Circulating and tumor-infiltrating arginase 1-expressing cells in gastric adenocarcinoma patients were mainly immature and monocytic Myeloid-derived suppressor cells.

Myeloid-derived suppressor cells (MDSCs) are a group of heterogeneous cells derived from immature myeloid cells (IMCs). MDSCs are known to play important roles in tumor immune evasion. While we know that there are a large number of circulating and tumor-infiltrating MDSCs existing in gastric cancer (GC) patients, the phenotypic characteristics and arginase 1 (ARG1) expression levels of these MDSCs remain very unclear. In our study, flow cytometric analysis of circulating MDSCs from 20 gastric adenocarcinoma (GAC) patients found that ≥80% ARG1-expressing MDSCs were mainly early-stage MDSCs (HLA-DR−CD33+CD14−CD15−MDSCs). In addition, our investigation showed that tumor-infiltrating MDSCs from 6 GAC patients consisted of >35% ARG1-expressing naïve MDSCs (HLA-DR−CD33−CD11b−CD14−CD15−MDSCs), >15% early-stage MDSCs and >40% monocytic MDSCs (HLA-DR−CD14+MDSCs). This preliminary study describes the phenotypic characteristics and ARG1 expression levels of MDSCs from GAC patients and shows that circulating and tumor-infiltrating ARG1-expressing cells were mainly immature and monocytic MDSCs, which provides information to better understand the mechanisms that allow gastric cancer cells to evade the immune system.

A number of studies relating to gastric cancer (GC) have reported MDSCs phenotype markers, including CD11b + , CD33 + , CD14 + , CD15 + and HLA-DR -26,27 . However, there is limited research data on this point at present and there has been no clear explanation of the phenotypic features of MDSCs or the expression level of ARG1 in MDSCs.
In this study, we first analyzed by flow cytometry the phenotype and the ARG1 expression level in MDSCs from peripheral blood and cancerous tissue of gastric adenocarcinoma (GAC) patients. Next, we determined the relative expression levels of ARG1/NOS2 in those circulating and tumor-infiltrating MDSCs by qRT-PCR and Western blotting. These findings provide clues for the further elucidation of the exact mechanism by which gastric cancer cells induce the generation of MDSCs, and how the immune function of these key cells is activated.

Materials and methods
Clinical sample collection. The study was conducted in accordance with the 1964 Declaration of Helsinki.
The project was approved by the Medical Ethics Committee of the First Affiliated Hospital of Henan University of Chinese Medicine (Zhengzhou, China) (protocol number 2016HL-085), and all individual participants included in the study gave written informed consent. Backgrounds of gastric cancer patients and volunteers are described in Supplementary Table 1. All patients with GAC were first diagnosed and enrolled without radiotherapy or chemotherapy. From November 2016 to March 2017, 6 GAC tissue specimens were collected at the general surgery department of the First Affiliated Hospital of Henan University of Chinese Medicine. In each case, specimens of cancerous tissues and paracancerous tissues (5 cm from the edge of the cancer) were taken. Tissue specimens were collected in sterile test tubes containing RPMI1640 culture medium. Simultaneously, peripheral blood samples were collected from 20 GAC patients prior to surgery. 20 ml blood sample were placed into EDTA.K 2 anticoagulant test tubes and immediately sent to the laboratory for analysis.

