Introduction

Ovarian cancer is the eighth leading cause of cancer death in women worldwide, with an estimated figure of more than 200,000 deaths per year (1). In addition to conventional surgery and chemotherapy, the introduction of anti-VEGF and PARP inhibitors as therapeutic modalities has remarkably prolonged the duration of progression-free survival (PFS), even though progression remains inevitable in the majority of cases (2, 3). The efficacy of PD-1/PD-L1 inhibition therapy is limited in ovarian cancer, as evidenced by clinical trials, and immunotherapy has not yet become the standard treatment option (4–6).

Of all the histologic types of ovarian cancer, high-grade serous ovarian cancer (HGSOC) remains the most common and the most aggressive (7, 8). Although the tumor microenvironment (TME) of HGSOC varies widely, it is well characterized by four subtypes based on gene expression profiles: differentiated, immunoreactive, mesenchymal, and proliferative subtypes (9, 10). The immunoreactive subtype is characterized as tumors with high expression of IFNγ and PD-L1, a representative antitumor cytokine and an immunomodulatory molecule upregulated in response to IFNγ, respectively, and patients have a better prognosis (10–12). The mesenchymal type is characterized as epithelial–mesenchymal transition–high, immunosuppressive tumors with poor prognosis (10, 13, 14). The differentiated and proliferative subtypes consist of immune "desert" tumors (14). PD-1/PD-L1 inhibition therapy is relatively effective in treating PD-L1–high immunoreactive tumors. However, nonimmunoreactive phenotypes, which comprise a majority of HGSOC tumors, respond poorly to the treatment (4). To elucidate what contributes to the differences in the TME, and new immunologic approaches to treat nonimmunoreactive phenotypes other than PD-1/PD-L1 inhibition, are required.

B7-H3 is a transmembrane protein from the B7 family; it was first reported in 2001 (15). Developments in therapies targeting B7-H3, including antibody–drug conjugates, radioimmunotherapy, and chimeric antigen receptor T cells, which focus on its high selective expression in tumor tissues, highlight the potential of B7-H3 as a therapeutic target (16–18). B7-H3 seems to play complex roles in modulating the TME. Not only does it exert its immunomodulatory effects by directly acting on target cells as an immune checkpoint (15, 19, 20), it is also involved in the intracellular signaling of cancer cells, such as the STAT3 pathway (21–23). Its contribution to immunosuppressive TME, other than as an immune checkpoint molecule, remains to be fully elucidated.

Here, we demonstrate that B7-H3 expression is upregulated in the nonimmunoreactive tumors of HGSOC and that B7-H3 contributes to ovarian cancer progression via CCL2–CCR2–M2 macrophage axis–mediated immunosuppression, in both mouse syngeneic ovarian cancer models and HGSOC cases. Our results not only augment the current understanding of the immunologic heterogeneity of HGSOC but also provide a foundation for stratification of the immunosuppressive HGSOC phenotype with a B7-H3–high, M2 macrophage–rich TME and facilitate the development of new treatment strategies other than PD-1/PD-L1 inhibition that can specifically benefit patient groups with extremely poor prognosis.

Materials and MethodsStudy approval

This study was approved by the Kyoto University Graduate School and Faculty of Medicine Ethics Committee (reference number G531) and conforms to the Declaration of Helsinki. Informed consent was obtained from all participants via an opt-in approach (wherein participants signed a printed informed consent document) or an opt-out approach (wherein participants were informed about the study through the website). All animal studies were approved by the Kyoto University Animal Research Committee.

Human samples

Formalin-fixed, paraffin-embedded (FFPE) primary site tumor specimens (n = 62) and fresh primary site tumor samples (n = 28) were collected from patients with HGSOC, who underwent primary surgery at the Kyoto University Hospital (Kyoto, Japan) between 1998 and 2015. Patients without primary site resection surgery or those who received chemotherapy prior to surgery were excluded. Blood samples from patients with HGSOC (n = 23) and healthy female donors (n = 10) were also collected between 2012 and 2020. FFPE samples were stored at room temperature, and fresh tumor samples and blood samples were stored at −80°C until use.

Bioinformatics analysis

Gene expression data of 30 HGSOC specimens from the Kyoto University Hospital (Kyoto, Japan) that were partially deposited in the Gene Expression Omnibus (RRID: SCR_005012, accession numbers: GSE39204 and GSE55512) were classified into the four transcriptome subtypes (24), according to the Classification of Ovarian Cancer (CLOVAR) subtype signatures (10). Differentially expressed genes (DEG) between the immunoreactive (n = 8) and nonimmunoreactive (n = 22) subtypes were extracted using the Samroc method as described previously (25). The gene expression profiles from The Cancer Genome Atlas-Ovarian Cancer (TCGA-OV) RNA-sequencing dataset (n = 263) from TCGA Data Portal (RRID: SCR_003193, illuminahiseq_rnaseqv2_Level_3_RSEM_genes_normalized_data files obtained on October 19, 2015) was used for comparison. Single-sample gene set enrichment analysis (26), obtained from GenePattern software (RRID:SCR_003201, version 3.5.0), was performed to evaluate the IFNγ signature activity scores using the previously reported gene set (27) using R software (RRID: SCR_001905, version 3.6.1).

