Breast cancers co-opt normal mechanisms of tolerance to promote immune evasion and metastasis

Abstract

Normal developmental processes, such as those seen during embryonic development and postpartum mammary gland involution, can be reactivated by cancer cells to promote immune suppression, tumor growth, and metastatic spread. In mammalian embryos, paternal-derived antigens are at risk of being recognized as foreign by the maternal immune system. Suppression of the maternal immune response toward the fetus, which is mediated in part by the trophoblast, is critical to ensure embryonic survival and development. The postpartum mammary microenvironment also exhibits immunosuppressive mechanisms accompanying the massive cell death and tissue remodeling that occurs during mammary gland involution. These normal immunosuppressive mechanisms are paralleled during malignant transformation, where tumors can develop neoantigens that may be recognized as foreign by the immune system. To circumvent this, tumors can dedifferentiate and co-opt immune-suppressive mechanisms normally utilized during fetal tolerance and postpartum mammary involution. In this review, we discuss those similarities and how they can inform our understanding of cancer progression and metastasis.

INTRODUCTION

Pregnancy and postpartum breast/mammary gland involution exemplify unique immunological conditions whereby a critical immune balance is necessary. During pregnancy, cells at the maternal-fetal interface must support fetal growth and development while also developing immune tolerance to the fetus and simultaneously maintaining the ability to mount an immune response if infection occurs. It has been long appreciated that the maternal immune system can recognize paternal antigens in the partially allogeneic fetus, yet it is critical that the fetus is not rejected by the maternal immune system (1). This dynamic immune state is mediated in part by the uterine microenvironment, which adapts as the immunologically distinct fetus develops. Although it has been postulated that pregnancy creates a chronic state of immunological suppression, a successful pregnancy in fact cannot exist without select immune cells and factors (2–4), highlighting the need for appropriate and well-controlled communication and balance between the developing fetus and immune cells at the maternal-fetal interface. Similarly, after lactation, the mammary gland must remodel to regenerate a lactation-competent epithelium for subsequent rounds of pregnancy; this process, termed postpartum mammary gland involution, involves death of the milk-secreting mammary epithelial cells, which results in release of self-antigens. Therefore, activation of immune-suppressive mechanisms that promote tolerance occurs in the mammary microenvironment. Select immune cells are also required for proper execution of involution (5, 6). In addition, tumor progression is mediated not only by the tumor itself but also by the tumor microenvironment (TME), which comprises immune cells, extracellular matrix (ECM) proteins, and soluble factors, highlighting the ongoing cross talk between tumor and immune cells.

The immune system comprises two main components: the innate and adaptive immune systems. Innate immunity refers to cells that generate a quick response to immune threats; these cells include natural killer (NK) cells, macrophages, dendritic cells (DCs), and neutrophils, as well as others. Innate immune responses are generally independent of antigens, which are molecular structures that are recognized as foreign and bound by antibodies or cells of the adaptive immune system. The adaptive immune system comprises cells, including B-cell and T-cell lymphocytes, that destroy immune threats, either directly through cytotoxicity or indirectly via the production of antibodies and cytokines. The adaptive immune system also assists in remembering foreign antigens to protect against future threats. During an infection, antigen-specific B cells and T cells become activated and undergo clonal expansion. B cells produce foreign antigen-targeted antibodies, whereas T cells can expand into multiple effector subtypes, including cytotoxic T cells, T-helper cells, and memory T cells. Effector T cells include two primary subtypes: Th1 and Th2 cells, which are proinflammatory and anti-inflammatory, respectively (7). Immune activation is heavily regulated by regulatory immune cells like Th2 cells, as well as T-regulatory (Treg) cells. Tregs function to suppress the expansion of cytotoxic T cells, reduce cytokine production, and promote cytotoxic T-cell exhaustion to result in resolution of the immune response (8–11). The balance between immune-suppressive regulatory cells, cytotoxic cells, and cytokines controls immune attack versus an equilibrium state.

Successful maternal-fetal tolerance, postpartum mammary gland involution, and tumor progression are all dependent on mechanisms of immunoediting, whereby dynamic cross talk between multiple cell types affects immune cell interactions and impacts the overall immune response. Because tumor growth and metastasis, which largely occurs through the blood and lymphatic vasculature, are highly reliant on avoiding destruction by the immune system, tumors evolve multiple mechanisms of immune tolerance to avoid immune clearance (12). During tumorigenesis, there is initial immune activation and elimination of tumor cells, followed by immune and tumor equilibrium, and finally tumor cell escape from the immune system (referred to as the “three Es” of immunoediting; reviewed in Refs. 13 and 14). Importantly, tumor immunoediting has been implicated in poor clinical response to multiple immunotherapy regimens (reviewed in Ref. 15). The equilibrium phase of tumor immunoediting reflects the balance between immune suppression and activation that is observed during maternal-fetal tolerance and postpartum mammary gland involution. During fetal development, physiological maternal immune suppression is tightly regulated by an “immune clock of human pregnancy,” where the composition and function of immune cells are temporally controlled (16). Likewise, during mammary gland involution, composition, and function of immune cells in the mammary tissue are initially increased, followed by a resolution phase (6, 17, 18). In this review, we will discuss normal mechanisms of immunoediting such as those in fetal tolerance and mammary gland involution that can be co-opted by pathologies, with a focus on breast cancer.

Trophoblast Mimicry

After fertilization, a human zygote travels down the fallopian tube to the uterus and develops into a blastocyst. The blastocyst further develops into two distinct compartments: the embryoblast, which comprises the inner mass of cells that will become the embryo, and the trophoblast, which comprises the outermost layer of cells. The trophoblast mediates attachment to the uterine wall—a process called implantation—and later forms part of the placenta. After initial implantation, the trophoblast begins to invade the maternal endometrial decidua (a thick mucous membrane described in Parallels to the Decidua section) and cytotrophoblast cells surround the embryo. The cytotrophoblast cells then differentiate into two distinct subtypes. The first is the syncytiotrophoblast, which mediates nutrient uptake and hormone production. The second is the extravillous trophoblast (EVT), which invades through the syncytiotrophoblast to attach to the uterine wall (reviewed in Ref. 19). Notably, tumor cells share multiple phenotypes with the trophoblast, including proliferation, migration/invasion, and vasculature recruitment—a concept sometimes referred to as “trophoblast mimicry” (20, 21). Importantly, the trophoblast also mediates immune protection for the developing embryo—another phenotype similarly observed during tumorigenesis.

Trophoblasts alter the cytokine milieu and directly express immune-altering proteins, such as major histocompatibility complex (MHC; also known as human leukocyte antigen/HLA). HLA proteins are expressed at the cell surface and can present antigens to T cells via interaction with the T-cell receptor (TCR). Trophoblasts repress expression of class I HLA-A, -B, -C, and class II HLA-D antigens, which stimulate immune rejection while they upregulate immune-suppressive HLA-G (22). HLA-G can bind to KIR2DL4 and LILRB1 receptors on decidual NK cells to trigger cell senescence-associated secretory phenotype to promote immune tolerance (23). HLA-G can also ligand the LILRB2 receptor to activate immune-suppressive Tregs and myeloid-derived suppressor cells (MDSCs) to maintain fetal tolerance (24, 25). Similarly, cancers exhibit altered HLA profiles that affect immune activation. Breast tumors and breast cancer cells express HLA-G in a hormone-dependent manner. HLA-G on tumor cells can activate Tregs, and HLA-G expression correlates with increased breast tumor size, lymph node status, and clinical disease stage (26). Like trophoblasts, breast tumors can express HLA-G with corresponding loss of all class I HLA genes (27). Tumors can also secrete extracellular vesicles with high amounts of HLA-G, which correlate with breast cancer progression and increased detectable stem cell-like circulating tumor cells (28). Thus, trophoblast mimicry by breast tumors can promote immune suppression in part via altered expression of antigen presentation receptors.

