Immunological Consequences of Epithelial-mesenchymal Transition in Tumor Progression

Abstract

Microenvironments that tumor cells encounter are different between primary tumor, during metastasis, and at the site of metastasis, suggesting potential differences in immune surveillance of primary tumor and metastasis. Epithelial-mesenchymal transition (EMT) is a key reversible process in which cancer cells transition into highly motile and invasive cells for dissemination. After which, only a tiny proportion successfully metastasize; supporting the notion of metastasis-specific immune surveillance. EMT involves extensive molecular reprogramming of cells conferring many clinically relevant features to cancer cells and affect tumor cell interactions within the tumor microenvironment. Here we will exclusively review the impact of tumor immune infiltrates on tumor cell EMT and the consequences of EMT in shaping the immune microenvironment of tumors. The utility of EMT as a model to investigate metastasis-specific immune surveillance mechanisms will also be explored. Finally, we will discuss potential implications of EMT for tumor immunogenicity, current immunotherapies and future strategies.

Introduction

Metastasis is the primary cause of cancer related mortality and can occur early through parallel progression along with the primary tumor or late after linear tumor progression (1). Epithelial-mesenchymal transition (EMT) is a reversible cellular process by which stationary epithelial cancer cells trans-differentiate into highly motile and invasive mesenchymal-like cells, giving rise to disseminating or circulating tumor cells to initiate tumor metastasis (2, 3). The disseminated tumor cells after reaching the target organ undergo the reverse phenotypic conversion from mesenchymal back to epithelial through a process known as mesenchymal-epithelial transition (MET) (4).

During EMT, cells downregulate the expression of multiple epithelial junctional proteins including E-cadherin, an adherens junction protein, leading to the dissolution of cell-to-cell contacts, loss of apico-basal polarity and acquire front-rear polarity (3, 4). Inhibition of E-cadherin expression is a hallmark of EMT and it is followed by the induction of proteins including N-cadherin, extracellular matrix components and the enzymes that can degrade them. Cells also undergo a robust reorganization of actin-cytoskeletal architecture and a dramatic change in the cell shape. Together, above key events result in a migratory and invasive capacity that defines the mesenchymal phenotype (3, 4). The change in the epithelial and mesenchymal gene expression that occurs during EMT is regulated by multiple transcription factor families that include Snail, Twist, Zeb and bHLH (5). The transcription factor involved and its role not only depend on the cell and tissue type, but also on the signaling pathway that initiates EMT (5). The multifunctional cytokine, TGF-β which is rich in tumor microenvironments (TME) and correlates with poor patient prognosis, has emerged as a potent inducer of EMT (6, 7). Initiation and progression of TGF-β induced EMT involves coordinated regulation of multiple signaling pathways by altering the expression or activation of their signaling components (6, 7). Several growth factors (EGF, HGF, FGF, IGF and PDGF) and developmental cytokines (Wnt, Notch and Hedgehog) are known to induce EMT (5). Similarly, inflammatory cytokines, including TNF-α, IL-6, IL-1 and IL-8, in the TME were also implicated in the induction of EMT (5, 8). All the non-TGF-β cytokines can induce EMT either through crosstalk with the TGF-β-dependent pathways or induce the expression of EMT transcription factors (8). In addition to imparting a migratory and invasive capacity, EMT was shown to endow resistance to chemotherapy, radiation, confer stem cell-like properties, and is known to promote immunosuppressive mechanisms in the TME (5, 9). Together, these abilities may allow cancer cells to successfully navigate the highly inefficient process of metastasis and link EMT to major clinical aspects that are responsible for cancer-related mortality. In contrast, the process of MET is less characterized. Even though, inhibition of TGF-β signaling or BMP-induced mir200 expression was shown to promote MET (10), the precise molecular mechanisms involved are still not clear. For a comprehensive understanding of all changes that occur during EMT, mechanisms that direct those changes, and pathways that mediate those mechanisms; please refer to several extensive and outstanding reviews already available in the literature (3, 5).

