Emerging Roles of Myeloid-Derived Suppressor Cells in Diabetes.

In several complications of diabetes, specific organ damage leads to the corresponding increase in the recruitment and activity of MDSC. MDSC play different functions in different complications. Among them, the complications caused by MDSC microangiopathies, such as diabetic retinopathy, diabetic nephropathy, and diabetic refractory wounds are prominent. These conditions will be discussed separately below. Diabetic nephropathy is the leading cause of the end-stage renal disease ( Li et al., 2016 ). The pathogenesis mainly lies in the fibrosis caused by the accumulation of extracellular matrix proteins in the glomerular mesangial interstitium ( Chang et al., 2014 ). The number of MDSC in the kidney was increased ( Xing et al., 2017 ), and the adoptive transfer of MDSC induced by cytokines reduced fibronectin levels in the glomerulus and resulted in a normal glomerular filtration rate ( Hsieh et al., 2018 ). PMN-MDSC in the kidneys of patients with T2DN might not be sufficient to maintain renal function, resulting in compensatory anti-inflammatory failure of the kidney ( Duran-Salgado and Montserrat, 2014 ; Islam et al., 2020 ). The increased number and enhanced anti-inflammatory ability of PMN-MDSC may be one of the therapeutic targets for diabetic nephropathy. Atherosclerosis (AS) is the leading cause of coronary heart disease, cerebral infarction and peripheral vascular disease ( Lusis, 2000 ). Diabetes accompanied by high fat and an inflammatory environment is a fundamental cause of the development of atherosclerosis ( Poznyak et al., 2020 ). The role of MDSC and their subgroups in AS is still controversial ( Wang et al., 2015 ; Fernández-Ruiz et al., 2019 ). It was demonstrated that MDSC increased about two-fold in the bone marrow of AS model mice, where M-MDSC increased in proportion and PMN-MDSC decreased, and the inhibitory activity of M-MDSC was enhanced ( Foks et al., 2016 ). This was consistent with the phenomenon observed in peripheral blood samples of AS patients ( Wang et al., 2015 ). Moreover, the frequency of PMN-MDSC was negatively correlated with low-density lipoprotein cholesterol ( Fernandez-Ruiz et al., 2019 ). It was proposed that the increase of MDSC in the bone marrow of depressed AS mice, especially PMN-MDSC, could increase neutrophil traps (NETS) and aggravate AS ( Yamamoto et al., 2018 ). The opposing effects of the two subgroups of MDSC remind us to observe the role of MDSC in AS and score clear subgroups. In the LDLr -/- model ( Foks et al., 2016 ) and the ApoE -/- murine model ( Wang et al., 2020 ) fed on the Western-type diet (WTD, containing 0.25% cholesterol and 15% cocoa butter) diet (0.25% cholesterol and 15% cocoa butter), MDSC were shown to significantly slow down the disease process of AS after adoptive transfer. Also, the therapy for MDSC was regarded as one of the hopes for the treatment of atherosclerosis. It was proved that oral HSP60 reduced the development of AS by increasing the number of M-MDSC and enhancing its function, while subcutaneous HSP60 caused the opposite response ( Hu et al., 2018 ). In addition, SBI-0206965, an inhibitor of autophagy, rapidly reduced MDSC and promoted the development of atherosclerosis ( Wang et al., 2020 ). Therefore, the immunomodulation of MDSC and its subgroups may be regarded as a potential treatment of atherosclerosis. The pathogenesis of diabetic retinopathy lies in the structural disorder of microvessels. The main sign of the development of the non-proliferative type to a more advanced proliferative type is the proliferation of ocular neovascularization ( Hendrick et al., 2015 ). Existing drugs are mainly focused on preventing neovascularization, such as the anti-VEGF drug ranibizumab ( Chatziralli and Loewenstein, 2021 ). In the ocular humor of patients with DR, both levels of IL-6 and CCL2, which are important proliferative factors of MDSC, and the number and activity of myeloid cells were observed to be increased ( Koleva-Georgieva et al., 2011 ; Koskela et al., 2013 ). MDSC play a role in stabilizing blood vessels in diabetic retinopathy ( Liyanage et al., 2016 ; Villacampa et al., 2020 ). MDSC reduced retinal neovascularization in oxygen-induced retinopathy ( Kataoka et al., 2011 ; Xu et al., 2018 ) and laser-induced choroidal neovascularization ( Espinosa-Heidmann et al., 2003 ; Sakurai et al., 2003 ; Nagai et al., 2007 ). These