Tumor-related stress regulates functional plasticity of MDSCs

Abstract

Myeloid-derived suppressor cells (MDSCs) impair protective anti-tumor immunity and remain major obstacles that stymie the effectiveness of promising cancer therapies. Diverse tumor-derived stressors galvanize the differentiation, intra-tumoral expansion, and immunomodulatory function of MDSCs. These tumor-associated ‘axes of stress’ underwrite the immunosuppressive programming of MDSCs in cancer and contribute to the phenotypic/functional heterogeneity that characterize tumor-MDSCs. This review discusses various tumor-associated axes of stress that direct MDSC development, accumulation, and immunosuppressive function, as well as current strategies aimed at overcoming the detrimental impact of MDSCs in cancer. To better understand the constellation of signals directing MDSC biology, we herein summarize the pivotal roles, signaling mediators, and effects of reactive oxygen/nitrogen species-related stress, chronic inflammatory stress, hypoxia-linked stress, endoplasmic reticulum stress, metabolic stress, and therapy-associated stress on MDSCs. Although therapeutic targeting of these processes remains mostly pre-clinical, intercepting signaling through the axes of stress could overcome MDSC-related immune suppression in tumor-bearing hosts.

1. MDSCs: An Introduction

Cancer-associated immunosuppressive myeloid cells were first characterized by their lack of T cell, B cell, and natural killer (NK) cell phenotypic markers and by their capacity to impair the effector functions of lymphoid cells. This population of immunosuppressive cells would later come to be known as myeloid-derived suppressor cells (MDSCs) due to their shared myeloid lineage and pronounced immunoinhibitory potential [1] . Currently, combined phenotypic and functional characterization remains the consensus strategy for the identification of bona fide MDSCs. However, MDSCs exhibit marked functional and phenotypic heterogeneity, which highlights their complexity, developmental trajectory, and close relationship with other myeloid subsets. The current classification of MDSCs encompasses both monocytic (M-MDSCs) and polymorphonuclear (PMN-MDSCs) subsets, as well as the recently identified early precursor MDSCs (eMDSCs) group in humans [2]. Although each MDSC subpopulation is identified by the differential expression of specific markers, these distinct subsets are unified in their ability to impair anti-tumor immune responses and promote tumor growth. Across multiple cancer types, elevated frequency of MDSCs has been linked to poor prognosis and impaired responsiveness to treatment [3–9]. The detrimental clinical impact of MDSCs is also apparent in patients receiving immunotherapy, as MDSCs have been reported to limit the efficacy of checkpoint blockade and chimeric antigen receptor (CAR) T cell therapies [10, 11]. Although significant research effort has been devoted to therapeutically targeting MDSCs, there are currently no approved clinical strategies to efficiently overcome the impact of MDSCs in cancer patients.

2. MDSC functions: An arsenal of immunosuppressive programs

MDSCs exert their immunoinhibitory effects through multiple distinct immunosuppressive mediators, including enzymatically produced reactive oxygen (ROS) and reactive nitrogen species (RNS), depletion of extracellular amino acids L-arginine, L-cystine and L-tryptophan, secretion of immunosuppressive cytokines, and production of chemokines and chemotactic factors, among others [12, 13] (Table 1). Collectively, these diverse immunosuppressive drivers enable MDSCs to impair multiple aspects of adaptive anti-tumor immunity by inhibiting T cell survival, proliferation, differentiation, and effector functions. Each immunosuppressive system can elicit multiple mechanistic effects based on the cell type(s) exposed. Additionally, multiple upstream axes of stress may converge on the induction of common immunosuppressive programs. Consequently, for clarity, the immunosuppressive drivers summarized below are categorized based on the proximal pathways impacted by each program, including suppressive mediators that induce depletion or sequestration of amino acids; regulatory systems that produce ROS or RNS; receptor-ligand based suppressive interactions; chemokines, cytokines; and other immunosuppressive agents.

Table 1:

Immunosuppressive mediators of MDSCs

	Identity (abbreviation)	Mechanism of action	Effect on antitumor immunity	Impact on stress axes	Citations
Amino acid depletion/sequestration	Arginase 1 (ARG1)	ARG1 catalyzes hydrolysis of L-arginine resulting in production of ornithine and urea	ARG1-induced L-arginine depletion impairs interferon gamma (IFNγ) production, CD3 zeta chain expression and cell cycle progression of T cells	Induces increased signaling through the metabolic stress axis	14–16
	Indoleamine-pyrrole 2,3-dioxygenase (IDO)	IDO catalyzes generation of L-kynurenine from amino acid L-tryptophan	Depletion of L-tryptophan by IDO impairs T cell activation and survival while promoting differentiation of regulatory T cells and tolerogenic dendritic cells	Induces increased signaling through the metabolic stress axis	21
	Cystine/glutamate xCT antiporter (XCT; SLC7A11)	XCT directs uptake of extracellular cystine via cystine/glutamate exchange	XCT decreases extracellular levels of cystine which impairs DNA synthesis and glutathione production by T cells resulting in stunted T cell proliferation	Induces increased signaling through the metabolic stress axis	18–20
	Methyl-glyoxal	Methyl-glyoxal induces glycation of intracellular L-arginine decreasing free L-arginine levels	Transfer of methyl-gloxyl from MDSCs to T cells depletes intracellular levels of L-arginine and impairs T cell metabolic fitness	Induces increased signaling through the metabolic stress axis	17
ROS/RNS production	Inducible nitric oxide synthase (iNOS; NOS2)	iNOS catalyzes conversion of L-arginine into citrulline and nitric oxide	iNOS depletes L-arginine and decreases production of IL-2; peroxynitrite derivatives induce nitrotyrosination resulting in impaired T cell receptor (TCR) activation and mitochondrial function	Increases signaling through the oxidative/nitrosative stress and metabolic stress axes	14–16, 22–26
	NADPH oxidase 2 (NOX2}	NOX2 catalyzes generation of superoxide from oxygen and NADPH	Superoxide generated by NOX2 induces macromolecular damage and depletes T cell stores of glutathione	Increases signaling through the oxidative/nitrosative stress axis	25, 26
Altered Receptor-ligand signaling	A disintegrin and metalloproteinase (ADAM) metallopeptidase domain 17 (ADAM17)	ADAM17 cleaves multiple targets including T cell homing receptor CD62L	Cleavage of CD62L by ADAM17 impairs homing of T cells to the tumor and lymph nodes	Increases signaling through the ER stress and hypoxic stress axes	28, 29, 35
	Programmed death-ligand 1 (PD-L1; CD274)	PD-L1 engages with programmed cell death protein 1 (PD1) expressed on lymphocytes	Binding of PD-L1 to PD-1 on T cells inhibits TCR activation, T cell proliferation, decreases cytokine production and impairs T cell survival.	Increases signaling through the chronic inflammatory stress axis	30–33
	Galectin-9 (GAL9; LGALS9)	GAL9 binds to T-cell immunoglobulin and mucin-domain containing-3 (T1M3) on lymphocytes	Interaction of GAL9 with TIM3 impairs development of T helper type 1 (Th1) immunity , enhances development of MDSCs and promotes resistance to anti-PD-1 checkpoint therapy	Increases signaling through the chronic inflammatory stress axis	30,34
Cytokines, chemokines and other suppressive mediators	Ectonucleoside triphosphate diphosphohydrolase-1 (ENTPD1; CD39) and 5’-nucleotidase (NT5E; CD73)	CD39/ENTPD1 and CD73/NT5E work serially to convert extracellular ATP into adenosine	Increased extracellular levels of adenosine promotes differentiation of tumor associated macrophages (TAMs), VEGF production and MHC I/II down regulation	Increases signaling through the chronic inflammatory stress axis	40
	S100 calcium-binding protein A8 (S100A8) and S100 calcium-binding protein A9 (S100A9)	S100A8 and S100A9 binding to toll ligand receptor 4 (TLR4) and activate nuclear factor kappa B (NFKB)	TLR4 induced NFKB activation directs expression of adhesion molecules and secretion of inflammatory cytokines	Increases signaling through the chronic inflammatory stress axis	38, 39
	Transforming growth factor beta (TGFβ)	TGFB binds and signals through TGF8 receptors on T lymphocytes and myeloid cells	TFGB signaling promotes development or regulatory T cells (T regs)and impairs proliferation of CD8+T cells	Increases signaling through the chronic inflammatory stress axis	36
	Tumor necrosis factor alpha (TNFα)	TNFa stimulates increased autocrine production of TNFa by MDSCs	Autocrine TNFa directs expression of iNOS in MDSCs which enables further suppression of target cells	Increases signaling through the chronic inflammatory stress axis	37

Multiple mediators direct the immunosuppressive capabilities of MDSCs. Table 1 documents some of the diverse mediators that enable MDSC function as well as the known mechanisms of actions and effects that these mediators have on antitumor immunity.