Mechanisms of tumor immune surveillance have been intensively investigated in the primary tumor setting where the competition between pro- and anti-tumor mechanisms dictates the outcome of tumor initiation and growth (11, 12). Accumulating evidence suggests the existence of metastasis-specific immune surveillance in epithelial malignancies (13, 14). Given the critical role of EMT in initiating and promoting tumor metastasis, we will exclusively review immunological consequences of EMT in tumor progression. We will outline the impact of immune microenvironment on tumor cell EMT and then describe how cells undergoing EMT may interact with immune cells in the primary tumor and during metastasis. Finally, we will discuss the implications of molecular reprograming that occurs during EMT on metastasis-specific immune surveillance, tumor immunogenicity, and immunotherapies.

Regulation of EMT by immune microenvironments

It is evident that inflammation plays a critical role at every stage of tumor development. Cancer cells produce various cytokines and chemokines that can recruit a diverse array of immune cells into the tumor including macrophages, neutrophils, dendritic cells, NK cells, mast cells, myeloid-derived suppressor cells, T and B lymphocytes, constituting the tumor associated inflammation (11, 12). Interactions of cancer cells with the immune microenvironment are critical for tumor progression beginning from tumor initiation to immune surveillance, promoting metastasis, and response to therapy (15). All tumor-infiltrating immune cells are capable of producing multiple inflammatory mediators to modulate tumor progression. Cancer cells, by engaging in a dynamic crosstalk with immune cells, exhibit EMT/MET plasticity to adapt to the changing microenvironment that they encounter in the primary tumor, during metastasis and at the distant site (Table. 1, Fig. 1) (16). These diverse interactions and the inflammatory mediators they produce, collectively and individually, can determine the course of tumor progression. In early stages, they can trigger neoplastic transformation by inducing genomic instability through production of DNA damaging agents. In later stages, they promote metastasis through multiple mechanisms including induction of EMT (15).

Table 1.

Summary of cancer-immune cell interactions and their functional consequences.

Cell Type	In Primary Tumor	During Transition	At Metastatic site
Effects of Tumor Infiltrating Immune Cells on Cancer Cell EMT
TAMs	Recruited by CCL217,18 Secretes: TGFβ, PDGF, EGF, TNFα, IL-1, and IL-611,17,18 Induces EMT11,17 Promotes angiogenesis and tumor growth11,16	Prevents anoikis24,25 Tumor cell survival21,23–25 Degrades ECM21,23 Aids intra- and extravasation22,23	Promotes metastasis17
Monocytes	Secretes: TGFβ, PDGF, EGF, TNFα, IL-1, and IL-611,17,21 Induces EMT?	Aids intra- and extravasation27	pMOs inhibit metastasis28
Neutrophils	Recruited by CXCL15 and HMGB132 Promote tumor growth?	NETs trap cancer cells30,31	Promotes metastasis30,31
MDSCs	Secrete: MMPs, CXCL5, CXCL12, VEGF, bFGF, HGF, TGFβ45 Induces EMT44 Promote tumor growth40–43		Establish premetastatic niche by versican secretion46
Platelets		Shields from NK cells38 Sustains EMT via TGFβ36 Aids in extravasation37	Promotes metastasis34
Complement	Induces EMT via C3a → TWIST48 Promotes tumor growth48	C3a sustains EMT48
T cells	Promotes tumor growth51 Induces EMT via Tregs?
Impact of EMT on Immune cell Functions
Macrophages	Promotes M1→ M2 via IL-4, GM- CSF, TGFβ27 M2 via TGFβ induced IRAK-M49 Promotes tumor growth49
Complement		Induction of CD59 on Tumor cells59 Resistance to CDC59	Promotes metastasis59
T cells	Induction of Tregs51 Evasion of CTL52	Tumor autophagy blocks CTL synapse formation52	Promotes metastasis51
NK cells	Secretion of soluble ligands69 Promotes tumor growth?	Induces activating ligands (NKG2DL69, Nectins71,72) Suppression of inhibitory ligands (Ecad)69 Secreted soluble NKG2DL leads to suppression?	Inhibits metastasis?68