Preparation of mononuclear cells suspension from cancerous or paracancerous tissue.
The cancerous and paracancerous tissue masses were cut into 2-4 mm 3 pieces. Fat, fibrous tissue and necrotic tissue was removed. Tissue masses were then transferred into a GentleMACS Dissociator. (Miltenyi Biotec, Bergisch-Gladbach, Germany). These techniques were carried out in accordance with the manufacturer's instructions. Following homogenization, the sample was centrifuged at 300 g for 7 min and the pellet was collected. The pellet was resuspended in 20 ml of RPMI1640 and any larger tissue masses were picked out using a sterile pipette tip. This suspension was then placed into a 70μm MACS SmartStrainer, and the column rinsed with 20 ml of RPMI1640 to gather the suspension on the filter. The column was then centrifuged at 300 g for 7 min, the supernatant discarded, and the cells were resuspended in preparation for subsequent experiments.
The prepared single tumor cell suspension was carefully added to 2 ml of human organ tissue mononuclear cells separation solution (HaoYang Biotechnology company, TianJin, China) and centrifuged at 300-500 g for 15 minutes. Centrifugation separated the preparation into four layers. The top layer was fetal bovine serum, the second layer consisted of loop ivory mononuclear cells, the third layer was transparent separation medium and the fourth layer consisted of histocytes. Cells in the second layer were collected into a test tube and added to 4-5 ml of phosphate buffered saline (PBS). This mixture was then blended thoroughly and centrifuged at 300-500 g for 30 minutes. Cells were washed twice with PBS and then collected.
Preparation of PBMCs. 10 ml of fresh peripheral blood, which had been anticoagulated with EDTA.K 2 , was mixed with RPMI1640 culture medium at a ratio of 1:1. Next, 4 ml of human peripheral blood lymphocyte separation solution was added to a 15 ml centrifuge tube, the diluted peripheral blood mixture was carefully layered onto the cell separation solution (at a ratio of 2:1) and the tube centrifuged at 1000 g for 15 minutes at 18 °C. Following centrifugation, the mixture had separated into the following four layers: the first layer (at the top) consisted of plasma, the second consisted of loop ivory lymphocytes, the third was transparent separation medium and the fourth consisted of erythrocytes. The cells in the second layer were collected and placed into a 15 ml tube. Cells were washed in AIM-V (Grand Island Biological Company, New York, NY, USA) culture medium and centrifuged at 600 g for 7 minutes. The supernatant was then discarded and the pelleted cells resuspended in AIM-V. The washing step was repeated two further times and resulted in the isolation of mononuclear cells.
Sorting HLA-DR -/low mononuclear cells by means of magnetic beads. To obtain HLA-DR -/low mononuclear cells from tumor tissue and peripheral blood, we took an appropriate amount of mononuclear cell suspension from cancerous or paracancerous tissue or peripheral blood mononuclear cells and washed in pre-cooled autoMACS ® Running Buffer (Miltenyi Biotec, Bergisch-Gladbach, Germany). Following resuspension, HLA-DR -/low mononuclear cells were isolated by negative sorting (anti-HLA-DR magnetic bead, LS column and a MACS Magnetic shelf) (Miltenyi Biotec). After sorting, the number of living cells was calculated by trypan blue staining. HLA-DR -/low mononuclear cells were then fluorescently labelled with anti-HLA-DR-PE(20 μl/10 7 cells) (BD Biosciences, Franklin Lakes, NJ, USA) and a CantoII flow cytometer(BD Biosciences) was used to analyze the purity of the HLA-DR -/low cells.
Sorting HLA-DR -CD33 + MDSCs by means of magnetic beads. To separate the CD33 + population, HLA-DRcells were co-incubated with CD33 microbeads, and HLA-DR -CD33 + MDSCs were collected using a Positive Selection program. Cell viability of sorted HLA-DR -CD33 + MDSCs was analyzed using trypan blue staining; the purity was then validated on a FACS Canto II flow cytometer using an APC-conjugated anti-CD33 antibody (BD Biosciences, USA).
Sorting CD8 + T cells by means of magnetic beads. To obtain CD8 + T cells, PBMCs from healthy donors were washed with precooled autoMACS ® Running Buffer (Miltenyi Biotec, Germany), incubated with www.nature.com/scientificreports www.nature.com/scientificreports/ anti-CD8 microbeads, and CD8 + T cells were separated on a autoMACS ® separator using the Depletion program following manufacturer's instructions. Sorted CD8 + T cells were labeled with CFSE (3 μM, Sigma) Analysis of myeloid cells by means of flow cytometry. In order to label peripheral blood MDSCs with a fluorescent antibody (BD Biosciences), 100 μl of peripheral blood was added to a test tube containing fluorescent labeled antibody or isotype control, mixed gently, and incubated for 20 minutes without light at 4 °C. Next, 400 μl of erythrocyte lysis buffer was added and mixed gently prior to incubation for 15 minutes without light at 25 °C. In order to label tumor-infiltrated MDSCs with a fluorescent antibody, 100 μl of prepared cancerous or paracancerous tissue mononuclear cell suspension (1 × 10 6 /ml) was added to a test tube containing fluorescent labeled antibody or isotype control, mixed gently, and incubated for 20 minutes without light at 4 °C.
In order to remove uncombined fluorescent antibodies, samples were centrifuged at 1000 g or 5 minutes and the supernatant was discarded. The pellet was then washed twice with PBS. The labeled cells were resuspended with 300 μl of flow cytometry (FCM) buffer solution prior to detection. Flow cytometry (Canto II type, BD FACSDiva TM analysis software) was then used to analyze the immunophenotype of circulating and tumor-infiltrating MDSCs that had been labeled with fluorescent antibody.
Detecting intracellular ARG1 in MDSCs subgroups by means of flow cytometry. Immune cells in peripheral blood or tumor tissue were labeled with anti-HLA-DR-PE/anti-CD33-APC/anti-CD11b-APC-Cy7/ anti-CD14-PerCP-Cy5.5/anti-CD15-PE-Cy7 fluorescent antibodies. At the same time an isotype control (mouse IgG2b-PE/mouse IgG1-APC/rat IgG2b APC-CY7/mouse IgG2a-PerCP-Cy5.5/mouse IgM-PE-CY7) (BD Biosciences) was prepared. Labeled cells were centrifuged at 300 g for 15 minutes at 18 °C and the supernatant was discarded. The pellet was then gently mixed with 100 μl of cytomembrane punching reagent 1 (Beckman Coulter Inc., Brea, CA, USA), which was mixed gently, and incubated for 15 minutes without light at 25 °C. Cells were then washed with normal saline (NS) and centrifuged at 300 g for 5 minutes. The supernatant was then discarded and the pellet mixed with 100 μl of cytomembrane punching reagent 2, and incubated for 5 minutes without light at 25 °C. Then, 5 μl of ARG1-FITC was added and incubated for 15 minutes without light at 25 °C, simultaneous with an isotype control (mouse IgG1 FITC). Next, 1 ml of NS was added and the mixture centrifuged at 300 g for 5 minutes. The supernatant was then discarded and the labelled cells resuspended in 500 μl of FCM buffer solution for detection.

