GPC2-CAR T cells have potent preclinical activity against orthotopic medulloblastoma xenografts

Abstract

Medulloblastoma (MB) is the most common malignant brain tumor in children. Patients with group 3 and 4 MB have poor clinical outcomes, underscoring the urgent need for new therapies. Glypican-2 (GPC2) is a recently discovered oncofetal antigen. Given its expression in brain tumors, we evaluated the preclinical activity of our GPC2-chimeric antigen receptor (CAR) against MB and compared it to two existing CARs targeting GD2 and B7-H3. Gene expression analysis of publicly available datasets was performed and validated with immunohistochemistry staining of patient samples. The MB cell line D283 and a newly generated cell line patient-derived xenograft, MAF1433, were used. Cytokine bead assays and single-cell RNA sequencing (scRNA-seq) were used for mechanistic studies. MB patient samples express up to moderate levels of GPC2. GPC2-CARs lead to significant in vivo tumor regression in orthotopic tumor models via intravenous or intraventricular administration route and had equivalent activity to the B7-H3-CAR against D283 and enhanced activity than GD2-CAR in both models in vivo. T cell kinetic studies revealed that GPC2-CAR T cells home to the area of the primary tumor, expand, and upregulate genes critical for cytotoxicity and T cell homing. These results provide a preclinical rationale for including children with GPC2⁺ MB in our upcoming clinical GPC2-CAR T cell trial.

Keywords: CAR T cells; GPC2; MT: Special Issue - Advancements in pediatric cancer therapy; glypican; immunotherapy; medulloblastoma; oncofetal antigen; pediatric oncology.

Graphical abstract

Figure 1

GPC2 expression in medulloblastoma (A) Expression of B4GLANT1 and ST8SIA1 (enzymes responsible for GD2 synthesis), CD276, and GPC2 in various primary tumors, derived xenografts, and cell lines from patients with embryonal tumor with multilayered rosettes (ETMR), high-grade glioma (HGG), atypical teratoid rhabdoid tumor (ATRT), ependymoma (EPD), and medulloblastoma (MB). One-way ANOVA test with ad-hoc multiple comparisons test: ns (not significant), p > 0.05; ∗p < 0.05; and ∗∗p < 0.01. (B) Examples of pathology scores 0–3+ in brain tumor samples stained with CT3 to detect GPC2 expression. (C) Summary of pathology scoring across different tumor histologies. ETMR, group 3, group 4, and Sonic hedgehog (SHH) MB, and HGG were tested. (D–F) Flow cytometry analysis for the detection of (D) GD2, (E) B7-H3, and (F) GPC2.

Figure 2

In vitro testing of GPC2-, B7-H3-, and GD2-CARs against medulloblastoma (A) Schematic of the GD2-, B7-H3-, and GPC2-CAR constructs used in this study. (B) Transduction efficiency of T cells transduced with the construct in (A). (C and D) Seventy-two-hour in vitro cytotoxicity assay of GD2-, B7-H3-, and GPC2-CAR against (C) D283 and (D) MAF1433. The OneGlo luciferase-based assay was used to quantify viable cells. One of two replicates is shown with technical triplicates per condition. One-way ANOVA test with ad-hoc multiple comparisons test. ns, not significant; ∗p < 0.05; ∗∗p < 0.01; and ∗∗∗p > 0.001. (E and F) Tumor rechallenge assay against (E) D283 and (F) MAF1433. Tumor rechallenges are indicated by the lines and arrows. The total green object intensity is measured to quantify GFP⁺ tumor viability. One-way ANOVA test with ad-hoc multiple comparisons test. ∗∗∗∗p < 0.0001. Means and standard deviations are plotted in subpanels (C)–(F).

Figure 3

GPC2-CAR T cells induce potent tumor regression against medulloblastoma in vivo (A) Experimental schema of in vivo studies. After stereotactic tumor injections, mice are given 14 days for tumor engraftment before undergoing IVIS BLI for randomization. CAR T cell injection occurs on day 0. Weekly IVIS BLI is performed for tumor tracking. (B) H&E staining demonstrates a representative orthotopic tumor location after stereotactic injection into the cerebellum. Scale bars, 1 mm. (C and D) IVIS BLI signals of tumor burden after CAR T cell therapy for (C) D283 and (D) MAF1433. (E) Comparison of intravenous (i.v.) and intraventricular (ICV) administration of GPC2-CAR T cells to treat MAF1433-bearing mice. Two-way ANOVA test with ad-hoc multiple comparisons test. ∗p < 0.05; ns p > 0.05. (F) Histological tumor status at the end of the study of mice from (E). (G) On day 6 and day 14, animals underwent serum analysis of cytokines, including IFN-γ, perforin (PRF), and granzyme B (GZMB). One-way ANOVA test with ad-hoc multiple comparisons test. ∗p < 0.05; ∗∗∗p < 0.001; and ns p > 0.05. Means and standard deviations are plotted in subpanel (E).

Figure 4

Single-cell RNA-seq analysis of tumor-infiltrating CAR T cells at peak infiltration (A) Tracking of T cells in MAF1433-bearing mice. T cells are luciferase-expressing and imaged on days 6, 8, and 10 post-CAR injection. (B) UMAP of the integrated tumor, tumor-infiltrating human and murine immune, and stromal cells from all samples. The independent clustering yielded 45 clusters. (C) Frequencies of cells based on manual annotations. (D) Violin plots show the expression levels of common genes across the different CARs, tumor cells, and the rest of the cells. (E–G) Ingenuity pathway analysis of differentially expressed genes comparing (E) GPC2- vs. B7-H3-CAR T cells, (F) GPC2- vs. GD2-CAR T cells, and (G) GD2- vs. B7-H3-CAR T cells.