B7-H3-mediated reversal of CAR-T cell exhaustion induces a notable antitumour response in ovarian cancer models

Abstract

Background: Functional CAR-T cell exhaustion in the immunosuppressive tumour microenvironment remains the main barrier to the success of CAR-T cell therapy for treating solid tumours. Mesothelin (MSLN) has emerged as an attractive target for CAR-T cell therapy for several solid malignancies, including ovarian cancer. In this study, we aimed to investigate the role and mechanism of lipid metabolites in anti-MSLN CAR-T cell exhaustion in ovarian cancer cells.

Methods: We engineered anti-MSLN CAR-T cells targeting ovarian cancer cells with high MSLN expression as a pivotal tool for in vitro and in vivo experiments. Moreover, liquid chromatography-tandem mass spectrometry (LC-MS/MS) revealed the critical role of oxylipin 12-HETE in the exhaustion of CAR-T cells. By employing structure-based high-throughput virtual screening (HTVS), we identified the inhibitor targeting B7-H3.

Findings: We demonstrated that GPR31-dependent 12-HETE accumulation in the ovarian cancer microenvironment drives CAR-T cell exhaustion via lipid peroxidation, impairing their antitumour efficacy. Genetic or pharmacological inhibition of the 12-HETE/GPR31 axis restored CAR-T cell cytotoxicity and proliferation, leading to significant tumour regression in murine models. Silencing B7-H3 relieved repression of FOXO3, leading to reduced 12-LOX expression and lower 12-HETE levels, which places B7-H3 upstream of this metabolic checkpoint. Through structure-based screening, we identified HI-TOPK-032 as a potent B7-H3 inhibitor that synergised with CAR-T cell therapy by reversing exhaustion markers (e.g., PD-1, TIM-3) and enhancing cytokine polyfunctionality. Combined HI-TOPK-032 and anti-PD-1 treatment achieved superior tumour control compared to monotherapies, particularly in B7-H3/12-LOX-high patient-derived xenografts, underscoring its precision therapeutic potential.

Interpretation: CAR-T cell therapy combined with HI-TOPK-032 is a promising novel strategy for treating MSLN-expressing solid tumours.

Funding: This study was funded by the National Natural Science Foundation of China (Grant number: 82503173), Beijing Hospitals Authority's Ascent Plan (Grant number: DFL20221201), Beijing Hospitals Authority Clinical Medicine Development of Special Funding Support (Grant number: ZYLX202120), Beijing Natural Science Foundation (Grant number: 7162063), Capital Medical University Laboratory for Clinical Medicine and Gynecological Tumour Precise Diagnosis and Treatment Innovation Studio.

Keywords: 12-HETE; Anti-PD-1; B7-H3; CAR-T cells; CD8 T cell; HI-TOPK-032; Ovarian cancer.

Fig. 1

TME mimicking nutrient availability impairs CD8⁺ T cell cytotoxicity. (A) Schematic of the overall experimental workflow, including generation of MSLN-directed CAR-T cells from human PBMC-derived CD8+ T cells via lentiviral transduction, establishment of MSLN- and luciferase-expressing ovarian cancer cell lines (SKOV3-ML and OVCAR3-ML), and the in vivo NSG mouse xenograft model treated with CAR-T cells. (B–C) Tumour-bearing NSG mice (n = 6 per group) were intravenously injected with MSLN-CAR-T cells (5 × 10⁶ cells in 200 μL per mouse) weekly starting from day 0 (tumour implantation). Tumour growth was assessed by bioluminescence imaging at the endpoint (day 21). Human CD8+ T cells isolated from peripheral blood (PBMC) and tumour infiltrating lymphocyte (TIL) were analysed using flow cytometry for proliferation (Ki-67⁺), cytokine production (IFN-γ, TNF-α), and exhaustion marker (PD-1) expression. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 compared to PBMC group; two-tailed Student's t-test. (D) Schematic of in vitro culture conditions: CAR-T cells were cultured alone in X-Vivo 15 medium or tumour cell-conditioned medium (CM), or in coculture with tumour cells. (E) IL-2 secretion (ELISA, left) and intracellular IFN-γ/TNF-α expression (flow cytometry, right) in CAR-T cells following monoculture in indicated media (n = 3–4). ∗∗∗P < 0.001 compared to X-Vivo15 group; two-tailed Student's t-test. (F) Proliferation (BrdU incorporation) of CAR-T cells cocultured with OVCAR3-ML or SKOV3-ML tumour cells. ∗∗P < 0.01, ∗∗∗P < 0.001 compared to X-Vivo15 coculture group; two-tailed Student's t-test. (G) IL-2 (ELISA) and IFN-γ/TNF-α (flow cytometry) levels in CAR-T cells cocultured with OVCAR3-ML cells. ∗P < 0.05, ∗∗P < 0.01 compared to X-Vivo15 group; two-tailed Student's t-test. (H) Proliferation (Ki-67) and apoptosis (Annexin V) of CAR-T cells cocultured with MSLN⁺ tumour cells (OVCAR3-ML, SKOV3-ML) or MSLN⁻ control cells (SK-HEP-1). ∗P < 0.05, ∗∗P < 0.01; ns, not significant; two-tailed Student's t-test. (I) Cytokine production (IL-2, IFN-γ, TNF-α) in CAR-T cells cocultured with SK-HEP-1 cells under different media conditions. ∗P < 0.05, ∗∗P < 0.01 compared to X-Vivo15 group; two-tailed Student's t-test. Data are presented as mean ± SEM from triplicate wells and are representative of at least three independent experiments.