EMT = Epithelial-Mesenchymal Transition, PDGF = Platelet derived growth factor, EGF = Epidermal growth factor, ECM = Extracellular matrix, HMGB1 = High mobility group box 1, MMPs = matrix metalloproteinases, VEGF = Vascular endothelial growth factor, bFGF = basic Fibroblast growth factor, HGF = Hepatocyte growth factor, IRAK-M = Interleukin-1 receptor associated kinase M, CDC = Complement Dependent Cytotoxicity, pMO = patrolling Monocyte, NET = Neutrophil Extracellular Trap, M1 = pro-inflammatory macrophage, M2 = anti-inflammatory macrophage, NKG2DL = Natural killer cell receptor D ligands, Ecad = Epithelial Cadherin. Numbers indicate references.

Figure 1. Cancer and immune cell interactions during tumor progression.

(1) Differential recruitment of various immune cell types is modulated by the tumor microenvironment (TME). (2) Tumor associated macrophages (TAMs), neutrophils, and Monocytes are converted to an anti-inflammatory phenotype. (3) M2 TAMs in conjunction with Myeloid derived suppressor cells (MDSCs) help drive EMT in the TME. These recruited cells also help suppress tumor infiltrating lymphocytes. (4) Upon successful EMT, cells can maintain this phenotype via autocrine complement protein C3a production and (5) blocks complement mediated cytotoxicity by upregulating CD59. (6) MDSCs can help metastasis formation by secreting versican to help the reverse process of mesenchymal to epithelial transition (MET) and seed the premetastatic niche at the distant site from the primary tumor. Increased PDL-1 and tumor autophagy aids in blocking cytotoxic T cell (CTL) (7) synapse formation and cytotoxicity. (8) This metastatic process can be limited by NK cell mediated cytotoxicity by targeting upregulated NKG2D ligands.

Tumor associated macrophages

Tumor Associated Macrophages (TAMs) are one of the major components of the immune cell infiltrates observed in the TME. They are derived from inflammatory monocytes that are recruited largely by MCP-1/CCL2, chemokines (17, 18). TAMs are implicated in multitude of tumor promoting functions including angiogenesis, immune suppression and EMT (11, 16). Genetic ablation of CSF1, a major lineage regulator of macrophages or deleting its direct effector, Ets2, results in macrophage depletion and marked reduction of metastasis in the PyMT model of breast cancer (19, 20). In the same model, recruitment of TAMs results in a TME rich in TGFβ, a potent inducer of EMT along with mitogenic growth factors like PDGF and EGF. This leads to induction of EMT by TAMs, thus promoting metastasis (11, 17). TAMS are also a source for TNFα, IL-6, IL-1 and MMPs which are known to enhance TGFβ-induced EMT and subsequent invasion (21). Consistently, analysis of primary tumors from patients with non-small lung cancer revealed a positive correlation between intratumoral macrophage densities, EMT markers, TGFβ levels, and tumor grade (16). Even though the precise mechanisms by which TAMs mediate tumor progression in vivo are still unknown; they are implicated in every step of the metastatic cascade. Following EMT, intravasation of cancer cells into vasculature is facilitated by perivascular TAMs (22). During this process cancer cells secrete CSF1 for the recruitment of TAMs and TAMs in turn produce EGF and activate EGFR signaling in cancer cells. Together with other cytokines in the TME, TAMs induce an EMT like phenotype with enhanced motility, invasion, and ECM degradation to promote intravasation of cancer cells (23). Once in circulation, cancer cells have to survive anoikis. Many inflammatory mediators derived from both immune cells and cancer cells, including TNFα, IL-6, and epiregulin, can promote cancer cell survival by activating pathways such as NF-kB and Stat3 (24, 25). In a mouse model of breast cancer TAMs were shown to promote cancer cell survival by physically interacting with them (23).

