Female C57BL/6J mice were purchased from Japan SLC (Shizuoka, Japan). All mice were bred and maintained under specific pathogen-free conditions and used for experiments at 6–8 weeks of age. All experimental procedures in this study were performed in accordance with the institutional guidelines for animal experiments at the Osaka University, Japan.

The EL4 cell lines were purchased from the American Type Culture Collection (ATCC) and maintained in RPMI-1640 medium (FUJIFILM Wako, Osaka, Japan) supplemented with 10% fetal bovine serum (FBS; Gibco, CA, United States) and 1% Antibiotic-Antimycotic Mixed Stock Solution (100 ×) (Nacalai Tesque, Kyoto, Japan). Cells were cultured according to the ATCC guidelines and used within 1 month of thawing from an early passage (<3 passages of original vial) lots.

The in vitro differentiation of bone marrow (BM) cells into MDSCs was performed as described previously (Xie et al., 2018). Briefly, BM cells from C57BL/6J mice were stimulated with 40 ng/ml recombinant Granulocyte-macrophage CSF (GM-CSF) (Peprotech, NJ, United States) for 4 days in the absence or presence of G-CSF (5 ng/ml, BioLegend, CA, United States) and/or GGsTop (50 mΜ, FUJIFILM Wako) to examine their effects on MDSCs.

Cells were pelleted and washed in phosphate-buffered saline (PBS) supplemented with 2% FBS (2% FBS/PBS). The cell suspension was first blocked with TruStain fcX (anti-mouse CD16/32) antibody (BioLegend) for 5 min, and then stained with the following antibodies (Abs) for 15 min at 4°C: APC anti-mouse CD11b, Pacific Blue anti-mouse Gr-1, PE anti-mouse F4/80, FITC anti-mouse CD11c, APC-Cy7 anti-mouse Ly-6C, FITC anti-mouse Ly-6G, Pacific Blue anti-mouse CD4, and FITC anti-mouse CD8α (BioLegend). Next, the cells were washed and resuspended in 2% FBS/PBS. Shortly before performing measurements, a 7-amino actinomycin D viability staining solution (BioLegend) was added to each sample to stain dead cells. Flow cytometry analysis was performed on a BD FACSCanto II flow cytometer (BD Biosciences), and results were analyzed using the FlowJo software (version10.7.0, BD Biosciences).

CD4⁺ T cells or CD8⁺ T cells were isolated from the spleens of C57BL/6J mice using the MojoSort magnetic cell separation system as previously described (Xie et al., 2018) and then labeled with the proliferation dye eFluor 670 (eBioscience, Thermo Fisher Scientific, CA, United States). The eFluor 670-labeled CD4⁺ T cells or CD8⁺ T cells were incubated with in vitro differentiated MDSCs at different ratios in a 96-well plate cultured with anti-mouse CD3ɛ Ab/anti-mouse CD28 Ab (BioLegend). After 3 days of incubation at 37°C in 5% CO₂, the proliferation of CD4⁺ and CD8⁺ T cells determined by the eFluor 670 fluorescence intensity was analyzed using flow cytometry.

PMN-MDSCs (Ly-6G⁺Ly-6C^int) and M-MDSCs (Ly-6G⁻Ly-6C^hi) were purified from in vitro-differentiated MDSCs cultured with or without the addition of G-CSF by fluorescence-activated cell sorting (purity >95%; JSAN, Bay bioscience Co., Ltd., Kobe, Japan). Total RNA was extracted from the in vitro differentiated MDSCs or sorted MDSC subsets and used to synthesize cDNA using the QuantiTect reverse transcription kit (Qiagen, Hilden, Germany) following the manufacturer’s instructions. qRT-PCR was performed using SYBR Premix Ex Taq (Tli RNaseH Plus; TaKaRa, Kusatsu, Japan) on a CFX96 touch real-time PCR detection system (Bio-Rad). The specific primer sequences are listed in Supplementary Table S1. Glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used as a reference gene, and relative expressions of other genes were calculated using the 2^−ΔΔCt method.

Total RNA was extracted from cells using the miRNeasy Mini kit (Qiagen) according to the manufacturer’s protocol. RNA libraries were prepared for sequencing using a TruSeq stranded mRNA sample prep kit (Illumina, CA, United States) according to the manufacturer’s instructions. Whole transcriptome sequencing was performed on RNA samples using the Illumina HiSeq 2500 platform in a 75 bp single-end mode. Volcano plot representation and pathway enrichment analysis of the differentially expressed genes were performed using BioJupies tools with adjusted p < 0.05 and fold change >2 (Torre et al., 2018). Raw data from this study have been submitted to Gene Expression Omnibus (GEO) under the accession number GSE183066.