Cell culture and transfection

The OV2944-HM-1 (HM-1) mouse ovarian cancer cell line was purchased from RIKEN BioResource Center (RRID: SCR_003250, catalog no. RCB1483; RRID: CVCL_E954; Tsukuba, Japan) in January 2003. HM-1 cells were cultured and maintained in minimal essential medium (MEM)-alpha (catalog no. 12571063, Thermo Fisher Scientific) supplemented with 10% (volume for volume) heat-inactivated FBS (catalog no. S1810, Biowest) and penicillin–streptomycin (100 IU/mL penicillin and 100 μg/mL streptomycin; catalog no. 26253-84, Nacalai Tesque) in an atmosphere of 5% CO2 at 37°C. The ID8 mouse ovarian cancer cell line was kindly provided by Katherine Roby of The University of Kansas Medical Center (Kansas City, KS; RRID: CVCL_IU14) in September 2009. Human ovarian cancer cell lines OVCAR3 (RRID: CVCL_0465) and OVCA420 (RRID: CVCL_3935) were kindly provided by Susan K. Murphy of Duke University (Durham, NC) in June 2007. Both ID8 and human ovarian cancer cells were cultured and maintained in RPMI1640 medium (Invitrogen) supplemented with 10% FBS and penicillin–streptomycin in an atmosphere of 5% CO2 at 37°C. All cell line–based experiments were performed before passage 20. All cell lines were regularly tested for Mycoplasma contamination (date of last check: April 12, 2021) using a Mycoplasma PCR Detection Kit (catalog no. G238; Applied Biological Materials). Human cell lines (OVCAR3 and OVCA420) were reauthenticated on April 14, 2021, by short tandem repeat analysis.

For CRISPR/Cas9-mediated Cd276 knockout (B7-H3 KO), target sequences were determined as described previously (16). Three single-guide RNAs (sgRNA) and a control sgRNA were synthesized by FASMAC. A mixture of the RNAs and Guide-it recombinant Cas9 (catalog no. 632641, Takara Bio Inc.) was transfected into the HM-1 and ID8 cells using the TransIT-X2 Dynamic Delivery System (catalog no. MIR6003, Mirus Bio). The cells were incubated with rabbit anti-CD276 (1:1,000 dilution, clone: EPNCIR122, catalog no. ab134161; RRID: AB_2687929, Abcam) for 20 minutes at 4°C and with Goat Anti-Rabbit IgG H&L (Alexa Fluor 647; 1:2,000 dilution, catalog no. ab150079; RRID:AB_2722623, Abcam) for 10 minutes at 4°C. Alexa Fluor 647–negative cell populations were sorted by flow cytometry using BD FACSAria II Cell Sorter (RRID: SCR_018934, BD Biosciences). The purity of Alexa Fluor 647–negative cells was >99%. Single cells were isolated by limiting dilution and cultured again on 96-well plates (catalog no. 655180, Greiner Bio-One) for 8 days under the same conditions described in the section above, followed by expansion. Finally, B7-H3 expression deficiency was confirmed using FACS analysis and quantitative PCR (qPCR).

B7-H3–knockdown HGSOC cells, OVCAR3-sh-B7-H3 and OVCA420-sh-B7-H3, and their control cells were generated by lentiviral-mediated transfection with short hairpin RNAs (shRNA) targeting B7-H3 or a nonsilencing control (sh1, catalog no. TRCN0000128599; sh2, catalog no. TRCN0000436667; sh control, catalog no. SHC002; Sigma-Aldrich). The sgRNA and shRNA sequences are listed in Supplementary Table S1.

Mice

Five- to 6-week-old female C57BL6 (strain code: 027) and B6C3F1 (C57BL6×C3/He F1, strain code: 031) mice were purchased from CLEA Japan and used in immunocompetent mouse experiments. Five- to 6-week-old female BALB/c-nu (Foxn1nu, strain code: 194) mice were purchased from CLEA Japan and used in immunodeficient mouse experiments. The C57BL6 background Cd276−/− (B7-H3 KO) mice were kindly provided by Brad St. Croix of the NCI (Bethesda, MD). Animals were maintained under specific pathogen–free conditions.

Tumor models

A total of 1 × 106 HM-1-B7-H3 KO cells or control cells were inoculated intradermally into the shaved abdominal lesion of B6C3F1 and nude mice. A total of 5 × 105 HM-1-B7-H3 KO or control cells were injected into the abdominal cavity of B6C3F1 or nude mice (intraperitoneal model). Similarly, 5 × 106 ID8 B7-H3 KO or control cells were injected into the abdominal cavity of C57BL6 or nude mice (intraperitoneal model). Intradermal tumor size was measured using calipers and calculated as follows: volume = length × width × height × π/6. In the intraperitoneal models, tumor growth was evaluated on the basis of changes in body weight, which reflect the amount of ascites and omental tumor weight. Survival analysis was also performed in these models. Intradermal tumor size and body weight were measured once or twice a week. Intradermal tumors of HM-1-B7-H3 KO or control cells at days 10, 12, and 25 were used for RNA sequencing, flow cytometry, IHC, and ELISA, respectively. Intraperitoneal tumors of ID8-B7-H3 KO or control cells at days 63 and 67 were used for ELISA and flow cytometry, respectively. For B7-H3 KO mouse studies, only Cd276+/+ and Cd276−/− littermates derived from Cd276 heterozygous intercrosses were used for comparison. The CCR2 antagonist (RS504393: 2 mg/kg body weight; catalog no. 17330, Cayman Chemical) treatment was initiated a day after tumor cell inoculation and was administered intraperitoneally daily. Mice were euthanized before becoming moribund.

