Overcoming resistance to immunotherapy by targeting GPR84 in myeloid-derived suppressor cells – Signal Transduction and Targeted Therapy

RNA sequences

Genbank

Sequencing data

GSE71706, GSE21927, GSE145370, GSE78220, GSE91061, EGAS00001002556.

The data from TCGA generated during this study are available at http://www.cbioportal.org.

Online analysis

PROMO (http://alggen.lsi.upc.es/)

JASPAR (http://jaspar.genereg.net/)

The detailed reagent materials were displayed in the KEY RESOURCES TABLE in Supplementary data.

Study approval

All mouse experiments were approved by the Henan Key Laboratory for Pharmacology of Liver Diseases (approval 00018170). All participants provided written informed consent. This study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval 2019-KY-256).

Animal models

Animals were maintained under pathogen-free conditions and provided care in accordance with the International Association for Assessment and Accreditation of Laboratory Animal Care policies and certification. Four-to-six-week-old female C57BL/6J mice were purchased from the Beijing Charles River Company (Beijing, China). C57BL/6J–GPR84/BRL mice were provided by BRL⁺ Medicine (Shanghai, China). OT-1 mice were kindly provided by the Bo Huang lab of the Institute of Basic Medicine, Chinese Academy of Medical Sciences.

The mouse model of orthotopic esophageal cancer was generated by challenge with 4-NQO added to drinking water.²⁶ A 5 mg/mL stock solution was prepared by dissolving 4-NQO in 5 mL of propylene glycol. Next, 3 mL of the stock solution was mixed with 147 mL of autoclaved water to a final concentration of 100 μg/mL) for drinking water. Drinking water was changed 1–2 times per week. The control group was provided 147 mL of drinking water mixed with 3 mL of propylene glycol. To construct mouse models of lung cancer and malignant melanoma for further study, LLC (5 × 10⁵) and B16 (2 × 10⁵) cells were subcutaneously injected. Once the tumor volume reached 50 mm³, the mice were randomly divided into two groups. A GPR84 antagonist was administered through drinking water at a dose of 50 mg kg⁻¹ day⁻¹. The control group was administered the same concentration of the solvent. Anti-PD-1 monoclonal antibody (BE0146, clone RMP1-14; Bio-XCell) or IgG isotype control (BE0089, clone 2A3; Bio-XCell) was administered at 10 mg/kg intraperitoneally once every three days, for a total of four injections.²⁷ Esophageal cancer was evaluated based on the amount of food consumed, water intake, or body weight of the mice. Mice were euthanized on days 0, 113, and 162; tissue sections stained with hematoxylin and eosin were examined to assess tumor development. On day 135 after 4-NQO treatment, WT and GPR84^−/− mouse models of esophageal cancer were sacrificed for further experiments. Lung cancer and malignant melanoma were assessed by evaluating tumor growth. To observe the effect of GPR84 deletion on MDSCs and CD8⁺ T cells in LLC and B16F0 models, tumor-bearing mice were sacrificed on the 20th day after LLC injection and the 22nd day after B16F0 challenged, respectively.

Clinical samples

Peripheral blood (10 mL) was collected from healthy donors aged 40–70 years and from patients with esophageal squamous cell carcinoma or lung adenocarcinoma. All patients were diagnosed via pathological examination. Mononuclear cell suspensions (peripheral blood mononuclear cells) were obtained by density-gradient centrifugation in lymphocyte separation solution. Labeled antibodies and MDSCs or CD8⁺ T cells were separately detected and purified according to experimental requirements.

Tumor and paired paracancerous tissues from patients with esophageal squamous cell carcinoma and lung adenocarcinoma cancer, confirmed by pathological diagnosis, were collected under sterile conditions, placed in RPMI 1640 medium containing 10% FBS, 10% penicillin, and 10% streptomycin, and immediately sent to the laboratory. All analyzed samples were confirmed by pathology. The Miltenyi CliniMACS (Cologne, Germany) system was used to obtain single-cell suspensions from tumor tissues, followed by flow cytometric detection or sorting of target cells. Samples used for immunohistochemistry and immunofluorescence were provided by the Department of Pathology, First Affiliated Hospital of Zhengzhou University, and the Henan Key Laboratory of Esophageal Cancer (Zhengzhou, China).