MDSCs (1 × 10⁶ cells) were homogenized in 200 µL of GGT assay buffer and centrifuged at 13,000 × g for 10 min following which the supernatant was collected. After reacting with the GGT substrate at 37°C for 40 min, fluorescence (365/460 nm) was measured using a microplate reader SpectraMax iD5 (Molecular Devices). GGT activity was calculated according to the protocol of the GGT Activity Colorimetric Assay Kit (BioVision, CA, United States).

The in vitro MDSC culture medium was collected on day 4. Next, glutamate concentration was measured using Glutamate Assay Kit-WST (Dojindo, Kumamoto, Japan) according to the manufacturer’s protocol. Briefly, a glutamate standard or MDSC culture medium was added to a 96-well microplate; the working solution was then added to each well. After incubating the microplate at 37°C for 30 min, the absorbance was measured at 450 nm using SpectraMax iD5 (Molecular Devices). The glutamate concentration in each sample was calculated using a standard curve.

Intracellular ROS was assayed using a ROS Assay Kit-Highly Sensitive DCFH-DA (Dojindo) according to the manufacturer’s protocol. Briefly, 2 × 10⁵ MDSCs were treated with 2′, 7′dichlorofluorescin diacetate (DCFH-DA) working solution and incubated for 30 min. Subsequently, the MDSCs were washed twice with Hanks’ Balanced Salt Solution, and ROS production was measured using a BD FACSCanto II flow cytometer (BD Biosciences).

EL4 lymphoma cells (4 × 10⁵ cells/mouse) were injected subcutaneously into the lower right flank of C57BL/6J mice. Seven days after inoculation with EL4 cells, mice were intraperitoneally administered a single dose of CPA (100 mg/kg, FUJIFILM Wako) to create a neutropenic mouse model as described previously (Hattori et al., 1990). Starting from the same day, experimental mice received intraperitoneal injections of PBS, GGsTop (5 mg/kg, FUJIFILM Wako), and recombinant G-CSF (200 μg/kg, BioLegend) for six consecutive days. Before administration of the above reagent, mouse retro-orbital blood (approximately 75 μL) was collected for flow cytometry analysis and determining blood cell counts using a Sysmex XT-2000i automated hematology analyzer (Sysmex) daily. The tumor volume was calculated periodically up to day 15 using the following formula: Tumor Volume (cm³) = 0.5 × (Length × Width²).

Significant differences were assessed using Student’s t-test, or a one- or two-way analysis of variance (ANOVA) using GraphPad Prism 7.0 (GraphPad Software). Statistical significance was set at p < 0.05.

Previous studies have reported that GM-CSF stimulates BM cells to differentiate into MDSCs (hereafter referred to as in vitro MDSCs). Freshly isolated BM cells were cultured for 4 days in a medium supplemented with or without G-CSF and GM-CSF to examine whether G-CSF affects the differentiation of in vitro MDSCs. G-CSF did not affect the percentage of MDSCs (CD11b⁺Gr-1⁺), dendritic cells (DCs; Gr-1⁻CD11c⁺), or macrophages (Gr-1^-F4/80⁺) differentiated from BM cells, as shown in Figure 1A. The proportion of M-MDSCs with the CD11b⁺Ly-6G⁻Ly-6C^hi phenotype increased in the presence of G-CSF, whereas PMN-MDSCs with the CD11b⁺Ly-6G⁺Ly-6C^int phenotype decreased (Figure 1B). However, since whole BM cells largely increased with stimulation by G-CSF, there was no change in the number of PMN-MDSCs, and M-MDSCs substantially increased (Figure 1C). These results suggest that G-CSF promotes the proliferation of MDSCs, especially M-MDSCs.