Cell treatments

OVCAR3 and OVCA420 human ovarian cancer cells (1 × 105) were cultured in 6-well plates (catalog no. 657160, Greiner Bio-One) as described in the “Cell culture and transfection” section above and treated with 20 ng/mL recombinant IFNγ (catalog no. 570202, BioLegend; ref. 28) or 0.1% BSA (catalog no. 7284, Sigma-Aldrich) as a control the following day. Cells were analyzed for the expression of B7-H3 (CD276), PD-L1 (CD274), and PD-L2 (PDCD1LG2) using qPCR and flow cytometry (as described below) after treatment for 24 and 48 hours, respectively.

For STAT3 inhibition, 1 × 105 cells were treated with the STAT3 inhibitor C188-9 (catalog no. S8605, Selleck Chemicals) for the mouse cell lines HM-1 and ID8 (29) or with Stattic (catalog no. ab120952, Abcam) for the human cell lines OVCAR3 and OVCA420 (30) at 1.25 and 5 μmol/L for 24 hours, respectively. Ccl2 and CCL2 expression levels were analyzed using qPCR as described below.

Tumor cell proliferation assay

HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells, or their respective control cells were seeded in 96-well plates at 1 × 103 cells/well (Greiner Bio-One) and incubated for 3 days in an atmosphere of 5% CO2 at 37°C. The number of viable cells in each well was examined every 24 hours by absorbance at 450 nm using Cell Count Reagent SF (catalog no. 07553-44, Nacalai Tesque) and iMark microplate reader (catalog no. 168-1130J, Bio-Rad). Absorbance values at 72 hours were compared using Microsoft Excel (RRID: SCR_016137).

Macrophage and monocyte selection

For tumor-derived macrophage selection, single cells from the tumors were prepared as described in the “Flow cytometry” section. F4/80-positive (F4/80+) macrophages were positively selected using anti-Mouse F4/80 PE antibody (clone: BM8, catalog no. 123109; RRID: AB_893498, BioLegend), EasySep Mouse PE positive selection kit II (catalog no. 17666, Stemcell Technologies), and EasySep Magnet (catalog no. 18000, Stemcell Technologies). The purity of F4/80+ cells was >90% as confirmed by flow cytometry. For mouse monocyte selection, single cells from the bone marrow of the femur and tibia of B6C3F1 or C57BL6 non–tumor-bearing mice were collected by mechanical dissociation. For human monocyte selection, peripheral blood mononuclear cells (PBMC) of healthy donors were separated using Leucosep (catalog no. 227290, Greiner) and Lymphosep (catalog no. 1692254, MP Biomedicals). Ly6C-positive mouse monocytes and CD14-positive human monocytes were positively selected by EasySep Magnet as described above using anti-Mouse Ly6C PE antibody (clone: HK1.4, catalog no. 128007; RRID: AB_1186133, BioLegend) and anti-Human CD14 PE antibody (clone: M5E2, catalog no. 301806; RRID: AB_314188, BioLegend). The purity of Ly6C-positive and CD14-positive cells was >90%, respectively, as confirmed by flow cytometry.

RNA extraction and reverse transcription qPCR

Human (OVCAR3 and OVCA420) and mouse (HM-1 and ID8) cells or macrophages isolated from the tumors of HM-1-B7-H3 KO or control cells (as described in the preceding section) were lysed in RLT buffer (catalog no. 1015750, Qiagen) containing 1% 2-mercaptoethanol (catalog no. 1610710, Bio-Rad). Intradermal tumors of HM-1-B7-H3 KO cells or control cells were treated in a bead homogenizer with RLT buffer and centrifuged at 16,000 ×g for 1 minute at 4°C. Total RNA was extracted from the lysates using RN easy Mini Kit (catalog no. 74104, Qiagen).

A ReverTra Ace qPCR RT Kit (catalog no. FSQ-101, TOYOBO) was used for cDNA synthesis. Gene amplification was performed using PowerUp SYBR Green Master Mix (catalog no. A25742, Thermo Fisher Scientific) on a StepOne Plus real-time PCR system (Applied Biosystems). Fifty nanograms of RNA was used as the template for PCR to detect the relative expression levels of CD276, CD274 (PD-L1), PTCD1LG2 (PD-L2), and CCL2 for human cells, and Cd276, Ccl2, Arg1, Il10, and Ifng for mouse cells; the primer sequences are listed in Supplementary Table S1. Relative expression levels were calculated using the 2−ΔΔCt method. GAPDH/Gapdh was used for normalization.

