Intratumoral IL-12 delivery empowers CAR-T cell immunotherapy in a pre-clinical model of glioblastoma

Abstract

Glioblastoma multiforme (GBM) is the most common and aggressive form of primary brain cancer, for which effective therapies are urgently needed. Chimeric antigen receptor (CAR)-based immunotherapy represents a promising therapeutic approach, but it is often impeded by highly immunosuppressive tumor microenvironments (TME). Here, in an immunocompetent, orthotopic GBM mouse model, we show that CAR-T cells targeting tumor-specific epidermal growth factor receptor variant III (EGFRvIII) alone fail to control fully established tumors but, when combined with a single, locally delivered dose of IL-12, achieve durable anti-tumor responses. IL-12 not only boosts cytotoxicity of CAR-T cells, but also reshapes the TME, driving increased infiltration of proinflammatory CD4⁺ T cells, decreased numbers of regulatory T cells (Treg), and activation of the myeloid compartment. Importantly, the immunotherapy-enabling benefits of IL-12 are achieved with minimal systemic effects. Our findings thus show that local delivery of IL-12 may be an effective adjuvant for CAR-T cell therapy for GBM.

Fig. 1. EGFRvIII-specific CAR-T cells control newly established orthotopic tumors but fail to control large ones.

A GL261 were transduced with a retroviral vector to express the murine version of EGFRvIII. The mutated portion of EGFRvIII was fused with the transmembrane domain of the mouse EGFR to obtain cells which express the epitope on the surface. Left panel: flow cytometry staining of wild type GL261 (blue) and GL261 transduced to express EGFRvIII (red). Right panel: immunohistochemistry staining for EGFRvIII on orthotopically implanted tumors (scale bar represents 100 μm). One representative tumor is shown of four mice. B Murine CAR construct. The C-terminal portion of murine CD34 was included as marker gene and separated by a T2A peptide from the CAR construct, which included an ScFv to graft specificity, a CD8 stalk, and CD28-CD3ζ as activation domains. MR1 was used as ScFv specific for EGFRvIII, while 4g7 was used as ScFv specific for human CD19, used as negative control CAR. C “Stress experiment”. Direct comparison of the effect on tumor control of intravenous CAR-T cell administration on day 11 or 17 post tumor implantation. D Representative MRI images (axial orientation) of a mouse receiving either TBI only or TBI followed by CAR-T cells at day 11 post tumor implantation. E Survival curves (n = 3 for TBI, n = 4 for TBI+CAR mice from one experiment, p = 0.0288 (*), Log-rank test) and F tumor volume quantification. G Representative MRI images (axial plane) of a mouse receiving either TBI only or TBI followed by CAR-T cells at day 17 post tumor implantation. H Survival curves (n = 3 for TBI, n = 4 for TBI+CAR mice from one experiment) and I tumor volume quantification. Source data are provided as a Source Data file.

Fig. 2. Combination of EGFRvIII-specific CAR-T cells and a single dose of locally administered IL-12:Fc results in effective control of late stage orthotopic tumors.

A Scheme of the IL12:Fc construct inserted in the mammalian expression vector pCEP4 and a schematic representation of heterodimeric IL-12 fused to the Fc portion of murine IgG3. B Experiment timeline. GL261 EGFRvIII⁺ cells were implanted in the right striatum at day 0. Mice received 5 Gy TBI on day 15 post implantation, while tumor engraftment was confirmed on day 17. On day 20, mice underwent surgery and received either PBS or 300 ng of IL-12:Fc at tumor site, followed by intravenous injection of 1 × 10⁶ CAR-T cells or non-transduced cells. Tumor growth was monitored weekly. C Tumor volume quantification. D Representative MRI images (axial view) of a mouse from each group. E Survival curves (NT+PBS n = 14, CAR+PBS n = 19, NT+IL-12:Fc n = 18, CAR+IL-12:Fc n = 20 from four independent experiments, (NT+PBS vs. CAR+PBS p = 0.0375, NT+PBS vs. NT+IL-12:Fc p = 0.0001, NT+PBS vs. CAR+IL-12:Fc p < 0.0001, CAR+PBS vs. CAR+IL-12:Fc p = 0.0005, NT+IL-12:Fc vs. CAR+IL-12:Fc p = 0.0176, Log-rank test). Source data are provided as a Source Data file.