Parallels to the Decidua

During fetal development, a thick mucous membrane called a decidua develops and lines the uterus. Maternal-fetal interaction is mediated by two distinct interfaces; the first is between the placenta and decidua basalis, and the second is between fetal membranes and the decidua parietalis. These interfaces facilitate immune tolerance to protect the fetus and to promote fetal growth and development. The placenta, which provides oxygen and nutrients to the fetus, is the primary barrier between mother and fetus and interfaces with the decidua basalis. Both placental and decidual cells are important for mediating maternal immune tolerance to the fetus.

After the maturing blastocyst is implanted in the uterus, it is initially exposed to a largely proinflammatory Th1 microenvironment. The immune milieu quickly shifts to an anti-inflammatory Th2 microenvironment, allowing the pregnancy to avoid immune clearance and continue (29, 30). The Th2-skewed state is maintained until after birth (16). Macrophages, NK cells, and DCs infiltrate the decidua and accumulate around the trophoblast (31, 32). Notably, when macrophages, NK cells, and DCs are depleted, placental development is impaired, supporting that these cells are critical for successful fetal development (4, 33–36). Women with a history of spontaneous miscarriages have higher ratios of proinflammatory to anti-inflammatory cytokines (37–39), which is thought to be largely mediated by NK cells (40, 41). Of all the immune cells present at the fetal interface, uterine NK (uNK) cells are the most abundant (42). As opposed to NK cells found in peripheral blood, uNK cells are less cytotoxic, as they do not express CD16—a critical antibody-dependent cellular cytotoxicity (ADCC) receptor (43). Instead, uNK cells are more immunomodulatory, as they secrete galectin 1 to promote tolerogenic DCs (34), and altered ratios of cytotoxic to immunomodulatory uNK cells have been linked to recurrent miscarriages (40).

Similar to the decidua microenvironment, Th2-skewed TMEs have been reported for numerous cancer types, including breast, melanoma, glioma, and lymphoma (44–46). Th2 cytokines, such as IL-4 and IL-10, induce macrophage polarization into a protumor state and suppress antitumor macrophage polarization, thus establishing an immune permissive TME that allows tumor progression (47). In breast cancer, Th2 high tumors are reported to be more proliferative and are correlated with higher tumor stage and grade (48, 49). During ADCC, NK cells are recruited to antibody-dense areas in a coordinated innate and adaptive response to tumorigenesis, which can be exploited therapeutically to induce a cytokine storm and leukocyte infiltration (50). However, soluble factors in the TME, such as TGF-β, adenosine, and PGE2, can suppress ADCC and render NK cells less cytotoxic (51), similar to the minimally cytotoxic uNK cells. Therapeutic approaches to increase ADCC by NK cells are thus currently being investigated and may prove fruitful in increasing immune clearance of tumors.

Mammary Gland Involution

The primary function of the mammary gland is to produce and secrete milk to nourish offspring after birth. As such, terminal differentiation of the mammary epithelium does not occur until postnatally and more specifically when lactation occurs. After lactation ends, postpartum mammary gland involution occurs, which is the highly conserved physiological process that remodels the breast tissue to a prepregnant state. The process of involution is essential for subsequent rounds of lactation for future offspring. A primary characteristic of postpartum involution is massive cell death, which is accompanied by immunosuppressive mechanisms so that the mammary gland can successfully regenerate for future lactation. Although pregnancy has long been thought to be protective against breast cancer in the long term, there is a transient increase in risk for all parous women that can last for upward of 30 yr that is dependent on age at first birth (52, 53).

Forced-weaned mice models have provided details regarding normal postpartum mammary gland physiology. Involution, after ceased lactation, occurs in two stages—reversible and irreversible (54, 55). The first reversible stage is triggered by milk stasis and upregulates cathepsins to mediate lysosomal cell death (56). If suckling resumes during this phase, which lasts for 48 h in mice, lactation can restart. However, in the absence of suckling, the irreversible second phase begins. During this phase, the mammary gland undergoes drastic changes, including anoikis-mediated cell death of ∼80% of the mammary epithelium and significant matrix remodeling (55, 57). The process of involution creates a wound healing-like environment and significant increases in proinflammatory cells, such as Tcells and macrophages, which aid in clearing apoptotic bodies and remodeling (58, 59).

The immune system plays a vital role in mammary gland involution and has been shown to promote an immunosuppressive environment. It is thought that breast cancers may take advantage of this microenvironment to facilitate tumor progression. During Phase I of involution, Signal Transducer and Activator of Transcription 3 (STAT3) upregulates cathepsin and facilitates lysosomal-mediated cell death (60). STAT3 is a transcription factor that is also linked to early breast cancer progression, chemoresistance, proliferation, and metastasis. During malignancy, STAT3 induces the production of VEGF and IL-10 in various immune cells, as well as production of the immunosuppressive factors IL-23 and TGF-β. Furthermore, NF-κB is activated during the early phase of involution, which aids in epithelial cell death and milk loss (61, 62). During Phase II of involution, apoptosis is mediated by intrinsic mitochondrial-mediated pathways (59, 63, 64). There is a peak of macrophage infiltration during Phase II, around involution day 6 in rodent models, to assist with clearing apoptotic bodies and remodeling (65). These macrophages are protumor, based on increased expression of Arginase-1 (Arg-1) and IL-13 (6). Furthermore, mammary epithelial cells become phagocytic to aid in this process (66, 67). During the second phase of involution, there is a peak in lymphatic vessel density, elevated levels of matrix metalloproteases (MMPs), deposition of fibrillar collagen, increased deposition of the oncofetal protein tenascin, and increased fibronectin and laminin fragments (6, 17, 58, 59, 68, 69). Each of these phenotypes has also been linked to stromal attributes of wound healing and inflammation, as well as tumor progression and dissemination (70–72). However, the immune system expands the innate/adaptive natural killer T (NKT) cell population in late-stage involution, which may help prevent oncogenesis via the antigen-presenting molecule CD1d. This suggests the complexity of mammary gland involution and the immune environment. NKT cells can promote an inflammatory environment via Th1 and Th2 cells, characterized by IFN-γ and IL-4. Increased IL-4 expression is significantly increased in breast tumors and promotes chemoresistance in breast cancer models (73, 74). Finally, rodent models have also shown that the mammary gland microenvironment created by postpartum involution is sufficient to drive mammary tumor growth and metastasis (70, 71). The ability of tumors to co-opt immune-suppressive programs observed during pregnancy and involution, combined with an inflammatory microenvironment in advanced-age pregnancies and subsequent involution, may increase tumor-promotional effects and warrants additional investigation.

Immune-Suppressive Leukocytes

T-regulatory cells.

In the decidua, T cells initially constitute ∼20% of total lymphocytes, and this population can increase up to 80% during a successful pregnancy (75) and Tregs are an essential immunosuppressive component in pregnancy (76). Approximately 14% of all decidual CD4⁺ T cells are CD25⁺ Treg cells, which peak in the first trimester and then gradually decrease during the remainder of pregnancy (77). Multiple murine studies have confirmed that CD4⁺ CD25⁺ Treg cells expand during pregnancy, and that depletion of these Treg cells causes immunological rejection of the fetus (78, 79). Syngeneic and allogeneic murine models have also revealed that mice prone to pregnancy loss exhibit lower CD4⁺ CD25⁺ Treg cell frequencies compared with mice with normal pregnancy success rates (80). Adoptive transfer of CD4⁺ CD25⁺ cells from other pregnant mice prevents fetal rejection, supporting that pregnancy-induced Treg cells promote allogeneic fetus immune tolerance (80). Interestingly, adoptive transfer of Tregs from nonpregnant mice did not have the same effect on fetal rescue, suggesting that prior exposure to paternal antigens is critical for transferred tolerance. Treg-mediated tolerance transfer is only effective before pregnancy day 4, highlighting the need for early immune tolerance in successful fetal survival and development. Furthermore, depletion of Treg cells increases pregnancy loss in allogeneic mice, but not in syngeneic mice (81), highlighting that immune recognition of paternal antigens is an important component of Treg-mediated immune suppression. Similarly, during involution, the T-cell population is enriched for Foxp3⁺ Tregs, reflective of an immune-suppressive microenvironment (18).