Cell lines and culture conditions

All cell lines were tested to confirm that they were free of mycoplasma or other pathogens. The murine lung-cancer cell lines LLC, B16, and B16-OVA were purchased from the American Type Culture Collection (Manassas, VA, USA). Cells were cultured at 37 °C in 5% CO₂ in Dulbecco’s modified Eagle medium supplemented with 10% FBS, penicillin, and streptomycin.

Isolation of tumor-infiltrating cells

Tumor tissues from mice and surgery patients were washed with PBS to remove impurities, cut into ~1-cm³ pieces, and placed in a gentleMACS™ C tube (Miltenyi). Next, 4.7 mL of serum-free RPMI 1640 medium was added to the tissues, followed by mechanical dissociation using a tri-enzyme solution (325 μL). The solution was filtered, and cell pellets were collected for subsequent experiments.

Isolation of immune cells

Mice in which 4-NQO had induced the formation of esophageal tumors were selected; these tumors were used to prepare single-cell suspensions, which were divided into three groups. Group 1 was labeled with CD3, CD8, and CD49b; group 2 was labeled with CD3, B220, and CD11c; and group 3 was stained with Gr1, CD11b, and F4/80. Single-cell suspensions from patients with esophageal cancer were prepared to isolate CD11b⁺ cells using model-based analysis of chromatin immunoprecipitation sequencing, as CD11b⁺ cells express CD33, 7-AAD, HLA-DR, CD11b, and GPR84. After incubation with antibodies at 4 °C for 20 min, cells were washed with PBS to remove unbound antibodies. Cells were resuspended in 500 μL of RPMI 1640 medium containing 10% FBS, then aliquoted into a sterile flow-type loading tube by passing through a 200-mesh strainer. For sorting, gating parameters were selected based on different target cells. After fixing the sorting position, the analysis was stopped, and sorting was initiated. Cells of interest were collected in a flow tube containing serum-free RPMI 1640 medium.

Flow cytometry

Single-cell suspensions prepared from peripheral blood, tumor tissue, or mouse spleen tissue were counted and 1 × 10⁶ cells were used for staining. After vortexing, labeling reactions were incubated at 4 °C in the dark for 15 min, followed by washing with 1 mL PBS and data acquisition with fluorescence-activated cell sorting (BD FACSCanto II). Export FCS files and FlowJo 10.4 software were used for data analysis.

MDSCs induction

Bone-marrow cells were obtained from 6-week-old C57BL/6 mice and divided into control, GM-CSF combined with IL-6, and GM-CSF combined with G-CSF treatment groups. The working concentration of cytokines was 50 ng/mL. After culturing for 72 h, cells were harvested and labeled for flow cytometry with antibodies against Gr1, Ly6G, Ly6C, 7-AAD, and GPR84 to determine MDSCs proportions and GPR84 expression. MDSCs from different groups were collected to analyze functional molecules, such as CYBB, NCF4, NOS2, and ARG1, by real-time quantitative PCR.

Sorting MDSCs and CD8⁺ T cells

After obtaining single-cell suspensions, red blood cells were lysed and an isolation kit (Miltenyi) was used to purify MDSCs or CD8⁺ T cells from clinical samples and tumors from the mouse models. The sorting efficiency of MDSCs or CD8⁺ T cells by flow cytometry was >90%.

Inhibitory effect of MDSCs on CD8⁺ T cells

The CFSE fluorescent probe labeling method was used to detect the proliferation of CD8⁺ T cells collected from clinical samples or tumor-bearing mice. Cells were mixed with CFSE (5 μmol/L) for labeling. Labeled cells were incubated with MDSCs from different sources (MDSCs:CD8 = 4:1), with individual CD8⁺ T cells used as controls. CD3/CD28 Dynabeads were added to stimulate CD8⁺ T cell proliferation. After 48 h, the fluorescence of fluorescein isothiocyanate (FITC) was analyzed by flow cytometry to measure CD8⁺ T cell proliferation. Supernatants were collected for IFN-γ detection by ELISA to measure differences in CD8⁺ T cell function between the control and coculture groups.

