003 ) HLA expression and immunologic properties of differentiated and undifferentiated mesenchymal stem cells . Exp. Hematol. , 31 , 890 - 896 . 68. Brahmer , J.R. et al. ( 2012 ) Safety and activity of anti-PD-L1 antibody in patients with advanced cancer . N. Engl. J. Med ., 366 , 2455 - 2465 . 69. Lee , Y. et al. ( 2014 ) PD-L1 is preferentially expressed on CD44+ tumorinitiating cells in head and neck squamous cell carcinoma . J. Immunother. Cancer , 2 , P270 . 70. Jinushi , M. ( 2014 ) Role of cancer stem cell-associated inflammation in creating pro-inflammatory tumorigenic microenvironments . Oncoimmunology , 3 , e28862 . 71. Wu , A. et al. ( 2010 ) Glioma cancer stem cells induce immunosuppressive macrophages/microglia . Neuro. Oncol., 12 , 1113 - 1125 . 72. Theocharides , A.P. et  al. ( 2012 ) Disruption of SIRPα signaling in macrophages eliminates human acute myeloid leukemia stem cells in xenografts . J. Exp. Med ., 209 , 1883 - 1899 . 73. Kuang , D.M. et  al. ( 2011 ) Peritumoral neutrophils link inflammatory response to disease progression by fostering angiogenesis in hepatocellular carcinoma . J. Hepatol. , 54 , 948 - 955 . 74. Hira , V.V. et al. ( 2015 ) CD133+ and nestin+ glioma stem-like cells reside around CD31+ arterioles in niches that express SDF-1α, CXCR4, osteopontin and cathepsin K. J. Histochem . Cytochem., 63 , 481 - 493 . 75. Smyth , M.J. et al. ( 1999 ) Perforin is a major contributor to NK cell control of tumor metastasis . J. Immunol. , 162 , 6658 - 6662 . 76. Bauer , S. et  al. ( 1999 ) Activation of NK cells and T cells by NKG2D, a receptor for stress-inducible MICA . Science , 285 , 727 - 729 . 77. Vivier , E. et al. ( 2008 ) Functions of natural killer cells . Nat. Immunol ., 9 , 503 - 510 . 78. Tseng , H.C. et al. ( 2010 ) Increased lysis of stem cells but not their differentiated cells by natural killer cells; de-differentiation or reprogramming activates NK cells . PLoS One , 5 , e11590 . 79. Jewett , A. et al. ( 2014 ) Natural killer cells as effectors of selection and differentiation of stem cells: role in resolution of inflammation . J. Immunotoxicol. , 11 , 297 - 307 . 80. Jewett , A. et al. ( 2013 ) Dual functions of natural killer cells in selection and differentiation of stem cells; role in regulation of inflammation and regeneration of tissues . J. Cancer , 4 , 12 - 24 . 81. Chester , C. et al. ( 2015 ) Natural killer cell immunomodulation: targeting activating, inhibitory, and co-stimulatory receptor signaling for cancer immunotherapy . Front. Immunol. , 6 , 601 . 82. Tseng , H.C. et al. ( 2015 ) Differential targeting of stem cells and differentiated glioblastomas by NK cells . J. Cancer , 6 , 866 - 876 . 83. Cheng, M. et  al. ( 2013 ) NK cell-based immunotherapy for malignant diseases . Cell. Mol. Immunol ., 10 , 230 - 252 . 84. Castriconi , R. et  al. ( 2009 ) NK cells recognize and kill human glioblastoma cells with stem cell-like properties . J. Immunol. , 182 , 3530 - 3539 . 85. Avril , T. et  al. ( 2012 ) Human glioblastoma stem-like cells are more sensitive to allogeneic NK and T cell-mediated killing compared with serum-cultured glioblastoma cells . Brain Pathol ., 22 , 159 - 174 . 86. Ames , E. et al. ( 2015 ) NK cells preferentially target tumor cells with a cancer stem cell phenotype . J. Immunol. , 195 , 4010 - 4019 . 87. Wang , B. et al. ( 2014 ) Metastatic consequences of immune escape from NK cell cytotoxicity by human breast cancer stem cells . Cancer Res. , 74 , 5746 - 5757 . 88. Tallerico , R. et  al. ( 2013 ) Human NK cells selective targeting of colon cancer-initiating cells: a role for natural cytotoxicity receptors and MHC class I molecules . J. Immunol., 190 , 2381 - 2390 . 