Myeloid-derived suppressor cells in cancer: therapeutic targets to overcome tumor immune evasion.

Due to their abundance and high immunosuppressive capacity, a large body of evidence suggests that tumor-associated myeloid cells have a profound impact on immunotherapy resistance [ 184 ]. In the last few years, an increasing number of preclinical studies and clinical trials have been conducted to validate the potential safety and benefits of inhibiting MDSCs alone or in conjunction with radiation, chemotherapy and immunotherapy for the treatment of cancer [ 185 ]. The main therapeutic strategies used in current studies to eliminate MDSCs and/or inhibit their immunosuppressive activity within the TME include (i) depletion of MDSCs populations; (ii) inhibition of MDSCs recruitment to tumor sites; (iii) attenuation of the inhibitory activity of MDSCs by targeting specific molecular pathways involved in MDSCs-mediated immune escape processes; and (iv) promotion of MDSCs differentiation, such as differentiation into M1 macrophages or dendritic cells (Fig.  6 ). Fig. 6 Primary therapies targeting MDSCs. MDSCs are eliminated, and/or their immunosuppressive activity is inhibited by different strategies, including (1) depletion of MDSCs populations, (2) inhibition of MDSCs recruitment, (3) inhibition of MDSCs immunosuppressive activity, and (4) induction of MDSCs differentiation Primary therapies targeting MDSCs. MDSCs are eliminated, and/or their immunosuppressive activity is inhibited by different strategies, including (1) depletion of MDSCs populations, (2) inhibition of MDSCs recruitment, (3) inhibition of MDSCs immunosuppressive activity, and (4) induction of MDSCs differentiation Intensive studies have explored new ways to target and deplete MDSCs. In mouse models, apoptosis in MDSCs can be mediated by targeting antibodies to the surface markers Gr-1 or Ly6G, inducing Fas-FasL or targeting the TNF-related apoptosis-induced ligand (TRAIL) receptor [ 186 – 188 ]. Similarly, in models of female malignancy estrogen and its receptor alpha signaling cause MDSCs amplification and enhanced immunosuppressive activity through altered pSTAT3 signaling, which supports the idea that more specific anti-estrogen drugs could complement emerging immunotherapies [ 189 ]. In addition, the use of the p38 MAPK inhibitor GW856553 in murine HCC models with cirrhosis effectively inhibited the enhancer reprogramming of M-MDSCs development and immunosuppression induced by activated hepatic stellate cells [ 190 ]. Furthermore, activation of the therapeutic liver-X nuclear receptor and its transcriptional target apolipoprotein E signaling by the application of the selective agonist GW3965 directly inhibited the survival of MDSCs in murine models and in patients treated in a first-in-human dose escalation phase 1 trial [ 191 ]. Some cytotoxic chemotherapies, such as carboplatin and paclitaxel, can also reduce the number of circulating MDSCs in tumor patients [ 192 ], some of which are now thought to support the antitumor effects of certain regimens. In addition, the combination of chemotherapy with other therapies to target MDSCs has demonstrated preclinical and clinical antitumor effects (Table  1 ). Table 1 Clinical trials targeting MDSCs in tumors Target Intervention Conditions Phase Status Number Enrolled NCT Adenosine A2B Receptor antagonist; Chemotherapy PBF-1129 and Nivolumab NSCLC I Recruiting 30 NCT05234307 Chemotherapy; Electrothermal therapy Gemcitabine; Focused Ultrasound; Gemcitabine and Focused Ultrasound Breast Cancer I Recruiting 48 NCT04796220 Chemotherapy Tadalafil Astrocytoma I Completed 18 NCT04757662 Endocrine Therapy; Chemotherapy Abemaciclib; Fulvestrant; Aromatase Inhibitors Breast Cancer II Active, not recruiting 18 NCT04352777 Anti-PD-1; Chemotherapy Nivolumab; Nivolumab and Gemcitabine NSCLC II Terminated 3 NCT03302247 Chemotherapy Fludarabine; Busulfan; Methotrexate Leukemia I Active, not recruiting 20 NCT02916979 CXCR1/2 antagonist; Anti-PD-1 SX-682 and Pembrolizumab Melanoma I Recruiting 77 NCT03161431 PDE-5 inhibitor Tadalafil Head and Neck Squamous Cell Carcinoma II Completed 40 NCT01697800 Anti-PD-1; ATRA Pembrolizumab with ATRA Melanoma I/II Active, not recruiting 26 NCT03200847 Anti-PD-1; ATRA ATRA and Atezolizumab NSCLC I Recruiting 18 NCT04919369 ATRA; Anti-CTLA-4 ATRA; Ipilimumab Melanoma II Active, not recruiting 10 NCT02403778 H2 receptor antagonist Ranitidine Cancer IV Completed 30 NCT03145012 TLR9 agonist; Anti-PD-1 CMP-001 and Nivolumab Melanoma; Lymph Node Cancer II Active, not recruiting 34 NCT03618641 Chemotherapy; Anti-VEGF Capecitabine; Bevacizumab Recurrent Glioblastoma I Active, not recruiting 12 NCT02669173 MDSCs, myeloid-derived suppressor cells; MNPs, mononuclear phagocytes; DCs, dendritic cells; ImCs, immature cells; MSCs, myeloid suppressor cells; PMN-MDSCs, polymorphonuclear-MDSCs; M-MDSCs, monocytic MDSCs; ARG1, arginase 1; MMP-9, metalloproteinase-9; PD-L1, programmed death ligand 1; HLA-I, human leukocyte antigens class I; CMPs, common myeloid progenitors; GMPs, granulocyte–macrophage progenitors; MB, myeloblasts; MDP, monocyte/macrophages and dendritic cell; IL-17A, Interleukin-17A; GM-CSF, granulocyte–macrophage colony-stimulating factor; G-CSF, granulocyte colony-stimulating factor; TNF-α, tumor necrosis factor-alpha; CCR2, C–C chemokine receptor 