Functional Diversity of Myeloid-Derived Suppressor Cells: The Multitasking Hydra of Cancer

Myeloid-derived suppressor cells (MDSCs) are gaining importance because of their key role in promoting resistance to current therapies in different types of cancer (1–3). In cancer, tumor-derived growth factors and inflammatory cytokines increase proliferation but prevent terminal differentiation of myeloid cells, resulting in increased frequencies of MDSCs (4). This is consistent with their initial characterization as immature cells of monocytic or granulocytic origin. They are distinct from monocytes and neutrophils because of their ability to suppress Ag-specific and non–Ag-specific T cells in many tumor models. Historically, MDSCs were considered inhibitory macrophages (5). However, MDSCs are phenotypically distinct from macrophages and dendritic cells (DCs) because of reduced expression of F4/80 (macrophage marker), MHC class II (Ag-presenting molecule) and CD11c (DC marker) (6). During their lifespan, some MDSCs differentiate to tumor-associated macrophages (TAMs) and DCs and others remain undifferentiated (7, 8) (Fig. 1). Although TAMs suppress T cells, they mainly promote an IL-4–dependent type 2 response, produce epidermal growth factor (EGF), and produce proteases to promote tumors (9). Therefore, MDSCs have a robust immunosuppressive activity among myeloid cells in cancer.

MDSCs were originally identified by the Gr-1 Ab (clone RB6-8C5), which recognizes both Ly6C (monocyte marker) and Ly6G (granulocyte marker) Ags (6). However, Gr-1 Ab binds with more affinity to Ly6G than to Ly6C. As a result, CD11b⁺Gr-1^lo cells are identified as monocytic MDSCs (M-MDSCs) and CD11b⁺Gr-1^hi cells as polymorphonuclear MDSCs (PMN-MDSCs) (10). Abs specific to Ly6C or Ly6G showed that M-MDSCs are CD11b⁺Ly6C⁺Ly6G⁻ and PMN-MDSCs are CD11b⁺Ly6G⁺Ly6C^lo (11). Recently, a consensus was reached about the criteria for classifying myeloid cells as MDSCs (11). Myeloid cells expressing Ly6C or Ly6G and capable of suppressing T cell proliferation or function are identified as MDSCs. Some studies report the function of total MDSCs (CD11b⁺Gr-1⁺), but others delineate subset-specific mechanisms of MDSCs (Table I). Evaluating the importance of MDSC subsets in dysregulating antitumor immunity is useful for defining effective treatment strategies. For example, in colon cancer, several reports show that M-MDSCs promote tumor growth by suppressing T cells (12–14), but recent studies show that PMN-MDSCs are also critical for tumor progression (15–17). In addition to M-MDSCs, an intermediate MDSC (I-MDSC) subset was found in spleens of tumor-bearing Apc^Min/+ mice (18), which have high circulating levels of GM-CSF (19). Compared with M-MDSCs, I-MDSCs express intermediate levels of Ly6C and Ly6G, suggesting a heterogeneous composition of cells within this subset (20). Consistently, an I-MDSC–like subset accumulated in mice with CT26 colon carcinoma overexpressing GM-CSF and in mice with breast tumor cells expressing high GM-CSF (6, 10). In lung cancer, patients with more M-MDSCs experienced disease relapse after treatment (21). However, in murine lung cancer, there is higher accumulation of PMN-MDSCs compared with M-MDSCs, and most studies report the suppressive potential of total MDSCs (22–24). Therefore, additional evidence will provide a complete picture of MDSC function in different tumor models.