In vitro cytotoxicity assay

CD8⁺ T cells were obtained from the spleens of 6-week-old OT-1 transgenic mice using magnetic bead sorting. B16F0-OVA cells labeled with CFSE were incubated with CD8⁺ T cells (5:1) and MDSCs from different sources were added (MDSCs:CD8 = 4:1) to interfere with the interaction. After incubation at 37 °C for 8 h, percentages of apoptotic tumor cells were measured by flow cytometry. Propidium (PI) was added before detected by flow cytometry. The double positive cells with PI and CFSE were apoptotic tumor cells.

Quantitative RT-PCR

Total RNA from MDSCs, T cells, B cells, NK cells, and macrophages was isolated using TRIzol Reagent (Invitrogen Life Technologies, Carlsbad, CA, USA). Reverse transcription was performed using a PrimeScript RT reagent kit, following the manufacturer’s instructions (Takara, Shiga, Japan). Gene expression was detected using SYBR Green qPCR Master Mix (Takara) on an MX3005P qPCR system (Agilent Technologies, Santa Clara, CA, USA). Relative mRNA expression was calculated using the 2^−ΔΔCt method.

Immunohistochemistry and immunofluorescence

Deparaffinized sections were subjected to antigen retrieval, followed by endogenous oxidase inactivation. After blocking in PBS containing 2% bovine serum albumin, primary antibodies for GPR84 or control anti-rabbit IgG were added and incubated overnight at 4 °C. Samples were then stained with anti-rabbit antibody labeled with horseradish peroxidase. Slides were analyzed by Pannoramic Scanner, and H-score was used to calculate expression levels for the statistical analysis of overall survival.

Dewaxing, antigen retrieval, and blocking for immunofluorescence detection were similar to those used for immunohistochemistry. Primary antibodies were added and incubated overnight at 4 °C. After washing with PBS, secondary antibodies were added for 1 h in the dark. Nuclei were counterstained with DAPI and slides were analyzed by fluorescence microscopy and Case Reviewer software.

Hematoxylin-eosin staining

Sections (3 µm) from normal esophageal tissues or 4-NQO challenged esophageal cancer tissue were placed in a 60 °C oven for 1.5 h. After dewaxing, hematoxylin staining was performed at room temperature for 10 min, washed with running water, and then placed in 0.7% hydrochloric acid and ethanol for 10 s. Eosin staining was performed at room temperature; then, sections were passed through increasing concentrations of ethanol and xylene for rehydration.

Dual-luciferase reporter assay

HEK 293T cells were transfected with the Renilla luciferase plasmid pRL-SV40, firefly luciferase plasmid pGL3-WT or MUT, and C/EBPβ overexpression plasmid in a 1:1:1 ratio for 24 h according to the manufacturer’s instructions. Cell lysates were analyzed using the Dual Luciferase Reporter Assay Kit (E2910, Promega) on a microplate reader. Firefly luciferase activity was normalized to Renilla luciferase activity for each sample.

ROS detection

MDSCs purified from healthy volunteers and patients with esophageal carcinoma were resuspended in 1 mL serum-free RPMI 1640 medium mixed with the fluorescent probe 2,7-dichlorofluorescin diacetate (DCFH-DA) (10 μmol/L) and incubated at 37 °C for 20 min, with inversion and mixing every 5 min. Cells were washed three times with serum-free RPMI 1640 medium. The presence of ROS was evaluated based on FITC fluorescence using flow cytometry.

ARG1 activity detection

Arginase is a manganese-containing enzyme that catalyzes the conversion of arginine to urea and ornithine. MDSCs from different sources were collected, cell lysates were added, and arginase activity was detected simultaneously using a urea standard working solution. Substrate buffer was added, incubated for 2 h, and urea reagent was used to stop the arginase reaction. The absorbance of each well at a wavelength of 430 nm was measured using a multi-functional microplate reader and calculated using the formula:

Lysosomal inhibition

MDSCs (2 × 10⁶/well) were treated with G-SCF and GM-SCF for 48 h. Then, the selective lysosomal inhibitor NH4Cl (250 μM) or chloroquine (20 μM) was added, followed by incubation for 24 h at 37 °C with 5% CO₂. After incubation, the cells were washed three times with FACS buffer (0.5% FBS in 1 × PBS). The primary antibody, LAMP1 (ab208943, Abcam, 1:500) was added and incubated for 30 min on ice in the dark. Cells were washed three times by centrifugation at 1500 rpm for 5 min and resuspended in FACS buffer. Fluorochrome-labeled secondary antibodies Alexa Fluor®R555 (FcMACS, 1:1000) and streaming antibodies (CD11b, Gr1, and PD-L1) were added and incubated for 30 min on ice in the dark. Cells were then washed three times by pelleting at 1500 rpm for 5 min and resuspended in FACS buffer. Cells were analyzed using a FlowSight Imaging flow cytometer (Amnis). Data were analyzed using IDEAS software (Amnis).