89. Jewett , A. et  al. ( 2012 ) Natural killer cells preferentially target cancer stem cells; role of monocytes in protection against NK cell mediated lysis of cancer stem cells . Curr. Drug Deliv., 9 , 5 - 16 . 90. Pietra , G. et  al. ( 2009 ) Natural killer cells kill human melanoma cells with characteristics of cancer stem cells . Int. Immunol. , 21 , 793 - 801 . 91. Fernández , L. et al. ( 2015 ) Activated and expanded natural killer cells target osteosarcoma tumor initiating cells in an NKG2D-NKG2DL dependent manner . Cancer Lett. , 368 , 54 - 63 . 92. She , M. et al. ( 2012 ) Resistance of leukemic stem-like cells in AML cell line KG1a to natural killer cell-mediated cytotoxicity . Cancer Lett. , 318 , 173 - 179 . 93. Bendelac , A. et al. ( 2007 ) The biology of NKT cells . Annu. Rev. Immunol. , 25 , 297 - 336 . 94. Ni , C. et al. ( 2015 ) Thymosin alpha1 enhanced cytotoxicity of iNKT cells against colon cancer via upregulating CD1d expression . Cancer Lett., 356(2 Pt B) , 579 - 588 . 95. Massi , D. et  al. ( 2014 ) PD-L1 marks a subset of melanomas with a shorter overall survival and distinct genetic and morphological cha-r acteristics . Ann. Oncol., 25 , 2433 - 2442 . 96. Chen , B.J. et al. ( 2013 ) PD-L1 expression is characteristic of a subset of aggressive B-cell lymphomas and virus-associated malignancies . Clin. Cancer Res. , 19 , 3462 - 3473 . 97. Wolchok , J.D. et  al. ( 2013 ) Nivolumab plus ipilimumab in advanced melanoma . N. Engl. J. Med ., 369 , 122 - 133 . 98. Zhi , Y. et  al. ( 2015 ) B7H1 expression and epithelial-to-mesenchymal transition phenotypes on colorectal cancer stem-like cells . PLoS One , 10 , 1 - 15 . 99. Yang , Y. et al. ( 2015 ) B7-H1 enhances proliferation ability of gastric cancer stem-like cells as a receptor . Oncol. Lett., 9 , 1833 - 1838 . 100. Lee , Y. et  al. ( 2016 ) CD44+ cells in head and neck squamous cell carcinoma suppress T cell-mediated immunity by selective constitutive and inducible expression of PD-L1 . Clin. Cancer Res. , 22 , 3571 - 3581 . 101. Gupta , H.B. et al. ( 2016 ) Tumor B7-H1 regulates cancer stem cell generation and virulence . J. Immunol., 196 (suppl. 1) , 72 .3. 102. Xie , T.X. et  al. ( 2004 ) Stat3 activation regulates the expression of matrix metalloproteinase-2 and tumor invasion and metastasis . Oncogene , 23 , 3550 - 3560 . 103. Wei , D. et al. ( 2003 ) Stat3 activation regulates the expression of vascular endothelial growth factor and human pancreatic cancer angiogenesis and metastasis . Oncogene , 22 , 319 - 329 . 104. Wong , A.L. et al. ( 2015 ) Phase I and biomarker study of OPB-51602, a novel signal transducer and activator of transcription (STAT) 3 inhib-i tor, in patients with refractory solid malignancies . Ann. Oncol., 26 , 998 - 1005 . 105. Rosenberg , S.A. et al. ( 2004 ) Cancer immunotherapy: moving beyond current vaccines . Nat. Med ., 10 , 909 - 915 . 106. Xu , Q. et al. ( 2009 ) Antigen-specific T-cell response from dendritic cell vaccination using cancer stem-like cell-associated antigens . Stem Cells , 27 , 1734 - 1740 . 107. Dhodapkar , M.V. et al. ( 2011 ) Vaccines targeting cancer stem cells: are they within reach? Cancer J ., 17 , 397 - 402 . 108. Zhao , F. et al. ( 2015 ) Cancer stem cell vaccine expressing ESAT-6-gpi and IL-21 inhibits melanoma growth and metastases . Am. J.  Transl. Res., 7 , 1870 - 1882 .

This is a preview of a remote PDF: https://academic.oup.com/carcin/article-pdf/38/2/107/24346374/bgw115.pdf

Sultan, Mohammad, Coyle, Krysta Mila, Vidovic, Dejan, Thomas, Margaret Lois, Gujar, Shashi, Marcato, Paola. Hide-and-seek: the interplay between cancer stem cells and the immune system, Carcinogenesis, 2017, 107-118, DOI: 10.1093/carcin/bgw115