2; CCR5, C–C chemokine receptor 5; TGF-β, transforming growth factor beta; CD62L, L-selectin2; ADAM17, a disintegrin and metalloproteinase domain 17; TACE, TNF-α-converting enzyme; HMGB1, high mobility group box-1; ROS, reactive oxygen species; VEGF, vascular endothelial growth factor; TCR, T-cell receptor; MHC, major histocompatibility complex; RNS, reactive nitrogen species; NO, nitric oxide; Trp, tryptophan; TME, tumor microenvironment; CAT-2B, cationic amino acid transporter protein; IDO1, indoleamine-2,3-dioxygenase 1; AhR, aryl hydrocarbon receptor; PD-1, programmed cell death 1; VISTA, V-domain Ig suppressor of T-cell activation; Gal-9, galactose lectin-9; AML, acute myeloid leukemia; TIM-3, T-cell immunoglobulin and mucin structural domain 3; STING, stimulator of interferon genes; TIGIT, T-cell immunoglobulin and ITIM domain; HNSCC, head and neck squamous cell carcinoma; NK, Natural killer; NKG2D, natural killer group 2D; IFN-γ, interferon-γ; STAT3, signal transducer and activator of transcription 3; NF-Κb, nuclear factor-κB; IL-10, interleukin-10; IL-12, interleukin-12; Treg, regulatory T; LTB4, leukotriene B4; EV, extracellular vesicles; HSP90α, heat shock protein 90α; NLRP3, NOD-like receptor protein 3; CCN4, Cell Communication Network Factor 4; HCC, Hepatocellular carcinoma; CAFs, cancer-associated fibroblasts; MIF, migration inhibitory factor; SLC7A11, solute carrier family 7 member 11; CSF1, colony-stimulating factor 1; SLC7A2, Solute carrier family 7 member 2; PI3K, phosphatidylinositol 3-kinase; AKT, protein kinase B; RIP3, receptor-interacting protein kinase 3; ENTPD2, ectonucleoside triphosphate diphosphohydrolase 2; HIF-1, hypoxia-inducible factor-1; ATP, adenosine triphosphate; 5'-AMP, 5'-Adenosine monophosphate; PIWIL1, Piwi Like RNA-Mediated Gene Silencing 1; FAO, fatty acid β-oxidation; AR, androgen; CCRK, cell cycle-related kinase; EZH2, Enhancer of zeste homolog 2; BC, breast cancer; GMPs, granulocyte-monocyte progenitors; GPs, granulocyte progenitors; BM, bone marrow; CCL20, C–C motif chemokine ligand 20; BCSC, breast cancer stem cells; NO, nitric oxide; CEBPB, CCAAT/enhancer-binding protein beta; AMPK, AMP-activated protein kinase; MAPK, mitogen-activated protein kinase; AMPKα, AMP-activated protein kinase alpha; eMDSCs, early myeloid-derived suppressor cells; JAK, Janus kinase; SOCS3, suppressor of cytokine signaling-3; PIAS3, protein inhibitor of activated STAT-3; SMAD3, Smad family member 3; KAT6A, lysine acetyltransferase 6A; TRIM, tripartite motif‐containing; Enpp1, ectonucleotide pyrophosphatase/phosphodiesterase 1; NET, neutrophil extracellular traps; PCa, Prostate cancer; ICB, immune checkpoint blockade; CRPC, castration resistant prostate cancer; IL-23, Interleukin-23; CHD1, chromodomain-helicase-DNA-binding protein 1; PTEN, Phosphatase and tensin homolog; LC, Lung cancer; NSCLC, non-small cell lung cancer; IL-7, Interleukin-7; Gprc5a, G-protein–coupled receptor, family C, member 5A; PTGES, PGE synthase; PGE2, prostaglandin E2; GALNT3, polypeptide N-acetyl-galactosaminyltransferase 3; PDH, pyruvate dehydrogenase; CAFs, Cancer-associated fibroblasts; PDAC, pancreatic ductal adenocarcinoma; AML, acute myeloid leukemia; CAC, colitis-associated cancer; FGFR1, fibroblast growth factor receptor-1; TRAIL, TNF-related apoptosis-induced ligand; SHP2, phosphatase 2; Aza, azacytidine; PDE-5, Phosphodiesterase-5; CIK, cytokine-induced killer; Gal-3, alactose lectin-3; Hv1, voltage-gated proton channels; COX-2, cyclo-oxygenase 2; BTK, Bruton’s tyrosine kinase; UPR, unfolded protein response;ATRA, All-trans retinoic acid Clinical trials targeting MDSCs in tumors Adenosine A2B Receptor antagonist; Chemotherapy Chemotherapy; Electrothermal therapy Gemcitabine; Focused Ultrasound; Gemcitabine and Focused Ultrasound Anti-PD-1; Chemotherapy CXCR1/2 antagonist; Anti-PD-1 Head and Neck Squamous Cell Carcinoma Anti-PD-1; ATRA Anti-PD-1; ATRA ATRA; Anti-CTLA-4 TLR9 agonist; Anti-PD-1 Melanoma; Lymph Node Cancer Chemotherapy; Anti-VEGF Capecitabine; Bevacizumab MDSCs, myeloid-derived suppressor cells; MNPs, mononuclear phagocytes; DCs, dendritic cells; ImCs, immature cells; MSCs, myeloid suppressor cells; PMN-MDSCs, polymorphonuclear-MDSCs; M-MDSCs, monocytic MDSCs; ARG1, arginase 1; MMP-9, metalloproteinase-9; PD-L1, programmed death ligand 1; HLA-I, human leukocyte antigens class I; CMPs, common myeloid progenitors; GMPs, granulocyte–macrophage progenitors; MB, myeloblasts; MDP, monocyte/macrophages and dendritic cell; IL-17A, Interleukin-17A; GM-CSF, granulocyte–macrophage colony-stimulating factor; G-CSF, granulocyte colony-stimulating factor; TNF-α, tumor necrosis factor-alpha; CCR2, C–C chemokine receptor 2; CCR5, C–C chemokine receptor 5; TGF-β, transforming growth factor beta; CD62L, L-selectin2; ADAM17, a disintegrin and metalloproteinase domain 17; TACE, TNF-α-converting enzyme; HMGB1, high mobility group box-1; ROS, reactive oxygen species; VEGF, vascular endothelial growth factor; TCR, T-cell receptor; MHC, major histocompatibility complex; RNS, reactive nitrogen species; NO, nitric oxide; Trp, tryptophan; TME, tumor microenvironment; CAT-2B, cationic amino acid transporter protein; IDO1, indoleamine-2,3-dioxygenase 1; AhR, aryl hydrocarbon receptor; PD-1, programmed cell death 1; VISTA, V-domain Ig suppressor of T-cell activation; Gal-9, galactose lectin-9; AML, acute