Multiple mediators that coordinate the suppressive effects of MDSCs hinge on depletion or sequestration of amino acids. Limiting the available pool of accessible amino acids impairs multiple aspects of T cell biology, reinforces the hostile metabolic milieu within the tumor microenvironment (TME) and further amplifies signaling through the metabolic stress axis (discussed in 3.5 below). A well-defined aspect of MDSC-mediated immunosuppression entails metabolizing the amino acid L-arginine via arginase 1 (ARG1) or inducible nitric oxide synthase (iNOS). Both ARG1 and iNOS utilize L-arginine as a substrate to produce L-ornithine and urea or nitric oxide and citrulline, respectively. L-arginine starvation impairs T cell proliferation, decreases production of T cell effector molecules, inhibits expression of CD3ζ, and arrests cell cycle progression to collectively induce severe T cell dysfunction [14–16]. Furthermore, the nitric oxide produced by iNOS decreases the stability of interleukin 2 (IL-2)-encoding mRNA and hinders the release of IL-2 by activated leukocytes [15]. Intracellular levels of arginine can also be depleted by MDSC-derived metabolite methyl-glyoxal, which jointly inhibits antitumor T cell responses by restricting amino acid availability and disrupting glycolysis in CD8⁺ T cells [17]. In addition to directing L-arginine starvation, MDSCs actively sequester L-cystine via elevated expression of cystine/glutamate antiporter xCT (XCT; SLC7A11) and induce depletion of tryptophan via indoleamine-pyrrole 2,3-dioxygenase (IDO) [18, 19]. XCT enables internalization of extracellular L-cystine by MDSCs which both limits L-cystine availability and impairs the production of derivative L-cysteine by other immunostimulatory antigen presenting cells. In the absence of L-cystine and L-cysteine, activated T cells are incapable of synthesizing antioxidant glutathione and become unable to successfully replicate DNA, resulting in impaired proliferation and T cell activation [20]. IDO catalyzes reduction of extracellular L-tryptophan to produce kynurenines. Depletion of L-tryptophan stunts T cell proliferation, while kynurenine exposure promotes regulatory T cell expansion [21]. Collectively, the MDSC immunosuppressive mediators that result in amino acid sequestration or depletion contribute to the severe metabolic stress encountered within the TME.

An additional common feature of multiple MDSC immunosuppressive mediators is the production of RNS or ROS. In addition to promoting depletion of L-arginine, nitric oxide generated by iNOS can further react with superoxide to generate highly reactive peroxynitrite, a major RNS. Peroxynitrite reacts with multiple T cell macromolecules, including components of the T cell receptor (TCR), proteins within the mitochondrial electron transport chain, and DNA. Reaction with peroxynitrite and formation of nitrotyrosine adducts ultimately impairs TCR signaling and induces T cell dysfunction, which may culminate in T cell apoptosis [22–24]. Production of superoxide via NADPH oxidase 2 (NOX2) has also been implicated in mediating MDSC immunoregulatory functions and antigen-specific T cell immunosuppression [25]. ROS and RNS released by MDSCs blunt TCR signaling, inhibits T cell survival, and promotes ROS and/or RNS-responsive transcriptional programs [25, 26]. Furthermore, ROS-induced activation of nuclear transcription factor erythroid-derived 2-like 2 (Nrf2) further supports MDSC immunosuppressive function via activation of the oxidative stress axis (discussed in 3.1) [27].

Receptor-ligand based suppressive mediators act as an additional means by which MDSCs inhibit antitumor immunity by impairing ligand-receptor interactions and inducing cell contact-dependent signaling. Mediators that act through impairment of receptor ligand interactions include the disintegrin and metalloproteinase metallopeptidase domain 17 (ADAM17), which induces cleavage of the homing receptor L-selectin (CD62L) on T cells. Cleavage of CD62L impairs interactions between CD62L and O-glycan containing ligands and ultimately inhibits migration of T cells to tumor sites [28, 29]. Cell contact-dependent suppressive mediators include checkpoint receptors, such as programmed death-ligand 1 (PD-L1; CD274), which induces T cell exhaustion and promotes T cell dysfunction [30–33]. Additional receptor-dependent mediators include, galectin-9 (LGALS9) that binds T cell localized T-cell immunoglobulin and mucin-domain containing-3 (TIM-3; HAVCR2) to induce T cell exhaustion, resistance to programmed cell death protein 1 (PD-1) targeted therapy, and further expansion of MDSC populations [30, 34]. Interestingly, both the endoplasmic reticulum (ER) stress and hypoxic stress axes have been reported to provoke ADAM17 upregulation in cancer cells, suggesting that a similar process may occur in MDSCs [35].

Finally, production of chemokines, cytokines, and other suppressive mediators by MDSCs provokes further impairment of antitumor immunity. MDSCs promote differentiation of uncommitted T cells into regulatory T cells via secretion of transforming growth factor beta (TGFβ) [36]. Secretion of tumor necrosis factor alpha (TNFα) by MDSCs provokes an inflammatory autocrine loop that triggers further accumulation of MDSCs and iNOS expression. The propensity for proinflammatory TNFα to induce a feedforward signaling loop underlines the capacity for MDSC suppressive mediators to perpetuate signaling through the chronic inflammatory stress axis [37]. In addition to cytokines, MDSCs secrete chemotactic factors such as S100 calcium-binding proteins A8 and A9 (S100A8 and S100A9) that promote MDSC trafficking to the tumor site and further stimulates release of inflammatory cytokines via toll ligand receptor 4 (TLR4) [38, 39]. Furthermore, expression of both CD39 (also known as ectonucleoside triphosphate diphosphohydrolase-1 (ENTPD1)) and CD73 (also known as 5’-nucleotidase (NT5E)) by MDSCs enables conversion of extracellular ATP to adenosine. Accumulation of extracellular adenosine in turn activates differentiation of M-MDSCs into highly immunosuppressive tumor-associated macrophages (TAMs) and decreases the expression of major histocompatibility complexes I and II (MHC-I, MHC-II) on tumor-infiltrating dendritic cells [40]. Collectively, the cytokines, chemokines and additional suppressive mediators produced by MDSCs may further promote inflammation and thus drive signaling through the chronic inflammatory stress axis (discussed in 3.3).

Understanding and therapeutically targeting the immunosuppressive mediators that underwrite MDSC suppressive function remains an active area of immunotherapy and oncoimmunology research. The diversity of immunosuppressive mediators underlines the functional heterogeneity of MDSCs. Additionally, the potential for immunosuppressive mediators to further stimulate signaling through various stress axes underscores the complex relationship between cellular stress and MDSC suppressive function. Altogether, MDSCs indisputably play a pivotal role in mediating cancer-induced immune dysregulation. The multifarious immunosuppressive mediators that MDSCs use to alter antitumor immunity offer both an opportunity and challenge for therapeutic targeting.

3. The axes of stress: Tumor-associated stress signals drive MDSC development and function

MDSCs are in nature immature myeloid cells that exit the bone marrow prematurely as the result of emergency myelopoiesis (Figure 1). In the context of infection, transitory acute stress, and physical injury, the induction of emergency myelopoiesis is essential and provides a consistent supply of myeloid cells to resolve infection or injury. However, during periods of chronic unresolved stress, like that found in cancer, emergency myelopoiesis results in MDSC expansion. MDSCs arise from common myeloid progenitors (CMPs) activated in response to multiple drivers of myelopoiesis, including granulocyte-macrophage colony-stimulating factor (GM-CSF) and granulocyte colony stimulating factor (G-CSF). After their exit from the bone marrow, tumor-associated signals dramatically impact the differentiation and immunoregulatory function of M-MDSCs and PMN-MDSCs [41–43]. Infiltration into tumor beds and exposure to tumor-linked stressors further increases the immunosuppressive capabilities of MDSCs through a multi-signal process mediated by the collective actions of signal transducer and activator of transcription 3 (STAT3), nuclear factor kappa B (NF-KB), 5’-prime-AMP-activated protein kinas alpha (AMPKα), hypoxia-inducible factor 1 alpha (HIF1α), and CCAAT/enhancer-binding protein (C/EBP), among others [44]. How tumor-associated ‘axes of stress’ impact MDSC development, expansion, and immunosuppression remains an area of active investigation. Our summarized understanding of these axes of stress and how they shape MDSC biology is detailed below.

