1. Introduction
Until recently, myeloid-derived suppressor cells (MDSCs) composed a taboo in the field of cancer immunology, since it is a vast and heterogeneous population of immature cells of the immune system [
1,
2,
3,
4]. These cells derive from hematopoietic stem cells (HSCs) residing in bone marrow (BM), which give rise to the immature myeloid cell (IMC) population [
2]. Normally, under the right combination of growth factors, the IMC population gives rise to all of the terminally differentiated myeloid cells such as neutrophils, macrophages, and dendritic cells (DCs) [
2]. However, a malfunction in the maturation process of this ancestral population favors the maintenance of a pool of MDSCs [
5]. MDSCs can arise under different circumstances in cancer. When there is need for more myeloid cells, a program called emergency myelopoiesis is activated in the BM, giving rise to MDSCs from the IMC population [
6,
7]. In the periphery, a similar procedure is initiated, called extramedullary myelopoiesis [
8]. The precursor cells, due to tumor-derived factors, might migrate out of the bone marrow into the blood, peripheral tissue, and lymph nodes. These cells would then proliferate and become MDSCs through activation at extramedullary sites [
9]. A novel hypothesis also suggests that MDSCs may arise as a part of reprogramming of the existing differentiated myeloid cells (monocytes and polymorphonuclear cells) [
9,
10,
11]. In any case, the development of MDSCs is governed by multiple signals found in their microenvironment (e.g., colony stimulating factors, growth mediators, and cytokines) that retain the ability of these cells to survive and stay undifferentiated [
9]. Once the MDSC population is established in the immune system, it is then free to execute its numerous functions, e.g., cancer progression [
5].
Given the fact that the MDSC population is actually comprised of a bounty of different cells, it is difficult to determine their actual phenotype. Nonetheless, it is evident that there are two distinct subpopulations within the major MDSC population. To begin with, a monocytic population (M-MDSC) is distinguished in mice by the expression of the surface markers CD11b and Ly6C, along with a polymorphonuclear subpopulation (PMN-MDSC) characterized by means of CD11b and Ly6G [
2]. As far as the characterization of the equivalent population in humans is concerned, the exact combination of markers still poses a challenge [
12,
13]. Regardless, some phenotypes were proposed for both the M-MDSC and the PMN-MDSC subpopulations. M-MDSCs were established as CD14
+CD15
−CD11b
+CD33
+HLA-DR
−Lin
−, as well as CD14
+CD15
+CD11b
+CD33
+HLA-DR
−Lin
−, whereas the PMN-MDSC subpopulation was designated as CD14
−CD15
+CD11b
+CD33
+HLA-DR
−Lin
− or CD11b
+CD14
−CD66b
+ [
13,
14,
15]. Recently, another MDSC subtype was proposed, called early-stage MDSC (eMDSC), which lucks the markers for both monocytic and granulocytic populations, baring the phenotype of Lin
−HLA-DR
−CD33
+CD11b
+CD14
−CD15
− [
13,
15,
16,
17,
18,
19]. These cell populations not only exist as free cells in the peripheral blood, but also as enriched cell populations in the tumor microenvironment (TME) [
20]. In the latter, MDSCs acquire a far more suppressive ability, with the M-MDSC population and the classical activated monocytes (M1) rapidly evolving into tumor-associated macrophages (TAMs), while the neutrophils tend to transform in a more suppressive subpopulation, the tumor-associated neutrophils (TANs) [
1,
15,
21].
Despite this generic discrimination between the two MDSC populations, a bias still exists regarding the accuracy of their nomenclature. This issue arises during the characterization of tumor-infiltrating myeloid (TIM) cells [
22]. Apart from MDSCs, other myeloid cells like macrophages (M1 and M2/TAMs), neutrophils (and TANs), and DCs reside in the tumor tissue [
22,
23]. Some of these cells share a common phenotypes like PMN-MDSCs and neutrophils, or even functions [
13], thus complicating the phenotypic characterization. To overcome this crucible, researches have been focused on using single-cell transcriptomics to characterize myeloid populations. Zilionis et al. demonstrated that the myeloid landscape within tumors is much more diverse and complex than originally thought [
22]. Based on the expression pattern of certain genes (chemokines, chemokine receptors, and lineage-specific molecules), they identified many different populations exerting both inflammatory and anti-inflammatory functions [
22]. Chevrier et al. also demonstrated, by single-cell transcriptomics, that within the myeloid population in renal cell carcinoma (RCC), the resident myeloid cells like macrophages display both anti-tumor and pro-tumor markers [
23]. Both teams also proved that the tissue-resident myeloid populations differ in the expression pattern of certain markers in comparison with their peripheral blood counterparts [
22,
23]. These results are indicative of the complexity and diversity of the myeloid cell populations present in both the periphery and in the afflicted tissue, proving that phenotypic characterization of TIM cells and MDSCs is a rather difficult task.