MDSCs also express Tie2 and CD31 that are markers of endothelial cells (5, 6). Consistently in a colorectal cancer (CRC) tumor model (MC26), they acquire vascular endothelial growth factor (VEGF) receptor and VE-cadherin expression and incorporate into tumor endothelium (25). However, it is unknown if these MDSCs perform the function of endothelial cells. MDSCs from bone marrow (BM) of mice with 4T1 tumors induced bone metastases, suppressed T cells, and differentiated to osteoclasts to cause bone degradation (26). Absence of F4/80 showed that these osteoclasts are not macrophages. The ability of MDSCs to differentiate to other cell types is an indication of their plasticity but also suggests a stem cell–like feature, providing clues about how MDSCs multitask. Stem cells have the ability to differentiate into other cell types. As cells progress to a terminally differentiated state, some genes are silenced to acquire a specific function. In MDSCs, it is likely that the transcriptional programs determining whether MDSCs eventually differentiate into DCs or macrophages or neutrophils are all functional, allowing them to perform multiple functions in response to growth factors or cytokines. Apart from immunosuppression, other mechanisms also contribute to the protumor capabilities of MDSCs, as previously reviewed (27) (Table I). MDSCs thrive in a tumor environment rich in growth factors, cytokines, and chemokines (28). GM-CSF is a growth factor promoting MDSC accumulation, and it activates the transcription factor CEBP/β to control the expansion of MDSC subsets and its ability to suppress Ag-specific T cells (10). IL-6–activated Stat3 also contributes to the suppressive potential of MDSCs (29). A combination of growth factors and cytokines can lead to waves of gene expression changes in MDSCs locally in tumors or systemically to focus their cumulative immunosuppressive strength in preventing cytotoxic T cell expansion and simultaneously perform other tumorigenic functions. The tumor environment could shift the functional direction of each MDSC subset. However, the component of the tumor environment determining this dynamic balance in MDSC function is unclear at present, and understanding the underlying mechanisms is valuable. This concept suggests why MDSCs occupy an important niche in promoting tumor development in a variety of cancers and why it is a hurdle for effective immunotherapy.

To consider new strategies for targeting MDSCs in cancer, we aimed to gain better understanding of the functional diversity of MDSCs by evaluating all mechanisms by which MDSCs support tumors. In this article, we offer our perspective about the diverse roles of MDSC subsets in modulating T and B cell immunity and other tumorigenic functions.

The immunosuppressive function of MDSCs is critical for their classification and distinction from mature myeloid cells. IL-4Rα on MDSCs is linked to their immunosuppressive potential due to IL-4R–dependent activation of arginase and inducible NO synthase (iNOS) (12). Arginase prevents non–Ag-specific T cell proliferation by depleting essential nutrients such as arginine (12, 30). iNOS in M-MDSCs produces NO to suppress T cells by nitrating the TCR, which blocks downstream signaling events required for activating T cells (22, 31). IL-4Rα on MDSCs mediates suppression of both Ag-specific and non–Ag-specific T cells in breast tumor (4T1) and colon cancer (CT26) models (32). In murine melanoma (B16 tumors overexpressing IDO), M-MDSCs rather than PMN-MDSCs from tumors suppress non–Ag-specific T cells, express IL-4Rα and arginase, and correlate with tumor growth (30). Loss of autophagy-related genes in M-MDSCs increases MHC class II expression to promote CD8 T cell cytotoxicity and CD4 T cell proliferation in murine melanoma (33). In general, M-MDSCs are more suppressive than PMN-MDSCs (4). PMN-MDSCs primarily suppress Ag-specific T cells by producing reactive oxygen species (ROS) (22). NADPH oxidase (NOX2) produces ROS in MDSCs from colon cancer (CT26 and MC38), lung cancer (Lewis lung carcinoma [LLC]), mammary cancer (DA3), and fibrosarcomas (MethA and C3 sarcomas) (23). These findings demonstrate that M-MDSCs and PMN-MDSCs use different mechanisms for suppressing T cells (4). However, in EL-4 tumor-bearing mice, both MDSC subsets from spleen suppressed Ag-specific CD8 T cells similarly, and their suppressive function was independent of IL-4Rα (22). Similarly, peripheral blood MDSCs from different tumor models, such as 4T1, CT26, and TS/A (mammary carcinoma), were not dependent on IL-4Rα for suppressing Ag-specific T cells (34).