Figure 1: MDSCs originate via emergency myelopoiesis and undergo suppressive polarization in response to cellular stress events.

MDSCs arise from myeloid precursors in response to cytokines such as G-CSF or/and GM-CSF, among others (Signal 1). Upon bone marrow exit and infiltration to tumor beds, MDSCs are exposed to multiple axes of cellular stress, including signals related with the oxidative/nitrosative stress axis, the hypoxic stress axis, the chronic inflammatory stress axis, the ER stress axis, the metabolic stress axis, and therapy-associated stress axis. These conditions polarize MDSCs into highly immunosuppressive subsets (Signal 2).

3.1. ROS and RNS Stress Axes

The elevated metabolic rate of tumor cells within the TME results in the accumulation of electrophilic ROS and RNS [45]. Additionally, chemotherapy and radiation treatment may also induce ROS and RNS accumulation within the TME [46]. ROS and RNS react with proteins, lipids, and DNA to generate adducts that impair macromolecular function and induce cellular damage. In order to survive, cells have developed a series of ROS and RNS mitigating antioxidant programs to prevent lethal oxidative or nitration-related damage. The primary ROS/RNS-responsive signaling pathways are regulated by several proteins, most notably the transcription factor Nrf2. After exposure to ROS or RNS, Nrf2 undergoes dissociation from its cytosolic negative regulatory partners ubiquitin-binding protein p62 (p62, SQSTM1) or Kelch-like ECH associated protein 1 (Keap1) and translocates to the nucleus [47]. Once in the nucleus, Nrf2 heterodimerizes with small Maf (sMaf) proteins to mediate transcription of genes containing antioxidant response elements (AREs) [48]. Beyond direct activation by ROS or RNS, several signaling pathways have also been reported to induce Nrf2 activation, including glycogen synthase kinase 3 beta (GSK3β), protein kinase C (PKC), extracellular signaling related kinase (ERK), and protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK; EIF2AK3) [48–55]. Nrf2 activation and subsequent binding to ARE sequences direct the expression of enzymes such as heme oxidase 1 (HO-1; HMOX1), nicotinamide adenine dinucleotide phosphate (NADPH) quinone dehydrogenase 1 (NQO1), catalase (CAT), and superoxide dismutase (SOD), which detoxify cellular ROS or RNS. Nrf2 also induces expression of proteins like glutathione synthase (GST) and thioredoxin (TXN) that synthesize or act as antioxidants to buffer the toxic effects of intracellular ROS and/or RNS [56, 57].

Exposure to high levels of ROS and/or RNS triggers Nrf2-induced gene expression, which critically enables MDSC survival within the hostile TME (Figure 2A). Due to the intrinsic high production of ROS and RNS via iNOS and NOX2, MDSCs require Nrf2 activation to avert self-inflicted macromolecular damage from their immunosuppressive mediators. In addition to promoting survival, Nrf2 signaling contributes to MDSC expansion and function. While MDSCs that lack Nrf2 show impaired survival and immunosuppressive capacity, mice with genetic ablation of Nrf2’s negative regulator Keap1 exhibit enhanced expansion of potently regulatory MDSCs [58, 59]. Beyond targeting Nrf2 itself, inhibiting Nrf2-regulated genes, such as HO-1, has also been reported to reduce the immunosuppressive capacity of MDSCs [60]. Nrf2 activation also induces expression of XCT, which allows MDSC to uptake L-cystine and induce L-cystine deprivation-dependent T cell dysfunction. In addition to inducing pro-survival antioxidant responses, Nrf2 activation promotes the expression of pathways that support MDSC function. Induction of peroxisome proliferation activated receptor gamma (PPARϒ) by Nrf2 directs upregulation of ROS production in MDSC, while regulation of C/EBPα by Nrf2 enhances MDSC proliferation, expansion, and the accumulation of circulating immature myeloid precursors [56]. Additionally, signaling via Nrf2 may also occur in response to ER stress wherein mediator PERK directs activating phosphorylation of Nrf2. The capacity for signaling through one stress axis to induce signaling through another axis underlines the interdependent nature of the stress axes signaling. These results underline the key role of Nrf2 in regulating MDSC biology, promoting MDSC survival, and enhancing MDSC immunosuppressive function in the TME [61].

Figure 2: Tumor-associated stress axes drive expansion, survival, and immunosuppressive polarization of MDSCs in tumor-bearing hosts.

Signaling through the oxidative/nitrosative stress axis (2A), promoted by elevated intratumoral levels of ROS, therapy-induced ROS, and intrinsically ROS/RNS produced by MDSCs, induces activation of Nrf2 to transcriptionally drive oxidative homeostasis, prosurvival programs, and secondary mediators AHR, PPARγ, and CEBPα. Tumor-linked intratumoral hypoxia prompts hypoxic stress axis (2B) signaling, which induces activation of the HIF family transcription factors. HIF transcription factors induce increased expression of checkpoint markers, promote trafficking of MDSCs to tumor beds, and promote MDSC survival and differentiation. Chronic inflammatory stress axis (2C) activation is induced by multiple chronic inflammatory mediators that collectively drive activation of STAT3 and NFKB. STAT3 and NFKB jointly coordinate MDSC survival and expansion, enhance MDSC suppressive activity, and induce expression of both iNOS and ARG1. Activation of the ER stress axis (2D) occurs in response to misfolded protein accumulation, increased biosynthetic demands, and chemotherapies. ER stress induces signaling through mediators PERK and IRE1α, which direct prosurvival signaling, including Nrf2 activation and maintenance of mitochondrial homeostasis along with enhanced expression of checkpoint receptors (PD-L1). Signaling through the metabolic stress axis (2E) in response to intratumoral amino acid depletion, glucose restriction, and metabolic waste accumulation induces activation of AMPK, GCN2, and lipid uptake receptors. These enable increased uptake of glucose, adaptation of low amino acid levels, and elevated use of lipids as a source of energy. Anticancer therapies including chemotherapy, lymphodepletion, and radiation can induce activation of the therapy-associated stress axis (2F). This results in induction of emergency myelopoiesis, which drives MDSC rebound and upregulation of checkpoint markers such as PD-L1. Collectively, the diverse axes of stress act as pivotal mediators of MDSC function, expansion and survival.

3.2. Hypoxic Stress Axis:

Successful angiogenesis within the tumor paves the way for increased nutrient supply and future metastasis. However, tortuous and incomplete neovascularization of tumor tissue results in restricted blood flow and lower oxygen levels. In order to survive under this hypoxic environment, both tumor and infiltrating immune cells activate pro-survival pathways to overcome the oxygen deficit. Oxygen restriction triggers activation of hypoxia-inducible factors (HIFs), which act as transcription factors that induce pro-survival transcriptional changes. The role of family members HIF-1α and HIF-1β has been comprehensively assessed in both tumor and myeloid subsets [62]. In well oxygenated cells, HIF transcription factors remain inactive in the cytosol due to prolyl hydroxylation and subsequent proteasomal degradation. During oxygen depletion, HIF members escape prolyl hydroxylation and undergo nuclear translocation. Upon entry into the nucleus, HIFα and HIFβ dimerize and bind DNA at hypoxia response elements (HREs) typically located in the promoter or enhancer regions of target genes [63]. Target genes transcriptionally activated by HIF include those implicated in cell proliferation, cell survival, apoptosis, adhesion, metabolism, and angiogenesis. Collectively, transcriptional promotion by HIF1α and HIF1β drives adaptation to intra-tumoral hypoxia, induces resistance to apoptosis, stimulates angiogenesis, and enables metastasis [64].