During tumorigenesis, Tregs suppress the immune system and allow tumor escape and growth. As reported in mechanisms of maternal-fetal tolerance, Treg depletion can cause immune-mediated rejection of leukemia, melanoma, plasmacytoma, and mastocytoma (58, 59). More recent studies have mirrored these results in murine mammary cancers, where Foxp3⁺ Treg cells accumulate with tumor progression and Treg ablation induces robust antitumor immunity in both primary and metastatic tumors (82). Furthermore, Treg ablation results in increased breast tumor invasion, collagen deposition, an increase in the percentage of cancer stem cells, as well as increased inflammatory cytokines IL-4 and IL-5 and reduced antitumor macrophages, consistent with a Th2 phenotype (83). In human breast tumors, Foxp3⁺ Treg cells are activated by DCs, reducing activation of effector T cells and promoting immune escape and tumor progression (84). Treg cells can also upregulate cytokine and chemokine receptors, including CCR8 (85). In addition, more aggressive breast cancers, including triple-negative breast cancer (TNBC) and HER2⁺ breast tumors, have increased numbers of Treg cells compared with less aggressive estrogen receptor-positive (ER⁺) subtypes. Higher Treg infiltration also correlates with a higher proliferative index, larger tumor size, and more lymph node metastases (86). Furthermore, Treg infiltration in breast cancer metastases correlates with poor patient survival (87). In addition, Tregs can affect breast cancer metastatic organotropism. Specifically, highly immune-suppressive Treg accumulation occurs during primary breast tumorigenesis, but tumors can then transcriptionally reprogram Tregs to render them ineffective at blocking metastasis to lymph nodes, but not to lungs. Tregs normally suppress NK-cell activation in lymph nodes, but this effect is lost after tumor-mediated reprogramming (88). Patients with breast cancer thus have high Treg:NK ratios in sentinel lymph nodes (88). These data support immune regulation of site-specific metastasis and, overall, highlight the importance of Tregs in tumor immune permissiveness.

Cytotoxic T cells.

During pregnancy, cytotoxic T cells that recognize paternal alloantigens in the fetus decrease to facilitate immune tolerance. These T cells become exhausted, and placental trophoblasts induce T-cell death to deplete fetus-directed clones (89). For example, the male H-Y peptide, which is present on a male fetus, can induce T-cell deletion and hyporesponsiveness in a mechanism mediated by a Fas/FasL interaction between the maternal T cells and the fetus (90). Exhausted and senescent CD4⁺ and CD8⁺ T cells are present at the maternal-fetal interface and decline in mothers who undergo preterm labor (91), supporting that T-cell exhaustion permits sustained fetal tolerance. Indeed, memory CD8⁺ T cells generated from a pregnancy are uniquely prone to exhaustion by re-stimulation with fetal antigens during subsequent pregnancies in a PD-1 and LAG-3-mediated mechanism (92). Similarly, during involution, CD4⁺ and CD8⁺ T cells in the mammary gland and in the draining lymph node are PD-1⁺ and accompanied by PD-L1⁺ lymphatic endothelial cells and macrophages, suggesting active immune checkpoint signaling.

This is mirrored in malignancies, whereby exhaustion of T cells can be a defining feature in many cancers (93). CD8⁺ T cells in the TME exhibit increased expression of PD-1 and LAG-3 and have impaired effector function with decreased production of IL-2, IFN-y, and TNF-α. The exhausted T-cell state allows for the tumor to escape cytotoxic T cells, and it can also predict clinical outcomes. An exhausted CD8⁺ T-cell phenotype correlates with poor survival in subset of patients with breast cancer (94). However, in a Phase II clinical trial, it was reported that patients with ER⁺ tumors were more likely to benefit from immune checkpoint inhibition if they had an exhausted CD8⁺ T-cell immune signature in both blood and/or tumor samples (95), suggesting that this patient population may be susceptible to select therapeutic interventions. PD-1 blockage can cause a paradoxical increase in PD-1 expression on Tregs and enhance their suppressive activity (96), further suggesting the need for advanced therapeutics for patients with exhausted T cells in the TME, particularly in an advanced, metastatic setting. In addition, in mouse models of postpartum breast cancer (PPBC), where tumor cells are exposed to postpartum involution, the tumors in postpartum mice take on characteristics of the immune milieu observed during postpartum involution including enrichment for PD-1/PD-L1 cells.

The ratio of CD8⁺ T cells to Treg cells has been implicated in breast cancer progression, with high CD8⁺/FOXP3⁺ ratios associated with decreased relapse rates, increased response to chemotherapy, and increased patient survival (97–100). In addition, CD4⁺ T cells have multifaceted roles in breast cancer. CD4⁺ “helper” T-cell infiltration corresponds to worse breast cancer patients’ overall and relapse-free survival, as does the intratumor CD4/CD8 ratio (101). However, multiple studies have also shown antitumor roles for CD4⁺ T cells. In breast cancer xenograft models, inhibition of naive CD4⁺ recruitment reduces the number of Treg cells and correspondingly decreases tumor growth (102). CD4⁺ T-helper cells have also been reported to block spontaneous breast cancer development in murine models by inducing epigenetic reprogramming of cancer cells to cause terminal differentiation, thus reducing invasive capacities (103). A subpopulation of CD4⁺ cells known as follicular helper T cells has a gene signature that positively correlates with patient survival and response to chemotherapy (104). Overall, a more robust characterization of CD4⁺ helper T cells in tumorigenic contexts is needed. Taken together, these data support that increased Treg infiltration and exhausted cytotoxic T cells are characteristic of maternal-fetal tolerance, postpartum mammary involution, and tumor-induced immune suppression.

Immune-Suppressive Myeloid Cells

Macrophages.

Macrophages can be recruited to the endometrium to contribute to a successful pregnancy (105). Approximately 20% of hematopoietic cells in the decidua are macrophages (106, 107). Decidual macrophages are critical in the clearance of apoptotic cell debris during pregnancy, which prevents the release of maternal self-antigens and paternal alloantigens, as well as a corresponding immune response that could damage the fetus (108). Decidual macrophages suppress T-cell alloreactivity—the ability to recognize allogeneic MHC complexes (109). Although decidual macrophages were previously considered a single population, three distinct populations have recently been characterized: 1) CCR2⁻ CD11cLO, 2) CCR2⁺ CD11cHI, and 3) CCR2⁻ CD11cHI (110). The CCR2⁻ CD11cHI population resides proximal to EVTs and expresses heme oxygenase-1 (HMOX1), which has immune-suppressive function that is discussed in the Heme oxygenase-1 section. The CCR2⁺ CD11cHI population expresses interleukin-1β (IL-1β) and prostaglandin synthase 2 (PTGS2/COX2) at the gene and protein level (110). Decidual macrophages can thus promote the secretion of prostaglandin E2 (PGE2), which is converted from arachidonic acid (AA) by COX-2. In addition, in the context of malignancies, macrophage polarization is frequently referred to in terms of tumor-suppressive “M1-like” or tumor-promotional “M2-like” phenotypes, although it is now appreciated that this is a plastic and dynamic spectrum comprised of many unique populations rather than simply two separate states. Notably, the CCR2⁻ decidual macrophage subtypes constitute a majority (>80%) of decidual macrophages and are considered more “M2-like” (i.e., protumor) (110). Decidual macrophages can also decrease the activation of T cells and alloreactivity via inhibition of IL-2 (109). Tissue-resident macrophages are also critical for additional developmental processes, as mice with loss of the colony-stimulating factor (CSF-1) signaling axis exhibit markedly reduced fertility capacity, as well as lactational defects due to incomplete mammary gland development (111).