actions of MDSC were different from those noted during tumor angiogenesis ( Albini et al., 2018 ; Zhang T. et al., 2018 ; Yang et al., 2018 ; Lin et al., 2019 ; Rahma and Hodi, 2019 ; Dysthe and Parihar, 2020 ). The difficulty in wound healing of diabetic patients is caused by chronic inflammation, vascular endothelial injury, hypoxia, autonomic nervous dysfunction and decreased neuropeptide signal transduction ( Noor et al., 2015 ). The diabetic foot ulcer is the most crucial reason for the amputation of patients with non-traumatic conditions ( Everett and Mathioudakis, 2018 ). MDSC distribution could change due to the addition of the wounds. During the period from the inflammatory to the proliferative phase, a new round of proliferation of bone marrow MDSC was stimulated by the wound. MDSC led to peripheral distribution and targeted the wound ( Mahdipour et al., 2011 ; Li et al., 2021 ). It was shown that the CD11b + Ly6C hi cell group on the wound was rapidly transformed into CD11b + Ly6C low cells within 1–2 days of wound formation. Subsequently, an additional CD11b + Ly6C hi cell group flowed into the wound on the 3rd-4th day sequentially ( Kimball et al., 2018 ). Nevertheless, MDSC were retained at a higher density explicitly in the assumed granulation tissue area of the wound. They were associated with endothelial cells at the injury site and their frequency was higher than that of non-diabetic mice ( Torbica et al., 2019 ). The role of MDSC in diabetic wounds was examined, suggesting that immature myeloid cells could impair diabetic wound healing, while the use of G-CSF in db/db wounds could accelerate wound healing ( Wicks et al., 2015 ). However, the positive effect of MDSC on wound healing was supported by various studies. The fact that the adoptive transfer of MDSC derived from the spleen ( Mahdipour et al., 2011 ) or the bone marrow ( Tong et al., 2014 ) of mice to diabetic wounds could assist wound healing has become an essential supporting basis. The number of blood vessels was analyzed in diabetic and non-diabetic mice by injecting bone marrow-derived MDSC into the wounds of diabetic and non-diabetic mice. The results demonstrated that MDSC benefited angiogenesis, whereas the diabetic microenvironment impaired their ability ( Mahdipour et al., 2011 ). Besides, MDSC derived from the bone marrow of diabetic mice indicated decreased proliferation and differentiation, decreased chemotactic function, lower expression of VEGF and higher MMP-9 levels as determined by in vitro studies ( Torbica et al., 2019 ). In addition, the recruitment of MDSC in wounds might be impaired by damaging the SDF-1/CXCR4 axis ( Tong et al., 2014 ). Abnormal differentiation of MDSC typing can also be noted in diabetic wounds. Bone marrow cells can be activated or polarized into different states related to Th1 and Th2 cytokines by the local microenvironment. These myeloid cells are termed classically activated (M1) or alternatively activated (M2) cells. The diabetic microenvironment affected the pedigree commitment of these progenitor cells, inhibited granulosa cell differentiation and promoted monocyte differentiation ( Mahdipour et al., 2011 ). The M1 group on the diabetic wound was mainly composed of M-MDSC. In the subsequent stage of healing, macrophages were transformed into M2 cells in the non-diabetic wound and retained a large amount of pro-inflammatory M-MDSC in the diabetic wound ( Mahdipour et al., 2011 ; Bannon et al., 2013 ). The transformation of MDSC to PMN-MDSC (like overexpressing Hoxa3) resulted in significant induction of neovascularization in diabetic wounds. In view of these studies, it is suggested that MDSC may promote wound healing by their conversion to CD11b + Ly6C low cells, which in turn promote angiogenesis. However, the effect of the complex diabetic microenvironment on MDSC, such as the imbalance between the different types of MDSC, may impair wound healing ( Figure 2 ). MDSC is rapidly transformed into macrophages following recruitment to the wound. This process exerts an anti-inflammatory effect and promotes wound healing. MDSC remains in diabetic wounds and affects angiogenesis, whereas diabetes hinders the recruitment of MDSC from the wound; the existence of an imbalance ratio of PMN/M-MDSC keeps the wound in an inflammatory microenvironment hindering wound healing. Therefore, MDSC promotes wound healing, whereas excessive M-MDSC impairs wound healing.