myeloid leukemia; TIM-3, T-cell immunoglobulin and mucin structural domain 3; STING, stimulator of interferon genes; TIGIT, T-cell immunoglobulin and ITIM domain; HNSCC, head and neck squamous cell carcinoma; NK, Natural killer; NKG2D, natural killer group 2D; IFN-γ, interferon-γ; STAT3, signal transducer and activator of transcription 3; NF-Κb, nuclear factor-κB; IL-10, interleukin-10; IL-12, interleukin-12; Treg, regulatory T; LTB4, leukotriene B4; EV, extracellular vesicles; HSP90α, heat shock protein 90α; NLRP3, NOD-like receptor protein 3; CCN4, Cell Communication Network Factor 4; HCC, Hepatocellular carcinoma; CAFs, cancer-associated fibroblasts; MIF, migration inhibitory factor; SLC7A11, solute carrier family 7 member 11; CSF1, colony-stimulating factor 1; SLC7A2, Solute carrier family 7 member 2; PI3K, phosphatidylinositol 3-kinase; AKT, protein kinase B; RIP3, receptor-interacting protein kinase 3; ENTPD2, ectonucleoside triphosphate diphosphohydrolase 2; HIF-1, hypoxia-inducible factor-1; ATP, adenosine triphosphate; 5'-AMP, 5'-Adenosine monophosphate; PIWIL1, Piwi Like RNA-Mediated Gene Silencing 1; FAO, fatty acid β-oxidation; AR, androgen; CCRK, cell cycle-related kinase; EZH2, Enhancer of zeste homolog 2; BC, breast cancer; GMPs, granulocyte-monocyte progenitors; GPs, granulocyte progenitors; BM, bone marrow; CCL20, C–C motif chemokine ligand 20; BCSC, breast cancer stem cells; NO, nitric oxide; CEBPB, CCAAT/enhancer-binding protein beta; AMPK, AMP-activated protein kinase; MAPK, mitogen-activated protein kinase; AMPKα, AMP-activated protein kinase alpha; eMDSCs, early myeloid-derived suppressor cells; JAK, Janus kinase; SOCS3, suppressor of cytokine signaling-3; PIAS3, protein inhibitor of activated STAT-3; SMAD3, Smad family member 3; KAT6A, lysine acetyltransferase 6A; TRIM, tripartite motif‐containing; Enpp1, ectonucleotide pyrophosphatase/phosphodiesterase 1; NET, neutrophil extracellular traps; PCa, Prostate cancer; ICB, immune checkpoint blockade; CRPC, castration resistant prostate cancer; IL-23, Interleukin-23; CHD1, chromodomain-helicase-DNA-binding protein 1; PTEN, Phosphatase and tensin homolog; LC, Lung cancer; NSCLC, non-small cell lung cancer; IL-7, Interleukin-7; Gprc5a, G-protein–coupled receptor, family C, member 5A; PTGES, PGE synthase; PGE2, prostaglandin E2; GALNT3, polypeptide N-acetyl-galactosaminyltransferase 3; PDH, pyruvate dehydrogenase; CAFs, Cancer-associated fibroblasts; PDAC, pancreatic ductal adenocarcinoma; AML, acute myeloid leukemia; CAC, colitis-associated cancer; FGFR1, fibroblast growth factor receptor-1; TRAIL, TNF-related apoptosis-induced ligand; SHP2, phosphatase 2; Aza, azacytidine; PDE-5, Phosphodiesterase-5; CIK, cytokine-induced killer; Gal-3, alactose lectin-3; Hv1, voltage-gated proton channels; COX-2, cyclo-oxygenase 2; BTK, Bruton’s tyrosine kinase; UPR, unfolded protein response;ATRA, All-trans retinoic acid Reducing the recruitment of MDSCs to tumor sites is one of the main therapeutic approaches to re-establish the immune microenvironment and improve the success of immunotherapy. Blocking chemokines and their interactions with ligands is an effective target for reducing the transport of MDSCs [ 193 ]. Src homology-2-containing protein tyrosine phosphatase 2 (SHP2) inhibitors (e.g., SHP099) have antitumor effects on Models with KRAS-mutant and EGFR-mutant NSCLC [ 194 , 195 ]. However, both SHP2 inhibitors and other RAS/ERK pathway inhibitors cause the recruitment of MDSCs by inducing NF-kB-dependent CXCR2 ligand production [ 196 ]. Therefore, SHP2 inhibitors need to be combined with CXCR1/2 (e.g., SX682) inhibitors and improve survival in multiple NSCLC models. Adjuvant epigenetic treatment with low-dose DNA methyltransferase and the histone deacetylase inhibitors entinostat and 5-azacytidine (Aza) after primary tumor resection within mouse models inhibited tumor cell dissemination by reducing the transport of MDSCs by downregulating CCR2 and CXCR2 and by encouraging MDSCs differentiation [ 197 ]. PMN-MDSCs transport was significantly inhibited in mouse tumor models following the application of SX-682 (CXCR1 and CXCR2 inhibitor), which enhanced the ability to respond to programed death-axis ICB and adoptive transfer of engineered T cells [ 198 ]. The CSF1/CSF1 receptor (CSF1R) pathway is another clear target for reducing MDSCs recruitment. CSF1/CSF1R-targeted drugs have been studied in a variety of tumor types [ 199 – 201 ]. In combination with anti-PD-L1 blockers in a mouse model of HCC, CSF1R inhibitors (e.g., PLX3397) significantly inhibited the recruitment of MDSCs, TAM infiltration and M2 polarization, leading to reversal of the immunosuppressed state of the HCC microenvironment [ 202 ]. In addition, the combined application of CSF1R and CXCR2 inhibitors in multiple mouse tumor models significantly reduced the recruitment of TAMs and PMN-MDSCs to the tumor site and significantly reduced tumor growth [ 203 ]. Furthermore, inhibition of CCR2 with PF-04136309 or RS504393 blocked the recruitment of macrophages and M-MDSCs in a mouse model of pancreatic cancer. However, this leads to an increase in the number of neutrophils and PMN-MDSCs within the tumor. Combining PF-04136309 with the CXCR2 inhibitor SB225002 or the CXCL8 neutralizing antibody was able to further increase chemotherapeutic efficacy [ 204 ]. This suggests the possibility of some functional compensation between M-MDSCs and PMN-MDSCs. Blockade of the MDSCs-secreted factor prokineticin (Bv8) can also inhibit the transport of MDSCs to tumors through anti-angiogenic effects in a mouse model of pancreatic cancer [ 205 ]. To target MDSCs recruitment, SX-682 is currently being used in connection with pembrolizumab in a phase I trial for the treatment of melanoma ( NCT03161431 ). Inhibition of the immunosuppressive effect of MDSCs is another therapeutic strategy to target MDSCs. Phosphodiesterase-5 (PDE-5) inhibitors can suppress the function of MDSCs by reducing the levels of iNOS and arginase. In mouse models, PDE-5 inhibitors (e.g., sildenafil and tadalafil) activate antitumor immunity and prolong the survival of tumor-bearing mice [ 206 , 207 ]. In mouses model of hepatocellular carcinoma, systemic treatment with PDE-5 inhibitors may also prevent the accumulation of MDSCs in the TME induced by cytokine-induced killer (CIK) cell-derived immunotherapy in HCC through ARG1 and iNOS blockade, increasing the antitumor function of CIK cell therapy [ 208 ]. Blockade of Bv8 inhibits the immunosuppressive properties of MDSCs by inducing increased expression of IDO, ROS1 and iNOS in a mouse breast cancer model [ 209 ]. In the tumor-bearing mouse models, the extracellular generation of ROS is one of the mechanisms by which MDSCs exert their immunosuppressive function. Combination treatment with an agonist anti-OX40 antibody (aOX40) and a galactose lectin-3 (Gal-3) inhibitor induces a decrease in ARG1 and an increase in iNOS and reduces M-MDSCs-mediated immune suppression, thereby increasing CD8 + T-cell recruitment [ 210 ]. In mouse models, inhibition of voltage-gated proton channels (Hv1) by 5-chloro-2-guanidinobenzimidazole or ZnCl2 and its consequent pH reduction inhibit NADPH oxidase 2-mediated ROS production, thereby reducing the immunosuppressive effects of MDSCs [ 211 ]. Ibrutinib reduces monocyte and granulocyte MDSCs-mediated NO production, mRNA expression of immunosuppressive cytokines and T-cell suppression in a mouse model of neuroblastoma through inhibition of Bruton’s tyrosine kinase (BTK) [ 212 ]. In order to inhibit the immunosuppressive activity of MDSCs, the clinical studies have reported fewer circulating MDSCs, lower expression of arginase and iNOS in these cells, and higher levels of tumor-specific T cells in patients with head and neck cancer and multiple myeloma treated with the PDE-5 inhibitor tadalafil [ 213 , 214 ]. In addition, tadalafil (PDE-5 inhibitor) has been clinically tested in patients with head and neck cancer in a phase II trial ( NCT01697800 ). In patients with HNSCC and melanoma treated with cisplatin, the expression of ARG1 and cyclo-oxygenase 2 (COX-2) was significantly reduced in M-MDSCs, and the ability of M-MDSCs to block activated T-cell responses in isolation was markedly reduced [ 215 ]. These findings suggested that platinum-based chemotherapeutic agents may enhance the efficacy of immunotherapy by overcoming M-MDSCs-mediated immunosuppression. A final therapeutic strategy to re-establish T-cell activity and immunotherapeutic success is the induction of MDSCs to differentiate into mature, non-suppressive myeloid cells. Although moderate activation of unfolded protein response (UPR)-related signaling helps immune cells differentiate and function physiologically, persistent and maladaptive initiation of UPR drivers facilitates immunodeficiency [ 216 , 217 ]. Increased PKR-like endoplasmic reticulum kinase signaling is characteristic of the UPR in MDSCs of a tumor-bearing mouse model, whose deletion converts MDSCs into cells that activate CD8 + T-cell immunity in tumor beds [ 218 ]. All-trans retinoic acid (ATRA), a derivative of vitamin A, has been found to be highly active against MDSCs [ 219 ]. In vivo administration of ATRA stimulates MDSCs from tumor-bearing mice to differentiate into mature myeloid cells [ 220 ]. ATRA-induced MDSCs differentiation involves ATRA specifically upregulating the expression of glutathione synthase and glutathione in MDSCs, thereby neutralizing ROS and driving myeloid differentiation [ 221 ]. In preclinical breast cancer models, ATRA was found to enhance the efficacy of antiangiogenic therapy for breast cancer by depleting MDSCs [ 222 ]. In addition, by promoting the expression of differentiation genes in peritoneal PMN-MDSCs, the receptor tyrosine kinase inhibitor sunitinib promotes the differentiation of MDSCs into mature polynuclear MDSCs in a mouse model of endometriosis [ 223 ]. Finally, adjuvant epigenetic therapy with low-dose Aza, entinostat, DNA methyltransferase and histone deacetylase inhibitors within mouse models was used to inhibit tumor cell dissemination by disrupting the premetastatic microenvironment through the promotion of MDSCs differentiation to a more mesenchymal macrophage-like phenotype [ 197 ]. To induce differentiation of MDSCs, high plasma concentrations (> 150 ng/mL) of ATRA in patients with metastatic renal cell carcinoma promote differentiation of MDSCs to APC precursors, thereby eliminating MDSCs-mediated immunosuppression [ 224 ]. Currently, combinations of ATRA and ICIs is used in several clinical trials, including pembrolizumab, atezolizumab, and ipilimumab ( NCT03200847 , NCT04919369 , and  NCT02403778 , respectively).