Monocytes: are a diverse set of cells with several subtypes with distinct function in the TME. A monocyte subpopulation, named metastasis associated monocytes (MAMs), was identified that preferentially migrate to metastatic sites rather than to primary tumors in breast and colorectal cancers (26). Similar to TAMS, MAMs are also derived from inflammatory monocytes, recruited by CCL2, and acquire pro-metastatic phenotype. MAMs were shown to promote cancer cell extravasation and survival at metastatic sites. Importantly, neutralizing CCL2 blocked recruitment of MAMs and inhibited cancer cell extravasation (27). Apart from the classical inflammatory monocytes which differentiate into TAMs or MAMs, a recent study also demonstrated a critical role for non-classical “patrolling” monocytes (pMO) in tumor metastasis (28). This study showed that pMOs accumulate in the microvasculature of lung and inhibit lung metastasis in multiple mouse models. pMOs reduce metastasis by interacting with cancer cells in the vessels and later recruit and activate NK cells (28). Unlike TAMs which promote EMT and metastasis, it would be interesting to study the effects of MAMs and pMOs on cancer cells. For instance, it would be fascinating to see if MAMs promote MET at metastatic sites to facilitate successful colonization by cancer cells.

Neutrophils

Studies thus far demonstrate both pro- and anti-metastatic effects of neutrophils during tumor progression. Depletion of neutrophils promoted lung metastasis in a mouse model of breast cancer (29). Consistent with this, neutrophils isolated from a tumor bearing mice showed cytotoxicity against cancer cells in vitro and adoptive transfer of these neutrophils blocked experimental lung metastasis (29). Neutrophils also produce unique structures called neutrophil extracellular traps (NETs) which are composed of extruded DNA and antimicrobial proteins. After surgical stress or infection, cancer cells have been shown to become trapped in NETs which formed in liver and lung capillaries promoting the development of micrometastasis (30, 31). On the other hand, cancer cells were also shown to recruit neutrophils through CXCL15 or HMGB1 secretion (32). Recruited neutrophils were implicated in enhancing angiogenesis, intravasation of cancer cells and suppression of cytotoxic CD8 T lymphocytes; thus promoting metastasis (11, 33). Interestingly, TGFβ has been shown to induce a switch from anti-metastatic to a pro-metastatic phenotype in neutrophils in a mouse model of mesothelioma (32). Therefore, it is possible that the pro-metastatic functions of neutrophils are regulated by specific environmental factors in a similar manner to TAMs; and like TAMs, neutrophils may also modulate EMT, at least in the context where they are known to promote metastasis.

Platelets

Also known as thrombocytes are small, enucleated cellular structures and are second most abundant in circulation after erythrocytes (34). The primary role of these cells is to stop bleeding (hemostasis) after tissue or vascular injury (34). Increased platelet numbers have been associated with decreased patient survival in a number of tumor types including, breast, lung, pancreatic, and brain suggesting a role for platelets in tumor progression (35). In circulation, platelets form platelet-cancer cell aggregates to aid and shield migrating cancer cells by multiple mechanisms and promote metastasis. In colon and breast cancer, platelets promote extravasation of cancer cells by inducing EMT, through direct contact and release of TGFβ (36). Platelet-specific ablation of TGFβ production or cancer cell specific inhibition of NF-kB activation protected mice from tumor metastasis (36). In melanoma, platelet-derived ATP was shown to activate a purinergic receptor, P2Y, on endothelial cells to increase vascular permeability and promote cancer cell extravasation (37). In this case, genetic ablation of P2Y suppressed metastasis (37). Formation of platelet-cancer cell aggregates may also protect circulating cancer cells from NK and T-cell mediated immune surveillance (38). In addition, platelet-derived cytokines including: PDGF, VEGF, and TGFβ can promote cancer cell survival, angiogenesis, and EMT in the primary TME and promote metastasis (34). Given the dynamic molecular changes that occur during EMT, it is reasonable to expect potential differences between epithelial and mesenchymal phenotypes in their ability to interact with platelets. Investigating these differences may help in the targeting of platelet-cancer cell interactions for metastatic control.