Stem cell therapy has been a research hotspot in recent years. As a newly discovered subgroup of MDSC, f-MDSC are CD33 + IL-4R α + fibrous cells differentiated by umbilical cord blood progenitor cells cultured with FDA-approved cytokines (rh-GM-CSF and rh-G-CSF). These cells are cultured for 4 days and display a fibroblast-like shape. They are also characterized by cytoplasmic elongation, nuclear nucleoli, phagocytic extension and high adhesion to plastic ( Mazza et al., 2014 ). F-MDSC have been shown to produce IDO following their interaction with activated T cells in NOD/SCID mice in order to promote Tregs differentiation and reduce blood glucose to normal levels for therapeutic purposes ( Zoso et al., 2014 ). Human umbilical cord mesenchymal stem cells (Huc-MSCs), which are widely used in NOD mice, can inhibit the differentiation of MDSC by secreting the soluble factors COX2/PGE2 and IFN-β and enhance their inhibitory ability on immune cells so as to achieve effective therapeutic effects against diabetes ( Qi et al., 2020 ). Based on the aforementioned findings, the metastasis of MDSC, f-MDSC, or the transfer of Huc-MSCs may possess the enhanced inhibitory ability on immunity and demonstrate the therapeutic effect on diabetic mice ( Yin et al., 2010 ; Xia et al., 2011 ; Torbica et al., 2019 ; Ren et al., 2021 ). MDSC-related stem cells also may be a promising treatment for diabetes. The use of antibodies against the molecular markers of MDSC was initially intended to deplete the MDSC of the subject. A one-time intravenous injection of anti-Gr-1 only temporarily reduced MDSC in the body, followed by a long-term increase in Gr1 + CD11b + cells, which was similar to the findings noted in the tumor microenvironment ( Ma et al., 2012 ). When the anti-Gr-1 antibody (0.25 mg/mice) was injected intravenously in NOD mice, MDSC depletion lasted only 4 days and the number of MDSC was significantly higher than that in the control group noted from the 7th to the 17th day following treatment ( Hu et al., 2012 ). These findings suggest that short-term injection of anti-Gr-1 antibody may induce a long-term compensatory increase. Rapamycin is a specific inhibitor of mTOR. It can trigger the inhibition of mTORC1, which leads to the increase of Tregs and MDSC. Rapamycin reduced the phosphorylation of S612 (insulin receptor substrate-1) in adipose, muscle and liver tissues, which inhibited the degradation of IRS-1 and improved insulin sensitivity in mice. Concomitantly, rapamycin also adjusted the classification of MDSC, increased the number of PMN-MDSC in fat and liver tissues and in the blood and reduced M-MDSC, thereby reducing inflammation ( Pederson et al., 2001 ; Makki et al., 2014 ). It was indicated that the mTOR inhibitor INK128 could inhibit the differentiation of M-MDSC into M1 pro-inflammatory macrophages, thus reducing inflammation and promoting diabetic wound healing ( Li et al., 2021 ). Knockout of the C3 complement gene significantly promoted the immunosuppressive ability of MDSC. In STZ-induced T1D mice, MDSC were highly activated and suppressed T cells in order to regulate Treg cells via TGF-β secretion. A similar effect was achieved by using the complement activation inhibitor FUT-175 ( Gao et al., 2013 ). IL-17 -/- mice resisted STZ-induced diabetes by increasing the percentage and number of MDSC in the spleen and enhancing their immunosuppressive ability ( Tong et al., 2015 ). Therefore, it was expected that IL-17 inhibitors could also play an anti-diabetic effect. In addition, K118 is an inhibitor of PISHIP1, which was shown to target SHIP1 in MDSC derived directly from visceral adipose tissues and increase the number of MDSC and improve blood glucose control and insulin sensitivity. K118-treated mice demonstrated no harmful side effects in the lung, small intestine, or other organs and in the bone mineral density ( Srivastava et al., 2016 ). Although the number of activated CD4 + T and CD8 + T cells was decreased, MDSC was determined by the positive expression of IL4aR and Arg and the T cell inhibitory effect of MDSC was not directly measured in that study. Dietary polyunsaturated fatty acids rely on STAT3 signaling to increase MDSC and ROS production to enhance their immunosuppressive ability, which was almost entirely reversed by application of the STAT3 inhibitor JSI-124 ( Yan et al., 2013 ). In tumors, MDSC use fatty acid oxidation as their energy supply and the tumor microenvironment can cause upregulation of the expression of enzymes critical to fatty acid oxidation, thus