Hypoxic stress and induction of HIF-mediated transcription play a significant role in directing expression of MDSC suppressive mediators and recruitment of MDSC to tumors (Figure 2B). Indeed, local regions of intra-tumoral hypoxia have been reported to be infiltrated predominantly by immunoregulatory populations, including MDSCs [65]. Activation of HIF1α and induction of HIF-regulated gene transcription are higher in tumor-infiltrating MDSCs, as compared to splenic MDSCs [65]. Multiple genes transcriptionally regulated by HIF1α act as mediators of suppression in MDSCs, including ARG1, iNOS, checkpoint receptors, chemokines, chemokine receptors, and factors that regulate M-MDSC differentiation into macrophages [66–68]. Additionally, hypoxia indirectly regulates MDSC functionality as exposure to exosomes from hypoxic tumors increases MDSC production of ROS and RNS [69]. In addition to driving expression of suppressive mediators, hypoxia directs the recruitment of MDSCs to tumors via HIF-induced transcription of CC motif ligand 26 (CCL26) and subsequent CX3C motif chemokine receptor 1 (CX3CR1)-mediated trafficking [70]. Therefore, blockade of CX3CR1 impairs hypoxia-induced recruitment of MDSCs to the TME and suppresses tumor growth [70]. Hypoxic stress signaling also plays a role in regulation of myeloid differentiation. HIF1α activation in tumor infiltrating M-MDSCs promotes their differentiation into TAMs [66]. Moreover, in the tumor milieu, HIF activation in tumor cells induces expression of ectonucleoside triphosphate diphosphohydrolase 1 (CD39L1; ENTPD1). Conversion of extracellular ATP to adenosine monophosphate (AMP) by ENTPD1 impairs the differentiation of immature myeloid cells and ensures maintenance of MDSCs in their characteristic immature state [71]. Signaling through the hypoxic stress axis is also reported to regulate chronic inflammatory responses and thus further promote immune suppression [72]. Collectively, the hypoxic stress axis regulates multiple aspects of MDSC biology, including MDSC function, recruitment, and differentiation.

3.3. Chronic Inflammatory Stress Axis

Chronic inflammation is a well-documented contributor to oncogenesis and a primary driver of dysregulated myelopoiesis. Multiple conditions may give rise to chronic inflammatory states, including autoimmune disease, metabolic dysregulation and chronic infection, among others. Notably, these programs of chronic inflammation coincide with an increased risk for cancer [73]. Inflammatory signals are propagated via soluble signaling molecules such as cytokines, chemokines, damage-associated molecular patterns (DAMPs), pathogen-associated molecular patterns (PAMPs), and alarmins. Despite the diversity of inflammatory drivers, multiple inflammatory signals converge on the activation of the NFKB and STAT3 transcription factors and downstream NFKB/STAT3-regulated pathways. Consequently, both NFKB and STAT3 are major mediators that link chronic inflammation to MDSC suppressive function (Figure 2C). Interestingly, chronic inflammation plays a dual role in cancer. Inflammation drives tumorigenesis, but also paradoxically stimulates anti-tumor immunity. For example, exposure to DAMPs or PAMPs promotes MDSC expansion, while additionally triggering dendritic cell maturation and production of protective type I IFNs [74].

Diverse proinflammatory signals, including S100A8/A9, toll ligand receptor agonists, interleukin 1 beta (IL-1β), and interleukin 6 (IL-6) encourages MDSC development, expansion, and immunosuppressive function via STAT3 and NFKB activation [75–78]. Consequently, targeting STAT3 and NFKB signaling can significantly decrease the accumulation, expansion, and suppressive potential of MDSCs in tumor-bearing hosts [76, 79–82]. Indeed, genetic or pharmacological targeting of NFKB members p65 or p50 in MDSCs significantly decreases their suppressive capacity [83, 84]. Additionally, inhibition of STAT3 is reported to reduce MDSC expansion and promote MDSC apoptosis in tumor models [81, 85]. Beyond playing pivotal roles in promoting MDSC expansion, NF-KB and STAT3 signaling also regulate expression of immunosuppressive mediators, including NOS2, IDO, and ARG1 [79, 86]. Targeting upstream inflammation instead common downstream signaling through STAT3 or NFKB also appears to impact tumor-MDSCs. Mice deficient in IL-1β demonstrated reduced infiltration of MDSCs to tumor beds, while blocking IL-6 impaired MDSC immunosuppressive function [87, 88]. Thus, production of inflammatory mediators by MDSCs can also act as a feedforward loop that ensures continued signaling through the chronic inflammatory stress axis and perpetuation of the expansion of MDSCs in cancer.

3.4. ER Stress Axis

The endoplasmic reticulum (ER) serves as a nexus of protein synthesis, modification, maturation, trafficking, and degradation in the cell [89–91]. Additionally, the ER contributes to intracellular calcium flux, mitochondrial homeostasis, and lipid catabolism. Several conditions, including the accumulation of misfolded proteins, disturb ER homeostasis leading to a phenomenon known as ER stress. In order to adapt to ER stress, cells activate the unfolded protein response (UPR). UPR induction is characterized by the activation of ER localized signal transducers, most notably, the inositol-requiring enzyme 1 (IRE1α; ERN1), the PKR-like ER kinase (PERK; EIF2AK3), and the activating transcription factor 6 (ATF6) [89].

During ER stress, IRE1α dissociates from its inhibitory partner binding immunoglobulin protein (Bip; GRP78) and mediates downstream target phosphorylation and RNA degradation via its kinase and ribonuclease domains, respectively. Activation of regulated IRE1α-dependent decay (RIDD) enables splicing of the transcription factor X-box binding protein 1 (Xbp1) that subsequently induces downstream transcriptional activation of pro-survival and immunosuppressive genes [92]. Similar to IRE1α, PERK dissociates from Bip during ER stress and induces phosphorylation of downstream protein and lipid targets. PERK driven phosphorylation of Nrf2 promotes antioxidant responses, while phosphorylation of the eukaryotic initiation factor 2α (eIF2α) inhibits bulk translation and simultaneously induces alternative translation of stress-related proteins. Activation of alternative protein synthesis by phospho-eIF2α induces translation of transcripts containing upstream open reading frames (uORFs) that typically remain translationally repressed in the absence of stress. Ultimately, this process induces translation of uORF-containing mRNA transcripts coding for the activating transcription factor 4 (ATF4), which directs the expression of multiple stress-responsive mediators, including C/EBP homologous protein (CHOP; DDIT3), Growth arrest and DNA damage-inducible protein 34 (GADD34; PPP1R15A), and activating transcription factor 5 (ATF5) [93, 94]. ATF6 acts as a transcription factor which, after Golgi localized activating cleavage, promotes transcription of pro-survival genes. Additionally, the ATF6 arm of the UPR has been reported to act as a molecular off switch that terminates IRE1α mediated signaling [95].

Intrinsic induction of ER stress or exposure to tumor-related ER stress augments the immunosuppressive potential of MDSCs (Figure 2D) [94, 96]. Accordingly, systemic administration of the pharmacological ER stress-inducing agent, Thapsigargin (Tg) to tumor-bearing animals elevated the expansion of splenic MDSCs and enhanced the immunosuppressive capacity of MDSCs via ARG1, iNOS, and NOX2 induction [96]. Interestingly, activation of PERK boosted the suppressive potential of tumor-infiltrating MDSCs by activating Nrf2-mediated maintenance of mitochondrial homeostasis. Consequently, we reported that conditional deletion of PERK in MDSCs impaired mitochondrial function and triggered cytoplasmic release of immunostimulatory mitochondrial DNA (mtDNA). Release of mtDNA in MDSCs provoked stimulator of interferon genes (STING)-dependent production of type I interferons (IFNs) and functional reprogramming of MDSCs into cells capable of activating protective T cell immunity [50]. Similarly, elimination of the ER stress-responsive factor CHOP impaired the capacity of tumor MDSCs to block anti-tumor T cell responses both in vitro and in vivo [97].

The role of other UPR mediators in modulating MDSC function remains less clear. While most of the work investigating IRE1α signaling in myeloid cells has not focused on MDSCs, reports from macrophage and dendritic cell based studies underlined the capacity for IRE1α/Xbp1 signaling to impair dendritic cell and TAM function, regulate cytokine production, and impact protease secretion [94, 98, 99]. Indeed, impairing signaling through the IRE1α to Xbp1 axis in tolerogenic dendritic cells transforms them into cells capable of activating protective T cell immunity in ovarian tumor models [100]. Within MDSCs, IRE1α/Xbp1 signaling contributes to alterations in lipid metabolism and development of PMN-MDSCs with elevated immunosuppressive function [101]. The role of ATF6 in directing the immunosuppressive function of MDSCs remains poorly understood. Although it has been suggested that the ATF6 arm of the UPR plays a less significant role in cancer and immunity [93], ATF6 has been implicated in promoting colon dysbiosis and tumorigenesis in murine models of colorectal cancer [102]. Additionally, the effects of ER stress on MDSC biology goes beyond the regulation of MDSC function. ER stress has been demonstrated to impact MDSC turnover and survival in tumors via regulation of TNF-related apoptosis-induced ligand receptor (TRAIL-R; DR5) [103]. As noted, signaling through the ER stress axis may trigger subsequent signaling through the oxidative/nitration stress axis due to PERK mediated phosphorylation of Nrf2. Collectively, these findings underline the complex role of ER stress in modulating MDSC function and survival in the TME.