Myeloid-derived suppressor cells (MDSCs)

Abnormal differentiation of the myeloid compartment in tumor-bearing mice and cancer patient’s results in the accumulation of immature immunosuppressive myeloid cells called MDSCs, reflecting their origin and function (39). MDSCs contribute to tumor progression involving a variety of immune suppression dependent and independent mechanisms. MDSCs are known to produce a plethora of soluble factors including: MMPs, CXCL12, CXCL5, VEGF, bFGF, HGF, and TGFβ to promote angiogenesis, cancer cell invasion, and metastasis. Clinical relevance of MDSCs was demonstrated in multiple cancers where the number of circulating MDSCs in patients was correlated with advanced disease stage and metastasis (40–43). In a spontaneous mouse model of Melanoma, MDSCs recruited to the tumor site produced HGF and TGFβ to induce EMT in melanoma cells. Depletion of MDSCs suppressed melanoma metastasis by inhibiting cancer cell EMT (44). MDSCs are also known to promote metastasis by inducing cancer cell stemness in ovarian cancer (45). Intriguingly, MDSCs are implicated in the formation of premetastatic niches where MDSCs reach the niche before the cancer cells and condition it to promote cancer cell seeding by secretion of immunosuppressive factors including: S100A8/A9, bFGF, IL-10, and IL-4. Once cancer cells reach this metastatic niche, MDSCs are implicated in promoting MET in cancer cells by secreting versican (46). Lack of clear understanding of the precise mechanism by which MDSCs are recruited to the premetastatic niche makes this a somewhat mysterious process.

Indirect mechanisms

Aside from above described direct effects on cancer cells, resulting from cell-to-cell interactions with immune cells, the inflammatory cytokines produced by all immune cell types can modulate EMT through indirect mechanisms. The transcription factor Snail, an important regulator of E-cadherin expression during EMT, is protected from degradation in response to TNFα signaling. Thus stabilized Snail aids in completing EMT and promotes cancer cell migration and metastasis (25). Similarly, other EMT transcription factors like Twist and Kiss are also regulated by pro-inflammatory cytokines (47) including complement component C3a (48). Activation of Stat3 was implicated in Twist induction and NF-kB-mediated induction was shown for Twist and Kiss expression (25, 47). EMT-induced cancer cell invasion requires extensive proteolysis of the extracellular matrix (ECM). In addition to cancer cells, inflammatory immune cells are also important source for ECM degrading proteases including MMP2 and MMP9. Again, cytokines like, TNFα, IL-6 and IL-1 are implicated in the induction of these proteases. After EMT, when metastatic cells enter circulation the same cytokines also promote the survival of tumor cells in circulation through activation of NF-kB and Stat3 mediated survival pathways (12, 25, 47).

Impact of EMT on cancer and immune cell interactions

The other most important aspect of the tumor-microenvironment crosstalk is the ability of cancer cells to modulate immune responses within the tumor. The most common theme in these interactions has been that cancer cells interfere with the antitumor responses by secreting soluble mediators that block the effector functions of the involved immune cells and reprogram them into cells of a regulatory phenotype. Robust morphological and molecular changes that occur during EMT does support the idea that cells undergoing EMT have the potential to modulate the function and phenotypes of both innate and adaptive immune cells in the TME (Table 1 and Fig. 1). However, only few studies have looked at the impact of EMT on the interactions between cancer and immune cells.