Immunostaining and evaluation

FFPE sections of the 62 HGSOC samples and 3 to 6 mouse tumors of HM-1-B7-H3 KO cells or control cells were immunostained as described previously (31). Epitope retrieval was performed at 120°C for 10 minutes using citrate buffer at pH 6.0 (B7-H3) or Tris-EDTA buffer at pH 9.0 (αSMA, CD206, CD8, and IFNγ). The kits, antibodies, and dilutions used are listed in Supplementary Table S2. Tumor cell B7-H3 expression was evaluated as described previously (32) and scored as follows: 0, negative or very weak; 1, weak; 2, moderate, and 3, strong (16). Scores of 0/1 and 2/3 were defined as B7-H3 low and B7-H3 high, respectively. Tumor-infiltrating CD206+ cells (200×) were counted from five fields, and the average number of cells was calculated. For the evaluation of CD8+ and IFNγ+CD8+ T cells, the number of cells in the representative CD8+ cell–rich field was counted at 200× magnification. The sensitivity and specificity of the IFNγ antibody was confirmed using appendix (catalog no. CS802695, OriGene) and tonsil (kindly provided by Kyoto University Department of Otolaryngology, Head and Neck Surgery, Kyoto, Japan) tissue as positive controls (Supplementary Fig. S1A and S1B). Slides were analyzed using a BZ 9000 fluorescence microscope (RRID: SCR_015486, Keyence).

Flow cytometry

IFNγ- or BSA-treated OVCAR3 and OVCA420 cells, HM-1-B7-H3 KO cells and ID8-B7-H3 KO cells or their respective control cells, and OVCAR3-sh-B7-H3 and OVCA420-sh-B7-H3 or their respective control cells were collected from the culture plates using trypsin (catalog no. T4549, Sigma-Aldrich) and incubated for 20 minutes at 4°C with the corresponding antibodies listed in Supplementary Table S2. Tumors of HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells, or their control cells were minced into thin pieces and dissociated in 100 μg/mL of DNase I (catalog no. 11284932001, Roche) and 1 mg/mL of collagenase IV (catalog no. CLS4, Worthington) in RPMI1640 medium. The tissues were incubated for 1 hour at 37°C with agitation. Single cells were prepared through a 70 μm nylon mesh strainer (catalog no. 542070, Greiner Bio-One). FcR blocking for leukocytes was performed with FcR Blocking Reagent (for mouse samples: catalog no. 130-092-575, RRID: AB_2892833; for human samples: catalog no. 130-059-901, RRID:AB_2892112, Miltenyi Biotec) for 10 minutes at 4°C. Referring to ref. 33, tumor-infiltrating immune cells were stained to detect CD4+ T cells (CD3+CD4+CD8−), CD8+ T cells (CD3+CD4−CD8+), M2 macrophages (CD45+F4/80+CD206+), M1 macrophages (CD45+F4/80+CD80+ or CD45+F4/80+CD86+), and dendritic cells (DC, CD45+CD11c+MHC class 2+) with the antibodies listed in Supplementary Table S2 for 30 minutes at 4°C. Gating strategies for each cell subset are shown in Supplementary Fig. S2. For IFNγ staining, single cells from the tumor were incubated for 4 hour with eBioscience Cell Stimulation Cocktail (catalog no. 00-4970-93, Invitrogen) and for 2 hours with Brefeldin A (catalog no. B7651, Sigma-Aldrich). BD Cytofix/Cytoperm Fixation/Permeabilization Kit (catalog no. 554714, BD Biosciences) was used for intracellular staining. Nonviable cells were stained with 7-amino-actinomycin D solution (catalog no. 51-68981E, BD Biosciences) or DAPI solution (catalog no. D1306, Life Technologies) and gated out. MACS Quant Analyzer 10 (Miltenyi Biotec) and FlowJo (RRID: SCR_008520, FlowJo LLC) were used for analysis.

T-cell proliferation assay

For T-cell selection, single cells from the spleens of B6C3F1 mice bearing tumors of HM-1-B7-H3 KO cells or control cells harvested at day 12 were collected by mechanical dissociation. Red blood cells were lysed with Pharm Lyse buffer (catalog no. 555899, BD Biosciences). T cells were negatively selected using mouse Pan-T-cell isolation kit II (catalog no. 130-095-130, Miltenyi Biotec), an LS column (catalog no. 130-042-401, Miltenyi Biotec), and QuadroMACS Separator (catalog no. 130-090-976, Miltenyi Biotec). The purity of CD3+ cells was >90% as confirmed by flow cytometry.

T cells were labeled using 10 mmol/L carboxyfluorescein succinimidyl ester (CFSE; catalog no. 600121, Cayman Chemical). T cells (1 × 105) were cocultured with F4/80+ macrophages (obtained as described in the “Macrophage and monocyte selection” section above) at 2:1, 4:1, and 8:1 ratios with RPMI1640 medium supplemented with 10% FBS, 2 mmol/L of l-glutamine (catalog no. 25030-081, Gibco), and 50 μmol/L of 2-mercaptoethanol for 72 hours in an atmosphere of 5% CO2 at 37°C. During incubation, T cells were activated with Dynabeads Mouse T-activator CD3/28 (1:100 dilution, catalog no. 11453D, Thermo Fisher Scientific). Activated T cells without macrophages were used as a positive control. After 72 hours, T-cell proliferation was examined using flow cytometry according to the decrease of the CFSE fluorescence intensity. MACS Quant Analyzer 10 (Miltenyi Biotec) and FlowJo (FlowJo) were used for analysis. The culture supernatants were collected and used for ELISA.