increasing the inhibitory ability of MDSC. These findings were confirmed both in mice and humans ( Hossain et al., 2015 ). Therefore, a new theoretical basis has been deduced for dietary recommendations to increase the intake of polyunsaturated fatty acids for T2D. Nevertheless, as a sensitive cell group, the effect of comprehensive food intake on MDSC may change. In a recent experiment that examined atherosclerotic subjects, the number and proportion of MDSC in the bone marrow of high-fat diet mice were both decreased following co-administration with omega-3 polyunsaturated fatty acids, flavanols, and phytosterols ( Moss et al., 2021 ). However, the number and function of MDSC in the blood are not currently known under these feeding conditions. Additional research is thus required to explain these findings. Brazilian propolis is a resin mixture of African honey bee saliva and wax mixed with plant exudates ( Kumazawa et al., 2003 ; Bankova et al., 2014 ). It has been widely used in folk medicine due to its anti-inflammatory, anti-viral, analgesic and metabolic effects ( Tsuchiya et al., 2018 ; Al-Hariri and Abualait, 2020 ). Following intraperitoneal injection, Brazilian propolis was shown to induce visceral adipose tissue and intraperitoneal MDSC production in mouse models, which exerts an anti-inflammatory effect and improves the severity of T2D ( Kitamura et al., 2018 ). Oral administration of PEE should be recommended to assess the induction of the stimulation of human MDSC and the potential side effects. Intestinal flora plays a vital role in the pathogenesis of diabetes and has been a research hot spot in recent years ( Dolpady et al., 2016 ; Ghaisas et al., 2016 ; Jia et al., 2017 ; Henschel et al., 2018 ; Mullaney et al., 2018 ). The IgM purified from the serum of normal mice can maintain the normal Bacteroides: Firmicutes ratio and reverse the pathogenesis of diabetes in NOD mice following administration by intraperitoneal injection. The number of Tregs and MDSC in mice treated with IgM was significantly increased. Oral feeding exhibited a certain effect on this process ( Chhabra et al., 2018 ). This also should be used as another convenient and feasible way to increase the number of MDSC. Worm infection and its antigens, such as soluble (TCS) or excretory/secretory (TCES) antigens derived from Taeniasolium can increase the number of MDSC in T1D mice. Intravenous injection of dichloromethylene diphosphonate (clodronate) encapsulated in liposomes depleted macrophages but increased the number of MDSC and their subtypes ( Espinoza-Jiménez et al., 2017 ). Various factors can affect the proliferation and recruitment of MDSC and related studies have been performed in the tumor microenvironment. Among them, HIF-1α, which is decreased due to the instability of the HIF protein in the diabetic microenvironment, is a strong chemokine of MDSC and regulates their function. HIF-1 can bind to the erythropoietin gene promoter during hypoxia and form heterodimerization of HIF-1α and HIF-1β. Insulin signaling upregulates HIF-1α through phosphorylation of PI3K and MAPK. However, diabetic patients had impaired insulin signaling due to insulin resistance. The levels of HIF-1 α decreased and the lack of HIF-1α further weakened the function and survival of ß cells, forming a vicious circle ( Cheng et al., 2010 ). The induction of hyperglycemia enabled the stimulation of the degradation of HIF-1α by 2-methyl Glyoxal and inhibited its transcriptional activity. 2-methyl Glyoxal inhibited the formation of the HIF-1α-HIF-1β dimer ( Bento and Pereira, 2011 ). In obese diabetic subjects, a decrease in succinic acid was caused by fatty acid metabolism and an increase in HIF-1α protein hydrolysis ( Dodd et al., 2018 ). HIF-1α was shown to be effective in increasing the number and function of MDSC in the tumor microenvironment. HIF-1α also activated glucose transporter-1 (Glut-1) to promote glycolysis, thus exerting the effect of glycolysis on MDSC ( Choi, 2017 ). HIF-1α yet induced the expression of nucleoside diphosphate hydrolase 2 ENTPD2/CD39L1 in order to consume extracellular ATP, which in turn promoted the maintenance of MDSC ( Chiu et al., 2017 ). It is known that metformin can reduce the phosphorylation levels of STAT3 and inhibit the expression of CD73/CD39 on MDSC by activating AMPK and inhibiting the HIF-1α pathway to exert an inhibitory effect on MDSC ( Li L. et al., 2018 ; Xu et al., 2019 ). However, it is unclear whether metformin has a special regulatory effect on MDSC in a diabetic microenvironment.