RNA sequencing

Fresh GPR84⁺ and GPR84⁻ MDSCs were sorted from esophageal cancer tissue and RNA was extracted using a Qiagen Allprep kit (Hilden, Germany). Similarly, LLC tumors from WT or GPR84^−/− mice administered anti-PD-1 or isotype mAb were collected and RNA was extracted using the same method. Approximately 1 µg of RNA was used by Novogene for library construction and sequencing. Messenger RNA profiles were generated based on single-read deep sequencing in triplicate using an Illumina HiSeq2000 (San Diego, CA, USA).

mRNA sequencing data analysis

Sequence files from Illumina HiSeq that passed quality filters were aligned against the mouse transcriptome using Bowtie2 aligner4. Gene-level count summaries were analyzed for significant changes using DESeq. Individual p-values were adjusted for multiple testing by calculating Storey’s q-values using the fdrtool trimmer. For each gene, the q-value had the smallest false discovery rate at which the gene was significant. Biological processes were analyzed according to the guidelines provided by the Gene Ontology Consortium, in which each gene ontology term defines a set of genes. The entire list of genes, sorted based on q-values in ascending order, was subjected to a non-parametric variant of GSEA, in which the parametric Kolmogorov-Smirnov p-value was replaced with the exact rank-order p-value. Heatmaps depicting the expression levels were generated using an in-house hierarchical clustering software that implements Ward clustering. The colors qualitatively correspond to fold changes in expression. The mRNA-sequencing data reported in this study have been deposited in Genome Sequence Archive (Genomics, Proteomics & Bioinformatics 2021) in the National Genomics Data Center (Nucleic Acids Res 2022), China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA003874) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human. The project number is PRJCA014498. And in this project, the data from LLC mice has been submitted in OMIX (https://ngdc.cncb.ac.cn/omix/submitList) as OMIX002872.

TCGA database analysis

The mRNA sequencing data from patients with esophageal squamous cell carcinoma and lung adenocarcinoma carcinoma were downloaded from TCGA. Eight molecules related to the MDSCs phenotype or function (such as CD33) were selected for correlation analysis with GPR84. The results are displayed in a heat map generated using the corrplot package in R. Patients were divided into GPR84^high and GPR84^low groups based on GPR84 mRNA expression levels. Kaplan–Meier analysis was performed to calculate the difference in overall survival between GPR84^high and GPR84^low patients.

Single-cell RNA sequencing data analysis

The single-cell RNA sequencing data of ESCC were downloaded from GEO database(GSE145370). we processed the unique molecular identifier (UMI) count matrix using the R package Seurat (version 3,2,2). As the quality control(QC), we first excluded genes detected less than ten cells and cells that have <200 nonzero count genes. We next filtered out the cells if their proportions of mitochondrial gene expression were larger than 10%. We used canonical correlation analysis (CCA) to aggregate each sample, respectively. After QC, 102611 cells and 18570 genes were included in downstream analysis. For clustering, we first run PCA and selected top 20 PCA to find clusters with resolution of 0.5, and then we performed run T‐distributed Stochastic Neighbor Embedding (tSNE) and the clusters were visualized by DimPlot in Seurat. The visualization of GPR84, ITGAM were used FeaturePlot in Seurat. All analysis were conducted by R(version 3.6.3).

Statistical analyses

Statistical analyses were performed using GraphPad Prism 7 software (GraphPad, Inc., La Jolla, CA, USA) or R project. The Kaplan–Meier method was used to evaluate the effect of GPR84 on the overall survival of patients with esophageal cancer or lung adenocarcinoma cancer. Paired or unpaired t-tests were used to analyze the differences in GPR84 or other immunosuppressive molecule expression in different MDSCs. One- or two-way ANOVAs were performed to compare multiple conditions. The normality of all samples was tested using the Shapiro-Wilk test. Spearman correlation analysis was used to evaluate the correlation between GPR84 expression and MDSCs functional molecule expression.