Fig. 2

12-HETE is secreted by tumour cells and accumulates in ovarian cancer. (A) Immunohistochemical staining of oxidised phospholipids (OxPLs, detected by E06 antibody) in human high-grade serous ovarian cancer (HGSOC) compared with normal ovarian tissues (scale bar: 250 μm). (B) OxPL staining in SKOV3/OVCAR3 xenografts (NOD/SCID mice, day 28) vs. spleen and intestine tissues (scale bar: 250 μm; n = 5 mice/group). (C) Lipid peroxidation (BODIPY™ C11 assay) in CD8⁺ tumour-infiltrating lymphocytes (TILs) vs. splenic CD8⁺ T cells from ID8 tumour-bearing C57BL/6 mice (day 28; n = 6). ∗∗∗P < 0.001, Student's t-test. (D) Differential oxylipin profiles between human HGSOC (n = 3) and normal ovarian tissues (n = 3) by LC-MS/MS. Left: heatmap; right: volcano plot. (E) 12-LOX protein expression in HGSOC and normal tissues by Western blot (normalised to GAPDH). Representative image. (F) 12-LOX protein levels in ovarian cancer cell lines (SKOV3, OVCAR3, A2780) vs. normal IOSE80 cells by Western blot (normalised to GAPDH). Representative image. (G) Schematic of plasma and tumour interstitial fluid (TIF) isolation from OVCAR3-ML cell-derived xenograft (CDX) model (day 18 post-implantation). (H) Extracellular 12-HETE levels in X-Vivo 15 vs. conditioned medium (CM, 24-h culture). ∗∗∗P < 0.001, Student's t-test. (I) 12-HETE concentrations in plasma vs. TIF from OVCAR3-ML or SKOV3-ML CDX models. ∗∗∗P < 0.001, Student's t-test. Data are presented as means ± SEM. Each experiment was repeated at least twice independently.

Fig. 3

12-HETE promotes CAR-T cell exhaustion. (A–B) IFN-γ and TNF-α production (flow cytometry) in MSLN–CAR-T cells treated with 0, 0.5, 1, or 1.5 μM 12-HETE. (C) Secreted IL-2 (ELISA) from MSLN–CAR-T cells treated with the indicated doses of 12-HETE. (D) GzmB and apoptosis (Annexin V) in CD8+ TILs cocultured with OVCAR3-ML pretreated with 12-LOX inhibitor CAY10698 or DMSO (left). GzmB in CAR-T cells cocultured with 12-LOX-overexpressing OVCAR3-ML (right). Data are means ± SEM of three independent experiments. P-values were calculated by one-way ANOVA (A–C, D right) or Student's t-test (D left). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; ns, not significant.

Fig. 4

GPR31⁺ TILs is in a more functional state of exhaustion. (A) Expression of GPR31 and TOX in human ovarian cancer. PD-1⁺ CD8⁺ TILs were analysed by flow cytometry. Left: Expression of GPR31 in PD-1⁺TOX⁺ vs. PD-1⁺TOX⁻ CD8⁺ TILs from individual tumours (n = 6 patients). Right: Proportion of GPR31⁺ cells in each subset. ∗∗∗P < 0.001 by paired Student's t-test. (B) C57BL/6 mice were implanted with ID8 ovarian cancer cells and tumours were harvested on day 14 or 28 post-implantation. Expression of GPR31 and PD-1 in CD8⁺ TILs was measured by flow cytometry. ∗∗P < 0.01, ∗∗∗P < 0.001; unpaired Student's t-test.