Programmed death ligand 1 (PD-L1) belongs to the B7 family of ligands that regulate T cells. Although blockade of PD-L1 reversed T cell suppression by MDSCs under hypoxic conditions (35), PD-L1 on in vitro–generated MDSCs or MDSCs from EL-4 tumor-bearing mice did not suppress T cells (22, 36). When coinhibited with other mediators of suppression, such as TGF-β or NO, PD-L1 appears to mediate T cell suppression (30). This suggests that other coinhibitory ligands in the B7 family, such as B7-H3, B7-H4, and V-domain Ig suppressor of T cell activation (VISTA), which are known to inhibit T cell proliferation, could mediate the suppressive function of MDSCs (37). Moreover, cyclooxygenase 2 (COX-2)–induced PG 2 (PGE₂) in MDSCs promotes immunosuppressive function by upregulating IDO among other mediators of T cell suppression (38). IDO is an important mechanism of immunosuppression preventing antitumor immunity (39). Therefore, more than one mechanism contributes to the suppressive function of MDSCs. A better understanding of mechanisms contributing to MDSC-mediated suppression of antitumor cells is vital for improving anticancer therapy and for the design and development of new therapeutics that can have a lasting effect on the survival and overall quality of life of cancer patients.

In murine colon cancer, MDSCs (Gr-1⁺ myeloid cells) but not macrophages increase IL-17 production by CD4⁺ T cells to promote tumor progression (40). Similarly, in a carcinoma model, MDSCs differentiated lL-17–producing CD4 T cells, whereas TAMs promoted TGF-β–producing T cells, which are likely T regulatory cells (Tregs) (41). We showed that I-MDSCs promote the development of IL-17–producing CD4 T cells, which is essential for intestinal tumor progression (20). CD4 T cells differentiate into Th1 cells, Th2 cells, Th17 cells, or Tregs, depending on the cytokines they are exposed to. The cytokines IL-1β, IL-23, and IL-6 differentiate CD4 T cells to disease-causing (pathogenic) Th17 cells in experimental autoimmune encephalitis compared with Th17 cells generated with IL-6 and TGF-β (42). Total MDSCs from ovarian cancer produced IL-1β, IL-23, IL-6, and TGF-β, which upregulate endogenous NO synthase 2 (NOS2) in T cells to promote IL-17 production (43). Addition of TGF-β to the cytokine mixture (IL-1β, IL-23, and IL-6) inducing Th17 cells dramatically decreased IL-17 production by CD4 T cells; however, these Th17 cells produced more IL-17 than unstimulated T cells. Therefore, MDSCs producing TGF-β promoted some IL-17 production, and the absence of TGF-β possibly increased IL-17 production (43). TGF-β also reduced expression of IL-23R, which suggests that it dampens the responsiveness of Th17 cells to IL-23 (42) and likely reduces the survival of Th17 cells in vivo. This is consistent with the necessity of IL-23 for maintaining pathogenic Th17 cells (44) and with the ability of IL-23 in increasing IL-17–producing CD4 T cells in the brain and spinal cord to promote disease severity in experimental autoimmune encephalitis (45). Therefore, MDSCs producing less TGF-β could promote Th17 cells with more disease-causing potential. These studies shed light on the ability of MDSCs to generate Th17 cells and suppress antitumor T cells to accomplish the overall aim of promoting tumor growth (Fig. 1). Moreover, MDSC-like cells regulate Th17 cells by recruiting them to tumors. In murine skin cancer, immature Gr-1⁺ myeloid cells lacking suppressive function recruited CD4 T cells via CCL4 to promote carcinogen-induced skin tumors by IL-17 (46). In castration-resistant prostate cancer (CRPC), PMN-MDSCs produced abundant IL-23, which activated the Stat3-RORγ pathway in tumor cells to promote survival (47). IL-23 produced by these PMN-MDSCs could also promote Th17 differentiation; however, it is unknown if this mechanism occurs concurrently in CRPC. Together, these studies illustrate the potential of MDSCs to alter the dynamic balance of the tumor immune system to create a self-promoting cycle.

In mismatch repair-proficient CRC patients, Tregs were associated with reduced cytotoxic response (48). Tregs were increased in CRC patients before surgery, suppressed Ag-specific CD4 T cells, and were associated with tumor recurrence (49). In a colon tumor model, M-MDSCs induced Tregs in vitro, and tumor growth was dependent on MDSC-induced Tregs (50). This study showed that MDSCs induce Tregs and suppress Ag-specific CD4 T cells simultaneously to promote tumors. In a B cell lymphoma model, induction of Tregs was dependent on the arginase activity of MDSCs (51). These tumor-derived MDSCs inhibited Ag-specific CD4 T cell proliferation but promoted the expansion of Tregs from naturally occurring Tregs and not from naive T cells. In this case, inhibition of T cell suppression allowed Treg expansion, suggesting that a shift in MDSC function could favor Treg expansion versus suppression of CD4 T cells. In murine melanoma, IDO-induced Tregs were critical for recruitment of MDSCs and their suppressive function (30). Together, these studies show that MDSCs promote Tregs, and Tregs in turn support MDSC function to promote an immunosuppressive environment conducive for tumor progression.