During mammary gland involution, macrophages are the predominant immune cell infiltrate. Although macrophages were thought to play a secondary phagocytic role to mammary epithelial cells during involution, they are required for proper clearance and tissue remodeling (67). Involution-associated macrophages are mostly immunosuppressive M2-like and secrete Arg-1 and IL-23 (6). This phenotype is conserved in mice, rats, and human mammary tissues (6). ECM proteins that are upregulated during involution, such as procollagen, laminin, and fibronectin, are also sufficient to recruit immunosuppressive macrophages in vitro (6). Recently, involution-associated macrophages also were found to express more podoplanin, a lymphatic cell marker, and incorporate into lymphatic vessels (112)—a phenotype that can subsequently promote breast cancer lymphangiogenesis and lymphatic metastasis (112).

In parallel, tumors can co-opt macrophages and induce immune-suppressive, protumor function. Often termed tumor-associated macrophages (TAMs), these immune cells can support tumor cell growth, invasion, and metastasis. Mechanistically, this is done in a similar manner to macrophages during fetal tolerance—via suppression of cytotoxic T-cell function, as well as degradation of ECM and promotion of angiogenesis. In the context of tumorigenesis, a large meta-analysis revealed that TAM infiltration correlates with worse outcomes in breast and other cancer types (113). TAM infiltration is also associated with tumor progression in murine models and transgenic expression of macrophage CSF-1 induces high macrophage tumor infiltration, accelerated mammary tumor growth, and increased lung metastasis (114).

Cancer cells can impact macrophage phenotypes via cell-derived factors. An example of this is the Janus Kinase (JAK)/STAT pathway, a known oncogenic survival pathway, and the Notch pathway, which are also activated during postpartum involution. In human breast cancer and murine mammary cancer cell lines, conditioned media stimulates the activation of STAT5 and STAT3 in macrophages via GM-CSF (115). Breast cancer cells can also activate signaling of Jagged1, a Notch ligand, which correlates with an M2-like polarization via increased STAT3 activation (116). HER2⁺ breast cancer cells can also signal to macrophages through the Wnt pathway, which contributes to priming the tumor for early dissemination. Specifically, using an MMTV-HER2⁺ murine model, it was reported that macrophages infiltrate early lesions of HER2⁺ breast cancer, in part to HER2⁺ activation of NF-κB and upregulation of CCL2 to attract macrophages and signaled Wnt activation on the macrophage (117). Activated macrophages correlate with a mesenchymal tumor phenotype and increased metastatic burden. Importantly, M2-like macrophages may play a role in epithelial-to-mesenchymal transition (EMT), a key step for tumor cells to leave their primary site, intravasate, and survive in circulation (118–121). Thus, therapeutic approaches to target TAMs and overcome resistance to chemotherapy and other targeted therapies are an active area of interest (122).

Myeloid-derived suppressor cells.

MDSCs are immature myeloid cells that promote an immunosuppressive environment via interaction with the immune milieu, affecting both the innate and the adaptative immune response. MDSCs can expand during states of chronic infection or inflammation and, as such, they have both physiologic and pathologic roles, including in pregnancy and cancer. The role of MSDCs in maternal-fetal tolerance has only recently been characterized (reviewed in Ref. 123). MDSCs accumulate during tumor metastatic spread in pregnant mice; they impacted NK-cell activity, and depletion of MDSCs reduced metastasis (124). MDSCs can be derived from different precursors to generate unique MDSC subtypes termed GM-MDSCs and PMN-MDSCs. Peripheral granulocytic MDSCs (GM-MDSCs) increase during normal physiologic pregnancies compared with nonpregnant controls (125). Levels were highest during early gestation and dropped a few days postpartum to nonpregnant levels (126). Furthermore, the human placenta was further enriched for maternally derived GM-MDSCs, which were localized to the decidua and intervillous space (127, 128). MDSC function during pregnancy was organ dependent and, in brief, was found to inhibit T-cell production, inhibit Th1 activity, induce Th2 activity, induce Treg production, reduce NK cytotoxicity, and inhibit perforin release (125, 127–131). Finally, during involution, myeloid cells express the IFN-y receptor, which is expressed by immune-suppressive MDSCs (18).

MDSCs can be protumorigenic by having roles linked to immune evasion, EMT, angiogenesis, and metastases (reviewed in Ref. 132). In breast cancer, MDSCs were found to suppress antitumor immune response via IDO, and MDSC expression correlated with lymph node metastases in patients (133). Furthermore, MDSCs are induced by PGE2, a product of COX2 activity, via EP2/4. Inhibition of this pathway led to maturation of effector T cells and suppresses Treg induction (134). IDO is also significantly upregulated in tumor-infiltrating MDSCs compared with periphery MDSCs (133). Monocytic MDSC infiltration is also linked to reduced immunotherapy efficiency and worse patient outcomes. Circulating peripheral MDSCs in patients with breast cancer are elevated in all stages of disease and increase with higher stages. MDSCs are correlated with worse metastatic burden and overall survival in patients with breast cancer (135–137). This phenotype was found to be associated with potent immunosuppressive activity on T cells. Finally, MDSCs directly impact breast cancer cells and contribute to invasion and metastasis via the PTEN/AKT pathway and subsequent recruitment of MMPs, which are known to allow for tumor cell invasion and metastasis (138). Overall, current studies have shown that MDSCs are increased in breast cancer and contribute to an immunosuppressive environment that may impact patients’ response to current immunotherapy. This suggests that targeting MDSCs may be a potential therapeutic for patients with breast cancer.

Cytokines and Chemokines

Cytokines are small, immune regulatory proteins that modulate the strength and duration of an immune response. Chemokines are cytokine family members with specialized promigratory functions. The suppression of proinflammatory cytokines, such as IL-2, tumor necrosis factor α (TNF-α), and interferon-γ (IFN-γ), alongside increased production of anti-inflammatory cytokines, such as IL-4, IL-6, and IL-10, is necessary for the progression of a successful pregnancy. Cytokines and chemokines are detectable in the serum of pregnant women, and these cytokine profiles have characteristic trajectories as pregnancies progress (139). Functionally, IL-15 is required for differentiation of uNK cells, as well as microenvironment remodeling (3). Distinct from peripheral blood NK cells, decidual NK cells produce IL-8 and interferon-inducible protein-10 to regulate trophoblast invasion, as well as angiogenic factors to promote blood vasculature growth at the maternal-fetal interface (34). Multiple cytokines such as IL-4, IL-6, and IL-10 have been linked to miscarriages and preterm delivery (140). Women with preeclampsia have high Th1/Th2 cytokine ratios with elevated IL-6, IL-8, and TNF-α, suggesting a pathogenic proinflammatory state (141). Decidual cells can directly produce proinflammatory cytokines such as IL-6 (142), which is reported to contribute to preeclampsia by causing endothelial cell dysfunction and increased macrophage infiltration during EVT invasion (143). Decreased Tregs are also observed alongside the altered cytokine milieu during preeclampsia, as Tregs can permit fetal tolerance in a heightened inflammatory state (78, 144). In addition, trophoblasts can recruit and modulate immune cells such as monocytes via secretion of chemokines including growth-related oncogene-α, monocyte chemoattractant protein-1, and macrophage inflammatory protein-1β, as well as cytokines including IL-6, IL-8, and TNF-α (145). However, trophoblast-educated monocytes have damped responses to stimulus with lipopolysaccharide (LPS) (145), supporting that trophoblast-secreted factors suppress immune cell function. Trophoblast-secreted factors can also differentiate monocytes into CD14⁺/CD16⁺ macrophages that produce more IL-1β and IL-10 and have increased phagocytosis abilities (146). Decidual CD14⁺ macrophages also express anti-inflammatory and immunosuppressive cytokines, such as IL-10, to mediate fetal tolerance and dampen the inflammatory fetal microenvironment (147). Furthermore, cytokines and chemokines are important during postpartum mammary gland involution. During the first phase of involution, there is an increase in IL-1α, IL-1β, and IL-13 (58, 59). This initial wave of proinflammatory cytokines and neutrophils is followed by an influx of monocyte and macrophage chemoattractants, including CCL6, CCL7, CC8, and CXCL14, followed by increases in CSF-1R and CD68-positive macrophages (58, 59).