Fig. 5

Elimination of GPR31 alleviates the dysfunction of T cells. (A) Tumour growth in GPR31 knockout (GPR31−/−) vs. wild-type (GPR31WT) female mice 28 days after ID8-luc tumour implantation (n = 6 mice/group). ∗∗∗P < 0.001; unpaired Student's t-test. (B) Flow cytometry analysis of IFN-γ and granzyme B (GzmB) expression in CD8⁺ T cells isolated from tumours of GPR31WT and GPR31−/− mice on day 28 post-implantation. ∗P < 0.05, ∗∗P < 0.01; unpaired Student's t-test. (C) Tumour growth in GPR31−/− mice treated with either IgG control or α-CD8 depleting antibody. ∗∗P < 0.01, ∗∗∗P < 0.001; two-way ANOVA with Sidak's post-test. (D) CFSE-labelled GPR31WT or GPR31−/− CD8⁺ TILs were activated in vitro for 48 h and then treated with vehicle or 1 μM 12-HETE for 24 h. Cells were restimulated with PMA/ionomycin for 4 h, and TNF-α and IFN-γ were measured by flow cytometry. ∗P < 0.05, ∗∗P < 0.01; two-way ANOVA. (E) GzmB expression in CAR-T cells cocultured with OVCAR3-ML cells transfected with control or 12-LOX overexpression plasmid, with or without GPR31 knockdown (si-GPR31 vs. si-NC). E:T = 2:1, 24 h coculture. ∗∗P < 0.01; two-way ANOVA. (F) IL-2 secretion (ELISA) in supernatant from CAR-T cells treated as in (E). ∗∗P < 0.01, ∗∗∗P < 0.001; two-way ANOVA. (G) Proliferation (Ki-67) and apoptosis (Annexin V) in CAR-T cells under the same conditions as (E). ∗P < 0.05, ∗∗P < 0.01; ns, not significant; two-way ANOVA.

Fig. 6

12-HETE induces lipid peroxidation in CAR-T cells in a GPR31-dependent manner. (A) Lipid peroxidation levels in activated CAR-T cells treated with ethanol (vehicle) or increasing concentrations of 12-HETE (0.5–1.5 μM) for 24 h, measured by BODIPY™ 581/591 C11 fluorescence. ∗∗P < 0.01, ∗∗∗P < 0.001 vs. vehicle; one-way ANOVA with Dunnett's post-test. (B) Lipid peroxidation in activated human CAR-T cells treated with 12-HETE (0, 1, or 1.5 μM) assessed via MDA level and GSH/GSSG ratio. ∗P < 0.05, ∗∗P < 0.01; one-way ANOVA. (C) MDA content and GSH/GSSG ratio in si-NC- or si-GPR31-treated CAR-T cells, with or without 1 μM 12-HETE supplementation. ∗P < 0.05, ∗∗P < 0.01; two-way ANOVA with Sidak's test. (D) Lipid peroxidation (BODIPY™ 581/591C11 fluorescence) in CAR-T cells treated with DMSO (Ctrl), 12-HETE (1.5 μM), vitamin E (α-tocopherol, Toco; 200 μM), or 12-HETE + Toco. ∗∗∗P < 0.001; ns, not significant; one-way ANOVA. (E) TNF-α and IFN-γ production (flow cytometry) in CAR-T cells under the same treatments as (D) after PMA/ionomycin restimulation. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001; one-way ANOVA. Data are presented as mean ± SEM; n = 3–8 per group from 2 to 3 independent experiments.

Fig. 7

B7-H3 inhibits 12-LOX expression through the transcriptional regulator FOXO3 (A–B) 12-LOX mRNA and protein expression levels, FOXO1, FOXO4, and FOXO3 protein expression in B7-H3 non-targeting control (B7-H3NC) and B7-H3-knockdown (B7-H3KO-sh526/528) SKOV3 and OVCAR3 cells, as measured by qRT–PCR and Western blot. ∗P < 0.05, ∗∗P < 0.01; two-way ANOVA. (C–D) 12-LOX mRNA and protein expression, FOXO1, FOXO4, and FOXO3 protein expression in B7-H3 non-targeting control (B7-H3NC) and B7-H3-overexpressing (B7-H3OE) SKOV3 and OVCAR3 cells. ∗P < 0.05, ∗∗P < 0.01; two-way ANOVA. (E) Representative immunofluorescence images showing subcellular localisation of B7-H3 (green) in ovarian cancer tissues, captured by STED super-resolution microscopy. Nuclei are stained with DAPI (blue). Scale bar: 10 μm. (F) KEGG pathway analysis of upregulated differentially expressed transcripts (DETs) in B7-H3KO-sh528 OVCAR3 cells compared to B7-H3NC OVCAR3 controls. (G) Immunohistochemical staining of FOXO3 in tumour tissues from B7-H3-knockdown or non-target control (NC) NOD/SCID mouse xenograft models. Scale bar: 50 μm. (H) 12-LOX mRNA levels in SKOV3 and OVCAR3 cells expressing constitutively active FOXO3 (FOXO3-activated) or DNA-binding domain truncated FOXO3 mutant (FOXO3-Mut). ∗∗P < 0.01; unpaired Student's t-test. (I) Protein levels of p-AKT, AKT, FOXO3, and 12-LOX in SKOV3 and OVCAR3 cells transfected with si-AKT or non-targeting control (NC) siRNA.