Next, we investigated whether G-CSF affected the immunosuppressive activity of in vitro MDSCs. Suppression of CD4⁺ T cell proliferation was observed in in vitro MDSCs that were not treated with G-CSF, and the immunosuppressive activity was impaired with decreasing numbers of MDSCs. Moreover, the immunosuppressive activity of in vitro MDSCs was significantly enhanced by G-CSF treatment (Figure 1D). Similar results were obtained in the CD8⁺ T cell suppression assay (Figure 1E). These observations revealed that T cells, especially CD4⁺ T cells, were less proliferative when co-cultured with G-CSF-conditioned in vitro MDSCs, suggesting that G-CSF strongly enhanced the immunosuppressive activity of in vitro MDSCs.

To explore the mechanism by which G-CSF enhances the immunosuppressive function of MDSCs, we first analyzed the expression of the main immunosuppressive molecules of MDSCs: arginase 1 and iNOS as well as ROS by qRT-PCR or flow cytometry analysis. We found that the mRNA levels of Arg1 and Nos2 were significantly upregulated by G-CSF and that ROS levels determined by DCFH-DA also increased (Figures 2A,B). Besides, G-CSF upregulated the expression of Arg1 and Nos2 in both PMN-MDSCs and M-MDSCs, indicating that the G-CSF-induced upregulation of those genes was because of the enhanced transcription of the genes but not because of the alteration of the population of MDSC subsets (Figures 2C,D).

Next, we performed an RNA-seq analysis, comparing the gene expression profiles of in vitro MDSCs differentiated with or without G-CSF. There were 228 genes differentially expressed with fold change >2 and adjusted p < 0.05 when comparing both groups. Among these genes, 89 genes were expressed at higher levels in MDSCs differentiated in the presence of G-CSF than in those differentiated without G-CSF, and 139 genes showed lower expression. Increased expression of Arg1, Nos2, and Mpo, which are the immunosuppressive molecules of MDSCs, was observed (Figure 2E), consistent with the above results. Additionally, pathway enrichment analysis revealed that the MDSCs differentiated with G-CSF displayed the activation of the cell cycle, metabolism, and DNA replication, which would account for the enhanced proliferation of MDSCs differentiated with G-CSF (Figure 2F).

Next, among the top 20 upregulated genes by G-CSF, genes associated with poor prognosis (Logrank p < 0.05) in more than four cancers, including Nos2, proprotein convertase subtilisin/kexin type 9 (Pcsk9), PRKC apoptosis WT1 regulator (Pawr), and gamma-glutamyltransferase 1 (Ggt1), were identified as candidate factors involved in the immunosuppressive ability of MDSCs using the GEPIA web server (Tang et al., 2017) (Supplementary Figure S1). GGT1 is a key enzyme involved in several metabolic processes, such as glutathione metabolism and cyanoamino acid metabolism (Verma et al., 2015), which were also enriched and active in MDSCs differentiated with G-CSF (Figure 2F). Furthermore, GGT is expressed on the cell surface, acts as a glutathione hydrolase that increases extracellular cysteinyl glycine and glutamate, and is reported to be associated with the progression of several tumors (Wang et al., 2017; Ince et al., 2020). Thus, to identify novel factors involved in the immunosuppressive ability of MDSCs, we focused on the Ggt1 gene.

The upregulation of the mRNA of Ggt1 by G-CSF in MDSCs was verified by qRT-PCR (Figure 3A). Besides, Ggt1 mRNA expression was more upregulated in M-MDSCs than in PMN-MDSCs upon G-CSF stimulation, suggesting that M-MDSCs can play an important role in increasing the G-CSF-promoted immunosuppressive effects (Figure 3B). To elucidate whether GGT1 is involved in the enhanced immunosuppressive function of MDSCs by G-CSF, we used GGsTop, a highly selective GGT inhibitor. The GGT activity of MDSCs was enhanced by G-CSF, which was inhibited below the detection limit in the GGsTop alone and G-CSF/GGsTop combination groups (Figure 3C). There was no change in the number of MDSCs in the presence of GGsTop (Figure 3D), suggesting that the inhibition of GGT did not affect the G-CSF-induced proliferation of BM cells. Meanwhile, the addition of GGsTop did not affect the proportion and number of M-MDSCs or PMN-MDSCs (Supplementary Figures S2A,B). We further investigated whether GGsTop cancels the enhancement of the immunosuppressive activity of in vitro MDSCs by G-CSF. Consistent with Ggt1 expression, the immunosuppressive activity of in vitro MDSCs was significantly increased by G-CSF, while the combination of GGsTop and G-CSF attenuated the immunosuppressive activity to the same level as that in the control group (Figures 3E–G). Additionally, GGsTop alone did not affect the immunosuppressive activity of in vitro MDSCs. Furthermore, we examined the effect of GGsTop alone or the G-CSF/GGsTop combination on Arg1, Nos2, and ROS expression. The G-CSF/GGsTop combination group downregulated the expression of Arg1 and Nos2. Unexpectedly, GGsTop alone upregulated their expression (Supplementary Figure S2C). As Arg1 and Nos2 upregulation by GGsTop alone did not increase the immunosuppressive ability of MDSCs (Figures 3E,F), suggesting that the upregulation of Arg1 and Nos2 did not play a decisive role in increasing the MDSC immunosuppressive ability caused by G-CSF. In addition, GGsTop did not abrogate the expression of ROS caused by G-CSF, suggesting that ROS is not a key factor either (Supplementary Figure S2D). These results suggest that inhibition of GGT1 activity eliminates G-CSF-induced enhanced immunosuppressive function of MDSCs, but has no effect on its ability to promote proliferation.

