To fully identify myeloid cells as MDSCs, their functional activity needs to be tested; this is achieved usually by an in vitro suppression assay, which measures their ability to suppress the function of immune cells in tumor-bearing hosts. Activated MDSCs are implicated in the direct suppression of NK and B cells along with their primary target, T cells [13–15]. Immune suppression by MDSCs involves several mechanisms such as the increase of arginase 1 (Arg1) and inducible nitric oxide synthase (iNOS) production, the increased production of reactive oxygen and nitrogen species (ROS and peroxynitrite, PNT), and other immune-suppressive factors. Upregulation of Arg1 in MDSCs leads to the depletion and conversion of l-arginine, an essential amino acid needed for T cell proliferation, to urea and l-ornithine [16]. Under limiting amounts of l-arginine, upregulation of iNOS in MDSCs leads to the production of nitric oxide (NO) which reacts with superoxide and generates PNT [17]. Production of PNT by MDSCs causes the nitration and nitrosylation of the T cell receptor (TCR), thus disrupting potential CD8⁺ T cell-antigen interactions leading to T cell tolerance [18]. Also, PNT reduces the binding of antigenic peptides to MHC molecules on tumor cells and blocks T cell migration by nitrating T cell-specific chemokines, such as CCL2 [19, 20]. Another mechanism for suppression of T cell responses is the production of ROS by MDSCs. It has been shown that tumor-derived factors generate MDSCs that produced high levels of ROS which contribute to the suppressive activity of these cells [21]. Although these are the major mechanisms of immunosuppression by MDSCs, there are several other factors involved including transforming growth factor-β (TGF-β), interleukin-10 (IL-10), cyclooxygenase-2 (COX-2), indoleamine 2,3-dioxygenase (IDO), and many others. The main immunosuppressive cytokines produced by MDSCs are TGF-β and IL-10. Although TGF-β has shown to be produced by tumor cells, the production of this cytokine by MDSCs inhibits cytotoxic T cell responses in tumor-bearing mice [22]. MDSCs produced high amounts of IL-10 which impairs antitumor responses by inhibiting T cell activation [23]. In addition, MDSCs expressed several enzymes that are involved in the depletion of essential nutrition factors for T cell function. Upon stimulation by pro-inflammatory molecules, COX-2 is activated and induces Arg1 activity in MDSCs [16]. Other studies had implicated IDO expression in MDSCs which inhibit T cell function by depleting l-tryptophan and inducing T cell apoptosis within the tumor microenvironment [24]. MDSCs also block T cell function by depleting cysteine and impairing T cell activation in tumor [25]. In addition, MDSCs have the ability to induce the expansion of regulatory T cells (Treg) [26, 27]. Most of the mechanisms of suppression found to be implicated in MDSC function do not act simultaneously and are dependent on type of MDSC, type of tumors, and location of the cells.

Several recent studies have provided evidence that the ratio of PMN-MDSCs to M-MDSCs is really important because these cells utilize different mechanisms to suppress T cell responses [28]. For instance, M-MDSCs have the ability to suppress T cell activation in an antigen-specific and nonspecific manner. This suppression of T cell responses by M-MDSCs is associated with the increased expression of iNOS and production of NO [28, 29]. On the other hand, PMN-MDSCs are capable of suppressing immune responses primarily in an antigen-specific manner, which induces CD8⁺ T cell tolerance in the host. Immune suppression by PMN-MDSCs is associated with the increased expression of Arg1 along with high levels of ROS and PNT [8, 18]. Although PMN-MDSCs are more abundant than M-MDSCs in a tumor-bearing host, they are less immunosuppressive than M-MDSCs when compared on a per cell basis [30]. Location was also determined to be important in dictating the strength of suppression by MDSCs. Recent years have provided ample evidence indicating that MDSCs in the tumor microenvironment are more suppressive than MDSCs in peripheral lymphoid organs and peripheral blood [31, 32]. There is clear evidence now suggesting that the increased suppressive activity of MDSCs is regulated by the low levels of oxygen (hypoxia) found in tumor tissues [33]. However, more studies are needed to better understand what other tumor-associated factors and mechanisms are implicated in the potent suppressive activity by MDSCs within the tumor microenvironment (Fig. 10.2).

One of the main questions in the MDSC field is how is their expansion, accumulation, and activation regulated? Several years ago we proposed a two-signal model describing that MDSC accumulation requires two distinct types of signals [34]. This model is divided into two phases: (1) the expansion phase associated with the inhibition of their terminal differentiation and (2) the activation phase that is responsible in the conversion of immature myeloid cells into immunosuppressive MDSCs. However, we assert that these two phases partially overlap but are governed by different sets of transcription factors and intermediates (Fig. 10.3).