Fig. 8

HI-TOPK-032 inhibits B7-H3 expression and promotes antitumour effect of MSLN-CAR-T cells in vitro. (A) Workflow of high-throughput virtual screening (HTVS) for identifying B7-H3 inhibitors. (B) Western blot and flow cytometry analysis of B7-H3 expression in SKOV3 and OVCAR3 cells after 48 h treatment with HI-TOPK-032. ∗P < 0.05, ∗∗P < 0.01 vs. DMSO control; one-way ANOVA. (C) Chemical structure of purpureaside C (HI-TOPK-032) and predicted binding mode within the human B7-H3 extracellular IgV domain (in silico docking). Hydrogen bonds are indicated with dashed lines. (D) ELISA measurement of 12-HETE in supernatant and qRT–PCR analysis of 12-LOX mRNA in SKOV3 and OVCAR3 cells treated with HI-TOPK-032 (0, 0.5, 1 μM) for 48 h. ∗P < 0.05, ∗∗P < 0.01; one-way ANOVA. (E) Schematic of repeated antigen stimulation model for inducing CAR-T cell exhaustion. OVCAR3-ML cells were pretreated with DMSO, HI-TOPK-032 (0.5/1 μM), or anti-B7-H3 antibody (5 μg/ml) and cocultured with CAR-T cells (E:T = 3:1) for three rounds (48 h per round). Exhaustion markers PD-1 and Tim-3 on CD8⁺ CAR-T cells were assessed by flow cytometry after each stimulation (n = 3). ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001 vs. DMSO group; two-way ANOVA.

Fig. 9

HI-TOPK-032 promotes the antitumour effect of CAR-T cells in vivo. (A) Schematic diagram of adoptive transfer of HI-TOPK-032 at different doses (DMSO as control, every 2 days)- or anti-B7-H3 antibodies (IgG as control, three times a week) with/without treated MSLN-CAR-T cells into systemic OVCAR3-ML tumour-bearing NSG mice. (B–D). Tumour growth was as sessed using bioluminescence imaging (B–C; n = 6 per group), and survival curves (D; n = 5 per group) were monitored. (E–F) The percentage and number of CD8⁺ T cells in tumours were assessed using flow cytometry; and percentages of CD8⁺ T, Ki67⁺, PD-1⁺, IFN-γ⁺ and TNF-α⁺ in tumour-infiltrating CD8⁺ T cells were estimated by flow cytometry intracellular staining (n = 3). All results are representative of three independent experiments. Data are presented as means ± SEM and were analysed by unpaired Student's t test (A to F, H to I, B and K) and log-rank test (J). ∗P < 0.05, ∗∗P < 0.01 and ∗∗∗P < 0.001; ns, P > 0.05.

Fig. 10

HI-TOPK-032 potentiates anti-PD-1 therapy in vivo. Adoptive transfer of HI-TOPK-032 (DMSO as control, every 2 days)–and anti–PD-1 antibodies (IgG as control, three times a week) either alone or together with treated MSLN-CAR-T cells into systemic OVCAR3-ML tumour-bearing NSG mice. (A) Representative images (B) of bioluminescence and survival curve (C) were monitored (n = 6 per group). The percentage and number of CD8⁺ T cells in tumours were assessed using flow cytometry (D); percentages of Ki67⁺, IFN-γ⁺ and TNF-α⁺ tumor-infiltrating CD8⁺ T cells were estimated by flow cytometric analysis (n = 3). All results are representative of three independent experiments. Data are presented as means ± SEM and were analysed by unpaired Student's t test (B and D) and log-rank test (C). ∗P < 0.05, ∗∗P < 0.01 and ∗∗∗P < 0.001; ns, P > 0.05.