3.5. Metabolic Stress Axis

First reported in 1927, the ‘Warburg hypothesis’ posits that tumors preferentially rely on aerobic glycolysis [104]. Broader characterization of the TME has demonstrated more global metabolic alterations than those initially suggested by Warburg. Indeed, the intra-tumoral milieu is characterized by depletion of anabolic building blocks including glucose, L-glutamine, and L-cystine due to rapid consumption by proliferating cancer cells [105, 106]. This depletion coincides with the accumulation of metabolic waste products such as lactate [107]. In order to survive this metabolically restricted milieu, infiltrating cells must activate adaptive responses, including those coordinated by the amino acid depletion sensing general control nonderepressible 2 kinase (GCN2; EIF2AK4), lipid uptake receptors or metabolizing enzymes, and AMP-activated protein kinase (AMPK; PRKAA1).

Low extracellular amino acid levels activate GCN2 kinase, which phosphorylates eIF2α and thereby reduces global protein translation. eIF2α phosphorylation enables amino acid depleted cells to adapt to low amino acid availability and induces expression of amino acid synthesis enzymes via downstream target ATF4 [108]. Additionally, metabolic stress prompts upregulation of receptors and pathways that enable the use of alternative metabolic substrates, including proteins that support the uptake and oxidation of fatty acids. Upregulation of fatty acid transporter 2 (FATP2; SLC27A2), oxidized low-density lipoprotein receptor 1 (LOX-1; OLR1), and enzyme carnitine palmitoyl-transferase 1A (CPT1A) enables uptake of lipids and fatty acid oxidation (FAO) in MDSCs [109, 110]. In this context, fatty acids act as an alternative fuel source that provides substrates for the mitochondrial tricarboxylic acid cycle (TCA) and subsequent ATP production in the absence of available glucose. Low levels of ATP and elevated levels of AMP in tumor-infiltrating cells induce activation of AMPK. AMPK promotes glucose uptake, FAO and autophagy, while blocking cell growth via repression of mammalian target of rapamycin complex 1 (mTORC1) signaling [111]. Collectively, activation of these metabolic sensors promotes a metabolic shift that enables survival and proper adaptation of MDSCs to the nutritionally depleted TME (Figure 2E).

Due to the competition for limited metabolic resources within the TME, infiltration into the tumor prompts metabolic and functional changes within MDSCs. Activation of amino acid-sensing GCN2 promotes immunosuppressive polarization and survival of MDSCs and TAMs resulting in enhanced tumor progression [12]. Interestingly, GCN2 activation produces functionally similar downstream responses to ER stress-induced PERK as both kinases share the common target, eIF2α. While GCN2 and PERK activation enable MDSCs to adapt to amino acid restrictions and ER stress, elevated expression of the FAO-pathway allows MDSCs to utilize fatty acids as metabolic building blocks. Upregulation of FATP2 and LOX-1 licenses tumor MDSCs to internalize extracellular lipids, including long chain fatty acids and oxidize low-density lipoproteins. Once lipid substrates have been internalized, expression of CPT1A enables MDSCs to increase lipid oxidation and funneling of FAO products into the TCA. These adaptations permit MDSCs to exploit metabolic resources and survive in the TME. Additionally, internalization and accumulation of lipids within MDSCs enhances their immunosuppressive capacity via activation of lipid responsive nuclear receptors such as peroxisome proliferator-activator receptors (PPARs) and liver X receptors (LXR) [110, 112, 113].

Interestingly, metabolic stress within cancer cells can also provoke changes in MDSCs. Elevated glycolysis in the tumor compartment promotes increased expression of GM-CSF and G-CSF and thus promotes MDSC expansion. Correspondingly, inhibiting glycolysis within tumor cells decreases GM-CSF and G-CSF production and results in impaired MDSC accumulation [114]. In other models, elevated glycolysis within cancer cells promotes MDSC accumulation via glycolysis-dependent production of IL-1β, IL-6, and GM-CSF [115]. Thus, adaptation to the metabolic stress axis enables MDSCs to survive within the TME and provokes enhanced acquisition of immunosuppressive function.

3.6. Therapy-associated Stress Axis

Many cancer therapies including chemotherapy, radiation, and myeloablative therapies induce significant local and systemic cellular stress. Therapy-associated stress propagates its effect through multiple mediators, including the accumulation of ROS, the expression of heat shock response proteins, the induction of ER stress, and the stimulation of metabolic stress [46, 116–118]. Importantly, the stress-associated signaling induced by multiple forms of therapy may impair future treatment responses by inducing chemotherapy and radiotherapy resistance [117, 119].

In the case of chemotherapy or radiation, elevated levels of stress signals within the TME provoke the development of immunosuppressive subsets, including MDSCs (Figure 2F). The rapid repopulation of MDSCs that occurs after lymphodepleting chemotherapy or radiation therapy exemplifies the potential for anti-cancer interventions to inadvertently induce compensatory ‘rebound’ MDSC expansion [120, 121]. Lymphodepletion triggers emergency myelopoiesis, which prompts the release of immature myeloid cells into circulation. In the context of cancer, these immature myeloid cells ultimately develop into MDSCs [122]. Although lymphodepletion prior to administration of chimeric antigen receptor (CAR) T cells or tumor-infiltrating lymphocytes (TIL) enhances the expansion, persistence, and anti-tumor efficacy of transferred T cell products, lymphodepletion also supports the expansion and rapid repopulation of MDSCs [123, 124]. Consequently, strategies aimed at impairing MDSC induction after lymphodepletion appear to enhance the therapeutic efficacy of transferred T cells [120, 125].

Other non-lymphodepleting therapies may also increase MDSC accumulation. Chemotherapies such as doxorubicin, melphalan, or cyclophosphamide induce both immunostimulatory inflammation and enhanced expansion of MDSCs. While chemotherapy induced inflammation can initially boost anti-tumor T cell responses, subsequent MDSC expansion ultimately impairs sustained anti-tumor immunity [126]. Additionally, surgical interventions have also been reported to increase tumor-related MDSCs, and under specific conditions, post-surgery MDSC expansion has been correlated with lower overall survival [127, 128]. Collectively, the therapy-associated axis of stress along with the other stress axes drive the accumulation, expansion, differentiation, survival, and immunosuppressive function of MDSCs (Figure 2) and the detrimental impact of MDSCs in cancer.

4. Therapeutic Targeting of MDSCs

Multiple approaches for overcoming the detrimental impact of MDSCs in cancer have been proposed. Current approaches generally fall within one of several overlapping categories: 1) depletion of MDSCs, 2) inhibition of MDSC immunosuppressive function, 3) blockade of MDSC development or recruitment, 4) MDSC reprogramming or repolarization (Table 2) [11, 129, 130].

Table 2:

Current strategies for therapeutically targeting MDSCs

Multiple therapeutic strategies for targeting MDSCs in cancer have been investigated. Current therapeutic approaches generally fall into one of four broader approaches: 1) MDSC depletion; 2) Intercepting MDSC recruitment or development; 3) Inhibition of MDSC immunosuppressive mediators; or 4) MDSC reprogramming. The table above details the standing and outcomes of current MDSC targeting approaches and the relationship between each strategy and signaling through cancer associated stress axes.

4.1. MDSC depletion

MDSC depletion has been reported to be achieved by conventional chemotherapies, small molecule inhibitor-based strategies, and antibody-oriented approaches. Although these strategies have a common goal, they do not focus on a single molecular target but rather aim to target multiple pathways.

Initial strategies aimed at depleting MDSCs leveraged the lymphodepleting effects of conventional chemotherapies. However, the transitory lymphodepletion induced by conventional chemotherapies ultimately provokes compensatory rapid repopulation and accumulation of MDSCs (rebound effect). Despite this issue, some conventional chemotherapy regiments have been reported to deplete MDSCs. Conflicting reports have been published on the effects of MDSC depleting chemotherapy agents such as fluorouracil (5-FU). In some models, depleting chemotherapies are reported to significantly decrease MDSC levels, while in other systems, only marginal depletion or increased MDSC expansion has been reported [131–133]. Conventional chemotherapy has also been reported to induce therapy-associated stress, ER stress, ROS and inflammation, indicating a potential for chemotherapy to stimulate signaling through multiple stress axes [116, 134]. Ultimately, multiple factors appear to impact the effect of chemotherapy on MDSCs, including cancer type, dose, and duration of treatment, indicating that chemotherapy-based MDSC depletion strategies may be difficult to apply clinically [133].