The functional relationship between MDSCs and B cells is interdependent, in which MDSCs regulate different B cell functions and vice versa. Regulatory B cells (Bregs) are a type of regulatory cell that primarily produce IL-10, accumulate in various cancers, and is one of the immunosuppressive mechanisms supporting tumor progression (52). In MCA205 fibrosarcoma, MDSCs promote Breg-like cells (B220⁺CD138⁺) that are dependent on IL-10 and TGF-β for producing IgA Abs (53). These Bregs are similar to regulatory plasma B cells that also produce Abs and suppress antitumor T cells (54). In a 4T1 breast cancer model, MDSCs appear to expand Bregs with high PD-L1 that produce IL-10 and IgA and also suppressed T cells by inducing apoptosis (55). MDSCs also nonselectively suppress B cells to disable another arm of antitumor immunity (54). In an orthotopic lung cancer model (LLC), MDSCs suppressed IL-4– and LPS-activated B cell proliferation in a T cell–dependent and –independent manner (56). Moreover, MDSCs restrict B cell trafficking to lymph nodes by downregulating l-selectin on B and T cells in 4T1 breast cancer (57). Together, these studies show that MDSCs regulate B cells in tumor systems.

CD19⁺CD20^lo Bregs expanded in mice with 4T1 tumors, in which B cells were depleted with CD20 Ab (58). In this study, Bregs promoted the generation of Tregs to support tumor growth and metastasis. On B cells, CD19 is expressed in early development stages and CD20 is expressed in more-differentiated B cells (44). This explains why depleting B cells with CD20 Ab deleted differentiated B cells while allowing less-differentiated CD19⁺ B cells to develop into Bregs (CD19⁺CD20^lo). In B cell–deficient mice with breast cancer or melanoma, MDSCs were poorly suppressive (59). In this model, tumor-evoked Bregs contributed to the suppressive potential of MDSCs, resulting in increased metastasis. These Bregs regulated the immunosuppressive function of MDSC partly through TGF-β signaling (60). Secretory IgM from B cells in mice with chronic lymphocytic leukemia and LLC tumors promoted the suppressive function of MDSCs (61). These studies show that in breast cancer and melanoma, Bregs contribute to MDSC function, and in fibrosarcoma, lung cancer, and breast cancer, MDSCs promote Bregs, suppress total B cell function, and reduce B cell trafficking. This illustrates the complex role of B cells in dysregulating immunity in cancer.

MDSCs produce IL-6, IL-1β, VEGF, and COX-2, all of which promote tumor cell proliferation (4, 62, 63). MDSCs cultured with murine 4T1, CT26, TS/A, or MC38 tumor cells produced more IL-6 compared with tumor cells themselves, showing that MDSCs are a source of IL-6 that supports tumor growth (62). This is consistent with the role of IL-6–induced Stat3 in inhibiting apoptosis to promote tumor cell survival (64). IL-1β promotes tumor cell proliferation by increasing angiogenesis in B16 melanoma and DA/3 mammary adenocarcinoma (65). This is accomplished by inducing chemokines to recruit inflammatory immune cells and promoting VEGF synthesis (66). In this manner, IL-1β activates endothelial cells by VEGF to vascularize tumors, resulting in tumor growth. In several cancers, VEGF is critical for developing tumor vasculature to supply additional growth factors to tumors and for migration of inflammatory and suppressor cells to tumors (67). Similarly, COX-2–induced PG 2 (PGE₂) increases angiogenesis and promotes proliferation of tumor cells (68). From these reports, it appears that besides regulating T cell function, MDSCs also produce cytokines, growth factors, and inflammatory mediators to promote tumor growth by preventing apoptosis, supporting angiogenesis, and expanding MDSCs.