The first phase is mostly driven by tumor-derived growth factors along with STAT3, IRF8, C/EBPβ, RB1, notch, adenosine receptor A2b, and NLRP3.

As part of our two-signal model, the acquisition of the suppressive activity found in MDSCs is mediated by factors mostly produced by the tumor stroma including pro-inflammatory molecules such as IFNγ, IL-1β, IL-13, toll-like receptor (TLR) ligands, and others. Each of these molecules is involved in the activation of several signaling pathways associated with many factors such as NF-κB, STAT1, STAT6, prostaglandin E₂ (PGE₂) and cyclooxygenase (COX-2), high-mobility group box 1 (HMGB1), and hypoxia-inducible factor-1α (HIF-1α), all of which are implicated in the suppressive activity of MDSCs. Recently, the endoplasmic reticulum (ER) stress pathway has been linked to the suppressive activity of MDSCs found in cancer.

Recent studies have suggested that ER stress responses are involved in the suppressive behavior of MDSCs in tumor-bearing hosts. The ER stress response pathway, also known as the unfolded protein response (UPR), is responsible for preserving ER homeostasis during extrinsic or intrinsic stress [77, 78]. ER stress response is comprised of three ER-localized transmembrane protein sensors: inositol-requiring enzyme 1α (IRE1α), activating transcription factor 6α (ATF6α), and double-stranded RNA-dependent kinase (PKR)-like ER-related kinase (PERK). Activation of these ER sensors leads to the upregulation of several transcription factors including the spliced X-box protein 1 (sXBP1), CCAAT-enhancer-binding protein homologous protein (CHOP), and others. Both MDSCs isolated from tumor-bearing mice and cancer patients overexpress several markers of ER stress, including sXBP1 and CHOP, and display an enlarged ER—a hallmark of ER stress [79]. Tumor-bearing mice treated with thapsigargin, an ER stress inducer, have an increased number of MDSCs with potent suppressive activity [80]. We showed that induction of ER stress with thapsigargin converts normal human neutrophils to a suppressive PMN-MDSC phenotype in an IRE1α-/XBP1-dependent manner [12]. Besides the role of IRE1α/XBP1 in MDSC function, recent findings have implicated CHOP in the suppressive activity of tumor-associated MDSCs. For example, CHOP-deficient MDSCs lose their suppressive function and acquire an immune stimulatory phenotype [81, 82]. Although CHOP-deficient MDSCs did not have the ability to suppress T cells that were stimulated in an antigen-nonspecific manner, they retained their ability to block antigen-specific T cells [83]. At this moment, only the IRE1α/XBP1 and PERK/CHOP signaling pathways have been implicated in the function of MDSCs. However, more studies are needed to understand the specific mechanisms of ER stress responses that regulate the suppressive activity of MDSCs.

Over the years, our understanding of myeloid cells in cancer has shed light to their important role in tumor development, progression, and metastasis. The main three groups of terminally differentiated myeloid cells are macrophages, DCs, and PMNs. It is now clear that the tumor microenvironment alters myeloid cell differentiation by arresting these cells in an immature stage with potent immunosuppressive activity. Another myeloid cell involved in immune suppression is the tumor-associated macrophage (TAM) that arises from a tumor-infiltrating monocyte or M-MDSC within the tumor microenvironment. However, the relationship of MDSCs with other myeloid cells is still not clear. The main concern is the lack of specific markers for MDSCs making it difficult to discreetly identify these cells separate from monocytes and neutrophils. This is why in many previous studies, MDSC-like cells with suppressive activity were called monocytes and neutrophils. Recent data has provided evidence about the specific nature of MDSCs in cancer [1]. Immunosuppressive activity is an intrinsic feature of MDSCs. For instance, MDSCs could be generated in vitro from bone marrow progenitor cells in the presence of tumor-secreted factors. Unlike MDSCs, mature neutrophils or monocytes in the presence of these tumor-associated factors cannot suppress T cell responses in vitro [2]. Several studies, involving genomic, proteomic, and transcriptomic analysis, had provided evidence that supports the differences between PMN-MDSCs and neutrophils in tumor-bearing mice [84–88]. Moreover, a recent study provided information that suggests that PMN-MDSCs are different to neutrophils from healthy donors or cancer patients based on their gene profile [12]. In humans, PMN-MDSCs can be distinguished from neutrophils by LOX-1 expression. In mice the role of LOX-1 is not clear [3]. Within the hypoxic tumor microenvironment, M-MDSCs differentiate into TAMs, and this is regulated by HIF-1α and STAT3 activation [9, 33, 37, 38]. Both myeloid cells have potent immune-suppressive function, especially within the tumor sites; however, M-MDSCs can be distinguished from TAMs by changes in monocyte-/macrophage-associated markers. The main changes observed during M-MDSC differentiation into TAMs include an increased in F4/F80 and CD115, intermediate expression of Ly6C, and low expression of IRF8 and S100A9 protein [4, 89, 90]. MDSCs possess several biochemical features that are not observed in neutrophils or monocytes including high Arg1 and iNOS expression and activity as well as high levels of ROS and PNT production. In addition, recent studies have shown that MDSCs have an increase in the ER stress response that is linked to their suppressive function.