Alterations in cytokine profiles are also observed during tumorigenesis (reviewed in Ref. 148) and NF-κB activity in myeloid cells is considered a master regulator of cytokine production in the TME. NF-κB can increase IL-1, IL-6, TNF, IL-8, IL-17, IFN-γ, and CCL-5 to promote cell proliferation (149, 150). IL-6 has a well-characterized role in breast tumorigenesis, where it is expressed by tumor cells and fibroblasts and is associated with tumor growth, metastasis, and therapeutic resistance (151). IL-6 is known to activate JAK/STAT signaling to upregulate proliferative genes, such as CCND1 (152). In murine mammary cancer models, IL-10 expression downregulates proinflammatory MHC class I and enhances NK-mediated tumor cell lysis, thereby suppressing tumor growth and metastasis (153). IL-10 can also affect DCs and macrophage function, although this can be tumor-suppressive or tumor-promotional depending on the context (154). Taken together, these data reveal multiple immune-altering cytokines found during fetal tolerance, involution, and tumorigenesis. However, for metastatic malignancies, we lack comprehensive characterizations of the cytokine milieu in distinct metastatic sites, which warrants additional investigation.

Immune-Suppressive Enzymes and Metabolites

Indoleamine 2,3-dioxygenase and tryptophan 2,3-dioxygenase.

Metabolic enzymes have reported roles in both fetal tolerance and tumorigenesis. The rate-limiting enzymes indoleamine 2,3-dioxygenase 1 and 2 (IDO1/2) and tryptophan 2,3-dioxygenase 2 (TDO2) in the tryptophan catabolism pathway produce kynurenine, which has immune-suppressive effects by decreasing cytotoxic T-cell viability and expanding T-regulatory cells (155–157). CD4⁺ CD24⁺ Treg cells treated with the proinflammatory cytokine IFNγ upregulate IDO, which is understood to be immunosuppressive in pregnancy (77). In pregnant women, immune-inhibitory decidual CD14⁺ macrophages express IDO at the maternal-fetal interface (147, 158). More specifically, IDO is expressed in the syncytiotrophoblast, in the decidua, and in the fallopian tubes. TDO2 is also expressed in decidual stromal cells in murine placenta (159), and the immune-suppressive tryptophan metabolite kynurenine is detectable in cervical mucus (158). Furthermore, DCs produce IDO, and IDO expression is induced by MCSC activity, which can inhibit the proliferation of T cells and contribute to an immune-suppressive state (160). In rodents, inhibition of IDO during pregnancy causes fetal loss, further supporting its critical function in fetal immune tolerance (161, 162). Interestingly, it has been reported that TDO2-KO and double IDO1-TDO2-KO females do not exhibit reduced fertility when mated with allogeneic males (159), although this is likely explained by the activation of compensatory immunosuppression mechanisms. Finally, kynurenine can prevent cytokine-mediated induction of NK-cell-mediated killing (163), and NK killing capacity can be restored by both IDO and PGE2 inhibitors, suggestive of an additional role for COX-2 (164). Given the roles of COX-2 during postpartum involution, kynurenine may also play a role during involution, but to the best of our knowledge, this has not been investigated.

IDO and TDO2 are implicated in tumor-mediated immunosuppression. IDO is expressed in aggressive TNBC and is frequently coexpressed with PD-L1 (165). These data suggest that IDO may be involved in PD-1/PD-L1 regulation and could have a role in immunotherapy resistance. IDO has been reported to promote angiogenesis in preclinical breast cancer models, and IDO expression correlates to blood vasculature in patient tissues (166). IDO- and TDO2-mediated production of kynurenine can reduce viability and function of cytotoxic CD8⁺ T cells, and TDO2 mRNA expression correlates with poor prognosis of patients with breast cancer (167). TIM-3/Gal-9 can induce NK-cell-mediated production of IFNγ to subsequently increase IDO1 expression, which then downregulates NK-cell degranulation and allows tumor cell to escape from immune clearance (168). Furthermore, a TDO2-mediated signaling axis promotes TNBC-cell survival in anchorage-independent conditions and metastasis (168). Importantly, IDO has an identified role in breast cancer chemoresistance (169), and pharmacological inhibition of IDO increases sensitivity to chemotherapeutic approaches in murine mammary carcinoma models (170) and in a recent Phase II clinical trial in breast cancer (171). The role of IDO in cancer progression may be partially attributed to its ability to increase the expression of M2-like macrophage markers and decrease M1-like markers in cultured THP-1 cells (172). In addition, TDO2 has recently been implicated in ovarian cancer, where gene expression correlates with increased disease stage and recurrence (173). However, additional studies are required to better assess the role of TDO2 in immune suppression and to determine if TDO2 is a therapeutic target in additional cancer types. Additional studies are also needed to test the efficacy of new drugs targeting both IDO and TDO2, which are in the development pipeline. Overall, these data support the role of tryptophan catabolism via TDO or IDO and kynurenine in fetal tolerance and tumor-induced immune suppression.

Cyclooxygenase-2.

PTGS2/COX2 catalyzes the rate-limiting step of AA to prostaglandin conversion. Multiple components of the PGE2 synthesis pathway have been identified in maternal uterine tissue postembryonic attachment, as well as in embryonic tissue itself, supporting that the inflammatory response is cooperatively mediated by both maternal and fetal tissues (174). This pathway is also implicated during involution. COX-2 expression on the mammary epithelium is high during pregnancy and during involution in mice and remains elevated after involution is complete when compared with nulliparous samples (175). PGE2 levels are also increased during involution in a manner that can be reversed by ibuprofen. Involution infiltrating monocyte phenotypes are driven by PGE2, which can also be reduced by ibuprofen (176). PGE2 is the only known prostaglandin that has demonstrated a role in tumor formation, progression, and metastases; it has been shown to act on tumor cells and the surrounding stroma, including macrophages. PGE2 secreted from melanoma cells inhibited macrophage production of proinflammatory cytokines, such as TNF-α and IL-12. Furthermore, pharmaceutical inhibition of COX2, one of the main enzymes responsible for prostaglandin synthesis, reduces tumorigenesis in multiple breast cancer models (reviewed in Refs. 177 and 178). It is hypothesized that COX2 inhibition reduced PGE2 production and its ability to signal mitogenic and antiapoptotic pathways. COX2 overexpression correlates with worse breast cancer stages and outcomes (179–181). Furthermore, COX-2 expression activates VEGF-C/D expression, which promotes lymphangiogenesis and lymphatic metastases in human breast cancer and mouse mammary carcinoma models (182). A similar phenotype is seen by TAMs in a manner dependent on tumor or macrophage PGE2 production (183). Finally, in PPBC models, tumors implanted into involuting mammary glands acquire increased COX-2 expression and fibrillar collagen deposition that was dependent on COX-2 (69), and tumor cells collected from postpartum mice had higher amounts of PGE2 (17). Collectively, these data support an important immune-suppressive role for COX2 in multiple contexts.

Heme oxygenase-1.