ELISAs

Mouse IFNγ levels in the cell culture supernatant of T cells cocultured with macrophages were measured using Mouse IFNγ Quantikine ELISA Kit (catalog no. MIF00, R&D Systems). HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells or their control cells, and OVCAR3-sh-B7-H3 and OVCA420-sh-B7-H3 or their control cells were seeded at a density of 1 × 105 with 500 μL of the corresponding medium (see “Cell culture and transfection” section) in 24-well plates (catalog no. 662160, Greiner Bio-One). Cell culture supernatants were collected after 24 hours with a 0.45 μm Millex filter (catalog no. SLHVR33RS, Merck Millipore) and used for ELISA. These supernatants were also used for migration and monocyte differentiation assays (described in detail below) as the tumor cell–conditioned medium (TCM). Human and mouse tumor tissues were treated in a bead homogenizer, and sonicated in 1× RIPA buffer (catalog no. 89900, Thermo Fisher Scientific) containing a protease inhibitor cocktail (catalog no. 03969-21, Nacalai Tesque) and a phosphatase inhibitor cocktail (catalog no. 07575-51, Nacalai Tesque), followed by centrifugation at 16,000 ×g for 5 minutes at 4°C. Human tumor lysates were used for B7-H3 and CCL2 detection at 75 and 200 μg as protein amount, respectively, and 20 μg as protein amount of mouse tumor lysates were used for CCL2 detection. Mouse CCL2 protein levels in culture supernatants and tumor lysates were measured using the Mouse CCL2/JE/MCP1 DuoSet kit (catalog no. DY479-05, R&D Systems). Human CCL2 protein levels in culture supernatants, tumor lysates, and serum samples were measured using Human CCL2/MCP-1 Quantikine ELISA Kit (catalog no. DCP00, R&D Systems). Human B7-H3 protein levels in tumor lysates were measured using Human B7-H3 Quantikine ELISA Kit (catalog no. DB7H30, R&D Systems). The absorbance readings at 570 nm were subtracted from the readings taken at 450 nm using an iMark microplate reader (Bio-Rad) to establish standard curves for calculation of the concentration. Microsoft Excel was used for analysis.

Immunoblotting

Nuclear proteins were collected from HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells or their control cells, and OVCAR3-sh-B7-H3 and OVCA420-sh-B7-H3 or their control cells using NE-PER nuclear and cytoplasmic extraction reagents (catalog no. 78833, Thermo Fisher Scientific). Protein (20 μg) of HM-1-B7-H3 KO cells or their control cells, and 10 μg protein of the other cells were loaded onto 8% acrylamide gels. Proteins were subsequently separated using SDS-PAGE and transferred onto polyvinylidene difluoride membranes (catalog no. 1620177, Bio-Rad). After blocking for 1 hour with Blocking One-P (catalog no. 05999-84, Nacalai Tesque), the membranes were then immunoblotted with the antibodies listed in Supplementary Table S2 at the indicated dilutions. The membranes were incubated with the primary antibody overnight at 4°C and with the secondary antibody for 1 hour at room temperature. Bands were visualized using ChemiDoc XRS+ Systems (Bio-Rad). Signals were quantified using Image Lab 2.0 (RRID: SCR_014210, Bio-Rad).

RNA sequencing

Total RNA from HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells or their control cells, or intradermal tumors of HM-1-B7-H3 KO cells or their control cells at day 10 of tumor growth were extracted as described in the “RNA extraction and reverse transcription qPCR” section and used for RNA sequencing. Sequencing was performed on an Illumina NovaSeq 6000 platform (RRID: SCR_016387, Illumina) to generate 100 bp paired-end reads. Raw reads were trimmed using Trim Galore (RRID: SCR_011847), and the resulting reads were aligned to the mouse reference genome GRCm38/mm10 using STAR (RRID: SCR_004463). Differential expression analysis was performed using DESeq2 (RRID: SCR_015687) between control (n = 3) and B7-H3 KO (n = 3) groups of the ID8 and HM-1 cell lines, and HM-1 intradermal tumors (processed as indicated in above sections). After excluding the genes showing low expression and low fold change between groups (“baseMean” > 1000; |log2FoldChange| > 2; Padj < 0.01), the remaining genes were considered as DEGs and listed in Supplementary Tables S3–S5. RNA-sequencing data have been deposited in the Gene Expression Omnibus database, under accession number GSE174137.