2. Development of MDSCs
It is considered a common notion that chronic inflammation promotes tumor development through various mechanisms. These mechanisms may include pro-angiogenic factors, matrix metalloproteinases (MMPs), damaged associated molecular pattern molecules (DAMPs), and activation of signaling pathways activated by a different combination of molecules existing in the peripheral blood and the TME [
24].
The TME network bares certain characteristics that, altogether, facilitate the development of cancer cells, as well as the induction and expansion of MDSCs. To begin with, the TME is as highly hypoxic environment where MDSCs can easily thrive. It has been demonstrated that MDSCs upregulate hypoxia-induced factor 1α (HIF1α), which in turn is translocated in the nucleus, binding to the hypoxia response element upstream of certain genes crucial for cancer progression [
2,
24]. Some of these targets are vascular endothelial growth factor (VEGF), inducible nitric oxide synthase (iNOS), MMPs, and arginase 1 (ARG1). The observed overexpression of HIF1α by MDSCs enables them to perform their immunosuppressive ability in the TME [
24]. Moreover, MDSCs express certain receptors for growth factors. The activation of these receptors is responsible for the development and function of MDSCs. The main pathway through which they are developed is the Janus kinase (JAK)/Signal Transducer and Activator of Transcription (STAT) pathway, with the members of the STAT protein family being the main downstream signal transductors activating transcription factors such as nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), interferon regulatory factor-8 (IRF-8), CCAAT-enhancer-binding proteins (C/EBPβ), and hypoxia-inducible factor 1-alpha (HIF-1α) [
2,
15,
24,
25]. The above signaling cascade can be activated by various factors situated in the TME, with the most common among them being growth factors secreted by cancer cells, pro-inflammatory cytokines, calcium binding proteins (calgranulin A/B; S100A8/A9), heat shock protein (HSP72), high-mobility group box 1 (HMGB1), and other molecules in the TME [
15,
24,
25].
2.1. VEGF-Driven MDSC Development
A most distinguishing example of a receptor involved in MDSC development is the VEGF receptor (VEGF2R). VEGF is a growth factor secreted by cancer cells and which enables neo-angiogenesis and metastasis. The circulating VEGF emitted in the TME acts as a chemo attractant for MDSCs [
5,
18,
26]. Specifically, VEGF attracts the MDSCs to migrate from the BM to the periphery, thus increasing their presence in blood circulation [
27]. Binding of the VEGF to its receptor is associated with increased production of reactive oxygen species (ROS) via the activation of the JAK2/STAT3 pathway. MDSCs themselves can produce VEGF, creating a positive autocrine feedback loop [
24].
2.2. G-CSF and GM-CSF-Driven MDSC Development
Other factors responsible for the development of MDSC are the granulocyte colony-stimulating factor (G-CSF) and the granulocyte–monocyte colony-stimulating factor (GM-CSF) [
5,
26]. G-CSF binds to the CSF3 receptor (CSF3R), activating a JAK/STAT downstream signal transduction. This signal activates the transcription factors MyC and C/EBP that promote the development and immunosuppressive functions of PMN-MDSC [
28]. Similar to G-CSF, GM-CSF has its own receptor, the GM-CSF2 receptor (CSF2R). Binding of its ligand to the CSF2R activates the JAK/STAT pathway, and high concentration of GM-CSF is correlated with generation and mobilization of M-MDSC and PMN-MDSC from the bone marrow and increased immune suppression [
24,
29].
2.3. Cytokine-Driven MDSC Development
Cytokines can facilitate the recruitment and activation of MDSCs in the TME, such as transformation growth factor β (TGFβ), interleukin-(IL)-4, IL-13, IL-28 (IFN-λ), IL-17, IL-10, and IL-1β [
17,
24,
30]. As it seems, by means of experiments conducted on tumor-bearing mice, IL-1β is a potent inducer of MDSCs [
24]. Not only does IL-1β aid in the accumulation of MDSCs, but it also commences the production of other molecules necessary for the expansion of MDSCs (e.g., VEGF, IL-6, GM-CSF) [
24]. In the TME, there is an abundance of TGF-β, which is able to generate MDSC populations, as well as prime other myeloid lineages to more suppressive ones [
5,
31]. Afterwards, MDSCs themselves are able to produce TGF-β, creating a feedback loop that sustains their antitumor immunity [
31]. S100A8/A9 is a pro-inflammatory stimulator that activates the STAT3/5 pathway responsible for keeping the immature myeloid cells from differentiating into their mature lineages, which makes it also responsible for activating MDSC’s suppressive mechanisms [
24,
25].
