Blocking MIF secretion enhances CAR T-cell efficacy against neuroblastoma

Abstract

Introduction: Chimeric antigen receptor (CAR) T-cell therapy is a promising and innovative cancer therapy. However, immunosuppressive tumor microenvironments (TME) limit T cell persistence and durable efficacy. Here, we aimed to identify and target immunosuppressive factors in the TME of neuroblastoma, a pediatric extracranial solid tumor, to improve CAR-T efficacy.

Methods: Immunosuppressive factors were identified using a multi-omics approach, including single-cell RNA sequencing (scRNA-seq) of 24 neuroblastoma tumors, published bulk-RNA sequencing datasets, and mass-spectrometry of patient-derived tumoroid models. Candidate targets were validated with functional assays in vitro and in vivo. Protein degradation of the top immunosuppressive target by PROTAC technology was used to evaluate the effect on CAR T-cell activity.

Results: ScRNA-seq revealed 13 immunosuppressive interactions in the TME of neuroblastoma, two effectors of which, Midkine (MDK) and Macrophage Migration Inhibitory Factor (MIF), were validated as candidate targets across multiple published datasets. Both factors were among the top 6 % of most abundantly secreted factors by patient-derived tumoroid models, substantiating their potential relevance in the TME. In vitro and in vivo functional assays confirmed MIF to be a potent inhibitor of CAR T-cell activation and killing capacity. To translate these findings into a potentially clinically applicable treatment, we explored MIF targeting by PROTAC technology, which significantly enhanced activation of CAR T-cells targeting GPC2 and B7-H3.

Conclusion: By defining the immunosuppressive effects of neuroblastoma's TME on CAR T-cell efficacy, revealing the pivotal role of MIF, we provide an analytic pipeline and therapeutic strategy for improving adoptive cell therapies for this pediatric malignancy and potentially other solid tumors.

Keywords: CAR T-cell therapy; Immunosuppressive tumor microenvironment; MIF; Neuroblastoma; PROTAC.

Fig. 1

Identification of MIF and MDK as immunosuppressive factors based on transcriptomic data. a, Schematic representation of analysis strategy of scRNAseq data, published earlier by our group, in order to determine significant correlations between the cytotoxicity score of immune cell subsets and predicted interactions of those immune cells with other cells in the tumor microenvironment. b, Dotplot representing the gene expression of 7 cytotoxicity genes to determine the cytotoxicity score of γδ-T cells, CD8 T cells and NK cells. c, Score of cytotoxicity genes in Fig. 1b in NK, CD8-T, and γδ-T cells. Tukey’s multiple comparisons test for one-way ANOVA. d, Heatmap representing the correlation between genes in specific cell subsets, which significantly interact with the indicated immune cell subset and significantly correlate with the cytotoxicity of the indicated immune cell subset. Genes with at least one significant correlation, with either NK, CD8-T, or γδ-T, were included. The black box indicates a negative sum of the correlations of the three immune cell subsets and represents the genes selected for further analysis. e, Correlation of 13 selected genes from Fig. 1d with cytotoxicity in the dataset with bulk-RNA data from 498 neuroblastoma tumors (r2.amc.nl/; Tumor Neuroblastoma - SEQC - 498 - RPM - seqcnb1; GSE497104243). f, Survival analysis using the 13 selected genes from Fig. 1d using the SEQC bulkRNA cohort. Left panel represents event-free survival (EFS) and right panel represents overall survival (OS). Survival when the expression of the gene is high is depicted on the x-axis, the Bonferroni p-value on the y-axis. g, Kaplan-Meier curve indicating event-free survival for high- or low expression of MIF (left panel, expression cutoff: 157.498) and MDK (right panel, expression cutoff: 216.718). Bonferroni p-value is depicted. h, Dotplot representing the expression of MIF and MDK in several cell subsets in scRNAseq dataset .