In addition to conventional chemotherapy agents, small molecule inhibitor-based approaches have also been investigated for their ability to deplete MDSCs. Targeting receptor tyrosine kinase (RTK) activation with small molecules such as Sunitinib, has been reported to decrease the numbers of circulating MDSCs in metastatic renal carcinoma patients and in cancer vaccination models [135, 136]. However, MDSCs have also been reported to act as a mechanism of resistance to Sunitinib, underlining potential limitations of this strategy [137]. Sunitinib has been reported to provoke both anti-inflammatory and antiangiogenic effects indicating the potential for RTK targeted treatments to mitigate signaling through the chronic inflammatory stress axis, but also to potentially promote signaling through the hypoxic stress axis in response to impaired angiogenesis [138, 139].

Small molecule agents that modulate epigenetic modifications have also been assessed for their effects on MDSCs. Histone deacetylases (HDACs) catalyze the removal of acetyl groups from histones and stimulate either enhanced or repressed DNA access depending on the site of deacetylation and presence of other epigenetic modifications. Mixed reports have been published on the capacity for HDAC inhibitors to eliminate MDSCs. In murine mammary tumor models, HDAC inhibitor treatment eliminated splenic, circulating, and intra-tumoral MDSCs [140]. By comparison, HDAC inhibition triggered greater expansion of MDSCs in a GM-CSF-dependent manner in other models, indicating that targeting epigenetic modifications may produce difficult to predict outcomes [141]. HDAC inhibitors have been reported to provoke ER stress, suggesting the potential for HDAC inhibitors to promote signaling through the ER stress axis [142].

Antibody-based approaches have additionally been reported to induce MDSC depletion. Targeting the angiogenesis promoting protein vascular endothelial growth factor (VEGF) with the humanized monoclonal anti-VEGF agent Bevacizumab has also been reported to reduce MDSC expansion [143]. However, in a similar manner to Sunitinib, MDSCs have been described as a mediator of resistance to VEGF-targeted therapies [144]. Moreover, targeting VEGF has been reported to exacerbate the effects of hypoxia, indicating that VEGF blockade may enhance, as oppose to inhibit, signaling through the hypoxic stress axis [145]. Collectively, the contradictory findings reported for MDSC depletion agents indicates that depleting MDSCs may not be as straightforward as it would be predicted and underlines the challenges of targeting a heterogeneous population of cells through depletion approaches.

4.2. Intercepting MDSC recruitment or development

Strategies that blunt MDSC development or recruitment include CC chemokine receptor type 5 – CC motif chemokine ligand 5 (CCR5-CCL5) blockade, colony stimulating factor-1 receptor (CSFR1) antagonists, and CC chemokine receptor type 2 - CC motif chemokine ligand 2 (CCR2-CCL2) inhibitors. CCR5-CCL5 signaling contributes to the intratumor migration of MDSCs; therein, blocking CCR5 signaling provides the opportunity to intercept trafficking of MDSCs to the tumor and peri-tumoral tissue. Consequently, CCR5 knockout mice exhibit decreased levels of tumor-infiltrating MDSCs [146]. In agreement, CCL5-null mice show decreased intra-tumoral levels of MDSCs and generate MDSCs with impaired suppressive capacity [147]. Consistent with these results, tumor-bearing mice treated with CCL5-targeting siRNA nanoparticles or CCR5 antagonist, Maraviroc, exhibited delayed tumor growth [147]. In a gastric cancer model, CCR5 blockade was reported to reduce intratumor and circulating MDSC numbers and to enhance the efficacy of anti-PD-1 therapy [148]. However, CCR5 is also expressed on other immune subsets, including T cells, macrophages, dendritic cells, and eosinophils. Thus, strategies that impair the CCR5-CCL5 axis may also inhibit recruitment of other leukocytes that contribute to protective anti-tumor immunity. Consequently, targeting CCR5-CCL5 signaling may downregulate signaling through the chronic inflammatory stress axis, but this may also provoke detrimental effects on the migration of antitumor immune subsets. Furthermore, CCL5 contributes to angiogenesis indicating that CCL5 blockade may decrease tumor neovascularization and promote associated signaling through the hypoxic stress axis [149, 150].

Engagement of CSFR1 with ligand colony stimulating factor 1 (CSF1) directs RTK signaling and subsequent differentiation, polarization, and recruitment of myeloid cells, including macrophages and M-MDSCs. Indeed, treatment of tumor-bearing mice with CSFR1-targeted receptor tyrosine kinase antagonist, PLX647, impairs M-MDSC recruitment, inhibits tumor growth and increases anti-tumor T cell activity [151]. Notably, inhibition of MDSC recruitment by PLX647 also enhances the efficacy of anti-PD-1 or anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) immunotherapy and boosts the effects of IDO inhibition [151]. Additionally, CSF1-CSFR1 signaling may in part contribute to the recruitment of MDSCs after radiation therapy. After administration of radiation, elevated serum levels of CSF1 were detected, and blockade of CSFR1 signaling reduced the post-radiation accumulation of MDSCs [152]. Conversely, treatment of tumor-bearing mice with CSFR1 inhibitor, JNJ-40346527, increased the intra-tumoral accumulation of PMN-MDSCs [153]. Interestingly, these findings were attributed to the effects of CSFR1 inhibition on cancer-associated fibroblasts (CAFs). CAFs exposed to CSFR1 inhibition were reported to produce CXC motif chemokine ligands 1, 2 and 7 (CXCL1, CXCL2, and CXCL7), which provoked PMN-MDSCs migration into the tumor [153]. Inhibition of CSFR1 with PLX647 has been reported to decrease inflammatory responses in vitro and in vivo, indicating that anti-CSFR1 oriented therapies may mitigate signaling through the inflammatory stress axis [154].

Targeting CCL2-CCR2 signaling is another potential strategy for intercepting MDSC recruitment. CCR2-CCL2 signaling is a key regulator of monocyte and basophil recruitment and coordinates metastatic and angiogenic processes that promote tumor progression [155]. However, two isoforms of CCR2, CCR2A and CCR2B, are reported to direct differential signaling in response to CCL2 binding [156]. Additionally, CCR2 isoforms bind to other ligand targets including CC motif chemokine ligands 8, 7 and 3 (CCL8, CCL7, and CCL3), whereas CCL2 alternatively complexes with CC chemokine receptor 4 (CCR4) and atypical chemokine receptors 1 and 2 (ACKR1, ACKR2) [155]. In addition to this complexity, the CCR2-CCL2 axis mediates effects on a range of other immune cells beyond MDSCs. CCR2-CCL2 signaling directs monocyte polarization to generate TAMs, promotes recruitment of neutrophils and NK cells to tumors, directs maintenance of pluripotency in stem cells, and induces Th2 polarization [157].

In the context of MDSCs, CCR2 deletion impairs homing of adoptively transferred MDSCs into tumors. Notably, CCR2 may play a role beyond recruitment alone as adoptively transferred CCR2-null MDSCs exhibit the ability to delay tumor growth compared to wildtype MDSC counterparts [158]. In mouse models of glioblastoma, treatment with CCR2 antagonists increased survival, impaired accumulation of MDSCs, and synergized with PD-1 blockade [159]. In both lymphoma and melanoma models, targeting CCR2 delayed tumor growth and reduced M-MDSC infiltration [160]. Additionally, other mouse model-based studies have demonstrated the role of CCR2-CCL2 in driving MDSC-mediated CD8 T cell suppression [161]. However, clinical trials targeting CCR2-CCL2 signaling have produced unclear results. While relatively well tolerated, human anti-CCL2 IgG1 blocking agent, CNTO888, did not result in objective tumor responses in patients with metastatic prostate cancer [155]. Furthermore, humanized IgG1 anti-CCR2 blocking agent, MLN1202, was reported as well tolerated in a phase 2 trial in patients with bone metastases; however, measures of clinical responses were not reported [155]. The relationship between CC2-CCL2 signaling and chronic inflammatory stress axis signaling appears to be complex. While targeting CCR2-CCL2 signaling can reduce the recruitment of MDSCs, it may also mitigate recruitment of antitumor macrophage and dendritic cells [162].