Impact on tumor associated macrophages

After the recruitment of macrophages into TME, the reciprocal interaction between macrophages and cancer cells involves modulation of macrophage phenotype by cells undergoing EMT. Studies have shown that cancer cells can skew macrophages more towards an M2 phenotype that is associated with TAMs, through production of various factors including, IL-4, GM-CSF, and TGFβ (27). Although, precise mechanisms by which macrophages acquire the tumor promoting TAM phenotype are not clear, recent studies, including ours, identified a role for TLR signaling. We demonstrated that tumor cell derived TGFβ induces the expression of IRAK-M, a negative regulator of TLR signaling, in macrophages promoting an M2 phenotype (49). Genetic ablation of IRAK-M in mice inhibited tumor growth by promoting an M1 like phenotype in TAMs (49). In another example, screening for cancer cell derived factors that promoted macrophage activation identified extracellular matrix component versican in mouse lung cancer cells (46). Versican is also upregulated in many human tumors. This study demonstrated that versican activated macrophages through TLR2, induced IL-6 and TNF-α to generate a microenvironment that facilitates metastatic outgrowth of Lewis lung carcinoma cells (46). A similar skewing of neutrophils to a more tolerogenic phenotype by cancer cells was also reported in the TME (32). However, in all the above studies differential impact of epithelial versus mesenchymal phenotypes was not assessed. Whereas, utilizing a Snail1-overexpression model, EMT cells were shown to induce differentiation of immature DCs into regulatory DCs, with low MHC class II expression (50).

EMT and cytotoxic T lymphocyte functions

To date, unlike innate immune cells, there is no evidence that T and B cells can directly modulate tumor cell phenotype including induction of EMT in spite of their contribution to the overall tumor progression. On the contrary, cells undergoing EMT were shown to induce the activation of immunosuppressive T-reg cells. Utilizing both TGFβ-induced as well as Snail1-overexpression models of EMT in melanoma cells, TSP1 (thrombospondin-1) produced during EMT was implicated in the induction of FOXP3 expression in CD4+ T cells (50). Inhibition of Snail1 or neutralizing TSP1 was sufficient to restore T-cell infiltration and induction of anti-tumor immune responses in the B16-F10 melanoma tumors (50). In MCF-7 human breast cancer cells, acquisition of EMT phenotype was associated with the inhibition of cytotoxic T lymphocyte (CTL)-mediated lysis (51). This inhibition in CTL-mediated lysis was attributed to the dysfunctional immunologic synapse between CTLs and cancer cells, along with the induction of autophagy in cancer cells. Interestingly, inhibition of autophagy in cancer cells restored susceptibility to CTL-mediated cytotoxicity (51). This is consistent with the fact that the extensive actin cytoskeletal remodeling that occurs during EMT is also critical for the formation of immunological synapse (52). Furthermore, this suggests that in addition to the recognition of a tumor cell, the formation of a successful immunological synapse is also critical for host immune surveillance. The observed differences in the ability of epithelial and mesenchymal-like cells to form immunological synapses may also contribute to the potential metastasis-specific immune surveillance. In a unique mouse model of melanoma in which tumor cells disseminate early even before the primary tumor is detectable, disseminated tumor cells were kept dormant at distant sites, in part, by cytostatic CD8+ T cells. Depletion of these cells restored metastatic out growth, demonstrating immune control of metastasis (14). A more recent study demonstrated an important molecular link between EMT and CTL dysfunction. This study provided evidences that microRNA-200 (miR-200), a suppressor of EMT, targets PD-L1 which is a ligand for the CTL check-point receptor PD-1 (53). Transcription factor ZEB1, an EMT activator induces PD-L1 expression on tumor cells by relieving the miR-200-mediated suppression of PD-L1, resulting in the suppression of CTL function and promotion of metastasis. These findings suggest that EMT phenotype may serve as a biomarker to identify subgroups of patients who may respond to checkpoint inhibitors such as PD-L1 and CTLA4 antagonists.