Reverse-transcription polymerase chain reaction (RT-PCR). An RNA Extraction Kit (Thermo Fisher
Scientific, Inc., Waltham, MA, USA) was used to extract RNA from Lin -HLA-DRcells derived from tumor tissue or peripheral blood. A reverse-transcription was then performed to reverse transcribe mRNA into cDNA. Primers were synthesized by Shanghai Biological Engineering Co. LTD. Primer sequences were determined using NCBI BLAST software (please refer to Table 1).
Next, we used a real-time fluorescence quantification kit to amplify cDNA on an ABI 7500 system (Applied Biosystems, Carlsbad, CA, USA) to calculate the expression level of each target gene relative to an internal reference gene using the 2 −ΔΔCT method.
Functional analysis of MDSCs. Sorted HLA-DR -CD33 + MDSCs were co-cultured with CD8 + T cells (MDSCs:T cells = 1:4) in vitro in RPMI-1640 complete medium supplemented with 2 mM L-glutamine, 10 mM HEPES, 20 mM 2-ME, 10% fetal bovine serum, penicillin/streptomycin, and 40 ng/ml hGM-CSF, at 37 °C in a humidified 5% CO 2 incubator. T cell proliferation was induce by anti-CD3/CD28 stimulation beads (Invitrogen,  Statistical analysis. Flow cytometry and real-time PCR were performed in triplicate, and each experiment was repeated at least three times. All western blot images and semi-quantitative qRT-PCR results are representative of at least three independent experiments. Data are presented as mean ± standard deviation (SD) of at least three independent experiments. T-tests were used to compare the differences between two groups, Paired/ Wilcoxon matched-pairs signed rank test or unpaired/Mann-Whitney tests were used to examine the differences within groups or between groups, respectively. All P values were 2 sided and considered to be statistically significant at p < 0.05 (*p < 0.05, **p < 0.01, as indicated in the figures and legends).
Ethics approval and consent to participate. The study was conducted in accordance with the 1964 Helsinki Declaration. All procedures performed and all clinical data obtained in the present study involving human participants were reviewed and approved by the Medical Ethics Committee of the First Affiliated Hospital of Henan University of Chinese Medicine (protocol number 2016HL-085). All individual participants included in the study gave written informed consent.
Open Access. This article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.