The key role of GGT1 is to hydrolyze glutathione. We speculated that metabolites of glutathione, such as glutamate and cysteinyl glycine, obtained by the action of GGT1, enhance the immunosuppressive function of MDSCs. Elevated glutamate concentrations are commonly observed in tumor patients, and studies indicate that glutamate could regulate the immunosuppressive activity of MDSCs on the proliferation of T cells (Morikawa et al., 2018; Wu et al., 2019). Therefore, we measured the concentration of glutamate to determine whether the level of glutamate is regulated by GGT1. Glutamate levels were increased on treating cells with G-CSF and were decreased on treatment with GGsTop both alone and in combination with G-CSF (Figure 3H). GGsTop alone or the G-CSF/GGsTop combination did not significantly reduce glutamate levels compared with the control, which explained why GGsTop alone did not inhibit MDSC-mediated T cell suppression. These observations revealed that elevated GGT activity induced by G-CSF increased glutamate levels, and that GGT1 is a novel factor involved in the enhanced immunosuppressive function of MDSCs induced by G-CSF.

Given that G-CSF enhances the immunosuppressive function and proliferation of MDSCs, it is possible that G-CSF causes side effects through MDSCs when used to prevent FN. We then tested whether G-CSF promoted MDSCs and tumor progression using the EL4 lymphoma-bearing neutropenic mouse model. In the EL4 tumor-bearing mouse model, administration of the anti-tumor drug CPA reduced white blood cells, including neutrophils, monocytes, and lymphocytes to 50% below the pretreatment value measured using an XT-2000i hematology analyzer, leading to neutropenia (Figures 4A–D). CPA strongly inhibited tumor growth (Figure 4F). In contrast, the accelerated recovery of white blood cells was observed in mice administered G-CSF after CPA. G-CSF dramatically increased neutrophil and monocyte counts from days 11–13 and facilitated the recovery of lymphocytes to control levels (Figures 4A–D), which would reduce the risk associated with CPA-induced neutropenia. Flow cytometry analysis revealed that the proportion of M-MDSCs and PMN-MDSCs in the G-CSF group increased significantly, suggesting that G-CSF promoted neutrophils and monocytes that were morphologically identified using an XT-2000i hematology analyzer to be mixed with most MDSCs (Figure 4E). Furthermore, tumor progression was observed after G-CSF administration, suggesting that G-CSF-induced MDSCs attenuated the anti-tumor effect of CPA (Figure 4F). Next, we explored whether GGsTop could inhibit the growth of tumors induced by G-CSF while maintaining myelopoiesis by G-CSF. As expected, GGsTop alone did not affect the dynamics of myelopoiesis (Figures 4A–E) and tumor progression (Figure 4F). The combination of GGsTop and G-CSF promoted the proliferation of white blood cells, neutrophils, and monocytes in the G-CSF group and inhibited tumor growth in the presence of CPA. Consistent with other studies (Jounaidi and Waxman, 2001; Kim et al., 2018), due to CPA drug toxicity and rapid tumor growth in the PBS and GGsTop groups, the biological significance of body weight was shown in the CPA-treated and control mice. Moreover, the results showed that GGsTop was well-tolerated, with no significant decrease observed in the body weight (Figure 4G). These results suggest that targeting GGT eliminates the effect of G-CSF in promoting tumor growth. GGsTop could be an attractive combination agent with G-CSF for the treatment of neutropenia in patients with cancer.