Regulation of complement mediated cytotoxicity

The complement pathway is recognized as a first line of defense in host immune surveillance against non-self-microbial and tumor cells (54). The deposition and activation of complement component proteins in tumor tissues, coupled with increased expression of inhibitory complement regulatory proteins on tumor cells illustrate the importance of complement pathway in host immune surveillance against cancer (55–58). In a recent study, we observed resistance to complement-dependent cytotoxicity and induction of CD59 expression, after TGFβ-induced EMT in lung cancer cells. CD59 is a potent inhibitor of membrane attack complex that mediates complement-dependent cytotoxicity (CDC). Inhibition of CD59 expression restored susceptibility to CDC of cells that have undergone EMT, in vitro and blocked experimental metastasis by these cells (59). On the contrary, complement activating components are also implicated in promoting tumor progression. For example, complement component C3a was shown to trigger EMT in ovarian cancer cells through the induction of Twist, an EMT transcription factor (48). One possibility for such paradoxical effects might be that EMT renders cancer cells resistant to complement-mediated cytotoxicity, after which the activating components of the complement might promote tumor progression by sustaining EMT.

EMT confers susceptibility to Natural killer (NK) cell mediated cytotoxicity?

NK cells were initially identified for their ability to kill tumor cells without prior sensitization (60, 61). An Epidemiological study showed that low NK cell activity in blood correlates with high incidence of malignancies, suggesting a critical role for NK cells in host’s immunosurveillance against cancer (62). Like in the case of T cells, in spite of their presence in the tumors there is little evidence that NK cells actively contribute to tumor progression, including induction of EMT. Consistently, tumor infiltration of NK cells was mostly associated with either better patient prognosis or had no influence at all. On the contrary, immunosuppressive TME, which may also include cells undergoing EMT, renders tumor infiltrating NK cells hyporesponsive with low cytotoxic activity (63). The other major obstacle for NK-mediated immunosurveillance is their limited access to cancer cells in the tumor bed. Multiple studies have shown that NK cells, when present, are preferentially localized to tumor stroma with little or no direct contact with cancer cells (63–66). Emerging data suggests that circulating NK cells are potent killers of cancer cells compared to organ specific (67) or tumor infiltrating NK cells (63). In agreement with this hypothesis, circulating NK cells were shown to be crucial for prevention of metastasis (68), but the mechanisms involved are not clear.

Recently, acquisition of mesenchymal-like phenotype has been shown to increase the expression of NKG2D ligands, a major class of NK cell activators, rendering cells undergoing EMT more susceptible to NK-mediated cytotoxicity (69). Consistent with this we observed a similar increase in susceptibility to NK-mediated killing in lung cancer cells after TGFβ-induced EMT. However, the mechanism was independent of the NKG2D receptor (unpublished data). This suggests that EMT cells may become susceptible through multiple mechanisms. Together, the above observations indicate that cells undergoing EMT, while contributing to the immunosuppressive microenvironment that inhibits NK-mediated immunosurveillance, when in circulation they become more susceptible to NK-mediated cytotoxicity. This is consistent with the notion of metastasis-specific immunosurveillance and may in part contribute to the inefficiency of the metastatic process.

Additionally, several epithelial cell adhesion molecules, whose expression is extensively modulated during EMT, are identified as potential activating/inhibitory ligands for NK cells. For example, E-cadherin is a known inhibitory ligand for NK cells (70) and down regulation of its expression is a hallmark of EMT. Therefore, it is tempting to suggest that modulation of E-cadherin expression could be another potential mechanism by which cells undergoing EMT may become more susceptible to NK-mediated cytotoxicity. Similarly, among other cell adhesion molecules Cadm1, which is identified as an activating NK ligand (71), is frequently down regulated in multiple different malignancies and the nectin protein receptor CD96 is implicated in promoting spontaneous metastasis (72). Like NKG2DL, it would be important to assess whether these non-classical ligands and their receptors are also modulated during EMT.