3. Accumulation of MDSCs to Tumor Sites
Tumor niches demonstrate an abundance of different cytokines and chemokines that are implicated in the recruitment of immunosuppressive cells. C–C motif chemokine ligand 2 (CCL2) was initially characterized as a cytokine, which, upon interaction with its correspondent receptor, CCR2, on circulating monocytes, could expedite chemotaxis to areas of inflammation [
21,
32]. It has been shown that in in vitro experiments of human cancer models, secretion of tumor-derived CCL2 attracts CCR2 expressing MDSCs towards the cytokine [
21], with evidence provided by Guan et al. positively correlating the amount of CCL2 to MDSC accumulation and immunosuppressive capacity [
33]. In a similar way, the chemokine interleukin-8 (IL-8, CXCL8) is a chemo attractant that was found to be released by cancer cells and to ulcerate cells of myeloid lineages upon binding to G protein-coupled receptors C–X–C motif chemokine receptor 1 and 2 (CXCR1 and CXCR2) [
32,
34], while the CCL3/CCR5 axis was shown to aid in the maintenance of immunosuppressive myeloid cells to the tumor niches [
21]. Guan et al. also highlighted that IL-17, a pro-inflammatory cytokine mainly secreted by Th17 cells, is over expressed by malignant cells and promotes the translocation of MDSCs from the periphery to the tumor site [
33].
In addition, the highly hypoxic TME contributes to the accumulation of MDSCs. In a more specific manner of speaking, HIF-1α was shown to induce the expression of ectonucleoside triphosphate diphosphohydrolase 2 (ENTPD2), commonly known as CD39L1, aiding their movement towards the TME [
35]. In the hypoxic TME, VEGF is thought to be the dominant chemoattractant for MDSCs, as it was shown that, both in mice and in non-small cell lung cancer (NSCLC) patients, hypoxia upregulates its expression and aids in MDSC accumulation [
36]. This process of VEGF-mediated attraction to the TME is possible by means of VEGFR expression on MDSCs [
37]. Primary tumor clusters can potentially gather MDSCs from the BM by releasing exosomes. The exosomal content is able to reprogram the target cell, leading to increased mobility of the progenitor myeloid populations to the tumor site [
18,
38].
6. MDSC Targeting for Cancer Therapy
Understanding the role and functions of MDSCs in cancer-related conditions is of vital importance. Every day, more and more therapeutic approaches are designed targeting MDSCs in an attempt to exterminate cancer cells [
32]. These approaches are immunotherapy, chemotherapy, or radiation therapy, or it could be a combination of the above [
84]. Some of these approaches are currently undergoing clinical trials to estimate the efficacy and the safety of their application in cancer patients (
Table 2). These trials involve both direct and indirect MDSC depletion and blood vessel shrinkage. Blocking of several mechanisms that are used by MDSCs to treat cancer were examined by various trials.
Other approaches include the maturation of MDSCs to their terminal lineages (macrophages, DCs, and neutrophils), blockade of VEGF signaling, and combinational treatments [
84,
85]. Treatment with vitamin E reduced the NO-mediated immunosuppressive function of MDSCs and enhanced NK cytotoxicity [
63]. Consequently, treatment with all-trans-retinoic acid (ATRA) induces the differentiation of MDSCs to DCs and macrophages reducing their immunosuppressive activity, while ameliorates the efficacy of anti-VEGFR2 immunotherapy [
63,
86]. Bennewith’s group demonstrated the importance of monitoring the presence of MDSCs in a mice model of metastatic breast cancer, indicating the role of MDSCs in correlation with increased metastatic potential [
87]. It constitutes a common notion that multiple signaling pathways are involved in cancer metastasis and angiogenesis. The JAK2/STAT3 is one that has drawn many researchers’ attention, since it mediates angiogenesis in both cancer cells and MDSCs [
65]. Since the immunosuppressive and angiogenic functions of MDSCs are regulated by the JAK2/STAT3 signaling cascade, it is only rational to assume that its inhibition could rescue immunosuppression and extravasation.