MDSC are immature bone marrow cells in healthy individuals that account for 0.5–1% of peripheral blood HLA-DR - cells ( Whitfield-Larry et al., 2014 ; Hassan et al., 2018 ). Recent studies have demonstrated that the total numbers of MDSC increased in T1D and T2D subjects (summarized in Table 1 ). Quantity changes of MDSC and subsets of MDSC in the environment of diabetes. NOD mice, (non-obese diabetic mice); STZ mice, (streptozotocin-induced diabetic mice); PMN-MDSC (polymorphonuclear myeloid-derived suppressor cells); M-MDSC, (monocytes myeloid-derived suppressor cells); T1D, (type 1 diabetes); T2D, (type 2 diabetes). In T1D, the non-obese diabetic (NOD) mouse model and STZ mouse model are commonly used in basic experiments ( Busineni et al., 2015 ; Chen et al., 2018 ). Compared with prediabetic (10–14 week old) NOD mice, newly diabetic NOD mice had an MDSC (mainly M-MDSC) expansion in bone marrow, peripheral blood and secondary lymphoid organs ( Whitfield-Larry et al., 2014 ), while MDSC reduced in pancreatic islets, which was similar to Fu’s research ( Fu et al., 2012 ) that indicated a negative correlation between MDSC in the islets and the progression of diabetes. The decrease of MDSC in the islets may be one of the reasons for the failure to salvage islet inflammation. After STZ injection, the number of MDSC in the peripheral blood increased significantly from day 3 and continued to increase to day 24, after which it remained stable at approximately twice of the normal control group ( Venneri et al., 2015 ). Afterward, in the fourth week, the proportion of MDSC in the peripheral blood experienced the change from low to high, which may be due to the differentiation of MDSC ( Hsieh et al., 2018 ). As for in the bone marrow, initially, the number of MDSC appeared to decrease and subsequently (about 2 weeks later) return to normal levels ( Furman, 2015 ; Hsieh et al., 2018 ; Kim et al., 2018 ). In addition, a significant increase was noted in the number of MDSC in other organs. Increased number of MDSC in the spleen, bone marrow, kidney and pancreatic lymph nodes (PLN) was observed 3 weeks following STZ treatment ( Gao et al., 2013 ; Hsieh et al., 2018 ). PMN-MDSC accounted for a large proportion of these cells in the bone marrow, whereas the PMN/M ratio was decreased ( Gao et al., 2013 ; Kim et al., 2018 ; Li et al., 2021 ). In the PLN, the ratio of M-MDSC was decreased 15 days following STZ treatment ( Carlos et al., 2017 ). MDSC in the peripheral blood of patients with T1D were also shown to be significantly higher than those of normal healthy volunteers. In an early study, MDSC (mainly M-MDSC) expansion was documented in the peripheral blood of T1D patients ( Whitfield-Larry et al., 2014 ). Similar results were obtained in the blood of patients with diabetic nephropathy ( Hassan et al., 2018 ). Also, it was noted that the frequency of M-MDSC in the group with HbA1c >7.5% was significantly higher ( Skyler, 2004 ). However, a slight decrease in the M-MDSC and a significant increase in the PMN-MDSC of CD14 − were also indicated in the peripheral blood of patients with T1D ( Hassan et al., 2018 ). In T2D, leptin-deficient ob/ob mice and leptin receptor-deficient db/db mice can be used as the rodent model of spontaneous type 2 diabetes ( Wang et al., 2014 ; Al-Awar et al., 2016 ; Todd, 2016 ). The frequency of MDSC in the spleen and the peripheral blood of db/db mice increased ( Wang T. et al., 2018 ), and the proportion of MDSC in the blood was positively correlated with the fast blood glucose value in the ob/ob mouse model. Despite these findings, the proportion of MDSC in the bone marrow did not change. The number of MDSC in the spleen, fat and liver of peripheral organs significantly increased in the T2D mouse models ( Xia et al., 2011 ). As for the T2D patients, MDSC exhibited a statistical increase in peripheral blood samples, based on different clinical sample sizes ( Wang T. et al., 2018 ; Fernández-Ruiz et al., 2019 ; Islam et al., 2020 ). Further evidence suggests that the absolute number of M-MDSC was increased in the peripheral blood of patients with T2D compared with that noted in obese normal glucose volunteers ( Friedrich et al., 2019 ). Therefore, both in the animal model of diabetes and in the blood samples of clinical diabetic patients, total MDSC cells showed an increase, while the performance of the two subsets of MDSC is not uniform, giving us a better understanding of the diabetic pathological process.