Among the different types of myeloid cells in cancer, there is a better understanding of the relationship between M-MDSCs and TAMs. Recent data from several groups have shown that M-MDSCs are TAM precursors within the tumor microenvironment. In addition, there are two polarized phenotypes of macrophages: the classically activated (M1 macrophages) with antitumor properties and the alternatively activated (M2 macrophages) with pro-tumor properties. The latter is more related to TAMs, but it is unclear if these two phenotypes are that of the same macrophage and what the molecular mechanism is that regulates them within the tumor microenvironment [91]. In addition, it is unknown if the polarization of macrophages is dictated by M-MDSCs before TAM differentiation. On the other hand, the nature of PMN-MDSCs and their relationship with neutrophils is a subject of discussion in different studies and an ongoing debate. Although PMN-MDSCs are widely accepted in the field, the concept of neutrophil polarization in cancer raises questions of whether these cells are pathologically similar to PMN-MDSCs [92]. Similarly to macrophages, the concept of tumor-associated neutrophils (TANs) consists of neutrophils with antitumor (N1 cells) or pro-tumor (N2 cells) properties within the tumor microenvironment [93]. Studies have implicated TGF-β and type 1 interferons (IFNs) as cytokines in the regulation of TAN plasticity in cancer [93–95]. Neutrophils by definition are short-lived and terminally differentiated cells, whereas PMN-MDSCs are immature cells. For this reason, it is hard to imagine that TANs could be polarized in the tumor microenvironment. Indeed, several studies have shown that these cells have the ability to suppress T cell responses in tumor-bearing mice, which supports these cells as PMN-MDSCs [96, 97]. However, the antitumor role of TANs has been described in cancer patients. Recently, Eruslanov et al. observed that TANs from early-stage lung cancer patients were not immunosuppressive, but rather stimulate T cell responses [98]. This study provides evidence for the role of TANs during early stages of tumor initiation. Only few studies have addressed the role of MDSCs in tumor initiation and how it regulates the immune responses at early stages. For instance, Ortiz et al. showed that the exposure of mice to cigarette smoke resulted in an accumulation of non-suppressive MDSC-like cells in various organs [99]. However, when cigarette smoke was combined with a carcinogen to promote the development of lung cancer, MDSCs presented potent immunosuppressive activity. These data suggest that MDSC function is controlled by the tumor microenvironment. Our understanding of the nature of MDSCs and their relationship with other myeloid cells during tumor development would open new therapeutic opportunities.

There is ample evidence that MDSC accumulation in peripheral lymphoid organs and tumor tissues correlates with a poor prognosis and clinical outcome in cancer patients [100]. In addition, MDSCs are implicated in resistance to anticancer therapies including the receptor tyrosine kinase inhibitor sunitinib as well as several chemotherapeutic drugs for lung cancer and multiple myeloma [101–103]. Since these cells have the ability to suppress antitumor immune responses, the success of cancer therapies, including immunotherapy, would depend on the immunosuppressive effect by MDSCs. Indeed, several clinical studies had shown an association with MDSC levels and response to immune checkpoint inhibitors, ipilimumab (anti-CTLA-4) and nivolumab (anti-PD-1) [104–107]. Today, the increased knowledge of the molecular mechanisms responsible for accumulation and function of MDSCs in cancer has allowed for the development of therapeutic strategies targeting this cell population. Some strategies for targeting MDSCs include (a) elimination of MDSCs, (b) blocking the accumulation of MDSCs, and (c) inactivation of MDSCs [108] (Fig. 10.4). Several of these strategies have been developed and are currently being tested in the clinic.