Results
The purity of HLA-DR -/low mononuclear cells. We found the purity of HLA-DR -/low mononuclear cells sorted by magnetic beads to be more than 95%, as determined by flow cytometry; The proportion of Lin -HLA-DR -/low cells was more than 60% (Fig. 1).Trypan blue staining showed the total number of living cells to be more than 90% and was sufficient for all subsequent experiments.

ARG1-expressing cells in peripheral blood of GAC patients were mainly early-stage MDSCs.
To study the phenotype and the ARG1 expression level of circulating myeloid cells from GAC patients, we collected peripheral blood samples from 20 GAC patients and 20 age-matched healthy volunteers. We used flow cytometry to analyze the phenotype and number of MDSCs and then qRT-PCR and western blotting to determine the expression level of ARG1 mRNA and protein, respectively, in different phenotypes of MDSCs. We first gated PBMCs based on the expression of HLA-DR and ARG1 (Fig. 2a,b). In the peripheral blood of healthy volunteers, HLA-DR -/low cells were the predominant cell type, and ARG1 expression levels were high (Fig. 2b). Compared to healthy volunteers, the number of HLA-DR -/low cells was significantly higher (7315 ± 2791/μl vs 4287 ± 1318/ μl) (p < 0.01), and the proportion of ARG1-expressing cells was not significantly higher (55.02 ± 7.53% vs 43.75 ± 11.15%) (p > 0.05) in the peripheral blood of GAC patients (Fig. 2c).

ARG1-expressing cells in tumor tissue were mainly immature and monocytic MDSCs.
In order to observe the phenotypic characteristics and the ARG1 expression level of tumor-infiltrating MDSCs from GAC patients, we collected cancerous and paracancerous tissues from 6 GAC patients to prepare mononuclear cell  www.nature.com/scientificreports www.nature.com/scientificreports/ suspensions. We then determined their immunophenotype by using BD FCSCantoTM-II flow cytometry. Data indicated that most of the mononuclear cells in cancerous and paracancerous tissues were HLA-DR -/low cells (92.27 ± 4.72% vs 92.87 ± 4.06%) (p > 0.05) (Fig. 3a). However, HLA-DR -/low cells with high expression levels of ARG1 were significantly more common in cancerous tissues than in the paracancerous tissue (17.03 ± 2.50% vs 4.97 ± 1.40%) (p < 0.01) (Fig. 3a,b).

ARG1 and NOS2 were highly expressed in circulating and tumor-infiltrating HLA-DR -/low mononuclear cells from GAC patients.
To further verify the results derived from flow analysis, we investigated ARG1 and NOS2 expression in HLA-DR -/low mononuclear cells in peripheral blood and cancerous tissues from GAC patients. First, HLA-DR -/low mononuclear cells were sorted from peripheral blood and a mononuclear cell suspension from tumor tissue. Then, the relative expression levels of ARG1 mRNA and NOS2 mRNA were detected by qRT-PCR, and the relative expression levels of ARG1 and NOS2 protein were detected by western blotting. The relative expression levels of ARG1 mRNA and NOS2 mRNAs in HLA-DR -/low mononuclear cells from the GAC patients' peripheral blood was 3.01 times and 6.4 times higher respectively, than those of healthy volunteers (p < 0.01) (Fig. 4a). The relative expression levels of ARG1 and NOS2 protein were increased by factors of 3.89 and 3.49, respectively, (p < 0.01) (Fig. 4b).