Fig. 2

Identification of MIF and MDK in secretome of neuroblastoma tumoroids by LC-MS. a, Pictures of cultures from selected tumoroids for secretome collection. Scalebar indicates 500 µM. b, Schematic overview of procedure to collect and process the secretome for analysis. Further details are provided in the methods section. Figure made with Biorender.com. c, Liquid chromatography–mass spectrometry (LC-MS) analysis of conditioned medium from neuroblastoma tumor organoids detailing their secretome. Red circles/black or pink font: targets identified in Fig. 1d. Blue circles/grey font: neuroblastoma reference proteins. Left green bar indicates top 100 most abundant proteins. iBAQ (intensity Based Absolute Quantification) indicates the protein’s non-normalized intensity divided by the number of measurable peptides indicating the relative abundance of the protein in the sample.

Fig. 3

The immunosuppressive effect of MIF and MDK on CAR-T cell activation. a, Flow cytometry analysis of healthy donor peripheral blood T cells after 4 day in vitro stimulation with anti-CD3/anti-CD28 beads, in the presence of rMIF or rMDK (10 ng/mL). Representative graphs of CellTrace Violet peaks in CD8+ population (left panel). Middle and right panel represent the normalized division index of the CD8+ and CD4+ population from three healthy donors, respectively. Division index was determined using FlowJo’s proliferation platform. Dunnett’s multiple comparisons test for one-way ANOVA (n = 3 healthy donors, blue indicates recombinant MIF, red indicates recombinant MDK). b, Granzyme B MFI (Median Fluorescence Intensity) analysis from same experiment as Fig. 3a. Representative graphs (left two panels) of granzyme B expression in CD8+ population. Combined data from three healthy donors (right panel). MFI was normalized to stimulated control. Dunnett’s multiple comparisons test for one-way ANOVA (n = 3 healthy donors, blue indicates recombinant MIF, red indicates recombinant MDK). c, Validation of MDK knock-down by western blot (intracellular; left) and ELISA (secreted; right) on SK-N-AS and SH-SY5Y. d, Validation of MIF knock-down by western blot (intracellular; left) and ELISA (secreted; right) on SK-N-BE2C and Kelly. e, Activation of GPC2 CAR T-cells after co-culture with shCtrl and shMDK SK-N-AS and SH-SY5Y models measured by IFN-γ ELISA using two effector:target ratios (E:T; 1:1 and 1:5) at 2 different timepoints (days 2 and 5). Two-way ANOVA with Šídák's multiple comparisons test. (n = 1 CAR donor with 3 technical replicates). f, Activation of GPC2 CAR T-cells after co-culture with shCtrl and shMIF SK-N-Be2C and Kelly models measured by IFN-γ ELISA using 1:1 and 1:5 E:T ratios at days 2 and 5. Two-way ANOVA with Šídák's multiple comparisons test. (n = 1 CAR donor with 3 technical replicates). g, Proliferation of GPC2 CAR T-cells in co-culture with SK-N-BE2C-shCTRL (grey) or SK-N-BE2C (blue) in 1:1 and 1:5 Effector:Target ratio. Left plot show proliferation (CFSE) plots of one representative assay. Two-way ANOVA with Šídák's multiple comparisons test. (n = 1 CAR donor with several technical replicates). h, % of CD25 + population in GPC2 CAR-T cells in co-culture with SK-N-BE2C-shCTRL (grey) or SK-N-BE2C (blue) in 1:5 Effector:Target ratio. Each CAR-T cell donor is connected by a line. Statistics represent a multiple paired t-test with False Discovery Rate correction. n = 3 CAR-T cell donors. i, % of CD107a+ population in GPC2 CAR-T cells in. Each CAR-T cell donor is connected by a line. Statistics represent a multiple paired t-test with False Discovery Rate correction. n = 3 CAR-T cell donors. j, Mean % of TIM3, LAG3, PD-1 and CD39 on GPC2 CAR-T cells day 5 of co-culture with SK-N-BE2C-shCTRL (grey) or SK-N-BE2C (blue). Statistics represent 2-way ANOVA with Šídák's multiple comparisons correction n = 3 CAR-T cell donors.