In vitro M2 macrophage generation

Mouse and human monocytes (obtained as described in the “Macrophage and monocyte selection” section) were cultured in 6-well plates (catalog no. 657160, Greiner Bio-One) at 3 × 106 cells/well and 6 × 105 cells/well, respectively, with RPMI1640 medium supplemented with 40 ng/mL of MCSF (mouse: catalog no. 576402, human: catalog no. 574802, BioLegend) and 40 ng/mL of IL6 (mouse: catalog no. 216-16, PeproTech, human: catalog no. 570802, BioLegend) for 6 to 7 days in an atmosphere of 5% CO2 at 37°C and used for migration assays. After incubation, >70% of the cultured mouse cells were F4/80+ and CD206+, and >85% of the cultured human cells were CD206+ as confirmed by flow cytometry.

Migration assays

Monocytes or the generated M2 macrophages (mouse: 1 × 105 cells, human: 5 × 104 cells, isolated as described in the “Macrophage and monocyte selection” and “In vitro M2 macrophage generation” sections, and pretreated with 2 μmol/L RS504393 as described in “Tumor models”) or DMSO was plated in the upper compartment of 8 μm cell culture insert (catalog no. 353097, Corning), and 500 μL of TCM of HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells or their control cells, and OVCA420-sh-B7-H3 or their control cells (generated as described in the “ELISAs” section) was added to the lower compartment. Plates were incubated for 1 hour in an atmosphere of 5% CO2 at 37°C, and migrated cells were counted using CountBright Absolute Counting beads (catalog no. 36950, Thermo Fisher Scientific).

Monocyte differentiation assay

Mouse bone marrow monocytes (isolated and cultured as described in the “Macrophage and monocyte selection” section) were plated in the upper compartment of 0.4 μm cell culture inserts (catalog no. 353095, Corning) at a density of at 1 × 105, and 500 μL of TCM of HM-1-B7-H3 KO cells, ID8-B7-H3 KO cells or their control cells (generated as described in the “ELISAs” section) supplemented with 20 μg/mL anti-mouse CCL2 (catalog no. AB-479-NA, RRID:AB_354366; R&D Systems) or goat IgG control (catalog no. AB-108-C, RRID:AB_354267; R&D Systems) was added to the lower compartment. Plates were incubated for 72 hours in an atmosphere of 5% CO2 at 37°C, and F4/80+CD206+ cells were then evaluated using flow cytometry, as described above.

Statistical analysis

At least three in vitro experiments and at least two in vivo experiments were independently performed. Data from one representative experiment are presented. For mouse experiments, cages of mice were randomly allocated to the experimental groups. Sample sizes were chosen to assure reproducibility of the experiments in accordance with the replacement, reduction, and refinement principles of the animal ethics regulation. Grouped data are shown as the average ± SEM in all figures except non-normally distributed data for clinical samples, which shows the median with 95% confidence intervals. The investigator was not blinded to the group allocation. A priori power analysis was not performed for human samples. Survival curves were constructed using the Kaplan–Meier method. Prognostic factors for PFS and overall survival (OS) of patients with HGSOC were assessed by univariate and multivariate analysis using the Cox proportional hazards regression model. Statistical analysis was performed using R software and GraphPad Prism 7 (RRID: SCR_002798, GraphPad Software), and the appropriate tests are indicated in the figure legends. Significance was set as *, P < 0.05; **, P <0.01; and ***, P <0.001.

ResultsB7-H3 expression is upregulated in the nonimmunoreactive TME of HGSOC

We first identified the genes acting antagonistic to the immunoreactive TME in HGSOC based on our RNA expression array data. CD276 (B7-H3) was the most upregulated gene in the nonimmunoreactive subtype (Fig. 1A). Analysis of TCGA-OV revealed that B7-H3 expression was higher in the proliferative and mesenchymal subtypes than in the immunoreactive subtypes (P = 0.004 and P < 0.001; Fig. 1B). Among the T cell–suppressive immune checkpoint molecules of the B7 family, only B7-H3 expression was lower in the immunoreactive subtype and higher in the nonimmunoreactive subtype (both P < 0.001; Fig. 1C).

Figure 1.

CD276 (B7-H3) expression is upregulated in the nonimmunoreactive TME of HGSOC. A, Microarray analysis of HGSOC clinical samples from Kyoto University (n = 30). The immunoreactive (n = 8) and nonimmunoreactive (n = 22) subtypes were compared using the Samroc method. The magenta dot represents CD276 (B7-H3). B, CD276 expression among the molecular subtypes of 263 HGSOC samples from TCGA-OV. Data are presented as the mean ± SEM; **, P < 0.01; ***, P < 0.001; N.S., not significant by one-way ANOVA with Tukey multiple comparison test. C, Gene expression comparison of T cell–suppressive B7 family molecules in the immunoreactive and nonimmunoreactive subtypes of TCGA-OV. Data are presented as the mean ± SEM after Z-score normalization. ***, P < 0.001 by one-way ANOVA. D, Pearson correlation analysis of the IFNγ signature activity scores and expression of T cell–suppressive B7 family molecules in TCGA-OV. Values after Z-score normalization are plotted. *, P < 0.05; ***, P < 0.001. (Green and blue dots represent the immunoreactive and nonimmunoreactive subtypes, respectively, in B–D.

