Modified method for differentiation of myeloid-derived suppressor cells in vitro enhances immunosuppressive ability via glutathione metabolism

Myeloid-derived suppressor cells (MDSCs), which accumulate in tumor bearers, are known to suppress anti-tumor immunity and thus promote tumor progression. MDSCs are considered a major cause of resistance against immune checkpoint inhibitors in patients with cancer. Therefore, MDSCs are potential targets in cancer immunotherapy. In this study, we modified an in vitro method of MDSC differentiation. Upon stimulating bone marrow (BM) cells with granulocyte-macrophage colony-stimulating factor in vitro, we obtained both lymphocyte antigen 6G positive (Ly-6G+) and negative (Ly-6G−) MDSCs (collectively, hereafter referred to as conventional MDSCs), which were non-immunosuppressive and immunosuppressive, respectively. We then found that MDSCs differentiated from Ly-6G− BM (hereafter called 6G− BM-MDSC) suppressed T-cell proliferation more strongly than conventional MDSCs, whereas the cells differentiated from Ly-6G+ BM (hereafter called 6G+ BM-MDSC) were non-immunosuppressive. In line with this, conventional MDSCs or 6G− BM-MDSC, but not 6G+ BM-MDSC, promoted tumor progression in tumor-bearing mice. Moreover, we identified that activated glutathione metabolism was responsible for the enhanced immunosuppressive ability of 6G− BM-MDSC. Finally, we showed that Ly-6G+ cells in 6G− BM-MDSC, which exhibited weak immunosuppression, expressed higher levels of Cybb mRNA, an immunosuppressive gene of MDSCs, than 6G+ BM-MDSC. Together, these data suggest that the depletion of Ly-6G+ cells from the BM cells leads to differentiation of immunosuppressive Ly-6G+ MDSCs. In summary, we propose a better method for MDSC differentiation in vitro. Moreover, our findings contribute to the understanding of MDSC subpopulations and provide a basis for further research on MDSCs.

as anti-programmed cell death 1 (PD-1) antibody (Ab) and anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) Ab have been reported for their revolutionary clinical outcomes [2][3][4]; they have become standard therapeutic agents for several types of cancer [5]. However, multiple studies have demonstrated that only a limited number of patients respond to anti-PD-1 or anti-CTLA-4 Abs [4,6]. Hence, to overcome this limitation, it is necessary to improve the therapeutic efficacy of ICIs.
Myeloid-derived suppressor cells (MDSCs) are an immature heterogeneous group of myeloid cells that expand in cancerous or inflammatory conditions and co-express CD11b and Gr-1 in mice with the ability to suppress T-cell proliferation [7,8]. MDSCs exacerbate tumors [9,10] and expand in patients with anti-PD-1 resistance or anti-CTLA-4 Abs [4,5,11]. This suggests that MDSCs are one of the causes of anti-PD-1 resistance or anti-CTLA-4 Abs [12]. Therefore, targeting MDSCs is a promising strategy for overcoming the limitations of ICIs.
Studies have reported that several inflammatory factors are involved in the differentiation, proliferation, and immunosuppressive function of MDSCs; for instance, cytokines such as granulocyte-macrophage colonystimulating factor (GM-CSF) and granulocyte colony-stimulating factor (G-CSF) induced MDSC differentiation [7,13,14]. However, little is known about the MDSC differentiation pathway and its mechanism. MDSCs are classified into CD11b + Ly-6G + Ly-6C int polymorphonuclear MDSCs (PMN-MDSCs) and CD11b + Ly-6G − Ly-6C hi monocytic MDSCs (M-MDSCs), based on their morphology and expression of surface markers [7]. In addition, PMN-MDSCs suppress T-cell proliferation using arginase and ROS, while M-MDSCs do so using arginase and nitric oxide [15]. Besides, such differences in the immunosuppressive activity of these two subpopulations and their corresponding origins are less characterized. In tumor-bearing mice, PMN-MDSCs constitute the majority of circulating MDSCs [16,17], whereas M-MDSCs exhibit stronger immunosuppressive activity [10]. Targeting PMN-MDSC development, recruitment, or function improved the efficacy of immunotherapy [18], which also indicated PMN-MDSC immunosuppression. Therefore, differentiating PMN-MDSCs in vitro is important to the understanding of immunosuppression. However, only few studies have reported the immunosuppressive effects of PMN-MDSCs in vitro. Rößner et al. showed that only the Ly-6G low cells enriched from MDSCs cultured in vitro were immunosuppressive, whereas the Ly-6G high population was not [19]. Here, we investigated the differentiation and immunosuppressive ability of MDSC subpopulations in vitro and modified a differentiation method, which can potentially be used for further MDSC studies.