Among the various functions of MDSCs, their immunosuppressive ability stands out as their definitive singular feature. M-MDSCs are most immunosuppressive per cell because of their ability to suppress Ag-specific or non–Ag-specific T cells. This was demonstrated in B16-IDO, C26-GM, CT26, and MCA26 tumor models and was associated with their ability to express both arginase and iNOS. In 4T1, C26-GM, or MCA203 fibrosarcoma tumors, CD11b⁺Gr-1^lo cells (M-MDSCs) are relatively more suppressive than CD11b⁺Gr-1^hi or CD11b⁺Gr-1^int subsets (6). PMN-MDSCs supplement this process by suppressing T cells by producing ROS, but their immunosuppressive strength is weaker than M-MDSCs (4, 6, 10). They suppress Ag-specific T cells and are more abundant than M-MDSCs in many tumor models (22). In murine colon cancer, total MDSCs promote Th17 cells, but in another colon tumor model, M-MDSCs expressing CSF-1R promote Tregs that are critical for tumor progression. It is unknown if PMN-MDSCs promote Tregs. PMN-MDSCs produce tumor-promoting factors such as IL-23 to support resistance by increasing tumor growth after surgical removal of prostate tumors (47). Because of IL-23 production, these PMN-MDSCs could promote Th17 cells. This suggests that MDSC subsets could differ in their ability to promote Tregs and Th17 cells. This also raises the question as to how these two processes coexist in the same cancer type. In cancer, Tregs promote tumors by suppressing T cells, and Th17 cells support tumor progression by promoting cell survival and angiogenesis. Occurrence of these processes in colon cancer demonstrates the multitasking capacity of MDSCs in regulating different arms of T cell immunity. These studies also suggest that different MDSC subsets promote Tregs versus Th17 cells in addition to their overall ability to suppress T cells. Although inhibition of suppressive function may not reduce their ability to promote Tregs or Th17 cells, more evidence is required to substantiate these assumptions conclusively.

By inducing epithelial-mesenchymal transition in 4T1 breast tumor cells, M-MDSCs promote the release and migration of tumor cells to metastatic niches (69). In this model, lung-infiltrating PMN-MDSCs upregulate genes associated with metastasis. MDSCs also suppressed cytotoxic CD8 T cells to promote lung metastases in 4T1 breast cancer (70). Consistently, lung metastases were reduced after resection of primary 4T1 breast tumors in Stat6-deficient mice, which was dependent on cytotoxic CD8 activity. A recent study showed that MDSCs suppress T cells by arginase activity to promote lung metastases independent of VEGF activity in a 4T1 breast cancer model (71). These reports suggest that MDSCs promote tumor metastasis by suppressing T cells and by promoting metastatic tumor cell transformation.

Taken together, M-MDSCs are highly immunosuppressive, induce the generation of Tregs, and promote metastasis. In a cancer setting, this is a potent resistance mechanism for preventing cytotoxic activity of antitumor cells both directly and through Tregs. PMN-MDSCs also suppress T cells, promote tumor growth, and metastasis. Therefore, M-MDSCs and PMN-MDSCs possess multiple mechanisms to promote tumorigenesis, supporting the idea that they are versatile and can adapt to suppress the host immune response in a demanding tumor environment (Fig. 1).

Immunotherapy is an effective strategy for treating most cancers. The success of checkpoint inhibition in increasing the survival of patients with aggressive lung cancer and melanoma has provided significant relief from high mortality (72). A subset of CRC patients with high mutation burden are also responsive to checkpoint inhibitor treatment (73). However, efficacy of this treatment is reduced because of the emergence of immunosuppressive mechanisms, including MDSCs and Tregs (3, 74).