A clear positive correlation was observed between the IFNγ signature, which is a major characteristic of the immunoreactive tumors and can be used as an index for predicting the therapeutic efficacy of immune checkpoint inhibitor (ICI) treatment (27), and the other B7 family molecules. By contrast, the IFNγ signature negatively correlated with B7-H3 expression (P = 0.020; Fig. 1D). Based on the response to IFNγ, we performed qPCR or flow cytometry and observed significantly increased expression of PD-L1 and PD-L2 and unchanged or slightly decreased expression of B7-H3 in HGSOC cell lines (Supplementary Fig. S3A and S3B). Moreover, PD-L1 expression negatively correlated with B7-H3 expression (Supplementary Fig. S3C). These results indicate that B7-H3 expression is associated with the nonimmunoreactive TME and that B7-H3 is involved in immunosuppression, albeit via a different mechanism than that of PD-L1.

B7-H3 mediates murine ovarian cancer growth via immune system regulation

To investigate the function of B7-H3 in tumor cells, we generated three B7-H3 KO HM-1 and ID8 mouse ovarian cancer cell lines using CRISPR/Cas9. Both cell lines express B7-H3 and are resistant to PD-1/PD-L1 inhibition therapy (34, 35). We confirmed B7-H3 KO by flow cytometry and immunostaining (Fig. 2A and B; Supplementary Fig. S4A). B7-H3 KO did not affect cell proliferation in vitro (Supplementary Fig. S4B). In immunocompetent mice, B7-H3 KO delayed the growth of HM-1 intradermal and intraperitoneal tumors, as well as ID8 intraperitoneal tumors (Fig. 2C and D; Supplementary Fig. S4C). B7-H3 suppression significantly prolonged the survival of mice with ID8 tumors, a trend that was also observed with HM-1 tumors (Supplementary Fig. S5A). In contrast, in immunodeficient mice, tumor growth and survival were not affected by B7-H3 expression in either HM-1 or ID8 cells (Fig. 2E and F; Supplementary Fig. S5B). These data suggest that B7-H3 is involved in tumor development via immune-related mechanisms. Next, to verify the role of B7-H3 in stromal cells, we injected ID8 cells into Cd276 KO (B7-H3 KO) mice and observed that B7-H3 expression did not significantly affect tumor development (Supplementary Fig. S6). Overall, these results indicate that B7-H3 expression in tumor cells is important for tumor progression.

Figure 2.

B7-H3 contributes to tumor progression via an immune-related mechanism in murine ovarian cancer models. A, Flow cytometry of B7-H3 KO HM-1 and ID8 and their respective controls. Blue, pink, and gray histograms represent B7-H3 KO cells, B7-H3 control cells, and the isotype control, respectively. B, Representative B7-H3 IHC (top) and immunofluorescence (bottom) images of controls and HM-1 B7-H3 KO tumors for the indicated markers. αSMA-positive fibroblasts represented by black and white arrowheads. Scale bar, 50 μm. Positive cells were stained brown in IHC, and red (αSMA), green (B7-H3), and blue (DAPI) in immunofluorescence images. C, Intradermal tumor volume (left, n = 6) and omental tumor weight at day 10 (right, n = 6) of immunocompetent mice injected with HM-1 B7-H3 KO or control cells. D, Changes in body weight (top, n = 5–6) and omental tumor weight at day 52 (bottom, n = 6) of immunocompetent mice intraperitoneally injected with ID8 B7-H3 KO or control cells. E, Intradermal tumor volume (left, n = 5) and omental tumor weight at day 10 (right, n = 6) of immunodeficient mice injected with HM-1 B7-H3 KO or control cells. F, Changes in body weight (top, n = 6) and omental tumor weight at day 52 (bottom, n = 6) of immunodeficient mice intraperitoneally injected with ID8 B7-H3 KO or control cells. C–F, Data are presented as the mean ± SEM; ***, P < 0.001; N.S., not significant, unpaired t test.

B7-H3 suppression reduces M2 macrophages and increases T cell–produced IFNγ in the TME

First, we confirmed a direct T cell–suppressive effect of tumor cell B7-H3 in vitro (Supplementary Fig. S7A–S7C), which has been reported in several studies (19, 20). Next, we evaluated tumor-infiltrating immune cells. Although there was no difference in the number of CD8+ T cells, CD206+ M2 macrophages decreased in HM-1 B7-H3 KO tumors (Fig. 3A; Supplementary Fig. S8A). Flow cytometry showed that F4/80+CD206+ M2 macrophages decreased in HM-1 B7-H3 KO tumors, whereas M1 macrophages, CD8+ T cells, CD4+ T cells, and DCs were unchanged (Fig. 3B). ID8 tumors similarly showed a decrease in M2 macrophages but no change in M1 macrophages (Supplementary Fig. S8B). These results suggest that B7-H3 KO tumors are not only associated with a decreased number of M2 macrophages within tumor tissues, but also alters macrophage polarization (i.e., a decrease in the immunosuppressive fraction).

Figure 3.