The heme-degrading enzymes heme oxygenase-1 and -2 (HO-1; HO-2) convert heme into carbon monoxide, ferrous iron, and biliverdin (which is quickly converted to bilirubin). HO-1 has been implicated in immune suppression during fetal development. HO-1 is expressed in the human placenta (184). Reduced HO-1 activity is associated with increased pregnancy complications in humans (185), and upregulation of HO-1 decreases the rate of fetal rejection in mice (186, 187). HO-1 prevents fetal loss in abortion-prone mice by increasing the Th2/Th1 ratio (187). HO-1 can maintain maternal DCs in an immature state during murine pregnancy, which promotes the expansion Tregs (188). Loss of HO-1 prevents macrophage differentiation and reduces the expression of the macrophage markers CD14 and macrophage colony-stimulating factor receptor (189), which likely affects pregnancy tolerance. In the context of bacterial infection, LPS causes upregulation of HO-1 in macrophages (190) and a corresponding decrease in inflammatory components (191). Collectively, these data suggest that HO-1 is immune suppressive during fetal development to promote maternal-fetal tolerance. Interestingly, Hmox1 (the gene encoding murine HO-1) mRNA is upregulated during involution in mouse mammary glands (Fig. 1A) (58, 59); thus, further investigation regarding its role in involution is warranted.

Figure 1.

Chi3l1 and Ho-1 during mammary gland involution. Microarray analysis of (A) Hmox1 and (B) Chi3l1 mRNA expression from mice that are virgin [V, (day 8 from the study by Clarkson et al. (59) and day 10 from the study by Stein et al. (58)] or undergoing postpartum mammary gland involution (I; days 1–4). *P < 0.05, **P < 0.005 via ANOVA with Tukey’s multiple-comparisons test. Data from the study by Clarkson et al. (pink) (59) and Stein et al. (black) (58).

Similar HO-1-mediated mechanisms of immune suppression appear to be co-opted by tumor cells. In breast cancer, HO-1 is detectable in the TME and correlates with poor patient prognosis and chemoresistance (192). In preclinical breast cancer cell models, HO-1 reduces chemotherapy-induced cell death (193–195). Inhibition of HO-1 blocks chemotherapy-elicited CD8⁺ T cell-mediated immune suppression in murine models (196). HO-1 inhibition can resensitize murine TNBC and melanoma models to anti-PD-1 immunotherapy (197, 198) and differentiate M2-like protumor macrophages from antitumor M1-like macrophages (199). TAMs also express HO-1 in an IL-6-mediated manner, and HO-1 activation in TAMs can facilitate tumor cell migration and metastasis (200, 201). TAM expression of HO-1 also induces CD8⁺ T-cell proliferation and cytotoxicity in lymphoma models (192). In normal CD4⁺ CD25⁺ Treg cells, Foxp3 promotes expression of HO-1, which causes downstream suppression of effector T cells (202, 203). HO-1 expression in DCs alters inflammatory function by preventing DC maturation and secretion of proinflammatory cytokines (204, 205). Collectively, these data suggest immune cells in the TME may utilize HO-1 to promote immune suppression in a manner similar to fetal tolerance, but additional studies are needed to assess the full function of HO-1 in tumorigenesis, particularly at distinct metastatic sites. Clinical studies to target HO-1 or its metabolites in metastatic cancers may be a promising future direction.

Granzymes.

Another family of enzymes involved in fetal tolerance is granzymes. Granzyme B (GrB) is a serine protease expressed in NK cells and cytotoxic T cells. It is secreted along with perforin, which forms pores in target cells and promotes apoptosis. Decidual CD8⁺ effector T cells have lower GrB and perforin compared with peripheral T cells (206), suggesting that they have suppressed effector function. In breast cancer, GrB and perforin are expressed in tumor cells, and high expression levels correlate with lymph node invasion (207). Granzyme B has also been characterized on cells in the TME, including CD4⁺ Foxp3⁺ Treg cells, but not naive Treg cells; further, TME-derived Treg cells promote cell death of NK and CD8⁺ T cell in a GrB-dependent manner (11). In PPBC models, ibuprofen reduces granzyme levels (208) and levels of granzyme B are altered during postpartum involution (59). Collectively, these data support the existence of a GrB-mediated immune-suppressive, protumor response, similar to the role of GrB in maternal-fetal and involution-mediated tolerance.

Immune Checkpoint Molecules

An additional conserved mechanism of immune suppression between pregnancy, mammary gland involution, and malignancy is altered immune checkpoint molecules, such as CTLA-4, PD-L1, PD-1, Tim-3, and LAG-3. The immune regulatory circuit at the maternal-fetal interface is heavily mediated by IFN-β and PD-L1, which polarize macrophages toward an M2-like phenotype and decrease inflammatory responses. Macrophages educated by trophoblasts acquire a unique transcriptional profile, characterized by increased expression of type I IFN-β. IFN-β further increases soluble PD-L1 from trophoblasts, and PD-1 inhibition reduces trophoblast-induced macrophage differentiation (209). Trophoblast-educated macrophages exhibit diminished responses to TLR4 ligation. Soluble PD-L1 is detectable in blood from pregnant women, and soluble PD-L1 increases increased during pregnancy progression (209). PD-L1, as well as the immune-regulatory molecule galectin-9, are increased in the blood of pregnant women, and, interestingly, women with male fetuses had significantly higher galectin-9, but not soluble PD-L1, in their blood (210). Furthermore, it has recently been reported that the decidual microenvironment alters the tolerance-defense balance in CD8⁺ T cells, as coculture with trophoblasts increased the exhaustion markers PD-1 and CD39 (211). Together, these immune checkpoint molecules promote fetal tolerance by diminishing the maternal immune response.

PD-L1 is expressed on mammary stem cells, and PD-L1 expressing mammary basal cells exhibited robust mammary regeneration (212). PD-L1 and PD-L1 were also found on lymphatic endothelial cells and myeloid cells during involution (5), and PD-L1 blockade during involution in a PPBC model reactivated antitumor responses, such as delayed tumor growth, increased CD8⁺ T cells, and decreased lymphatic endothelial cells (5). Malignancies exhibit immune suppression via immune checkpoint signaling. Indeed, clinical availability of immune checkpoint inhibitors targeting CTLA-4, PD-L1, and PD-1 in solid tumors has dramatically impacted patient treatment and improved patient outcomes. The CTLA4-blocking antibody ipilimumab was authorized in 2011, and antibodies targeting PD-L1 (atezolizumab and durvalumab) and PD-1 (pembrolizumab and nivolumab) were approved soon after. There are multiple, excellent reviews available on the state of immune checkpoint blockade therapies in cancer treatments (213, 214). Although immune checkpoint-blocking therapies are now approved for ∼50 different cancer types, including TNBC, there are still large numbers of patients whose tumors do not respond to immune checkpoint inhibition. This highlights the need for better understanding of the complexities of the immune system to increase our ability to predict for efficacy and toxicities of these regimens.

Immune-Suppressive Secreted Factors

Semaphorin 7a.

Semaphorin 7a (SEMA7A; also named CDw108) is a glycosylphosphatidylinositol-linked protein that can be cleaved and shed into the extracellular space, allowing autocrine, paracrine, and juxtracrine signaling (215, 216). SEMA7A can signal through two receptors: ß1 integrin and plexin C1 (217–219). SEMA7A was first identified on lymphocytes to aid in wound healing, but it is best known for its role in axonal guidance during development (216, 220, 221). SEMA7A has well-characterized immune modulation capabilities. SEMA7A is expressed in B and T cells, macrophages, DCs, and NK cells, along with immature monocyte and lymphocyte lineages (222–228). SEMA7A is best characterized in T lymphocytes, where SEMA7A^−/− mice exhibited hyperactivated T cells after stimulation, suggesting SEMA7A’s role in immunosuppression (226). To further support this, SEMA7A was found to be crucial for severe inflammation resolution. SEMA7A was linked to metabolic reprogramming of macrophages to an M2-like phenotype via IL-10 in a murine colitis model (229). However, SEMA7A also is upregulated on stimulated T cells and signals through ß1 integrin to release proinflammatory cytokines, such as TNF-α, IL-1β, and IL-6 (224, 230–232). SEMA7A promotes lymphangiogenesis by inducing macrophages to express podoplanin and become more epithelial-like (112). Furthermore, SEMA7A has been recently linked to regulating PD-1/PD-L1 checkpoint blockade (5). The varying immune activating and suppressing roles may be due to SEMA7A’s different receptors and tissue specificity. In the placenta, SEMA7A and plexin C1 expression are significantly increased and SEMA7A expression in the placenta was not found to be cell-specific, consistent with its role in a variety of cell types (233). The exact mechanism of why SEMA7A and plexin C1 are high in placenta tissue has yet to be uncovered; however, it is believed that SEMA7A promotes immune suppression via plexin C1 to prevent maternal-fetal immune responses.