Fig. 3. Intra-tumoral IL-12:Fc administration reinvigorates EGFRvIII-specific CAR-T cells.

A Experiment timeline. GL261 EGFRvIII⁺ cells were implanted in the right striatum at day 0. Mice received 5 Gy TBI on day 15 post implantation, while tumor engraftment was confirmed on day 17. On day 20, mice underwent surgery and received either PBS or 300 ng of IL-12:Fc at tumor site, followed by intravenous injection of 1 × 10⁶ CAR-T cells or non-transduced cells. On day 29 mice were euthanized and FACS analysis was performed on recovered lymphoid cells from brain and spleen. B Manual gating of flow cytometry data on CAR-T defined as CD45.1⁺TCR-β⁺CD34⁺CD8⁺ cells in a representative brain sample among glioma-bearing mice treated with CAR+IL-12:Fc. C Relative marker distributions, shown as scaled histograms of arcsinh-transformed marker expression, for all CAR-T cells (violet) and the selected population (red) by CellCNN analysis (left panel); boxplot showing the median frequency and 25^th and 75^th percentile of the selected population in CAR and CAR+IL-12:Fc group (right panel), n = 5 mice per condition, representative of n = 2 independent experiments. D t-SNE map showing the FlowSOM-guided metaclustering of live intratumoral CD8⁺CAR-T cells in the different treatment groups; each color represents a metacluster and is associated with a different immune population. Heatmap showing the median marker expression in each metacluster (value range: 0–1, black and white). E Frequency of CD8⁺CAR-T cells for each metacluster in each condition, n = 5 mice per condition, representative of n = 2 independent experiments. F Median expression of selected cell markers shown for LAG3^hiPD1^hi CAR-T cells and LAG3^lowPD1^low CAR-T cells in both CAR+PBS and CAR+IL12:Fc conditions, n = 5 mice per condition, representative of n = 2 independent experiments. Data are presented as mean values ± SEM. 2-tailed unpaired Mann–Whitney T test (C, E, F). Source data are provided as a Source Data file.

Fig. 4. IL-12:Fc reshapes the endogenous compartment within the glioma TME.

A Umap showing the FlowSOM-guided metaclustering gated on TCR-β⁺ T cells from live CD45⁺ cells and heatmap showing the median marker expression for each defined metacluster (value range: 0–1) and Pie charts represent relative frequencies for the three TCR-β⁺ T cell subclusters among total TCR-β⁺ T cells within the different conditions (NT+PBS n = 4, NT+IL-12:Fc n = −5, CAR+PBS n = 5, CAR+IL-12:Fc n = 5 mice, representative of n = 2 independent experiments). B Median expression of selected cell markers shown for all TCR-β⁺ T cell subclusters: CD4, CD8, and T_regs, (NT+PBS n = 4, NT+IL-12:Fc n = 5, CAR+PBS n = 5, CAR+IL-12:Fc n = 5 mice, representative of n = 2 independent experiments). C t-SNE map showing the FlowSOM-guided metaclustering on CD11b⁺CD45⁺ cells and heatmap showing the median marker expression for each defined metacluster (value range: 0–1) and Pie charts represent relative frequencies for the nine CD11b⁺CD45⁺ cell subclusters among total CD11b⁺CD45⁺ cells within the different conditions, (NT+PBS n = 4, NT+IL-12:Fc n = 5, CAR+PBS n = 5, CAR+IL-12:Fc n = 5 mice, representative of n = 2 independent experiments). Data are presented as mean values ± SEM. Ordinary One-way Anova with Dunnett’s multiple comparison test (B). Source data are provided as a Source Data file.

Fig. 5. Local administration of IL-12:Fc results in minimal systemic effects.

Serum cytokines and chemokines were measured by cytokine bead array (CBA, Legendplex) at day 4 and 11 post IL-12:Fc administration. A IL-12, B IFN-γ, C CXCL9, D CXCL10, E IL-6, and F GM-CSF (NT+PBS n = 4, NT+IL-12:Fc n = 5, CAR+PBS n = 4, CAR+IL-12:Fc n = 5 mice from one experiment). Plots show mean ± SEM. Ordinary Two-way Anova with Tukey’s multiple comparison test. Source data are provided as a Source Data file.