B7-H3 suppression in tumor cells decreases the number of intratumoral M2 macrophages and increases IFNγ production of CD8+ T cells. A, IHC of CD206 in the B7-H3 KO intradermal HM-1 tumors and controls (n = 3). CD206+ cells were stained brown. Scale bar, 50 μm. B, Flow cytometry of immune cells from HM-1 tumors. Density plots (top left) show the M2 macrophages at day 12. Each dot represents live CD45+ cells, and the boxed area represents F4/80+CD206+ M2 macrophages. The percentage of positive cells relative to the total cells is plotted (n = 6). C, qPCR for the indicated genes in F4/80+ macrophages (Mϕ) isolated from the HM-1 intradermal tumors (n = 5). D, T-cell proliferation in presence of HM-1 tumor–derived macrophages. Top histograms show the percentage of proliferating T cells cocultured (4:1) with macrophages from control or B7-H3 KO tumors. Bottom bar graph shows the percentage of proliferating T cells cocultured with various ratios of macrophages (n = 5). E, IFNγ levels in the supernatants of cocultures in D (n = 5). F, Flow cytometry of IFNγ+CD8+ T cells from HM-1 tumors. Each dot represents live CD45+CD3+ cells, and the boxed area represents IFNγ+CD8+ T cells. The percentage of IFNγ+CD8+ T cells relative to total CD8+ T cells is plotted (n = 6). A–F, Data are presented as the mean ± SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001; N.S., not significant, unpaired t test.

In macrophages from the B7-H3 KO tumors, we observed decreased expression of the M2 markers Arg1 and Il10 but unchanged expression of the M1 marker Ifng (Fig. 3C). T-cell proliferation assays with CD3/CD28-activated T cells cocultured with tumor-derived macrophages showed that macrophages from the B7-H3 KO tumors had reduced ability to inhibit T-cell proliferation compared with those from control tumors (Fig. 3D), and IFNγ production in coculture supernatants was increased (Fig. 3E). We also observed an increased number of IFNγ+CD8+ T cells in B7-H3 KO tumors (Fig. 3F). These results suggest that B7-H3 suppression in tumor cells reduces immunosuppression in the TME by not only attenuating direct T-cell inhibition, but also reducing the number of M2 macrophages, which may contribute to the increased IFNγ production by CD8+ T cells and enhanced antitumor immunity.

B7-H3 suppression downregulates CCL2 production in tumor cells, partly via the STAT3 pathway

Next, we elucidated the mechanism by which B7-H3 suppression in tumor cells reduced M2 macrophage infiltration. Using RNA sequencing, we found 90, 131, and 67 DEGs in the HM-1 and ID8 cell lines and HM-1 intradermal tumors relative to controls, respectively (Supplementary Tables S3–S5). We identified Ccl2 as a common downregulated gene in the KO groups (Fig. 4A). We also confirmed that changes in expression of several chemokines related to T-cell trafficking were not consistent between the cell lines and the B7-H3 KO tumors (Supplementary Fig. S9). A reduction in CCL2 protein in the B7-H3 KO groups was confirmed in both culture supernatants and tumor lysates from HM-1 and ID8 cells (Fig. 4B and C). Then, we investigated whether B7-H3 suppression also altered CCL2 in human HGSOC cell lines (Supplementary Fig. S10) and confirmed a decrease in CCL2 in the culture supernatants of the shB7-H3 cells (Fig. 4D).

Figure 4.

B7-H3 suppression downregulates CCL2 production, in part, via the STAT3 pathway in ovarian cancer cells. A, Differential gene expression based on RNA sequencing (RNA-seq) between controls (n = 3) and B7-H3 KO (n = 3) groups in HM-1 and ID8 cells and HM-1 intradermal tumors. CCL2 protein in the culture supernatants (B) and tumors derived from HM-1 B7-H3 KO (top), ID8 B7-H3 KO (bottom), and their respective controls (n = 5; C). D, CCL2 protein in the culture supernatants of the OVCAR3 shB7-H3 (top), OVCA420 shB7-H3 (bottom), and their respective controls (n = 5). Representative Western blot analysis of nuclear phosphorylated STAT3 (p-STAT3) in HM-1 and ID8 (E) and OVCAR3 and OVCA420 (F) cells. Bottom bar graphs show the signal intensity of p-STAT3 relative to the TBP control in HM-1 and ID8 cells (both n = 10) and OVCAR3 and OVCA420 cells (both n = 7). qPCR for expression of Ccl2 in HM-1 and ID8 cells (G) and CCL2 in OVCAR3 and OVCA420 cells (H) treated with 1.25 and 5 μmol/L of the STAT3 inhibitors C188-9 (G) and Stattic (H), respectively. DMSO was used as the control (n = 3). Data are presented as the mean ± SEM; *, P < 0.05; **, P < 0.01; ***, P < 0.001; N.S., not significant, unpaired t test in B, C, D (OVCA420), E, and F and one-way ANOVA with Tukey multiple comparisons test in D (OVCAR3), G, and H.

To investigate the mechanism by which B7-H3 suppression induced CCL2 downregulation, we examined STAT3 (21–23) and NFκB (36) pathways, which have been reported as downstream pathways of B7-H3. Western blots showed that phosphorylated STAT3 was decreased in the B7-H3–suppressed cells (Fig. 4E and F). In contrast, phosphorylated p65, a major factor in the classical NFκB pathway, was decreased in ID8 and OVCAR3 cell lines; thus, consistent results were not observed among cell lines (S