SEMA7A is well known for its role in wound healing and fibrosis—mechanisms that are co-opted in breast cancer to aid in progression and invasion. In breast tissue, SEMA7A is upregulated during postpartum involution and plays a role in fibrosis-like remodeling (69, 234–236). During involution, SEMA7A promotes macrophage-mediated lymphatic remodeling, anoikis resistance, and luminal progenitor cell survival (112, 237). SEMA7A promotes tumorigenesis in breast cancer, lung cancer, glioblastoma, and melanoma (112, 238–241). SEMA7A can promote drug resistance in lung and breast cancer models (241, 242). In addition, SEMA7A is upregulated in PPBC and drives multiple aspects of tumor progression such as proliferation, survival, migration/invasion, and lymphangiogenesis (112, 243). In breast cancer, SEMA7A promotes EMT via TGF-ß signaling, the release of proangiogenic cytokines, and tumor growth and metastases (239, 242–247). Finally, SEMA7A was reported to promote metastasis through the upregulation of Chitinase 3–like-1 (CHI3L1) by signaling via ß1-integrin to activate AKT and MAPK signaling in mammary carcinoma and melanoma models (239). Collectively, these data support that SEMA7A has important roles in multiple immune-suppressive contexts.

Chitinase 3–like-1.

CHI3L1 (also called breast regression protein 39/BRP-39 in mouse and YKL-40 in human) has reported immune-modulatory functions. CHI3L1 is a glycosyl hydrolase family 18 protein member, which binds to chitin but lacks the enzymatic activity necessary to degrade it. CHI3L1 is expressed and secreted by multiple cell types, including tumor cells, macrophages (248, 249), neutrophils (250), and activated T cells (251). Although the role of CHI3L1 during fetal development is not well characterized, it has been reported that pregnant women with gestational diabetes mellitus have an increased inflammatory status, including an enriched neutrophil/lymphocyte ratio, platelet/lymphocyte ratio, and high serum CHI3L1 levels (252). In the normal mammary gland, Chi3l1 (the gene encoding murine CHI3L1) is upregulated during postpartum involution, where it is thought to be expressed because of the massive death of mammary epithelial cells on weaning to suppress the immune system during this normal process (Fig. 1B) (58, 59, 253, 254). CHI3L1 suppresses E-cadherin but increases MMP-9 activity and cell motility in mammary epithelial cells (255). It is suggested that STAT3 is regulated by CHI3L1 during involution, suggesting a link between CHI3L1 and the immunosuppressed microenvironment during mammary gland involution (254, 256).

CHI3L1 has been implicated in multiple cancer types where it correlates with poor patient outcomes (239, 257–260). In TNBC, tumor-promotional M2-like macrophages exhibit increased secretion of CHI3L1 compared with tumor-suppressive M1-like macrophages (261). In lung cancer, CHI3L1 suppresses Th1 and cytotoxic T-cell responses to promote immunosuppression and metastasis (251). In the normal lung epithelium, the murine CHI3L1 homolog, BRP-39, prevents cell death of inflammatory cells via increased AKT signaling; CHI3L1 also increased Arg-1 activity and macrophage mannose receptor (MMR) expression, which are characteristic of M2-like macrophages (262). Cancer-associated fibroblasts in mammary tumors also upregulate CHI3L1, which then induces recruitment of macrophages and promotes tumor growth (263). Recently, CHI3L1 was reported to promote the expression of immune checkpoint proteins, including PD-L1, PD-L2, PD-1, LAG3, and TIM3, in melanoma. Genetic ablation of Chi3l1 (the murine gene encoding CHI3L1) increases CD8⁺ and CD4⁺ T-cell infiltration and promotes a Th1 phenotype via increased proinflammatory cytokine secretion, further supporting the role of CHI3L1 in protumor immune suppression (264). Collectively, these data suggest that tumor cells upregulate CHI3L1 to block the antitumor immune response, particularly via skewing the immune milieu toward a Th2 phenotype. This is similar to the Th1 to Th2 shift observed during fetal development, further demonstrating how tumor cells can co-opt mechanisms of immune suppression from maternal-fetal tolerance.

Prolactin inducible protein.

Another secreted factor involved in immune suppression is prolactin inducible protein (PIP; also called gross cystic disease fluid protein 15/GCDFP-15 and secretory actin-binding protein/SABP). PIP has multiple receptors, including actin, myosin, keratin, tropomyosin, immunoglobulin (Ig)-G, and human zinc-α2 glycoprotein (ZAG). PIP is a large component of amniotic fluid and breast milk (265), where it likely facilitates immune tolerance. PIP expression is positively regulated by both prolactin and androgen-bound androgen receptors and is negatively regulated by estrogens (266). PIP can also be promoted by the cytokines IL-1α and IL-6 (267, 268). Immune cells express PIP, including a specialized population of T cells, termed invariant natural killer T (iNKT) cells, that are involved in the recognition of lipid antigens and protect against autoimmunity. PIP expression on iNKT cells promotes DC maturation and differentiation in a Toll-like receptor (TLR)2-mediated manner, induces naïve T-cell maturation into Foxp3⁺ Treg cells, and reduces the number of Th1 and Th17 cells (269). Furthermore, PIP is also described as an antiapoptotic factor and is found to be regulated by STAT5A during mammary gland involution (270). Inactivation of STAT5A via glucocorticoids preserved the expression of PIP and inhibited apoptosis (270). For decades, PIP has been used to identify breast origin of metastatic tumors of unknown origin (271–273) and over 90% of breast cancers express PIP (273). PIP has mitogenic activity in breast and prostate cancers (274, 275). Surprisingly, in murine mammary carcinoma tumors, PIP expression reduces tumor growth and type 2 T-helper cells while simultaneously promoting protumor cytokines and metastasis (276), suggesting that PIP’s effects on tumorigenesis and immune alteration may be complex and at times conflicting. Overall, a more robust characterization of the role of PIP and its receptors during tumor-associated immune suppression is needed.

Invasion and Epithelial to Mesenchymal Transition

During fetal development, EVT cells must invade through the decidua to establish a blood supply by remodeling the spinal arteries. To facilitate this invasive process, EVT cells undergo EMT processes (277). During embryonic development, when epithelial cells invade surrounding tissue (for instance when trophoblasts invade the maternal uterine lining to seek a blood supply), a gene expression program is utilized to facilitate invasion. Our group discovered a common link between this invasive signature, steroid hormone receptor action, and immune suppression, all tightly controlled by small noncoding microRNA family miR-200. This miRNA family regulates trophoblast invasion in the uterus, and it is elevated in pathological conditions such as preeclampsia (278). Conversely, loss or silencing of miR-200 family members in carcinomas renders tumor cells more invasive by relieving repression of its direct targets and permitting expression of proteins mesenchymal-like markers such as ZEB that are not normally expressed in epithelial cells, and an increase in immune-suppressive factors including PD-L1 (279–281) and many of the factors discussed in this review. Since ZEB1 is a direct repressor of E-cadherin and other epithelial markers, these are often decreased or completely repressed when miR-200c is silenced.

