Reciprocal Relationship between Myeloid-Derived Suppressor Cells and T Cells

Myeloid-derived suppressor cells (MDSCs) play a major role in the regulation of immune responses in cancer and many pathological conditions associated with chronic inflammation. These cells have a common myeloid origin, relatively immature state, common genetic and biochemical features, and, most importantly, the ability to inhibit immune responses. MDSCs consist of two main subsets: polymorphonuclear cells (PMN-MDSCs) and monocytic (M-MDSCs) cells (1, 2). The phenotype of these populations is now well defined in mice, and recently these cells were defined in cancer patients as well (3). PMN-MDSCs consist of relatively immature and pathologically activated neutrophils (4), whereas M-MDSCs consist of pathologically activated inflammatory monocytes. A small proportion of MDSCs is represented by precursors of myeloid cells, with the ability to form colonies in semisolid medium. It appears that, at least in cancer, M-MDSCs may play a central role in the development of immune suppressive myeloid cells. In the tumor site, they differentiate to tumor-associated macrophages with potent immunosuppressive activity, and in the periphery they may give rise to PMN-MDSCs (5, 6). The phenotype of MDSCs, their mechanisms of expansion, and the specific mechanisms by which they exert their suppressive effects are described in many reviews (3, 7–10). Initial studies of MDSCs were almost exclusively performed in tumor-bearing mice or in cancer patients. Cancer still remains the main focus of MDSC research. However, in recent years, it became increasingly clear that MDSCs play a critical role in the regulation of different types of inflammation not directly associated with cancer. It also became clear that the interaction of MDSCs with different populations of CD4⁺ T cells is not one-directional and goes beyond the simple direct immunosuppressive activity of MDSCs on T cells. These issues are discussed in this review.

Ample evidence favors the important functional role of MDSCs in various pathologic conditions associated with noncancerous inflammation. The priming of mice with CFA resulted in an expansion of MDSCs. These cells could subsequently be stimulated by activated T cells to produce reactive oxygen species (ROS) and NO (11). Mycobacterium bovis bacillus Calmette–Guérin (BCG) vaccination recruited NO-producing MDSCs. These cells were unable to kill BCG or the nonpathogenic Mycobacterium smegmatis, and they impaired T cell priming in the draining lymph node. The elimination of MDSCs by all-trans retinoid acid increased the number of IFN-γ–producing CD4⁺ T cells after vaccination with BCG (12). In individuals who received hepatitis B virus vaccine, GM-CSF augmented the preservation of peripheral blood MDSCs, which could contribute to the lack of improved vaccine responses (13).

Most chronic infections cause an expansion of M-MDSCs. Goñi et al. (14) found that during Trypanosoma cruzi infection, suppression was mediated through IFN-γ–dependent NO secretion by MDSCs. In lupus-prone MRL-Fas^lpr mice, MDSCs had a suppressive effect on CD4⁺ T cell proliferation, which was restored by an arginase 1 (Arg1) inhibitor (15). The MyD88-dependent expansion of MDSCs induced T cell suppression and Th2 polarization in sepsis (16). The administration of cerulean, which induces gallbladder contraction and the release of insulin, to MyD88^−/− mice resulted in severe pancreatitis, whereas this effect was much smaller in MyD88^+/+ mice. The number of IL-10–expressing MDSCs in cerulean-treated MyD88^−/− mice was significantly smaller than in the control MyD88^+/+ mice, which was associated with a reciprocal increase in the infiltration of CD4⁺ T cells (17).

In an inflammatory bowel disease model, the repeated transfer of Ag-specific T cells led to an increase in the frequency of NO synthase 2 (Nos2)– and Arg1-expressing MDSCs in spleen and intestine. The cotransfer of MDSCs with specific CD8⁺ T cells into mice ameliorated enterocolitis and suggested a direct immunoregulatory effect of MDSCs on the induction of inflammatory bowel disease by Ag-specific T cells (18). In inflammatory bowel disease induced by resveratrol, MDSCs also attenuated T cell proliferation and reduced the IFN-γ and GM-CSF production by lamina propria–derived T cells (19).

Multiple sclerosis is a demyelinating disease associated with an inflammatory immune response in the CNS. In a Theiler’s murine encephalomyelitis virus mouse model of multiple sclerosis, the depletion of M-MDSCs increased the virus-specific CD4⁺ and CD8⁺ T cell responses during the early virus infection, which were associated with an increased expression of IFN-γ and IL-17 and a decreased expression of IL-10 in the CNS (20). The in vivo transfer of MDSCs ameliorated the experimental autoimmune encephalomyelitis (EAE), significantly decreased demyelination, and delayed disease onset through the inhibition of encephalitogenic Th1 and Th17 immune responses (21).