MDSCs from cancer patients suppressed both T cells proliferation and IFNγ production.
HLA-DR -CD33 + MDSCs from cancer patients or from healthy donors were co-cultured with CD8 + T cells-labeled with CFSE respectively, and CD8 + T cells were cultured alone. After 24 hour, CD8 + T cell proliferation was analyzed on a CantoII flow cytometer(BD Biosciences). As shown in Fig. 5, compared to MDSCs from healthy donors, MDSCs from cancer patients suppressed both T cell proliferation and IFNγ production (p < 0.01) (Fig. 5f,g). At the same time, compared to a positive T cell proliferation control, MDSCs from healthy donors also suppressed both T cell proliferation and IFNγ production (p < 0.01) (Fig. 5f,g).

Discussion
In 2000, a group reported the presence of CD11b + /Gr-1 + myeloid progenitor cells in mice; These cells had the ability to inhibit CD8 + T cells 28 . Subsequent studies have also found that in the peripheral blood of tumor patients, there is a type of immature myeloid cell that is devoid of lymphoid and mature markers, but with the ability to inhibit CD8 + T cell response 29,30 . These cells were identified as immature myeloid cells 31,32 . As research in this field has progressed, some scholars have proposed that it would be more accurate to name such cells, myeloid-derived suppressor cells 33,34 .
MDSCs are a group of heterogeneous cells derived from IMCs. Under normal conditions, IMCs can be differentiated into the precursor of dendritic cells, macrophages and/or granulocytes. However, in the presence of tumors, inflammation, trauma and other conditions, the mature differentiation of IMCs is blocked and they transform into MDSCs 1 .
MDSCs are derived through the action of a variety of cytokines secreted by various tumor and/or host cells, and as a result, the phenotype of MDSCs differs in different tumors. As yet, there are few in vivo studies of MDSCs in GC patients. Moreover, research has yet to elucidate the phenotypic characteristics of MDSCs under GC conditions, as well as the distribution and aggregation characteristics of MDSCs in patients, along with the activation of their immune suppression function. Furthermore, the relationship between the immune-avoidance of GC cells and MDSCs cannot be explained.
In related studies on gastric cancer, granulocytic MDSCs with CD15, CD66 and CD33 markers. and monocytic MDSCs with CD14 markers, have been found to increase significantly in the peripheral blood of patients. In other studies, Lin(CD3/CD19/CD56) -HLA-DR -CD33 + MDSCs, CD11b + CD15 + CD14 -MDSCs, CD15 + FSC hi SSC hi MDSCs and CD14 + HLA-DR -MDSCs were significantly increased in the PBMCs of GC patients 27,58 .  www.nature.com/scientificreports www.nature.com/scientificreports/ We divided the circulating myeloid cells into four phenotypes: HLA-DR -CD33 + , HLA-DR -CD11b + , HLA-DR -CD15 + and HLA-DR -CD14 + . First, we observed changes in the numbers of the four cell subtypes in peripheral blood under gastric cancer conditions. Compared with healthy volunteers, all four subtypes were significantly increased; the number of cells in each of the four subtypes was similar and did not differ significantly. Further analysis of the expression level of ARG1 in different subtypes showed that the subtypes with high expression of ARG1 in the peripheral blood of healthy volunteers were mainly HLA-DRmononuclear cells, while the number of HLA-DRmononuclear cells with high expression levels of ARG1 in GC patients was significantly higher than that in healthy volunteers.
Our study suggested that although there was no difference in the number of the four subtypes, there were significant differences in the expression of ARG1. The expression frequency of ARG1 in HLA-DR -CD33 + MDSCs was significantly higher than that in other MDSC subtypes, and the expression frequency of ARG1 in HLA-DR -CD14 + MDSCs was also higher than that in HLA-DR -CD15 + MDSCs and HLA-DR -CD11b + MDSCs, indicating that HLA-DR -CD33 + MDSCs were different subtypes compared with the other three MDSC phenotypes. HLA-DR -CD14 + MDSCs were not of the same type as HLA-DR -CD15 + MDSCs and HLA-DR -CD11b + MDSCs, but there was no significant difference in the expression frequency of ARG1 between HLA-DR -CD15 + MDSCs and HLA-DR -CD11b + MDSCs. Further immunophenotyping of HLA-DR -CD33 + MDSCs was performed. We identified two main subgroups with high expression of ARG1 in the peripheral blood of healthy volunteers: HLA-DR -CD33 + CD14 -CD15 -MDSCs and HLA-DR -CD33 + CD14 -CD15 + MDSCs. While ARG1 was highly expressed only in the HLA-DR -CD33 + CD14 -CD15 -MDSC subtype, accounting for more than 90%, the HLA-DR -CD33 + CD14 -CD15 + subtype hardly expressed ARG1 in peripheral blood of GAC patients. These data gave us a new understanding of HLA-DR -CD33 + MDSCs. The circulating HLA-DR -CD33 + CD14 -CD15 -MDSCs under gastric cancer conditions are therefore the main force that plays an immunosuppressive role and such subtypes are early-stage MDSCs 7 .
It has been reported that the proportion of HLA-DR -CD14 + MDSCs and HLA − DR − CD45 + CD11b + CD14 + MDSCs in GC tissues are significantly increased 20,59 . We therefore tried to find out whether other phenotypes of MDSCs exist in gastric cancer tissues. First, we analyzed the expression of ARG1 in HLA-DRmononuclear cells. Data indicated that the ARG1-expressing mononuclear cells are mainly the HLA-DRcells in GC tissue. Further immunophenotyping of HLA-DRmononuclear cells verified that the ARG1-expressing cells are derived mainly from HLA-DR -CD33 + CD11b -, HLA-DR -CD33 + CD11b + and HLA-DR -CD33 -CD11bsubtype MDSCs, and that HLA-DR -CD33 -CD11b + MDSCs barely express ARG1.
Further immunophenotyping of the first three MDSC subtypes verified that ARG1 was barely expressed in CD15 + granulocytic MDSCs, while ARG1 was highly expressed in over 35% of HLA-DR -CD33 -CD 11b -CD14 -CD15 -MDSCs. These subtype MDSCs lack a surface immunophenotype and have been tentatively named naïve MDSCs. Simultaneously, ARG1 was also highly expressed in more than 15% of HLA-DR -CD33 + CD14 -CD15 -MDSCs, such subtypes are most likely early-stage MDSCs. Also, ARG1 was highly expressed in over 40% of HLA-DR -CD14 + monocytic MDSCs. On the whole, ARG1-expressing MDSCs were mainly naïve MDSCs, early-stage MDSCs and monocytic MDSCs in gastric cancer tissue. This study confirmed there were significant differences between circulating MDSCs and tumor-infiltrating MDSCs in terms of phenotypic characteristics and ARG1 expression under GC conditions.
Previous studies have shown that there is a large amount of ARG1 and NOS2 in MDSCs under tumor conditions and that these factors can inhibit the function of T cells, thereby creating an important mechanism for immune-avoidance 60,61 . We used RT-PCR and western blotting to determine the relative expression levels of ARG1 and NOS2 in HLA-DRmononuclear cells in the peripheral blood and tumor tissues of GC patients. Data confirmed that there was a sufficiently high amount of ARG1 and NOS2 to cause T cell function inhibition in HLA-DR -/low mononuclear cells under gastric cancer conditions. Although our study highlights that phenotypic characteristics and the quantity of MDSCs, and the levels of ARG1, differ when compared between circulating and tumor-infiltrating MDSCs, our study was limited because there was not a large number of specimens and myeloid cell surface markers. In addition, this study did not focus on the specific mechanisms by which gastric cancer cells induce immunosuppression; these should be investigated in future research.
In Conclusion, our findings demonstrated for the first time that a large number of early-stage MDSCs can aggregate in the peripheral blood of GAC patients and express high levels of ARG1. We also verified that the specific subtypes of ARG1-expressing MDSCs in GAC tissue were mainly naïve MDSCs, early-stage MDSCs and monocytic MDSCs. There were significant differences in ARG1 expression and immunophenotype when comparing circulating MDSCs and tumor-infiltrating MDSCs. Therefore, we speculate that MDSCs aggregate in the peripheral blood and tumor tissues of GC patients and that this may represent an important mechanism by which gastric cancer cells induce immunosuppression. It is possible that this mechanism could explain the manner in which gastric cancer cells avoid the immune system.

Data availability
All datasets on which the conclusions of the paper depend are available to readers.