Altogether, diabetes is mainly divided into T1D and T2D. We summarized the variation in quantities, classification, activity and immunosuppressive ability of MDSC in T1D and T2D. The underlying roles in diabetes and potential MDSC-targeting diabetes treatment were assessed. The diabetes environment activates the one-time development of MDSC in the bone marrow and promotes the accumulation of MDSC in peripheral organs except for pancreatic islets. This is inseparable from the abundant MDSC recruiting factors, such as glycolysis products, inflammatory factors, CCL2, etc. MDSC exert a certain degree of immunosuppressive ability in T1D, while their ability to regulate inflammation and immunosuppression in T2D slow down the progress of diabetes. Therefore, many possibilities for the treatment of diabetes related to MDSC have been derived. In addition, the main pathology of diabetic complications focuses on the disorder of blood vessel formation and the inflammatory environment. The stable angiogenesis and immunosuppressive ability of MDSC should have a therapeutic effect on complications. However, the performance of MDSC subsets in diabetic complications differs widely. NETS developed by excessive PMN-MDSC in AS and pro-inflammatory excess of M-MDSC in diabetic refractory wounds aggravates the development of the disease. In-depth understanding of MDSC and its subsets, and intervention and adjustment according to different pathological characteristics, will gradually become the key to making good use of the double-edged sword of MDSC and personalized immunotherapy.

Diabetes has widespread incidence in almost every country and age group, which severely affects the worldwide economy and is regarded as an epidemic. According to the International Diabetes Federation, approximately 436 million patients worldwide had diabetes in 2019 and it was predicted that this number will exceed 700 million by 2045 ( Saeedi et al., 2019 ). Diabetes is mainly divided into type 1 diabetes (T1D) and type 2 diabetes (T2D). T1D accounts for less than 10% of the total incidence of the disease, while T2D accounts for more than 90% of diabetic cases ( Association, 2013 ). The immune system plays an essential role in the pathogenesis of diabetes: the insufficient central tolerance of thymus, too little variable number of tandem repeats (VNTR), quantitative and quality defects in Tregs lead to CD8 + T cells attacking islet β cells, while inflammatory cytokines and recruitment of macrophages, B cells and CD4 + T cells assist in attacking islet β cells, resulting in insufficient insulin secretion ( Donath et al., 2019 ; Chen et al., 2021 ; Petrelli et al., 2021 ; Roep et al., 2021 ). At present, the drugs available on the market are mainly oral hypoglycemic drugs and insulin infusion, but there is no treatment for the basic pathological process that leads to ß -cell failure and destruction. Although myeloid-derived suppressor cells (MDSC) have been reported to implicate in the incidence of diabetes, it has not been fully explored with regard to their interactions with diabetes ( Whitfield-Larry et al., 2014 ; Wang T. et al., 2018 ; Hassan et al., 2018 ; Grohová et al., 2020 ). Herein, we reviewed the quantity, subtypes and activity of MDSC in the diabetic microenvironment. We also introduced the drugs targeting MDSC to delay the progression of this disease.