MDSCs can interfere with T-cell immunity by preventing naive T cells from homing to lymph nodes in which they could become activated. Through transforming growth factor beta (TGF-β) signaling, MDSCs disrupt HBV‐specific T-cell trafficking by downregulating CCR5 on them [ 27 ]. In addition, MDSCs have been shown to mediate the downregulation of the cell adhesion molecule L-selectin2 (CD62L) on T cells through the expression of the surface metalloprotease a disintegrin and metalloproteinase domain 17 (ADAM17), also called TNF-α-converting enzyme (TACE) [ 28 ]. It has been shown that tumor-expressed high mobility group box-1 (HMGB1) also enhances MDSCs-mediated downregulation of L-selectin on naive T cells [ 29 ]. This results in decreased homing and antigen-driven activation of lymph node CD8 + T cells [ 30 ].

Although MDSCs primarily target effector T cells, recent reports have proven that MDSCs can also mediate immune escape through the inhibition of other tumor-killing immune cells, such as DCs and natural killer (NK) cells. DCs are the other major myeloid cells infiltrating into the TME. Although signals from the TME promote the influx of immature DCs, multiple factors, including adenosine accumulation, lactate accumulation, and hypoxic conditions, induce DC dysfunction [ 54 ]. In addition, it has been suggested that crosstalk between DCs and MDSCs may also be partly responsible for the decreased DC function. When bone marrow-derived MDSCs are co-cultured with DCs in vitro, the DC population decreases as the number of MDSCs increases [ 55 , 56 ]. Studies of MDSCs in melanoma patients have shown that high frequencies of M-MDSCs impair DC maturation by reducing antigen uptake, preventing migration of immature and mature DCs, skewing DC cytokine production toward an anti-inflammatory phenotype, and blocking the ability of DCs to induce IFNγ-producing T cells [ 57 ]. It was also found in a mouse model that increased interleukin-10 (IL-10) production by MDSCs in hepatocellular carcinoma inhibited the secretion of interleukin-12 (IL-12) by DCs [ 58 ]. In addition, PMN-MDSCs produce oxidatively truncated lipids that can be transferred to DCs, attenuating the ability of DCs to cross-present antigens [ 59 ]. One of the main mechanisms of MDSCs-induced NK cell incompetence is the reduction in natural killer group 2D (NKG2D) and interferon-γ (IFN-γ) expression in NK cells via TGF-β, which thereby inhibits cytotoxic potential under tumor conditions [ 60 ]. When MDSCs are adoptively transferred to tumor-bearing mice, the cytotoxic activity of NK cells can be inhibited by reducing the levels of perforin in NK cells [ 61 ]. Furthermore, MDSCs can also inhibit NK cell function through the NKp30 receptor or by downregulating the expression of CD247 on NK cells [ 62 , 63 ]. In addition, MDSCs-expressed IDO reduces NK cell activity by downregulating receptors such as NCR, NKG2D and DNAM-1 and decreasing IFN-γ secretion from NK cells [ 64 , 65 ]. This inhibition can be regulated by blocking signal transducer and activator of transcription 3 (STAT3)-mediated nuclear factor-κB (NF-κB) activation [ 66 ]. In addition to their ability to suppress immune T cells to destroy tumors, MDSCs may also be involved in tumor immune escape by stimulating other immune suppressor cells, such as macrophages and regulatory T (Treg) cells [ 67 ]. MDSCs not only are a source of tumor-associated macrophages but also may influence macrophage activation status, function, and polarization through association [ 68 ]. Driven by IL-6, the IL-6R/JAK/STAT3 pathway is activated in PMN-MDSCs, which in turn causes the synthesis and secretion of exosomal miR-93-5p, driving differentiation of M-MDSCs into M2 macrophages [ 69 ]. In the tumor microenvironment, MDSCs crosstalk with macrophages mainly through the production of IL-10, which promotes macrophage polarization toward the M2 phenotype. IL-10 produced by MDSCs also biases the differentiation of the helper T-cell population toward the Th2 phenotype, which in turn affects the development of cytotoxic T lymphocytes. Th2 cells also produce high levels of IL-4, which in turn promotes TAM development [ 70 ]. In addition, IL-10 production by MDSCs could also decrease the antigen-presenting potential of macrophages by affecting MHC II expression [ 71 ]. In a mouse model of colon cancer, increased secretion of IL10 and TGF-β by MDSCs after IFN-γ stimulation promoted the development of CD4 + CD25 + Treg cells [ 72 ]. In an A20 B-cell lymphoma model, MDSCs overexpressing low levels of MHC II have been reported to act as tolerogenic antigen-presenting cells (APCs) capable of antigenic uptake and presentation to tumor-specific Treg cells in an arginase-induced manner [ 73 ]. When adoptive transfer of MDSCs was performed from CD40-deficient mice, it failed to induce expansion and tolerance of tumor-specific Treg cells. This finding suggests, in part, a role for CD40/CD40L interactions in the crosstalk between MDSCs and Treg cells [ 74 ]. In mouse models of cancer, M-MDSCs inside tumors can generate CCR5 ligands to attract Tregs with high levels of CCR5 to infiltrate tumor tissue [ 75 ]. Additionally, in a mouse model of melanoma, Treg cells promoted MDSCs function by enhancing the levels of B7 family members of immunomodulatory ligands, such as B7-H1 (also known as PD-L1), B7-H3, and B7-H4, as well as the generation of IL-10 in MDSCs [ 76 ]. TME is a highly complex system. In addition to tumor cells and infiltrating immune cells, cancer-associated fibroblasts (CAFs) are also an important part of it. Under the influence of various cytokines and chemokines released by CAFs, MDSCs infiltrate and generate inside the tumor, thus inhibiting the anti-tumor activity of effector T cells. It has been reported that MDSCs may migrate to tumor sites induced by CAFs-activated STAT3-CCL2 signaling [ 77 ]. For example, in lung squamous cell carcinoma, CCR2 + monocytes are induced to migrate toward the tumor site by CAFs-secreted CCL2 and are then reprogrammed to M-MDSCs [ 78 ]. Another study described a similar role for CAFs-secreted CCL2 in recurrent bladder cancer [ 79 ]. In addition, STAT3 signaling was activated in recruited monocytes in hepatocellular carcinoma induced by IL-6 secreted by CAFs and promoted monocyte differentiation into M-MDSCs [ 80 ]. CAFs within hepatocellular carcinoma can also recruit M-MDSCs to hepatocellular carcinoma tissues by promoting macrophage migration inhibitory factor (MIF) secretion in a CD36-dependent manner [ 81 ]. The importance of CAFs-secreted IL-6 in the differentiation of MDSCs was reconfirmed in a study of esophageal squamous cell carcinomas, where it was observed that CAFs-derived exosome-packed microRNA-21 (miR-21) also generates M-MDSCs by activating STAT3 signaling [ 82 ]. Furthermore, IL-6 and IL-33, which are mainly expressed by CAFs, mediated the metabolism of the over-activated 5-lipoxygenase in MDSCs, and promoted the synthesis of leukotriene B4 (LTB4) in MDSCs to enhance the stemness of intrahepatic cholangiocarcinoma [ 83 ].