EMT plays an important role in mammary gland involution where it aids in tissue remodeling and in the termination of milk production. The main overlap between EMT and involution is TGF-ß, which is known as a central inducer of EMT and contributor to involution (282, 283). TGF-ß is found to be highly expressed during pregnancy and greatly reduced during lactation because it inhibits the synthesis of milk protein (284). TGF-ß3 was found to promote loss of tight junction and adherens reorganization during mammary gland involution, which in turn helped mammary epithelial cell phagocytosis (284). A decrease in lactogenic hormones induced TGF-ß1 and AA, the main substrate of COX-2, expression in mammary epithelial cells (285). This relationship induced a mesenchymal-like phenotype via upregulation of vimentin and TWIST1 gene expression (285). Interestingly, TGF-ß1 was found to induce expression of AA synthesis, but AA did not induce TGF-ß expression in mammary epithelial cells (285). In breast cancer, TGF-ß has been shown to play a multitude of roles in immunomodulation, cancer stem cells, and metastasis. Exosomes secreted from mesenchymal stem cells (MSCs) in breast cancer contain TGF-ß and promote M2-like macrophage phenotypes, contributing to breast cancer progression (286). These MSC-derived exosomes also contain PD-L1 and semaphorins, which contribute to an immunosuppressive microenvironment (286).

SNAI2, an EMT transcription factor, is required for proper mammary gland involution (287). SNAI2-deficient mice had lower levels of activated STAT3, a key molecule during Phase I of involution (287). STAT3 is a key regulator of EMT and is known to be co-opted by tumors to promote progression (reviewed in Ref. 288). However, the link between EMT and the immunosuppressive environment found in involution has yet to be studied. There are underappreciated immune-suppressive characteristics of cancer cells that have undergone EMT to facilitate invasion into surrounding tissues and corresponding metastasis (reviewed in Ref. 289). Many of the aforementioned immunosuppressive phenotypes (e.g., PD-L1 and COX-2) have been linked to the ability of cancer cells to escape the primary tumor, travel in an anchorage-independent manner, and seed a secondary site.

Approaches to manipulate EMT in TNBC models revealed altered immune modulatory factors, including the cytokines GM-CSF and M-CSF, the enzymes TDO2 and HO-1, the checkpoint protein PD-L1, and other secreted proteins including CHI3L1 (281, 290–292). It has been proposed that these factors could work in tandem to facilitate the characteristic immune-suppressed state often observed in mesenchymal-like TNBC and that they could be inhibited to create an antitumor-skewed microenvironment and promote disease clearance (289). Future studies should utilize our understanding of immune suppression observed with these mesenchymal tumor models.

Additional Therapeutic Implications

Despite breast cancers being widely regarded as immunologically quiescent compared with other cancer types, there is a large amount of interest in identifying therapeutic strategies that harness the ability to reactivate the immune system and initiate tumor clearance. These include strategies to improve presentation of tumor antigens, to reduce inhibitory cytokines and immune regulatory cells, and to activate and expand effector T cells, NK cells, and DCs (as reviewed in Ref. 293). Although most current immunotherapy approaches harness the power of the adaptive immune system, lessons from maternal-fetal tolerance suggest that innate immunity is also highly effective in mounting a robust immune response. This could be particularly useful in cancer cases that have acquired resistance to adaptive immunotherapeutic agents, such as those that target PD-L1, PD-1, and CTLA-4. In line with this, novel approaches that harness innate immunity to induce antitumor immunity are being investigated (294). An example of this is anti-CSF-1R therapy, which blocks CSF-1 from binding to CSF-1R and preventing myeloid cell differentiation into immunosuppressive MDSCs and M2 macrophages. Early clinical trials have generated promising reports that reprogramming myeloid responses can effectively enhance immunotherapy (295, 296). Additional clinical trials are ongoing to assess anti-CSF-1R therapy in multiple malignancies, both along and in combination therapy approaches.

Just as the maternal immune system can recognize paternal antigens as foreign and cause fetal rejection, tumors can also express antigens that may be recognized as foreign by the host immune system. Tumor cells exhibit genomic instability and consequently express neoantigens, sometimes referred to as tumor-associated antigens, which can arise from genetic rearrangements, alternative splicing, or viral infections (297). For example, TNBCs produce the largely tumor-specific antigens MAGE-A and NY-ESO-1, which are proposed immunotherapeutic vulnerabilities of these tumors with efforts underway to target these antigens to induce a tumor-specific immune response (298, 299). Interestingly, NY-ESO1 is also expressed in adult and fetal testes (300). T cells recognize neoantigens presented by MHC molecules and can initiate a robust downstream immune response. Indeed, vaccines targeting tumor neoantigens are an emerging and promising area of cancer immunotherapy, and tumor neoantigen-targeted vaccines are now in clinical trials for multiple cancer types (297, 301–303).

It was recently reported that decidual NK cells can selectively target bacteria in infected trophoblasts by transferring the antimicrobial peptide granulysin through nanotubes, allowing decidual NK cells to fight infection and simultaneously preserve maternal-fetal tolerance (304). Soon after this report, it was reported that nanotubes can mediate the transfer of mitochondria from immune cells to cancer cells, which boosts the cancer cell metabolically while simultaneously metabolically depleting the immune cells (305). Although nanotubes were identified over nearly two decades earlier (306), they have only recently been proposed as a novel target for cancer immunotherapy. This is a prime example of immune-related mechanisms reported in the context of fetal tolerance that have soon after been identified as having therapeutic potential during tumorigenesis.

In summary, identification of new therapeutic approaches to activate antitumor immune responses during cancer progression is an active area of interest. Indeed, there are numerous drugs currently in clinical trials targeting different components of the immune system in multiple cancer types. Emerging understandings of fetal tolerance and mammary gland involution should offer insights into next-generation approaches to cancer immunotherapy.

CONCLUSIONS

There are numerous parallels between immune suppression during fetal development, mammary gland involution, and tumorigenesis (summarized in Fig. 2, A–C). During pregnancy, tightly regulated, transient immune suppression is a critical process that protects the allogeneic fetus from the maternal immune system until birth. During involution, the mammary gland and breast tissues create an immunosuppressive “wound healing” environment to aid in tissue remodeling and prevent an immune response to dead mammary/breast epithelial cells that have terminally differentiated after lactation. Similar mechanisms can be activated and co-opted by cancer cells, creating a chronic state of immune suppression that promotes tumor progression. We propose that further understanding of the fine-tuned, reversible process of immune suppression that facilitates normal fetal tolerance during pregnancy and involution will allow us to develop novel approaches to prevent tumor immune escape and metastasis.

Figure 2.

Parallels between the immune-suppressive milieu at the maternal-fetal interface, in the involution mammary microenvironment, and during breast tumorigenesis. Immune cells and immune-suppressive factors (e.g., TDO2 metabolites, HO-1 metabolites, cytokines, and other secreted factors including CHI3L1, SEMA7A, and PGE2) are shown in (A) the maternal endometrial decidua, (B) during mammary gland involution, and (C) in the breast tumor microenvironment. CHI3L1, chitinase 3-like-1; HO-1, heme oxygenase 1; MDSC, myeloid-derived suppressor cell; NK, natural killer; PGE2, Prostaglandin E2; SEMA7A, semaphorin 7a; TAM, tumor-associated macrophage; TDO2, tryptophan 2,3-dioxygenase. This figure was created with BioRender.com.

GRANTS

This work was supported by The Department of Defense W81XWH-19-BCRP-EA (to J.K.R.), HERA Ovarian Cancer Foundation Outside-The-Box grant (to L.S.C.), NIH/NCI R01 CA211696-01A1 (to T.R.L.), American Cancer Society (ACS) RSG-16–171-010CSM (to T.R.L.), The Department of Medicine Outstanding Early Career Scholars Award (to T.R.L.), and The University of Colorado Cancer Center Men For the Cure Award (to T.R.L. and J.K.R.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.S.C., K.T.K., J.K.R., and T.R.L. conceived and designed research; L.S.C., K.T.K., J.K.R., and T.R.L. drafted manuscript; L.S.C., K.T.K., J.K.R., and T.R.L. edited and revised manuscript; L.S.C., K.T.K., J.K.R., and T.R.L. approved final version of manuscript.