4.3. Modulating MDSC effector functions

Approaches directed at targeting mediators of MDSC suppressive function include small molecule inhibitors that target ARG1, iNOS, IDO, STAT3, cyclooxygenase 2 (COX2), and phosphodiesterase type 5 (PDE5) [143, 144]. While some therapies that block MDSC immunosuppressive mediators have been reported to induce elevated anti-tumor T cell responses, others have failed to provide therapeutic benefit. Notably, treatment of metastatic melanoma patients with the IDO1 inhibitor, Epacadostat, failed to extend survival in a phase 3 clinical trial [163].

By comparison, treatment of triple negative breast cancer models with iNOS inhibitor nitro-L-arginine methyl ester (L-NAME), in combination with chemotherapy reduced syngeneic tumor growth and impaired growth of patient derived xenograft tumor models [164]. In agreement, combination of iNOS inhibitors and radiotherapy synergistically delayed tumor growth compared to radiotherapy alone in mouse models of orthotopic pancreatic ductal adenocarcinoma [165]. In a phase 2 clinical trial, administration of iNOS inhibitor, ASP9853, in combination with docetaxel, induced decreased tumor growth compared to docetaxel alone. However, toxicity issues including neutropenia limited further studies [166]. ARG1 inhibitors have also demonstrated preclinical efficacy in multiple mouse models and are reported to reduce tumor growth, restore anti-tumor T cell immune responses, and suppress metastasis [167–169]. An ongoing phase 1/2 clinical trial testing the use of arginase inhibitor, INCB001158, in combination with chemotherapy or PD-L1 checkpoint blockade has reported initial results indicating higher partial response rates in patients receiving INCB001158 [170]. Targeting ARG1, iNOS or IDO can partially mitigate signaling through the metabolic stress axis by rescuing depletion of L-arginine and L-tryptophan respectively. Additionally, targeting iNOS can ameliorate signaling through the oxidative/nitration stress axis by reducing production of nitric oxide and subsequent formation of peroxynitrite derivatives.

COX2 catalyzes the production of proinflammatory prostaglandins from arachidonic acid. Targeting COX2 has demonstrated efficacy in preclinical models. Inhibition of COX2 in mesothelioma was reported to reduce tumor-associated expansion of MDSCs and to prolong survival when combined with adoptive transfer of dendritic cells [171]. Moreover, COX2 inhibition in combination with anti-CD40 monoclonal antibody promoted maturation of MDSC-like cells and prolonged survival in glioma models [172]. FDA approved COX2 inhibitors in the form of nonsteroidal anti-inflammatory drugs (NSAIDs) have been in clinical use for decades. Consequently, repurposing these compounds for cancer therapy is an attractive option. However, caution may be warranted as COX-2 inhibitors have been reported to increase the risk of kidney failure, hepatotoxicity, and adverse cardiac events. By decreasing production of inflammatory prostaglandins, COX2 inhibitors can decrease signaling through the chronic inflammatory stress axis.

Phosphodiesterase type 5 (PDE5) catalyzes the breakdown of cyclic guanosine monophosphate (cGMP), which results in decreased vasodilation of vascular smooth muscle tissue. Approved PDE5 inhibitors typically prescribed for erectile dysfunction have also demonstrated utility in targeting MDSCs. In fact, PDE5 inhibition attenuated the suppressive activity of MDSCs in mouse models and restored T cell proliferation in patients with head and neck cancer [173]. In lymphoma mouse models, treatment with PDE5 inhibitor, Sildenafil, decreased tumor growth and induced a corresponding decrease in MDSC ARG1 levels [174]. Multiple clinical trials are now underway investigating the use of PDE5 inhibitors in a number of cancer types [175]. Overall, PDE5 inhibition appears to be a promising, well tolerated strategy that can modulate MDSC function and impact tumor growth. Inhibition of PDE5 has been reported to provide protection against hypoxia indicating that PDE5 targeted drugs may normalize dysregulated singling through the hypoxic stress axis [176, 177].

STAT3 is a key mediator of myeloid cell differentiation and contributes to the development of dendritic cell, macrophage, and MDSCs. Consequently, targeting STAT3 may impact the differentiation and function of other myeloid subsets beyond MDSCs that modulate anti-tumor immunity. In MDSCs, exposure to inflammatory cytokines or essential factor G-CSF induces STAT3 activation. Treatment of MDSCs isolated from head and neck squamous cell carcinoma patients with the STAT3 inhibitor, Stattic, or STAT3 small interfering RNA (siRNA), decreased expression of ARG1 and impaired MDSC suppressive capacity [82]. Additionally, inhibition of STAT3 promoted MDSC apoptosis, suggesting a second mechanism by which STAT3 inhibitors might regulate MDSC biology [81]. However, intact STAT3 signaling is required for full activation of CAR-T cells indicating that, while STAT3 inhibition may hold therapeutic promise, use of such inhibitors should be synchronized with other therapies, especially for patients receiving T cell-based immunotherapies [80]. STAT3 acts as an intrinsic driver of inflammation; consequently, targeting STAT3 can mitigate signaling through the chronic inflammatory stress axis [178].

4.4. Reprogramming MDSCs or Redirecting MDSC differentiation:

Reprogramming MDSCs into immunostimulatory cells capable of priming anti-tumor T cell immunity is an area of active research that holds strong therapeutic promise. The ultimate goal of reprogramming-oriented therapies would be to utilize reprogrammed MDSCs as immunostimulatory Trojan horses that could promote antitumor immunity in tumor beds. Alternatively, inducing differentiation of characteristically immature MDSCs may provide a similar opportunity to utilize tumor resident MDSCs as in situ immunostimulatory myeloid cells. Therapies aimed at reprogramming or altering the differentiation of MDSCs include chemotherapies, casein kinase inhibitors, and differentiation promoting vitamins such as all-trans-retinoic acid (ATRA), among others.

In addition to inducing MDSC depletion, some chemotherapies are reported to induce differentiation of MDSCs. Paclitaxel treatment is reported to promote differentiation of MDSCs into dendritic cells independently of toll ligand receptor 4 (TLR4) activation [179]. However, paclitaxel has also been reported to induce depletion of MDSCs making the ultimate outcome of paclitaxel treatment on MDSCs unclear [133]. As stated above, despite the potential beneficial effects of chemotherapy on MDSCs, chemotherapy has also been reported to trigger ROS, ER stress, and inflammation, thereby further stimulating signaling through multiple axes of stress.

Treatment with casein kinase 2 inhibitors has been reported to impair the differentiation of circulating precursor CMPs into MDSCs [180]. Casein kinase 2 regulates both DNA damage repair and cell cycle processes and is reported to be dysregulated in multiple cancer types. Consequently, targeting casein kinase 2 may provide the dual benefit of impacting both tumor cells and forestalling MDSC accumulation. Further research is needed to determine how casein kinase 2 inhibitors may impact MDSC accumulation in clinical trials. Interestingly, targeting casein kinase 2 may mitigate signaling through multiple stress axes as inhibiting casein kinase 2 is reported to downregulate critical hypoxia responsive factor HIF-1, to decrease responsiveness to ER stress and to modulate redox homeostasis [181–183].

The vitamin A derivative ATRA appears to provoke functional reprogramming of MDSCs. Through upregulation of glutathione synthase in MDSCs, ATRA directs increased accumulation of glutathione and subsequent differentiation of MDSCs into a dendritic cell like subset [184, 185]. ATRA has also been reported to reduce recruitment of MDSCs to the tumor site and thus enhance the efficacy of antiangiogenic therapy in solid tumors [186]. In a phase II clinical trial, cotreatment with CTLA-4 checkpoint blockade and ATRA decreased circulating MDSCs compared to checkpoint blockade alone. Additionally, ATRA reduced MDSC expression of immunosuppressive mediators IDO and PD-L1 [187]. When combined with CAR T cell therapy in xenograft models, ATRA further enhanced the efficacy of CAR T delivery and induced depletion of M-MDSCs [188]. However, the beneficial effect of ATRA appears to be heavily dependent on when the therapy is administered. Studies combining ATRA with therapeutic anti-cancer vaccines observed enhanced anti-tumor responses only when ATRA was delivered three days after vaccine administration. If simultaneously delivered with vaccination, ATRA treatment decreased anti-tumor immune responses, resulting in reduced survival [189]. Due to its ability to increase intracellular glutathione, ATRA can restore intracellular redox homeostasis and mitigate signaling through the oxidative/nitrosative stress axis. Additionally, in other models, ATRA treatment is reported to ameliorate ER stress and hypoxia, indicating the potential ability to decrease signaling through the ER stress and hypoxic stress axes [185, 190].