Mice
All C57BL/6J female mice used for this study were 6-8 weeks of age at the start of the experiments and were purchased from Shimizu Laboratory Supplies (Kyoto, Japan). All mice were bred and maintained under specific pathogen-free conditions. All animal experimental procedures in this study were performed in accordance with the institutional guidelines for animal experiments at Osaka University, Japan (Douyaku R03-7-2).

Bone marrow (BM) mononuclear cell isolation
BM mononuclear cells were flushed out vigorously from the femur and tibia with 2% fetal bovine serum (FBS; Gibco, Waltham, MA, USA)/ phosphate-buffered saline (PBS; Gibco). Cells were collected by centrifugation at 330×g for 5 min at 4 • C and then filtered through a 70-μm cell strainer. The erythrocytes were then removed by suspending in ammonium chloride-potassium (ACK) hemolytic buffer, followed by washing with 2% FBS/PBS.

Cell line
The EL4 lymphoma cell line was purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained in RPMI-1640 medium (FUJIFILM Wako, Tokyo, Japan) supplemented with 10% FBS and 100 units/mL penicillin-streptomycin. Cells were resuscitated and cultured following ATCC guidelines. The cells were used within one month of thawing from early passage (≤3 passages of the original vial) lots.

Murine tumor studies
The hair from the inoculation site of C57BL/6J mice was removed in advance. EL4 cells (2 × 10 5 cells/mouse) were inoculated subcutaneously into the lower right flank of mice. Four days after the inoculation, MDSCs that differentiated in vitro were harvested, suspended in PBS, and injected once intravenously into EL4-bearing mice (3 × 10 6 cells/ mouse). The tumor volume and body weight of the mice were measured periodically; the volume was calculated using the following formula: tumor volume (cm 3 ) = 0.523 × length (cm) × width (cm) 2 .

T-cell suppression assay
The spleen harvested from the mice was ground to splenocytes and passed through a 70-μm cell strainer. The cells obtained were suspended in ACK hemolytic buffer to remove any erythrocytes and washed with 2% FBS/PBS to get spleen mononuclear cells. Following the protocol for MojoSort mouse CD4/CD8 nanobeads (BioLegend), these spleen mononuclear cells were then isolated as CD4 + and CD8 + T cells. The T cells were suspended in 0.2% BSA/PBS, an equal volume of 10 μM Cell Proliferation Dye eFlour670 (Thermo Fisher Scientific, Waltham, MA, USA) was added, and the cells were incubated at 37 • C for 10 min for staining. The cells were washed and suspended in RPMI-1640 medium supplemented with 10% FBS, 2 mM GlutaMAX, 100 units/mL penicillinstreptomycin, 10 mM HEPES (Gibco), MEM non-essential amino acids (FUJIFILM Wako), 1 mM sodium pyruvate (Gibco), and 55 μM 2-mercaptoethanol (Sigma-Aldrich). T cells were seeded in 96-well plates at a density of 1 × 10 5 cells/200 μL per well. All wells were pre-coated with anti-CD3ϵ Ab (Clone 145-2C11; BioLegend), diluted with PBS to a concentration of 1 μg/mL, and stored at 4 • C overnight before use. MDSCs differentiated in vitro or separated were added to the wells at different Tcell ratios. For stimulation, anti-CD28 Ab (Clone 37.51; BioLegend) was also added to each well at a final concentration of 0.5 μg/mL. After 3 days of incubation at 37 • C in a 5% CO 2 atmosphere, the proliferation of CD4 + and CD8 + T cells was analyzed using flow cytometry (BD FACS-Canto II; BD Biosciences, San Jose, CA, USA).

Flow cytometry analysis
Cells were washed with 2% FBS/PBS, treated with TruStain FcX (anti-mouse CD16/32) antibody (Clone 93; BioLegend), and incubated at 4 • C for 5 min to block Fc receptors. Then, the cells were stained with fluorescently labeled antibodies (listed in Supplementary Table 1) at 4 • C under light-shielded conditions for 15 min. The cells were then washed and resuspended in 2% FBS/PBS containing 7-aminoactinomycin D as a viability stain (BioLegend) to remove dead cells during analysis. Flow cytometry analysis was performed using a BD FACSCanto II device (BD Biosciences), and the acquired data were analyzed using FlowJo software, version 10.6.2 (BD Biosciences).