In recent years, significant progress has been made in cancer immunotherapy, especially in the treatment of various types of solid tumors (e.g., melanoma, breast cancer, and non-small cell lung cancer). Nevertheless, there are still many patients who do not benefit due to drug resistance or relapse, which is most likely attributable to multiple immunosuppressive cells in the TME. As prognostic and predictive biomarkers, MDSCs play an essential role in development of tumor immune escapes. Patients may benefit from targeting MDSCs due to the diverse roles of MDSCs in TME. This review describes some of the mechanisms by which MDSCs participate in tumor immune escape through immunosuppression and summarizes the specific pathways by which MDSCs have been involved in various types of tumor immune escape in recent years. Unlike Treg cells or checkpoint molecules, MDSCs do not seem to exist in a steady state. This provides a unique opportunity to target MDSCs with potentially no side effects. However, current therapeutic strategies targeting MDSCs are only partially effective [ 225 ]. First, MDSCs are highly heterogeneous in different cancers, and identification of human MDSCs phenotypes is a challenge. Second, majority of studies in humans have focused only on circulating MDSCs, and very little is known regarding tumor-infiltrating MDSCs. Then, the complex nature of the TME and the multifunctional nature of MDSCs, where inhibitory mechanisms of MDSCs are unlikely to function simultaneously, make it difficult to identify the primary targets against MDSCs. Finally, targeting M-MDSCs leads to an increase in PMN-MDSCs and vice versa, so that targeting one type of MDSCs alone may not be effective. Therefore, it seems impossible to control or eliminate MDSCs via a single approach and thus induce a significant antitumor effect, and the combination of MDSCs-targeted therapy with other anticancer therapies should be the preferred strategy. In addition, there is still a need to further investigate the major mechanisms and upstream signals underlying the emergence, amplification and immunosuppressive functions of MDSCs in various tumors. Advances in this area should help rationalize the design of new strategies against MDSCs to enhance the clinical response to current immunotherapies and improve patient prognosis. Future studies should clarify how much efficacy and survival benefit combination therapies can provide to cancer patients. A large-scale clinical trial and in-depth preclinical study are needed to confirm these questions.

PD-L1 is a key negative regulator of the immune system that mediates immune escape in tumors [ 46 ], and PD-L1 expression on MDSCs is closely associated with immunosuppression. Consistent with the immunosuppressive activity of MDSCs, it has been proven that blocking MDSCs improves the antitumor effect of programmed cell death 1 (PD-1) inhibitors in mice, which can be in conjunction with enhanced CD8 + T-cell infiltration in tumors and reduced expression of immunosuppressive proteins such as arginase 1, S100A8, S100A9, and iNOS by MDSCs [ 47 ]. In addition, several other immune checkpoint molecules, including V-domain Ig suppressor of T-cell activation (VISTA), galactose lectin-9 (Gal-9) and CD155, are involved in immune suppression mediated by MDSCs. High VISTA expression on MDSCs in the peripheral blood of patients with acute myeloid leukemia (AML) strongly correlates with PD-1 expression on T cells [ 48 ]. VISTA expression is enhanced on tumor-infiltrating MDSCs and linked to areas of severe hypoxia in the TME, and antibodies targeting or genetically ablating VISTA under hypoxia alleviate MDSCs-induced T-cell suppression [ 49 ]. Gal-9 on MDSCs can interact with T-cell immunoglobulin and mucin structural domain 3 (TIM-3) expression on T cells to expand MDSCs and suppress T-cell reactions [ 50 ]. Moreover, Gal-9 from nasopharyngeal carcinoma cells upregulated the expression of several pro-inflammatory cytokines essential to MDSCs differentiation, including IL-1β and IL-6. The process is based on enhanced interferon gene (STING) protein catabolism resulting from direct interaction of Gal-9 carbohydrate recognition domain 1 with the STING C-terminus and subsequent enhancement of K48-linked ubiquitination of STING via the E3 ubiquitin ligase tripartite motif‐containing (TRIM) 29 [ 51 ]. T-cell immunoglobulin and the ITIM domain (TIGIT) is a suppressive regulatory factor that has been shown to have an immunosuppressive effect on antitumor immunity in a wide range of solid tumors and leukemias [ 52 ]. In head and neck squamous cell carcinoma (HNSCC), CD155 expression on MDSCs promoted MDSCs-mediated T-cell suppression, and in vitro blocking the TIGIT/CD155 pathway with anti-TIGIT antibodies substantially inhibited MDSCs immunosuppressive capacity and enhanced the antitumor immune response [ 53 ].