Elimination of MDSCs has shown to enhance antitumor efficacy of cancer immunotherapy including adoptive T cell transfer [109]. Conventional anticancer agents, in addition to their direct effect on cancer cells, have demonstrated the depletion of MDSCs in tumor-bearing hosts. Gemcitabine was the first chemotherapy agent reported to be capable of eliminating MDSCs in tumor models [110]. This study demonstrated that elimination of MDSCs by gemcitabine improves antitumor responses and enhances the effects of immune therapy resulting in tumor regression. Other chemotherapeutic drugs showing elimination of MDSCs include 5-fluorouracil, cisplatin, doxorubicin, and others [111–113]. Another strategy to eliminate MDSCs is by targeting TNF-related apoptosis-induced ligand-receptors (TRAIL-R) [79]. It was shown that upregulation of TRAIL-R, especially DR5, regulates MDSC survival in tumor-bearing host, and by using a TRAIL-R agonist, MDSCs were selectively eliminated leading to a decrease in tumor growth in a CD8⁺ T cell-dependent manner. Results from a phase 1 trial showed promising data that supports the use of TRAIL-R agonist in cancer patients [114]. Recently, a novel therapeutic approach to target MDSCs was developed by genetically fusing S100A9-derived peptides with the antibody Fc portion to generate a peptibody [115]. These peptibodies successfully depleted MDSCs from blood, spleen, and tumors of mouse models. However, future studies are needed to further elucidate the mechanisms of action that leads to MDSC elimination (Fig. 10.4).

MDSCs use several mechanisms to suppress antitumor immune responses, and targeting the suppressive machinery of MDSCs has been tested in cancer patients. Increased production of ROS and NO by MDSCs plays a major role in the suppression of CD8⁺ T cell responses. Inhibitors for the production of ROS using ROS scavengers N-acetylcysteine (NAC) and catalase are commonly used to test the function of mouse and human MDSCs ex vivo. NF erythroid 2-related factor 2 (NRF2) is a transcription factor involved in the activation of the antioxidant response and in protecting cells against damage caused by ROS. Upregulation of NRF2 by a synthetic triterpenoid was able to neutralize human MDSC activity by reducing production of ROS and dampening their suppressive function ex vivo [116]. However, a recent study from Beury et al. showed that Nrf2^−/− MDSCs had a decrease in their suppressive activity compared to Nrf2^+/+ MDSCs in two mouse models [117]. Scavengers of NO, such as carboxy-PTIO (C-PTIO), have been tested recently for MDSCs. Treatment with C-PTIO decreased the function of MDSCs and improved the efficacy of adoptive T cell transfer in tumor-bearing mice [118]. Besides blocking ROS and NO, another approach to inactivate MDSC function is to inhibit the catabolic enzymes upregulated by MDSCs including Arg1 and iNOS which are implicated in the suppression of T cell activation and proliferation. [63]. Inhibitors for Arg1 and iNOS have been used extensively in functional assays for MDSCs. Early studies by Rodriguez et al. demonstrated that a specific inhibitor for Arg1, N-hydroxyl-nor-l-Arg (nor-NOHA), could inhibit tumor growth and decrease Arg1 levels in MDSCs [63, 119]. Arg1 inhibitor, as well as iNOS inhibitor l-NG-monomethyl-l-arginine (l-NMMA), had shown to reduce MDSC function and improve T cell proliferation in vitro. Moreover, both Arg1 and iNOS levels in PBMCs of cancer patients have been shown to decrease in response to phosphodiesterase-5 (PDE5). Several PDE5 inhibitors had been used in the clinic to treat cancer and other conditions showing a decrease in MDSC numbers [120, 121]. Recently, a clinical report indicated that head and neck cancer patients treated with the PDE5 inhibitor tadalafil had a reduction in MDSCs and the expression of both Arg1 and iNOS, associated with an increase of T cells [122]. A parallel study demonstrated that treatment with tadalafil decreased not only MDSCs but also Tregs in blood and tumors of patients with head and neck squamous carcinoma [123]. Also, these patients treated with PDE5 inhibitor had an increase in tumor-specific CD8⁺ T cells. Nitroaspirin, a NO-releasing aspirin, has been shown to induce downregulation of both Arg1 and iNOS and PNT in MDSCs from a colon carcinoma model [124]. Since nitroaspirin was poorly effective as an adjuvant for adoptive T cell transfer, Molon et al. developed AT38 ([3-(aminocarbonyl)furoxan-4-yl]methyl salicylate) to inhibit reactive nitrogen species in MDSCs [20]. In this study, AT38 was shown to inhibit Arg1, iNOS, and PNT in MDSCs as well as inhibition of CCL2 nitration leading to an increase in CD8⁺ T cells within the tumor microenvironment. Another major mechanism of immunosuppression by MDSCs is the production of PGE2 through activation of COX-2 [125]. Several COX-2 inhibitors, including celecoxib, have demonstrated a reduction in PGE2 production and inhibition of Arg1 and iNOS in MDSCs from different mouse models [16, 17, 125]. Selective COX-2 inhibition by celecoxib reduces MDSC numbers and production of ROS and NO and improved a dendritic cell-based immunotherapy in a mesothelioma model [69] (Fig. 10.4).