In the light of the realization of the role of MDSCs in cancer progression, many studies have tried to unmask different approaches. Knowing that MDSCs promote vascularization, treatment options targeting angiogenic pathways are imperative. In fact, many studies have demonstrated the efficacy of such procedures. Previous work from our group demonstrated that anti-VEGF-based combinational chemotherapy had an impact on PMN-MDSC reduction in the peripheral blood of NSCLC patients [
27]. Ko et al. also demonstrated the reductive effect of sunitinib, a receptor tyrosine kinase inhibitor, on MDSCs in renal cell carcinoma patients [
85]. Consequently, Liu and co-workers investigated the effect of JAK2/STAT3 blockade on MDSCs. They discovered that inhibition of this signaling cascade in head and neck squamous cell carcinoma (HNSCC)-bearing mice minimized the production of VEGF and CK2, subsequently leading to the reduction of vessel formation [
65]. Moreover, Bauer et al. showed that treatment of MDSCs with all-trans retinoid acid (ATRA) hampered their accumulation and decreased the production of MMP9, indicating the possible effect of ATRA in vessel normalization and MDSC reduction [
88]. Other experimental approaches targeting MDSCs in the concept of cancer include the blockade of accumulation pathways like the CCL/CCR2 axis and IL-8/CXCR1/2 interactions, which may hinder the trafficking of MDSCs to the TME and inhibit extravasation [
32]. Overall, current studies regarding anti-VEGF treatment are centered on the production of tyrosine kinase inhibitors in order to arrest the activation of VEGFR-mediated vascularization and other secondary angiogenic pathways using the same signaling transducers [
89].
Although many clinical trials have been conducted in the hope of curing cancer, not all of them are crowned winners. Nonetheless, a research group demonstrated in a clinical trial (NCT02922764) (
Table 1) that depletion of MDSCs leads to cancer reduction. Tavazoie et al. researched the impact of liver-X nuclear receptor (LXR)/Apolipoprotein E (ApoE) signal on MDSCs, and what they found was astonishing. Treatment of MDSCs with agonist of LXR/ApoE, led to the depletion of MDSCs in in vitro as well as in in vivo models. Namely, the RGX-104 LXR agonist reduced the amount of PMN-MDSCs and M-MDSCs in the tumor site and in the periphery by initiating apoptosis, whereas in ApoE-deficient mice, there was no observable MDSC depletion, indicating that ApoE is necessary for the killing of MDSCs in cooperation with LXR agonism. Thus, Tavazoie et al. suggested that synergistically, LXR/ApoE are efficient in targeting and eliminating MDSCs in the context of cancer [
90].
However, taking into consideration that in the TME, MDSCs are the only tumor-promoting populations, many studies have focused on targeting TAMs and TANs. Indeed, TAMs are able to induce extravasation in tumor sites, thus targeting them is imperative to fight tumor growth and metastasis [
91]. In this concept, many approaches have been tested, including the common TIKs that are widely used, but the results are not that promising [
91]. Nonetheless, Penn et al. demonstrated that blockade of VEGF-C, which is secreted by TAMs, limited lymphagiogenesis and depleted TAMs [
92]. As far as TANs are concerned, it was shown that they exert an anti-VEGF function. Specifically, Seeger et al. showed that the presence of CD177
+ neutrophils in colorectal cancer patients is associated with poor prognosis [
93]. Finally, blocking of the Ang2/Tie axis was shown to reduce the tumor-promoting functions and recruitment of TAMs [
92,
94], and could be used as credentials denoting that blocking of secondary angiogenic mechanisms could improve anti-VEGF treatment [
93].
7. Conclusions and Future Perspectives
In conclusion, the role of MDSCs in cancer firstly lies in the establishment of an immunosuppressive environment, and secondly, in the remodeling of the TME in order to initiate tumor invasiveness and metastasis. During cancer progression, MDSC-mediated extravasation is exerted by many different pathways, but first and foremost, by the production of VEGF and MMPs. Nonetheless, secondary mechanisms also exist to support the angiogenic process, resulting in the composition of a complex signaling network. Considering that the MDSC population is vast and exerts its functions in different manners, it is relatively difficult to devise a therapeutic approach to diminish their numbers or arrest them in the peripheral blood and in the TME. Regardless, many studies have demonstrated the efficacy of MDSC targeting and the possibility of overcoming the boundaries in MDSC-reducing cancer treatment. Further studies are in order to reinforce the arsenal of knowledge about MDSCs and to better understand the possibilities of combinational treatments in response to MDSC-supported cancer progression.