MDSCs were shown to counter proinflammatory immune cells in the liver and adipose tissue during obesity. In obese mice, MDSCs suppressed proliferation of CD8⁺ T cells, induced their apoptosis, and skewed the differentiation of macrophages into insulin-sensitizing, alternatively activated M2 macrophages (22). Lysosomal acid lipase cleaves cholesteryl esters and triglycerides to generate free fatty acids and cholesterol in lysosomes. Lysosomal acid lipase deficiency causes an expansion of MDSCs, the loss of T cells, and an impairment of T cell function (23). MDSCs were essential for the IL-6–mediated protection of liver injury caused by an anti-CD137 Ab via inhibition of CD8⁺ T cell proliferation and IFN-γ expression (24).

MDSCs were implicated in the regulation of immune response during organ transplantation and graft-versus-host disease (GVHD). The data suggested that the expansion of MDSCs, together with regulatory T cells (Tregs), may be an important factor in the survival of cardiac allografts (25). The administration of recombinant G-CSF or IL-2 in mice resulted in the accumulation of MDSCs and Tregs in the peripheral lymphoid organs. This treatment significantly delayed MHC class II disparate allogeneic donor skin rejection (26). GVHD is the significant cause of morbidity and mortality following allogeneic bone marrow transplantation. It was shown that in minor histocompatibility, mismatched bone marrow transplantation is associated with the accumulation of MDSCs in blood, which peaked at week 3 and returned to the physiological level at week 12 (27). MDSCs, generated in vitro or in vivo, alleviated GVHD in murine allogeneic bone marrow transplantation models (28–30). The addition of functional MDSCs to the donor graft alleviated GVHD, whereas removal of MDSCs in vivo exacerbated GVHD. MDSC accumulation has been positively correlated with the severity of GVHD (31).

Recent reports have also implicated MDSCs in viral diseases. Patients with chronic hepatitis C virus showed a significant correlation between the MDSC levels, disease progression, and the response of patients to antiviral therapy. MDSCs suppressed T cell function in an Arg1-dependent manner (32). HIV and SIV infections induced a population phenotypically similar to M-MDSCs that expressed higher levels of STAT3 and NOS2 and a suppressed expansion of CD8⁺ T cells (33). In a large study of HIV-1–seropositive subjects compared with healthy controls, the presence of M-MDSCs in peripheral blood correlated with prognostic HIV-1 disease markers, including the HIV-1 load and CD4⁺ T cell loss. M-MDSCs from HIV-1⁺ subjects suppressed T cell responses in both HIV-1–specific and Ag-nonspecific manners (34). In a recent study, infections with an acute Armstrong or a chronic clone 13 strain of the lymphocytic choriomeningitis virus led to two distinct phases of innate immune response. Seven days after infection, there was an increase in immunosuppressive M-MDSCs and PMN-MDSCs in lymphoid organs and blood. This expansion was sustained only in the chronic clone 13 infection, whereas it occurred only transiently in acute Armstrong infection (35).

Thus, the role of MDSCs as an important negative regulator of immune responses is extended beyond cancer and observed in many pathological conditions. Although the immunosuppressive activity of MDSCs is the most prominent feature of these cells, ample evidence points to their role in the regulation of different populations of CD4⁺ T cells. Importantly, it appears that T cells can, in turn, regulate MDSC expansion and activity as well (Fig. 1). In this review, we discuss the interaction between MDSCs and specific subsets of CD4⁺ T cells.