MDSCs can induce immunosuppression by generating reactive oxygen species (ROS), which results in an oxidative stress response, promoting MDSCs expansion and suppressing the immune response of T cells [ 31 ]. Increased ROS levels also stimulate elevated expression of VEGF receptors on MDSCs, which facilitates MDSCs recruitment into the tumor microenvironment [ 32 ]. Therefore the combination of anti-VEGF and PD-1 blockade may exert better anti-tumor immune efficacy [ 33 , 34 ]. In addition, ROS can catalyze the nitration reaction of T-cell receptor (TCR)/CD8 molecules and prevent TCR/major histocompatibility complex (MHC)-peptide interactions [ 35 ]. MDSCs also overexpress inducible nitric oxide synthase and produce large quantities of reactive nitrogen species (RNS), mainly nitric oxide (NO) [ 36 ]. Rapid binding of O2 − to NO to form RNS can lead to nitration or nitrosylation of TCR/CD8 proteins, ultimately resulting in impaired recognition of TCR/MHC-peptide [ 37 ]. NO also drives P53 pathway activation in T cells to cause DNA damage, resulting in severe impairment of T-cell proliferation and survival [ 38 ].

MDSCs can impair T-cell function by reducing metabolites and factors critical to the immune system, such as L-arginine, cysteine and tryptophan (Trp) [ 39 ]. A variety of tumor microenvironment (TME)-derived factors induce the upregulation of cationic amino acid transporter protein (CAT-2B) and ARG1 expression in MDSCs. CAT-2B was able to transfer extracellular L-arginine into MDSCs, followed by degradation of L-arginine to urea and L-ornithine catalyzed by ARG1 [ 40 ]. In tumor patients, MDSCs have been found to deliver ARG1 to the extracellular environment to promote extracellular L-arginine depletion [ 41 ]. Thus, a reduction in the extracellular space arginine concentration can result in the loss of the CD3ζ chain and significant inhibition of T-cell proliferation [ 42 ]. Furthermore, MDSCs can take up cystine and metabolize it to cysteine via the xc- transporter, but because of the absence of the neutral amino acid transporter, MDSCs are unable to transport cysteine back to the extracellular level, thus affecting T-cell activation [ 43 ]. Furthermore, MDSCs also reduce Trp levels through indoleamine-2,3-dioxygenase 1 (IDO1) expression in external environment, thereby preventing T-cell development through the general control non-repressed 2 pathway [ 44 , 45 ]. The production of kynurenine and serotonin due to Trp depletion activates the aryl hydrocarbon receptor (AhR) to trigger IDO1 production and an anti-inflammatory reaction [ 45 ].

The population of myeloid cells is highly diverse. Myeloid cells include mononuclear phagocytes (MNPs) (encompassing macrophages, monocytes, and dendritic cells [DCs]) and granulocytes (mast cells, neutrophils, eosinophils, and basophils), which play a variety of different and specific roles in protecting the body in response to pathogenic stimuli. However, sustained stimulation by inflammation, chronic infection, or cancer (which involves relatively low-intensity signals) causes sustained myelopoiesis. Although the exact nature of these myeloid cells depends on the pathogenic stimulus in the host, they share several similar features: lack or reduced expression of mature myeloid cell markers, inability to differentiate into mature myeloid cells in the presence of tumor-derived factors, expression of Gr-1 and CD11b molecules in mice, high levels of reactive oxygen species, and activation of arginase I and other molecules. This endows these myeloid cells potential to suppress immune effects both in vitro and in vivo [ 1 ]. Reports on immunosuppressive myeloid cells were initially published sporadically beginning in the 1970s and 1980s, based on the fact that co-culture of activated T cells with bone marrow cells suppressed T-cell function [ 2 ]. At the beginning of the twentieth century, these myeloid cells were renamed immature cells (ImCs) or myeloid suppressor cells (MSCs). In 2007, the name MDSCs was proposed to unify the descriptions of these cell types [ 1 ]. The name is based on the fact that the cells originate from the myeloid lineage and are characterized mainly by their immunosuppressive activity. MDSCs have since been used as a catch-all term in a variety of settings, particularly in the field of cancer biology. There are two primary types of MDSCs called polymorphonuclear-MDSCs (PMN-MDSCs) and monocytic-MDSCs (M-MDSCs). These cells resemble neutrophils and monocytes phenotypically and morphologically and thus, phenotype and morphology alone are not enough to identify MDSCs. Besides the two main types of cells, MDSCs include a small population of cells (less than 3%) with myeloid colony-forming activity [ 3 ]. Murine MDSCs were initially defined as those that expressed the Gr1 and CD11b surface molecules. Therefore, in mice, PMN-MDSCs were defined as CD11b + Ly6C − Ly6G + , and M-MDSCs were defined as CD11b + Ly6C + Ly6G − [ 3 ]. In humans, PMN-MDSCs and M-MDSCs in human peripheral blood can be separated by density gradient centrifugation. Human PMN-MDSCs are often described as HLADR − CD11b + CD14 − CD15 + CD33 Mid  cells, and M-MDSCs are described as HLADR − CD11b + CD14 + CD15 − CD33 high  cells [ 4 ]. MDSCs mediate immune escape mainly by exerting immunosuppressive functions. Although MDSCs are involved in suppressing various immune cells, their primary target is T cells. In contrast depletion of MDSCs using specific antibodies enhances T cell infiltration, survival and cytotoxic efficacy driven by bispecific antibody or chimeric antigen receptor [ 5 ]. MDSCs exert immunosuppressive effects mainly by generating active ingredients such as arginase 1 (ARG1) [ 6 ], reactive oxygen species [ 7 ], and nitric oxide [ 8 ]. In the tumor microenvironment, MDSCs can display effective immunosuppressive and immune escape effects through a variety of mechanisms: depletion of metabolites essential for T-cell function, production of nitrogen species and reactive oxygen, blockade of lymphocyte homing, expression of ectoenzymes regulating adenosine metabolism, induction of immunosuppressive cells, and expression of negative immune checkpoint molecules [ 9 ]. In addition to their effects on immune responses, MDSCs promote tumor development by secreting vascular endothelial growth factor (VEGF) [ 10 ] and matrix metalloproteinase-9 (MMP9) [ 11 ] to