Microarray
Total RNA was extracted from cells using the miRNeasy Mini kit (Qiagen) according to the manufacturer's protocol. Total RNA (200 ng) was reverse-transcribed into double-stranded cDNA using AffinityScript multiple temperature reverse transcriptase (Agilent Technologies Inc., Palo Alto, CA). The resulting complimentary RNA (cRNA) was labeled with cyanine-3-labeled cytosine triphosphate (PerkinElmer, Wellesley, MA, USA) using a Low Input Quick-Amp Labeling kit (Agilent Technologies Inc.). One color experiment was performed by hybridizing four cRNAs onto an Agilent SurePrint G3 Mouse GE 8 × 60 K Microarrays (design ID 028005; Agilent Technologies Inc.). The Subio Platform and Subio Basic Plug-in (v1.12; Subio Inc., Aichi, Japan) were then used to calculate the between-sample fold change. Gene set enrichment analysis (GSEA) was used for the comparison of differentially enriched gene sets. The web tool "Calculate and draw custom Venn diagrams" [20] was used to create Venn diagrams.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)
Total RNA was extracted from MDSCs using ISOGEN (NIPPON GENE, Tokyo, Japan). After removing the genomic DNA with RNase-free DNase, a reverse transcription reaction was performed using the Su-perScript VILO cDNA synthesis kit (Thermo Fisher Scientific). Using the synthesized cDNA as a template, PCR was performed using SYBR Premix Ex Taq II (Thermo Fisher Scientific) to amplify the cDNAs of various genes. The StepOnePlus Real-time PCR system (Thermo Fisher Scientific) was used for PCR and analysis. The primer sequences used are listed in Supplementary Table 2 mRNA expression was standardized using the mouse glyceraldehyde 3-phosphate dehydrogenase (Gapdh) gene and expression comparisons were performed using the ΔΔCt method.

Statistical analyses
Significant differences were assessed using Student's t-test or one-or two-way analysis of variance (ANOVA) using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA). p < 0.05 was considered to be statistically significant.

Ly-6G + subpopulation of MDSCs derived from BM cells exhibits no immunosuppression
Previous studies have developed a method that differentiates BM cells into MDSCs in vitro upon GM-CSF stimulation (hereafter referred to as in vitro MDSCs) [21,22]. Following this method, we investigated the immunosuppressive ability of CD11b + Ly-6G + MDSCs and CD11b + Ly-6G − MDSCs. These MDSCs were enriched based on Ly-6G expression and co-cultured with CD4 + T cells. CD11b + Ly-6G + Ly-6C int cells (hereafter referred to as Ly-6G + cells) enriched from in vitro MDSCs showed negligible immunosuppression, indicating that these cells were not immunosuppressive. In contrast, CD11b + Ly-6G − cells (hereafter referred to as Ly-6G − cells) showed stronger immunosuppression than unseparated MDSCs (Fig. 1A). Since Ly-6G + cells were not immunosuppressive, we investigated whether their depletion from BM cells at the beginning of in vitro MDSC differentiation resulted in the differentiation of Ly-6G + MDSCs. Ly-6G + and Ly-6G − BM cells were enriched and cultured for 4 days upon GM-CSF stimulation (Fig. 1B), and the immunosuppressive activity of each group of MDSCs was investigated by co-culture with T cells. The proliferation of both CD4 + and CD8 + T cells was not suppressed by MDSCs derived from Ly-6G + BM cells, hereinafter referred to as 6G + BM-MDSC, suggesting that Ly-6G + BM cells could not acquire immunosuppressive ability during 4-day cultivation ( Fig. 1C and D). In contrast, in vitro MDSCs differentiated from unseparated BM cells (hereafter referred to as MDSCs (conv.)) and MDSCs cultured from Ly-6G − BM cells (hereafter referred to as 6G − BM-MDSC) showed immunosuppressive activity ( Fig. 1C and D). In addition, 6G − BM-MDSC were able to suppress CD8 + T cells more strongly at a ratio of 0.5:1 than MDSCs (conv.) (Fig. 1D). Taken together, our results show that in vitro MDSCs with stronger immunosuppressive activity can be obtained by differentiating MDSCs from Ly-6G − BM cells.