In an early study, Terabe et al. (36) demonstrated that MDSCs can be activated to produce TGF-β in response to IL-13 in tumor-bearing mice. More recently, in a mammary adenocarcinoma model, IL-4–expressing CD4⁺ Th2 cells promoted expansion of MDSCs and tumor-associated macrophages. This enhanced pulmonary metastasis through activation of the epidermal growth factor receptor signaling in malignant mammary epithelial cells (37). Immune-mediated liver injury in hepatitis is caused by activated IFN-γ–producing Th1 cells. The accumulation of Th1 cells in liver was associated with the accumulation of MDSCs and suppression of T cell proliferation. TGF-β1–deficient mice acutely develop liver inflammation caused by Th1 cells. The rapid accumulation of MDSCs in TGF-β1–deficient liver was abrogated when mice were either depleted of CD4⁺ T cells or rendered unable to produce IFN-γ, demonstrating that Th1 cells can induce MDSC accumulation (38). In humans, LPS has been associated with protection from allergic diseases such as asthma. However, in mouse models of allergic asthma, a low dose of LPS promoted Th2 responses and allergic disease, whereas a high dose has been associated with suppression of allergic airway inflammation. The adoptive transfer of LPS-induced CD11b⁺Gr-1^intF4/80⁺ cells suppressed allergen-induced airway inflammation, suggesting that these cells may have regulatory functions in asthma. These cells were found to blunt the ability of the lung dendritic cells to upregulate GATA-3 or to promote STAT5 activation in primed Th2 cells (39). TLR7 was shown to modulate the accumulation of MDSCs during influenza A virus infections in mice. A lack of TLR7 signaling led to a Th2-biased response and an accumulation of MDSCs in the lungs (40). Trauma induced STAT6-dependent MDSC accumulation in spleens. This process was dependent on Th2-type cytokine release (41). Taken together, these data suggest that in contrast to Th1 cells, Th2 cells are directly involved in the expansion and activation of MDSCs, apparently via STAT6 (Fig. 1). The exact role of the specific cytokines (IL-4, IL-13, or others) and the molecular pathways responsible for this phenomenon remain to be elucidated.

The interaction between MDSCs and Tregs in cancer is well documented. In an initial study, Huang et al. (42) showed that Gr-1⁺CD115⁺F4/80⁺ MDSCs induced the expansion of Foxp3⁺CD25⁺ Tregs in vitro. Additionally, the adoptive transfer of CD115⁺Gr-1⁺ MDSCs induced IL-10– and IFN-γ–dependent Foxp3⁺CD25⁺ Tregs in vivo and suppressed the antitumor response in a mouse colon carcinoma model (42). Another study from the same group demonstrated that CD40 expression by MDSCs was required for MDSC-mediated Treg induction and tolerance (43). The Lewis lung cancer model showed increased MDSC and Foxp3⁺ Treg accumulation in the tumor tissue. After in vivo depletion of MDSCs, the number of tumor-infiltrating Tregs was significantly decreased, and this reduced tumor growth and prolonged survival of tumor-bearing mice (44). The Arg1-dependent induction of Tregs by MDSCs was found in a B cell lymphoma model (45). MDSCs may attract Tregs via various chemokines. Tumor-infiltrating M-MDSCs had significantly higher levels of CCL3, CCL4, and CCL5 as compared with the other subsets of MDSCs in lymphoma-bearing mice. Tregs from CCR5 knockout mice had a diminished ability to migrate toward chemokines secreted by M-MDSCs (46). It was suggested that infiltration of tumors by Tregs could be coordinated by mast cells and MDSCs. One study showed that mast cells could mobilize MDSCs to tumor and induce the production of IL-17 by MDSCs. IL-17 increased the level of CCL18 and CCL22 in tumor microenvironment, which attracted Tregs to tumor (47). Conversely, Treg depletion downregulated the production of IL-10 and the expression of PD-L1 in MDSCs from melanoma-bearing mice and promoted the MDSC conversion into a less immunosuppressive phenotype (48). The depletion of CD4⁺CD25⁺ Tregs abrogated the suppression activity of CD80⁺Gr-1⁺ MDSCs from mice bearing ovarian carcinoma (49).

There are some data suggesting interaction between MDSCs and Tregs in cancer patients. CD14⁺HLA-DR^−/low MDSCs from hepatocellular carcinoma patients induce functional CD4⁺CD25⁺Foxp3⁺ Tregs when cocultured with autologous T cells. The induction of Tregs was cell contact–dependent and was abrogated when MDSCs and T cells were separated (50).

There are now some indications that MDSCs and Tregs can interact in conditions other than cancer. M-MDSCs accumulated in lungs of mice with evolving experimental allergic airway inflammation were able to downregulate T cell activation, recruit Tregs, and dramatically decrease Ag-induced airway hyperresponsiveness (51). The MDSC-mediated expansion of Tregs and T cell suppression required MHC-dependent Ag presentation in a murine type 1 diabetes model in which the animals received CD4- hemagglutinin-TCR transgenic T cells. A significant reduction in the incidence of diabetes was observed in recipients receiving MDSCs plus influenza hemagglutinin, but not OVA peptide. The protective effects of MDSCs required an induction of anergy in autoreactive T cells and the development of Tregs (52). The administration of MDSCs in mice with pancreatic islet transplants was associated with attenuation of CD8⁺ T cells in grafts and a marked expansion of Tregs in a B7-H1–dependent manner (53).