support tumor angiogenesis and expressing CXCR2 to promote the formation of pre-metastatic niche [ 12 ]. In the last two decades of research, the immune system has been shown to paradoxically inhibit and support tumor development. This process is known as cancer immunoediting and undergoes 3 main phases, namely, elimination, equilibrium and escape [ 13 ]. During the elimination phase, the innate immune system and the adaptive immune system team up to recognize and eliminate cells that have become transformed, evading tumor suppression mechanisms. The few surviving tumor subclones can enter the equilibrium phase, where tumor growth is limited and even stagnates over time. However, subclones of tumors with low immunogenicity can be selected by the adaptive immune system in combination with the genetic instability of tumor cells to evade immune surveillance [ 14 ]. This selection process may involve various types of immune modifications rather than the death of tumor subclones. These changes include the selection of tumor variants that are resistant to immune effectors (sometimes referred to as "immunoediting") and the progressive establishment of an immunosuppressive environment within the tumor. These modified tumor cells can then enter the escape phase, and their growth will no longer be restricted by immune surveillance, leading to the development of clinically detectable tumors. The mechanisms of immune escape include abnormalities in antitumor-associated immune cells, the selection of immune resistance to tumor cells, impaired transport of T cells, and the formation of an immunosuppressive tumor microenvironment [ 15 ]. All of these processes are involved in different stages of cancer immunoediting. The complexity of the composition and spatial structure of the tumor immune microenvironment has led to the involvement of MDSCs in cancer immunoediting by various forms [ 16 ]. In recent years, research has revealed the clinical significance of MDSCs. Various studies have documented the proliferation of MDSCs in several types of human tumors, such as cutaneous melanoma [ 17 ], hepatocellular carcinoma [ 18 ], breast cancer [ 19 ], prostate cancer [ 20 ] and lung cancer [ 21 ]. In addition, a number of studies have shown that MDSCs are important prognostic biomarkers for cancer development and potential targets for anticancer therapy [ 22 ]. MDSCs can suppress the immune response and protect tumor cells from attack by the host immune system, resulting in tumor immune evasion. Targeting MDSCs to activate tumor immunity and reverse immune escape may be a viable option in tumor patients. In this review, we discuss the biological role of MDSCs in tumor immune escape. In addition, we also review the specific mechanisms by which MDSCs are involved in tumor immune escape in various types of tumors and discuss in detail the approaches used to target MDSCs for cancer treatment.

MDSCs originate from hematopoietic stem cells, common myeloid progenitors (CMPs) and granulocyte–macrophage progenitors (GMPs) [ 19 ]. GMPs then differentiate into myeloblasts (MBs), monocytes/macrophages and dendritic cells (MDPs) in reaction to multiple tumor-induced growth signals, cytokines, and other factors [ 23 ]. In the early stages of development, these cells with certain biochemical characteristics of MDSCs do not exhibit immunosuppressive activity and can be referred to as MDSCs-like cells. Under continuous stimulation by tumor-secreted factors, MDSCs-like cells expand and transform into immunosuppressive PMN-MDSCs and M-MDSCs (Fig.  1 ). Fig. 1 Differentiation and accumulation of MDSCs. MDSCs are differentiated in the bone marrow from hematopoietic stem cells (HSCs) through common myeloid progenitors (CMPs) and granulocyte–macrophage progenitors (GMPs). In response to multiple tumor-induced cytokines, GMPs differentiate to form myeloblasts (MBs), monocytes/macrophages and dendritic cells (MDPs). Most MBs and MDPs can further differentiate into neutrophils and monocytes. However, under pathological conditions, immature myeloid cells expand and transform into immunosuppressive MDSCs. In cancer patients, neutrophils and monocytes, as well as pathologically activated MDSCs, coexist at any given time and accumulate more MDSCs during tumor progression Differentiation and accumulation of MDSCs. MDSCs are differentiated in the bone marrow from hematopoietic stem cells (HSCs) through common myeloid progenitors (CMPs) and granulocyte–macrophage progenitors (GMPs). In response to multiple tumor-induced cytokines, GMPs differentiate to form myeloblasts (MBs), monocytes/macrophages and dendritic cells (MDPs). Most MBs and MDPs can further differentiate into neutrophils and monocytes. However, under pathological conditions, immature myeloid cells expand and transform into immunosuppressive MDSCs. In cancer patients, neutrophils and monocytes, as well as pathologically activated MDSCs, coexist at any given time and accumulate more MDSCs during tumor progression In cancer patients, neutrophils, monocytes and pathologically activated MDSCs coexist at any stage. As the tumor progresses, MDSCs further accumulate in the tumor immune microenvironment (Fig.  1 ). The accumulation of MDSCs is a complex phenomenon whose process can be described by a model that requires two different but partially overlapping signal types. The first one is responsible for the proliferation of immature myeloid cells in connection with the suppression of their terminal differentiation. The second one is responsible for the pathological activation of these cells and the transformation of immature myeloid cells into MDSCs [ 24 ]. The development of MDSCs in the tumor context can be divided into 4 main steps. First, factors such as Interleukin-17A (IL-17A), granulocyte–macrophage colony-stimulating factor (GM-CSF), granulocyte colony-stimulating factor (G-CSF) and tumor necrosis factor-alpha (TNF-α) from the tumor site enter the blood and then stimulate bone marrow production. Next, directed by several key chemokine receptors, such as C–C chemokine receptor 2 (CCR2) and C–C chemokine receptor 5 (CCR5), myeloid cells rapidly proliferate during myelopoiesis from the bone marrow and possibly secondary lymphoid organs into the blood. Then, under the action of chemokines, MDSCs home to the tumor site and accumulate. The final step is retention at the tumor site [ 25 ].