Fig. 4

Reducing MIF secretion by the tumor increases the cytotoxicity of CAR-T cells. a, IncuCyte S3 experiment measuring tumor growth during co-culture with CAR-T cells. SK-N-BE2C neuroblastoma cells with shCtrl (grey) or shMIF (blue) were cultured with a control CAR-T cell targeting CD19 (triangles) or tumor antigen GPC2 (diamonds). Effector:Target ratio of 1:1. b, Normalized residual tumor cells at endpoint of the experiment in Fig. 4a. One-way ANOVA statistical test with Holm-Šídák's multiple comparisons test for significance. c, Schematic overview of procedure to study the effect of MIF-knock down on efficacy of CAR-T cells in mice. Figure made with Biorender.com. d, Average SK-N-BE2C shCtrl or shMIF tumor growth. Measuring of tumor size started when CD19 or GPC2 CAR-T cells were injected (5 × 10⁶, iv, at arrow indication). Experimental groups of n = 6–9. Statistics show two-way ANOVA using Tukey’s multiple comparisons test at t = 2.6 weeks, when mice in the control group were sacrificed. e, Progression free survival (PFS) of shCtrl or shMIF tumor bearing mice treated with CD19.CAR or GPC2.CAR. Statistical analysis shows p-values for a log-rank (Mantel-Cox) test.

Fig. 5

MIF PROTAC reduces MIF secretion, enabling more efficient activation of CAR-T cells. a, Schematic representation of PROTAC degraders, which causes ubiquitination and degradation of proteins of interest through the proteosome. Figure made with Biorender.com. b, MIF concentration after treatment with 0.1, 1, or 10 µM MD13 for 48 h. Left panel represents MIF concentration of tumoroid model AMC691B and right panel of tumoroid model AMC691T. Measured using Luminex. c, IFN-γ concentration of supernatant from co-culture of AMC691B without treatment (grey) or with 1uM MD13 treatment (blue) in combination with GPC2 CAR-T cell in a 1:5 Effector:Target ratio, as measured by ELISA. Left panel shows concentration at 24 h and right panel shows concentration at 48 h. Statistical analysis shows results for paired t-test. (n = 4 CAR T-cell donors). d, Left panel: Luminescence signal of luciferase transduced tumoroid model AMC691B after co-culture of 24 h. Normalized to untreated tumoroid only. Tumoroids were pre-treated with or without PROTAC for 48 h before co-culture. (n = 2 CD19-CAR T-cell donors, n = 4 GPC2-CAR T-cell donors). Right panel: Normalized GPC2 CAR-T cell killing. Data were normalized to the tumoroid only untreated or treated control, respectively. Statistical analysis shows results for paired t-test. e, IFN-γ concentration of supernatant from co-culture of AMC691T without treatment (grey) or with 1uM MD13 treatment (blue) in combination with GPC2 CAR-T cell in a 1:5 Effector:Target ratio, as measured by ELISA. Left panel shows concentration at 24 h and right panel shows concentration at 48 h. Statistical analysis shows results for paired t-test. (n = 4 CAR T-cell donors). f, Left panel: Luminescence signal of luciferase transduced tumoroid model AMC691T after co-culture of 24 h. Normalized to untreated tumoroid only. Tumoroids were pre-treated with or without PROTAC for 48 h before co-culture. (n = 2 CD19-CAR T-cell donors, n = 4 GPC2-CAR T-cell donors). Right panel: Normalized GPC2 CAR-T cell killing. Data were normalized to the tumoroid only untreated or treated control, respectively. Statistical analysis shows results for paired t-test. g, IFN-γ concentration of supernatant from co-culture of AMC691B without treatment (grey) or with 1µM MD13 treatment (blue) in combination with B7-H3 CAR-T cell in a 1:5 Effector:Target ratio, as measured by ELISA. (n = 3 B7-H3-CAR T-cell donors). h, IFN-γ concentration of supernatant from co-culture of AMC691T without treatment (grey) or with 1µM MD13 treatment (blue) in combination with B7-H3 CAR-T cell in a 1:5 Effector:Target ratio, as measured by ELISA. (n = 3 B7-H3-CAR T-cell donors).