The exact contribution of Th17 cells to tumor progression is not clear. Th17 cells were implicated in both tumorigenesis and in the eradication of established tumors. For instance, Th17 cells elicited neovascularization and promoted angiogenesis and tumor growth (54). Increased Th17 cell density within the tumors in patients with hepatocellular carcinoma correlated with microvessel density and poor prognosis (55). In contrast, it was reported that tumor-specific Th17 cells could mediate the destruction of advanced B16 melanoma (56). It appears that Th17 cells may play opposite roles depending on the stage of cancer. It has been shown that MDSCs could induce Th17 cell polarization from naive CD4⁺ T cells. The generation of Th17 cells by MDSCs was independent of MDSC/T cell contact but dependent on the cytokines secreted by MDSCs (57). Novitskiy et al. (58) found that the incubation of MDSCs with IL-17 increased the suppressive activity of MDSCs through the upregulation of Arg1, IDO, and cyclooxygenase-2. Consistent with that report, another study showed that MDSCs from IL-17R^−/− tumor-bearing mice expressed lower levels of Arg1, matrix metalloproteinase 9, and S100A8/A9 than from wild-type tumor-bearing mice, and they did not have an inhibitory effect on T cell proliferation (59). One study demonstrated rather different results. MDSCs reduced Th17 responses in an HLA-G⁺ xenotumor model. HLA-G induced the expansion of MDSCs and formation of the Th2-type cytokine environment rather than Th1 or Th17. However, no data were provided indicating whether those MDSCs were directly involved in the Th17 cytokine profile in the HLA-G⁺ tumor model (60).

MDSCs could drive a Th17 response that consequently contributes to the pathogenesis of experimental EAE. MDSCs from mice with EAE promoted Th17 cell differentiation under Th17-polarizing conditions. Th17 cell differentiation was mediated by IL-1 from MDSCs and required an IL-1 receptor on T cells. The depletion of MDSCs by gemcitabine reduced the frequency of Th17 cells in vivo and ameliorated EAE (61). Flagellin-induced MDSCs efficiently suppressed polyclonal T cell proliferation in a dose-dependent manner and substantially dampened released IL-17 protein by Th17 cells (62). However, in a clinical study, a negative correlation between increased circulating MDSCs and Th17 cells was found in the peripheral blood of patients with rheumatoid arthritis (RA). Compared with healthy controls, both the prevalence of circulating MDSCs and plasma Arg1 increased significantly in RA patients. However, no significant difference was observed in the mRNA level of NOS2 between RA patients and healthy controls. The frequency of Th17 cells in RA patients was significantly higher than in healthy controls but correlated negatively with the frequency of MDSCs and plasma Arg1 (63).

The complex nature of interaction between MDSCs and T cells contributed to the controversy associated with the role of Ags in the MDSC-mediated suppression of T cell responses. The fact that MDSCs can inhibit different types of T cell responses is widely accepted. It was demonstrated that MDSCs can inhibit Ag-specific CD8⁺ or CD4⁺ T cell responses (42, 64, 65). The suppression of MDSCs was mediated by cell-to-cell contact between MDSCs and T cells (64). Peroxynitrite (PNT) production by MDSCs during direct contact with T cells resulted in the nitration of the TCR and CD8 molecules, which induced conformational changes in these molecules and a loss of binding of the T cells. Ultimately, T cells are rendered nonresponsive to Ag-specific stimulation (66). PNT scavenger completely eliminated the MDSC-induced T cell tolerance, suggesting that ROS, and peroxynitrite in particular, could be responsible for MDSC-mediated CD8⁺ T cell tolerance. MDSCs are also reported to inhibit nonspecific immune responses. MDSCs from bone marrow or spleen from tumor-bearing mice significantly suppressed CD3/CD28-induced T cell proliferation (67–69). Human prostatic adenocarcinomas were reported to be infiltrated by terminally differentiated unresponsive cytotoxic T lymphocytes (70). A higher presence of nitrotyrosine in prostatic tumor–infiltrating lymphocytes suggested a local production of PNT. Thus, local PNT production could represent one of the important mechanisms by which tumors escape immune response.