MDSC have been defined for more than 30 years since the initial description of cancer patients ( Buessow et al., 1984 ; Young et al., 1987 ; Seung and Schreiber, 1995 ). Immature bone marrow cells undergo activation, proliferation and differentiation into MDSC. MDSC are defined based on the CD11b + Gr-1 + phenotype in mice, whereas monocytic MDSC (M-MDSC) are defined as CD11b + Ly6C hi Ly6G – cells with low side scatter. Polymorphonuclear MDSC (PMN-MDSC) that is granulocytic MDSC (G-MDSC), can be defined as CD11b + Ly6C lo Ly6G + cells as determined by the high side scatter in the flow cytometry plot ( Peranzoni et al., 2010 ; Damuzzo et al., 2015 ; Zhao et al., 2016 ). In view of the fact that the term PMN-MDSC is more capable of distinguishing from steady-state neutrophils ( Bronte et al., 2016 ), we use the term PMN-MDSC in the following text. The markers on the surface of human MDSC are complicated. At present, human MDSC are defined by the expression of the common myeloid markers CD33 or CD11b and the lack of the marker of mature myeloid cells, such as HLA-DR, which is CD11b + /CD33 + HLADR − . According to the molecular markers used, which are specific to the cluster, M-MDSC are defined as CD11b + /CD33 + HLA-DR - CD14 + /CD15 − /CD66b − and PMN-MDSC as CD11b + /CD33 + HLA-DR - CD14 − /CD15 + /CD66b + ( Gabrilovich and Srinivas, 2009 ; Zhao et al., 2016 ). In addition, early-stage MDSC (e-MDSC) lack molecular markers of any specific subtype, which are defined as CD11b + Gr-1 − F4/80 − MHCII − in mice ( Zhang W. et al., 2018 ) and HLA-DR – CD33 + Lin – (Lin: CD15, CD14, CD3, CD56, and CD19) in humans ( Bronte et al., 2016 ). The latter is a newly defined type of MDSC, which is different from the other two subtypes. The definition of MDSC depends on specific molecular markers and on their immunosuppressive ability. In particular, given that PMN-MDSC and neutrophils, M-MDSC and monocytes have the same source and differentiation pathway, they are almost indistinguishable in phenotype. In human peripheral blood, gradient centrifugation using 1.077 gl −1 density can help isolate neutrophils and PMN-MDSC ( Zhou et al., 2018 ). However, this method often leads to miscalculation due to the rise of activated neutrophils to low-density fraction and the preservation of specimens for too long and frozen damage. Specific surface markers of PMN-MDSC are still being studied. Mouse PMN-MDSC expressed higher levels of CD115 and CD244 than neutrophils ( Youn et al., 2012 ), human PMN-MDSC expressed lectin type oxidized LDL receptor 1 (LOX-1) ( Condamine et al., 2016 ). Although mouse M-MDSC also expressed F4/80, M-MDSC can still be isolated from macrophages and dendritic cells by detection of low levels of both MHC class II and the dendritic cell marker CD11c. Human M-MDSC expressed higher S100A8/A9 and lower HLA-DR ( Bronte et al., 2016 ; Kwak et al., 2020 ). The specific tests that can verify the inhibitory ability of MDSC on T cells have become the gold standard. In mouse tests, the immunosuppressive ability was determined by measuring T-cell proliferation or inhibition of interferon (IFN)-γ production following MDSC culture with antigen-specific and antigen-nonspecific T cells. In human tests, the validation was divided into three groups as follows: Following the addition of candidate MDSC population, the detection of T cell proliferation or IFN-γ production was performed; following removal of the MDSC population, the measurement of T cell proliferation was performed and following allotransplantation, the measurement of T cell proliferation or IFN-γ production was performed ( Bronte et al., 2016 ). During the process of assessing the immune ability, M-MDSC exhibited the highest immunosuppressive function, while PMN-MDSC had the weakest ( Zhao et al., 2016 ). MDSC play different roles in various pathological processes by direct cell contact and secretion of immunosuppressive factors during various pathological processes, including tumor progression, infection and autoimmune diseases. MDSC are prominent in immunosuppression, inflammation, and angiogenesis, which are also the characteristic features of diabetes. However, their role in diabetes has not been fully explored.