Existing literature demonstrates the value of targeting MDSCs, and possibilities to block MDSC function in cancer to improve treatment are reviewed previously (28, 74). Efforts to inhibit MDSCs are focused on depleting MDSCs, inhibiting their function, or diverting their differentiation to activating DCs or tumoricidal macrophages (4, 28). M-MDSCs retain the potential to differentiate to macrophages or DCs and therefore are amenable to re-education to immunostimulatory myeloid cells that could prime T cells to generate antitumor T cells (22, 75). The close link between CSF-1R⁺ MDSCs that are similar to M-MDSCs and melanoma and colon cancer progression suggests that blocking CSF-1R could target M-MDSCs by inhibiting their proliferation (1, 50). In a tumor-free condition, PMN-MDSCs convert to neutrophils that have a short life span and might not result in long-lasting antitumor immunity, unlike Ag-presenting cells such as DCs or macrophages. Studies assessing if MDSC depletion reduces tumor growth use the Gr-1 Ab that primarily depletes PMN-MDSCs. Using Abs that recognize alarmins S100A8 and S100A9, both M-MDSCs and PMN-MDSCs were depleted, resulting in greater decrease of EL4 tumor growth, indicating the importance of M-MDSCs in tumor progression (76). However, these alarmins do not distinguish between PMN-MDSCs and neutrophils. Neutrophils are required for maintaining host immunity, which highlights the importance of finding PMN-MDSC–specific targets. A different approach to targeting PMN-MDSCs is by identifying subsets within PMN-MDSCs that are directly related to tumor development. In non–small cell lung carcinoma patients, PMN-MDSCs expressing lectin-type oxidized low-density lipoprotein receptor-1 (LOX-1) were more immunosuppressive than total PMN-MDSCs, suggesting that LOX-1 could differentiate between PMN-MDSCs and neutrophils (77). This finding suggests that PMN-MDSCs could be specifically depleted using LOX-1 without affecting neutrophils. In CRC, both M-MDSCs and PMN-MDSCs promote tumors. However, the specific mechanisms by which MDSC subsets alter T cells to promote tumor progression in CRC require further study. Although immunosuppression and induction of Tregs by these subsets could be contributing to disease progression, the association of IL-17 and MDSCs with reduced overall survival and the role of MDSCs in promoting Th17 cells in colon cancer pinpoint the importance of the mechanism by which MDSCs regulate Th17 cells (15). By inhibiting the ability of MDSCs in suppressing T cells and promoting Tregs and Th17 cells, there is a better likelihood of promoting antitumor immunity. This has significant potential for defining novel therapies for colon cancer by identifying targets in MDSC subsets.

Several clinical trials of commonly used chemotherapeutic and immunotherapeutic agents for targeting MDSC function in various cancers are currently ongoing (74). Despite these intense efforts to divert MDSC function to antitumor activity, their adaptability to support tumor progression is a hurdle that researchers are fighting to overcome. With recent success of immunotherapy in bringing unprecedented improvement in survival, there is widespread interest in finding more translatable therapies to boost efficacy and reduce toxicity. This exemplifies the vital need to target MDSC mechanisms to improve the efficacy of checkpoint inhibitors and surgical removal of tumors. By studying various facets of MDSC function and functional relationships with other immune cell populations, it will be possible to optimally block the tumor-promoting function of MDSCs.

We illustrate the multifarious roles of MDSCs in a tumor-promoting environment. Considering that host immune response is protective, MDSCs are a singular example of how tumors dysregulate antitumor immunity using different mechanisms. The immunosuppressive function of MDSCs has the ability to overpower antitumor immunity and is a signature feature of MDSCs. When other mechanisms of MDSCs are taken into account, we get an overall picture of how MDSCs multitask to accomplish the goal of promoting tumor development. The feedback between MDSCs and tumor cells creates a tumor-promoting cycle, which can be interrupted to identify immunotherapeutic targets. Several questions arise from this analysis of studies about MDSCs. First, M-MDSCs play a critical role in regulating the differentiation of T cells into Tregs in a colon cancer model. The relative contribution of M-MDSCs versus PMN-MDSCs in regulating tumor-promoting Tregs is unclear. Second, the critical component or process in MDSCs that promotes regulatory cells, such as Tregs and Bregs versus proinflammatory T cells such as Th17 cells, is unknown. Third, the interaction between MDSCs and Bregs in breast cancer and melanoma is studied, but their role in CRC is unknown. Depending on the type of cancer and related treatment regimens, targeting MDSCs could improve existing treatments and can serve as indicators of disease progression. Knowledge about MDSC mechanisms in each cancer type could provide critical input for devising targeted therapies to improve existing treatments and reduce the intensity of refractory mechanisms.

The authors have no financial conflicts of interest.