4.5. Targeting MDSCs: Heterogeneity as a hurdle for MDSC targeted therapies

Although many of the strategies discussed above offer significant therapeutic promise, the conflicting results reported for some of these agents may impede translation of these approaches into the clinic. One potential reason for the conflicting results reported for MDSC-targeting therapies may be the functional complexity and developmental heterogeneity of MDSC subsets. The indistinct phenotypic and functional profiles of MDSCs makes the development of targeted therapies difficult. Indeed, many of the targetable features exhibited by MDSCs are also apparent in other immune cell subsets.

One strategy for overcoming this difficulty may be to target the axes of stress that drive MDSC development, accumulation, and immunosuppressive polarization. We found that pharmacological inhibition of the ER stress axis via tauroursodeoxycholic acid (TUDCA) or inhibition of PERK reprogrammed MDSCs into immunostimulatory cells capable of inducing T cell activation [50]. Consequently, therapeutically targeting the ER stress axis can direct MDSC reprogramming. Inhibition of the metabolic stress axis in MDSCs has also been demonstrated to promote reprogramming. Thus, blocking FAO in MDSCs with Etomoxir has been reported to reduce the immunosuppressive potential of MDSCs and delay tumor growth [191]. Moreover, inhibition of AMPKα in tumor-bearing mice transformed M-MDSC into a subset capable of eliminating tumor cells through increased production of nitric oxide, while impairing their immunosuppressive activity and differentiation into TAMs [192]. Ultimately, a comprehensive understanding of the diverse stressors that push MDSCs into their characteristic immunoregulatory state will provide the foundation for new MDSCs-combating therapies.

5. Summary and Conclusion:

Collectively, the role of stress axes in directing the development, expansion and immunosuppressive function of MDSCs has been well substantiated. Despite the diversity of stimuli that direct signaling through the axes of stress, common downstream mediators dictate the impact stress signals have on MDSCs. Overlapping axes of stress culminate in activation of key transcriptional regulators, including STAT3, NFKB, Nrf2, and HIF1, among others, that underwrite the phenotypic and functional changes that link stress to shifts in MDSC biology and the acquisition of immunoinhibitory potential. However, despite the convergence of stress axes signaling, effective strategies capable of overcoming the detrimental impacts of MDSCs in cancer remain limited. Consequently, developing MDSC-targeting therapies remains an unmet need. Understanding the forces such as the tumor-associated stress axes that provoke MDSC development may provide the insight needed to generate effective anti-MDSC therapies and meet this urgent clinical need.

Highlights:

• MDSCs are key inhibitors of protective anti-tumor immunity and anti-cancer therapies

• Axes of stress shape the expansion and immunosuppressive activity of MDSCs in tumors

• Understanding how stress drives MDSC function can enable new MDSC-targeting therapies

Abbreviations:

CD73; NT5E

5’-nucleotidase

ATF4

Activating transcription factor 4

ATF5

Activating transcription factor 5

ATF6

Activating transcription factor 6

ADAM17

ADAM metallopeptidase domain 17

ATRA

All-trans retinoic acid

AMPK; PKRAA1

AMP-activated protein kinase

AREs

Antioxidant response elements

ARG1

Arginase 1

ACKR1/2

Atypical chemokine receptor

Bip GRP78

Binding immunoglobulin protein

CAFs

Cancer-associated fibroblasts

CPT1A

Carnitine palmitoyl-transferase 1A

CAT

Catalase

C/EBP

CCAAT/enhancer-binding protein

CHOP; DDIT3

C/EBP homologous protein

CCL2

CC motif chemokine ligand 2

CCL3

CC motif chemokine ligand 3

CCL5

CC motif chemokine ligand 5

CCL7

CC motif chemokine ligand 7

CCL8

CC motif chemokine ligand 8

CCR2

CC chemokine receptor type 2

CCR4

CC chemokine receptor type 4

CCR5

CC chemokine receptor type 5

CAR

Chimeric antigen receptor

CSF1

colony stimulating factor 1

CSFR1

Colony stimulating factor 1 receptor

CMP

Common myeloid progenitor

cGMP

Cyclic guanosine monophosphate

COX2

Cyclooxygenase 2

CTLA-4

Cytotoxic T-lymphocyte-associated protein 4

CXCL1

CXC motif chemokine ligand 1

CXCL2

CXC motif chemokine ligand 2

CXCL7

CXC motif chemokine ligand 7

CX3CR1

CX3C motif chemokine receptor 1

DAMPs

Damage-associated molecular patterns

GADD34; PPP1R15A

Growth arrest and DNA damage-inducible protein 34

eMDSCs

Early precursor MDSCs

CD39L1; ENTPD1

Ectonucleoside triphosphate diphosphohydrolase 1

ER

Endoplasmic reticulum

ERK

Extracellular signaling related kinase

eIF2α

Eukaryotic Initiation factor 2 alpha

FAO

Fatty acid oxidation

FATP2; SLC27A2

Fatty acid transporter 2

5-FU

Fluorouracil

LGALS9

Galectin 9

GCN2; EIF2AK4

General control nonderepressible 2

GST

Glutathione synthase

GSK3β

Glycogen synthase kinase 3 beta

G-CSF

Granulocyte colony stimulating factor

GM-CSF

Granulocyte-macrophage colony-stimulating factor

HO-1; HMOX1

Heme oxidase 1

HDACs

Histone deacetylases

HIF

Hypoxia-inducible factors

HREs

Hypoxia response elements

IDO

Indoleamine- 2,3-dioxygenase

iNOS

Inducible nitric oxide synthase

IRE1α; ERN1

Inositol-requiring enzyme 1

IL-6

Interleukin 6

IL-1β

Interleukin 1 beta

Keap1

Kelch-like ECH associated protein 1

LXR

Liver X receptors

MHCI

Major histocompatibility complex I

MHCII

Major histocompatibility complex II

mTORC1

Mammalian target of rapamycin complex 1

mtDNA

Mitochondrial DNA

M-MDSCs

Monocytic MDSCs

MDSCs

Myeloid-derived suppressor cells

NOX2

NADPH oxidase 2

NQO1

NADPH quinone dehydrogenase 1

NK

Natural killer

Nrf2

nuclear factor erythroid-derived 2-like 2

L-NAME

Nitro-L-arginine methyl ester

NSAIDs

Nonsteroidal anti-inflammatory drugs

NF-KB

Nuclear factor kappa B

LOX-1; OLR1

Oxidized LDL receptor 1

PAMPs

Pathogen-associated molecular patterns

PDE5

Phosphodiesterase type 5

PD-1

Programmed cell death protein 1

PD-L1

Programmed death-ligand 1

PPARs

Peroxisome proliferator-activator receptors

PERK

PKR-like ER kinase

PMN-MDSCs

Polymorphonuclear MDSCs

PKC

Protein kinase C

ROS

Reactive oxygen species

RNS

Reactive nitrogen species

RTK

Receptor tyrosine kinases

RIDD

Regulated IRE1α-dependent decay

S100A8

S100 calcium-binding protein A8

S100A9

S100 calcium-binding protein A9

siRNA

Small interfering RNA

STAT3

Signal transducer and activator of transcription 3

sMaf

Small Maf

STING

Stimulator of interferon genes

SOD

Superoxide dismutase

TIM-3

T cell immunoglobulin and mucin-domain containing-3

TCR

T cell receptor

Th1

T helper type 1

Th2

T helper type 2

Tg

Thapsigargin

TXN

Thioredoxin

TRAIL-R

TNF-related apoptosis-induced ligand receptor

TLR4

Toll ligand receptor 4

TGFβ

Transforming growth factor beta

TCA

Tricarboxylic acid cycle

TUDCA

tauroursodeoxycholic acid

TAMs

Tumor-associated macrophages

TIL

Tumor infiltrating lymphocyte

TME

Tumor microenvironment

TNFα

Tumor necrosis factor alpha

IFNs

Type I interferons

UPR

Unfolded protein response

VEGF

Vascular endothelial growth factor

uORFs

Upstream open reading frames

Xbp1

X-box binding protein 1

XCT; SLC7A11

xCT cystine/glutamate antiporter