6G + BM-MDSC exhibit no ability to promote tumor progression
Previous studies have shown that MDSC transfer promotes tumor progression in vivo [23,24]. To evaluate the immunosuppressive ability in vivo, MDSCs cultured in vitro using each differentiation method were intravenously transferred into EL4 tumor-bearing mice ( Fig. 2A). Compared to that in the PBS control group, tumor progression was promoted in MDSC (conv.)-transferred mice and 6G − BM-MDSC-transferred mice. Notably, the transfer of 6G + BM-MDSC did not affect tumor progression (Fig. 2B), further supporting that 6G + BM-MDSC are not immunosuppressive. However, the body weight of the mice did not change post MDSC transfer, indicating the absence of toxic reactions (Fig. 2C). Together, these data suggest that, consistent with our findings in vitro, 6G + BM-MDSC (but not MDSCs (conv.) and 6G − BM-MDSC) showed no immunosuppressive ability in vivo also and failed to acquire immunosuppressive ability even in tumor-bearing mice.

Immunosuppressive activity of MDSCs may require ROS production and down-regulation of retinoblastoma 1 (Rb1)
To investigate the underlying mechanisms of the immunosuppressive ability of 6G − BM-MDSC, we performed microarray followed by GSEA. Comparing the gene expression profiles of MDSC (conv.) and 6G − BM- MDSC, we observed that several gene sets, such as those associated with inflammation and metabolism, were enriched in 6G − BM-MDSC (Fig. 4A). Next, to identify the genes responsible for the enhancement of the immunosuppressive activity of 6G − BM-MDSC, we compared these MDSCs with BM-derived dendritic cells (DCs), which are differentiated upon stimulation of GM-CSF for 8 days. We identified 1046 genes that were highly expressed in MDSC (conv.) and 6G − BM-MDSC compared to DCs. Moreover, six genes were highly expressed in 6G − BM-MDSC, and among these genes, glutathione S-transferase, that is, glutathione S-transferase θ 1 (Gstt1) and glutathione S-transferase θ 4 (Gstt4) were supposed to be the genes responsible for the enhancement of the immunosuppressive activity of 6G − BM-MDSC (Fig. 4B). Then, GSEA further confirmed that the glutathione metabolism gene set was enriched in 6G − BM-MDSC (Fig. 4A, C). Since glutathione S-transferase degrades glutathione, which is a ROS scavenger, these results suggest that ROS are critical for the enhanced immunosuppressive ability of 6G − BM-MDSC.
Since the bulk microarray analysis did not determine the differentially expressed immunosuppressive molecules, we measured their mRNA expression in each MDSC subpopulation. Compared to MDSC (conv.), 6G − BM-MDSC, and Ly-6G − /Ly-6G + cells separated from 6G − BM-MDSC, the expression of arginase 1 (Arg1) was higher in 6G + BM-MDSC (Fig. 4D). There was no significant difference in the expression of inducible nitric oxide synthase (Nos2) in each MDSCs. However, the expression of Nos2 tended to be higher in Ly-6G − cells in 6G − BM-MDSC but lower in Ly-6G + cells in 6G − BM-MDSC than in the others (Fig. 4E). Notably, the expression of NADPH oxidase 2 (Cybb), a ROS-producing enzyme, was significantly higher (p = 0.0315) in Ly-6G + cells in 6G − BM-MDSC compared to 6G + BM-MDSC (Fig. 4F). In addition, the expression of Rb1 tended to be down-regulated in Ly-6G + cells in 6G − BM-MDSC compared to that in 6G + BM-MDSC (Fig. 4G). Rb1 is a tumor suppressor gene [25]; the downregulation of Rb1 expression leads to the differentiation of PMN-MDSCs from M-MDSCs [26]. Therefore, these results suggest that the immunosuppressive ability of Ly-6G + cells in 6G − BM-MDSC requires upregulation of Cybb and possibly downregulation of Rb1. Taken together, our results suggest that high levels of glutathione S-transferase and Cybb expression would promote ROS production, resulting in the enhanced immunosuppressive ability of 6G − BM-MDSC.