Conclusions and potential clinical implications

Each of the studies described above has been carried out in isolation and in different models. However, the effects of different immune cell types in modulating various aspects of tumor progression are similar. As many of these immune cells work together in other contexts, it is likely that they also work together during primary tumor growth and metastatic seeding. Initiation and progression of EMT involves a robust reprogramming of gene expression, change in signaling and metabolic pathways, and reorganization of cytoskeleton. Given the wide spectrum of changes that occur during EMT, it is reasonable to speculate how EMT can have a broad range of consequences for cancer cells, host immunosurveillance and the efficacy of immune therapies as discussed below:

Tumor immunogenicity

Many conventional therapies including chemotherapy, radiation, and targeted therapies have been shown to rely on the induction of anti-tumor immune responses for their optimum efficacy (73). However, the triggering of an anti-tumor immune response depends on the immunogenicity of a tumor. Whereas, the immunogenicity is dictated by the cancer cell antigenicity and the multitude of other factors produced in the TME (74). Several mechanisms, including genetic and epigenetic changes are known to regulate both antigen expression and antigen presentation, which are two major criteria that regulate tumor immunogenicity. What is not fully considered in this process is the plasticity of cancer cells to undergo EMT. It is now well established that induction of EMT involves robust modulation of cell surface proteins, isoform switching by alternative splicing, immune modulatory cytokine secretion, and actin cytoskeletal remodeling (5, 75). Each of the above changes that occur during EMT is capable of generating neo-epitopes and modulates their presentation. As a result, EMT may alter tumor immunogenicity at a much faster time scale then genetic effects, which are inherently slower. Cataloging molecular changes during EMT, particularly ones that have the potential to modulate immunogenicity may identify novel antigens to design primary tumor- and metastasis-specific immunotherapeutic strategies.

Metastasis-specific immune surveillance

Studies thus far have focused on understanding how a tiny proportion of disseminating cells escape host surveillance and metastasize. Unfortunately, very little attention has been paid to the understanding of the mechanisms which successfully clear more than 99% of tumor cells. Granted that it is the escape of these less than 1% of cells is what results in the lethal metastatic disease, but not exploiting the effective tumor clearance mechanisms that are already employed by the host may be a missed opportunity. Since EMT is critical for metastasis, exclusive focus on evasive or resistance mechanisms that cells acquire after EMT may have promoted an unintended bias; that cells undergoing EMT must resistant to anything that these cells encounter. On the contrary, it is equally feasible that metastatic cells after EMT are also vulnerable to host immunosurveillance, as illustrated by the increased susceptibility to NK mediated cytotoxicity (69). In other words, when cancer cells exit the immunosuppressive primary TME, it is possible that they may pay a toll to metastasize by becoming more susceptible to host immunosurveillance, and thus contribute to the inefficiency of metastasis. However, this concept needs further and more careful investigation. If proven, identifying the molecules and mechanisms that regulate these potential EMT-induced vulnerabilities may be critical for any metastasis-specific prevention strategies. Given the potential for the presence of metastasis-specific immune surveillance mechanisms, in clinical trials, it may be important to somehow assess the efficacy of a given therapy on metastatic disease in tandem, even when there is no effect on the primary tumor.

Current and future immunotherapies

EMT has been implicated in conferring resistance to both conventional therapies such as chemotherapy, radiation therapy, and targeted therapies such as anti-EGFR small molecules. In addition, EMT based gene-signatures were shown to predict patient prognosis (76, 77). For example, a 20-gene signature that we derived from an EMT-associated secretory phenotype predicted patient survival in non-small cell lung cancer patients (78). More recently, primary tumors stratified based on a EMT score showed a strong enrichment for immune check point molecules in mesenchymal-like tumors compared to epithelial-like tumors, including significant upregulation of PD-L1 &2, PD-1, and CTLA4 (77). This suggests that EMT-based biomarkers can be valuable to select patients who will benefit from immune checkpoint blockade agents and other immunotherapies in cancer. Similarly, engineering chimeric antigen receptor T-cells or NK cells to specifically target mesenchymal-like cells may be a feasible metastasis-specific prevention strategy. Cell culture models of cytokine-induced EMT can be valuable tools in this endeavor particularly in identifying specific targets expressed by mesenchymal like cells.