The Ag-specific nature of MDSC-mediated immune suppression could be regulated by several factors: the type of MDSC involved, the local microenvironment, the state of T cell activation, and the retrograde signaling provided to MDSCs from T cells.

There is now enough evidence demonstrating that PMN-MDSCs and M-MDSCs use different mechanisms of immune suppression (71). The immunosuppressive activity of M-MDSCs is largely dependent on a high level of production of NO and different immunosuppressive cytokines and intermediates. There is a large body of literature indicating that these cells exert their suppressive activity in an Ag-independent manner (34, 72–74). In contrast, PMN-MDSCs are largely dependent on ROS, which require closer and more prolonged cell–cell contact and which are better provided during Ag-specific interaction (1, 2, 75, 76). This may explain the fact that PMN-MDSCs, in contrast to M-MDSCs, were implicated in Ag-specific T cell suppression. However, the type of MDSCs cannot fully explain the nature of immune suppression because several reports demonstrated that PMN-MDSCs could also inhibit the Ag nonspecific immune responses (74–79).

Several recent reports have demonstrated that MDSCs may exhibit different activities in peripheral lymphoid organs and in tumor tissues. We found that splenic MDSCs suppress only Ag-specific T cell response, whereas tumor MDSCs exerted a profound suppressive effect on both Ag-specific and nonspecific T cell responses. Splenic MDSCs displayed a significantly higher level of ROS than did tumor MDSCs, whereas tumor MDSCs had much higher levels of NO and Arg1 than did splenic MDSCs (6). A similar phenomenon exists in the peripheral blood and tumor MDSCs from patients with head and neck cancer. The data suggested that the tumor microenvironment converted MDSCs into nonspecific suppressor cells by upregulating Arg1 activity or NO production via HIF-1α (6). Recently Lesokhin et al. also demonstrated that CD11b⁺ MDSCs (mainly CCR2⁺CD11b⁺ M-MDSCs) from tumor tissues, but not from the spleens, were able to suppress the Ag nonspecific proliferation of CD8⁺ T cells induced by CD3/CD28 Abs in mouse melanoma model (74).

It was suggested that the state of T cell activation may determine the Ag-specific nature of immune suppression mediated by MDSCs (7). In most of the studies that investigated the nature of CD8⁺ T cell tolerance induced by MDSCs, T cells were activated by specific peptides. Therefore, this hypothesis needs to be formally tested. However, in recent study, the nonspecific activation of CD4⁺ T cells did not affect the Ag-specific suppression of these cells by MDSCs (80).

CD8⁺ T cell tolerance caused by MDSCs was mediated via MHC class I (66). MDSCs could induce Ag-specific CD4⁺ T cell tolerance via MHC class II (80). Because in most mouse tumor models expression of MHC class II on MDSCs was low (80), this mechanism apparently is operational only in few experimental systems. Similar variability in MHC class II expression was described in some human studies (81–84). This may explain some of the contradictory data regarding the effect of MDSCs on CD4⁺ T cell function. Ag-specific CD4⁺ T cells (but not CD8⁺ T cells) could dramatically enhance the immunosuppressive activity of MDSCs by converting them into powerful nonspecific suppressor cells. This effect was mediated through crosslinking of MHC class II on MDSC with subsequent upregulation of cyclooygenase-2 expression and PGE₂ production by MDSC (80), which were previously implicated in MDSC-mediated immune suppression (85–87). We suggest that activated Ag-specific CD4⁺ T cells may enhance the immunosuppressive activity of MDSCs and convert these cells into nonspecific suppressors, a mechanism that normally might serve as a negative feedback loop to control hyperactivated immune responses (Fig. 1). In cancer, this mechanism is hijacked by tumor cells and contributes to heightened immune suppression associated with tumor progression.

Recent years have brought understanding that MDSCs may play a critical role in regulation of immune responses, not only in cancer but also in many other pathologic conditions. It is clear that the interaction of MDSCs with T cells is not a one-directional where MDSCs inhibit T cell proliferation, cytokine production, or tumor cell killing. T cells can affect MDSC function in a major way by promoting their expansion and suppressive activity. Many questions regarding the molecular mechanisms of the complex interaction between MDSCs and T cells have remained unanswered. Understanding the nature of this interaction may help to develop more precise targeted therapy for many diseases.

The authors have no financial conflicts of interest.