Discussion
Our study shows that Ly-6G + cells in MDSC (conv.) phenotypically resemble PMN-MDSCs but are not immunosuppressive. Zhu et al. had earlier found that ex vivo-induced PMN-MDSCs fail to suppress T cells [27]. On the other hand, it is also known that PMN-MDSCs from tumor-bearing mice, which despite being less immunosuppressive than M-MDSCs, are clearly defined as immunosuppressive [16,28]. Here, using our newly developed in vitro MDSC differentiation method and cultivation of Ly-6G − BM cells, we obtained immunosuppressive Ly-6G + cells, that is, PMN-MDSCs. Although these Ly-6G + cells differentiate from Ly-6G − BM cells, the presence of non-immunosuppressive Ly-6G + cells, possibly neutrophils, inhibits their differentiation. Therefore, the removal of Ly-6G + cells before in vitro MDSC differentiation should be considered. This point might be supported by Haverkamp et al. who reported that M-MDSCs are the dominant suppressive population and that removal of the neutrophilic lineage might be important to differentiate MDSCs [29]. However, the mechanisms by which neutrophils inhibit the differentiation of PMN-MDSCs remain to be addressed. Rößner et al. found that only the Ly-6G low cells enriched from MDSCs cultured in vitro were immunosuppressive, but the Ly-6G high population was not [19]. All these findings indicate that the conventionally differentiated Ly-6G + cells that phenotypically correspond to PMN-MDSCs are not immunosuppressive, unlike PMN-MDSCs obtained from tumor-bearing hosts. Our findings complement the above idea and improve the current understanding of in vitro differentiation of PMN-MDSCs. However, it is still not known whether the tumor microenvironment plays a role in MDSC differentiation. Thus, the role of the tumor microenvironment in the early stages of in vitro MDSC differentiation requires further elucidation.
In the present study, correlating with their strong immunosuppressive ability, 6G − BM-MDSC and its Ly-6G − cells tended to show higher Nos2 expression. These results are consistent with previous findings that Nos2 plays an important role in the immunosuppressive ability of M-MDSCs with the Ly-6G − phenotype [15]. Moreover, we found upregulated Cybb expression in Ly-6G + cells in 6G − BM-MDSC compared to that in 6G + BM-MDSC, which might contribute to their immunosuppressive activity. Interestingly, the loss of RB1 protein in MDSCs is related to PMN-MDSC differentiation from M-MDSCs [26]. In our study, Rb1 expression was low in Ly-6G + cells in 6G − BM-MDSC, which differed from that of 6G + BM-MDSC. In conclusion, the downregulation of Rb1 expression might be a potential indicator of in vitro differentiation of immunosuppressive PMN-MDSCs. However, further studies are needed to define the immunosuppressive mechanisms of these Ly-6G + cells.
The analysis of gene expression profiles showed that glutathione metabolism was the pathway responsible for the enhanced immunosuppressive ability of 6G − BM-MDSC. We previously showed that glutamate, one of the metabolites of glutathione, would enhance the immunosuppressive function of MDSCs [14]. In addition, we found that glutamate signaling through metabotropic glutamate receptor 2/3 enhances the immunosuppressive ability of MDSCs [30]. Altogether, these results suggested that glutamate would be an important factor in enhancing the immunosuppressive ability of 6G − BM-MDSC.
Our study also suggested that the stimulation of BM cells by GM-CSF in the absence of Ly-6G + cells could lead to the differentiation of stronger immunosuppressive MDSCs. Kumar et al. summarized the therapeutic effects of GM-CSF in cancer immunotherapy [31]. They also mentioned the pathogenic effects, including reprograming macrophages to the tumor-promoting M2 phenotype and the induction of MDSC differentiation. Most importantly, they listed some clinical cases that used GM-CSF to rescue tumor therapy-induced neutropenia. We previously reported that G-CSF, which is used for the prevention and therapy for neutropenia, enhanced the immunosuppressive activity of MDSCs through the glutathione degradation pathway, leading to tumor progression [14]; therefore, GM-CSF might also do so. Although further in vivo cancer model studies are needed, our findings indicated that such GM-CSF treatment might lead to the differentiation of stronger immunosuppressive MDSCs. Therefore, our results contribute to the investigation of the immuno-toxic effect of GM-CSF treatment in tumor therapy.
To gain a more comprehensive and detailed understanding of the characteristics and functions of MDSCs, a method for differentiating MDSCs in vitro that resembles those in vivo is essential. Our findings suggest an in vitro method for differentiating PMN-MDSCs and more immunosuppressive M-MDSCs in the absence of Ly-6G + BM cells. To summarize, we provide a potential in vitro method to study in vivo PMN-MDSCs, which would be necessary for developing strategies in the future for accurately studying the MDSC subpopulations. We believe that our method would be helpful to overcome the limitations of ICIs and contribute to the understanding of cancer immunity.

Funding
This work was supported in part by JSPS KAKENHI Grant Numbers JP19H04049 and JP22H03533 (Grants-in-aid for Scientific Research (B)) (M.T.). This research was also partially supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from AMED under grant numbers JP22am121052 and JP22am121054.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.