To overcome the paucity of known tumor-specific surface antigens in pediatric high-grade glioma (pHGG), we contrasted splicing patterns in pHGGs and normal brain samples. Among alternative splicing events affecting extracellular protein domains, the most pervasive alteration was the skipping of ≤30-nt-long exons. Several of these skipped microexons mapped to L1-immunoglobulin cell adhesion molecule (IgCAM) family members, such as neuronal CAM (NRCAM). Bulk and single-nuclei short- and long-read RNA-seq revealed uniform skipping of NRCAM microexons 5 and 19 in virtually every pHGG sample. Importantly, the Δex5Δex19 (but not the full-length) NRCAM proteoform was essential for pHGG cell migration and invasion in vitro and tumor growth in vivo. We developed a monoclonal antibody selective for Δex5Δex19 NRCAM and demonstrated that "painting" pHGG cells with this antibody enables killing by T cells armed with an FcRI-based universal immune receptor. Thus, pHGG-specific NRCAM and possibly other L1-IgCAM proteoforms are promising and highly selective targets for adoptive immunotherapies.

Supplementary Figure 3. Mouse retina expression of GPC2 and GD2, retinal CAR T-cell infiltration, systemic toxicities, and antigen loss in the intraocular model (see related main Fig. 4).

Cell surface proteins are valuable for generation of targeted and immune-based cancer therapies. While large molecular datasets are available for multiple central nervous system (CNS) tumors, identifying such targets on a tumor-level scale to generate novel therapeutics for patients is limited by a lack of analytical pipelines. Here, we present CARTarT (CAR-T Target identification Tool), a no-code, web application that utilizes user-inputted bulk transcriptomic data to identify putative surface targets for pediatric tumors. CARTarT was built on R 4.3.1 and performs differential gene expression between tumors and normal tissue (GTEX) utilizing an empirical bayes method. Data is then filtered based on predicted-surfaceome expression, non-immune cell expression, and limited normal tissue expression to mitigate possible off-target toxicities. Targets are finally scored using a weighted scoring metric based on RNA-level log fold change, t-value, average tumoral expression, and prevalence in oncogene databases to prioritize putative targets. Targets with a score less than 0.20 were removed from our final table. A total of 919 samples from OpenPedCan were utilized to validate the tool. Amongst diffuse midline glioma specimens (n=196), we identified 100 putative surface targets. 60 (60%) of surface targets were previously identified oncogenic or cancer-associated genes. 17 surface targets (15%) are shown to have previously described immune or targeted therapeutics across cancer types, most notably identifying B7-H3, IL13RA2, EGFR, and GPC2. In addition, we identified 45 targets with potential across multiple pediatric neuro-oncologic malignancies (high grade glioma, medulloblastoma, ependymoma, atypical teratoid/rhabdoid tumor) including CHRNA5, GLP1R, B7-H3 and GPC2. These findings highlight the potential utility of CARTarT as a no-code solution for surface target identification and describe established and novel surface targets across numerous pediatric CNS tumors. We present CARTarT to the scientific community as an open-access opportunity to advance immune-based and targeted therapies for tumors.

Glioblastoma has a poor prognosis with limited therapeutic options. To date, almost all therapeutic agents that showed promise in in vitro assays, preclinical trials, or even in early clinical trials have failed to make a substantial impact on the survival of patients with high-grade glioma. One major obstacle for small molecules, therapeutic proteins, or immune cells remains the blood brain barrier, which often prevents the efficient delivery of agents to the tumor. To address this delivery issue, we developed a cellular vector where implanted modified post-mitotic Migratory Cortical Inhibitory Interneuron Precursors (MCIPs) migrate to high-grade glioma by chemoattraction to locally secrete a therapeutic protein. The inhibitory interneurons of the cerebral cortex originate predominantly in the ventral/subcortical portion of the telencephalic neural tube. During fetal brain development, MCIP migration is guided over long distances by chemorepulsant and chemoattractant factors. Remarkably, several MCIP chemoattractant factors are also secreted by high-grade gliomas. Indeed, our in vitro and in vivo data show that MCIPs robustly migrate to the majority of glioblastoma cell lines evaluated. As a proof of principle, we modified the MCIPs to secrete bispecific T-cell engagers (BiTEs), linking the EGFR tumor antigen to CD3 on T-cells to create an adaptor molecule that induces an anti-tumor response of resident and supplied T-cells by bridging a tumor antigen and the T-cell receptor. We find that implanted BITE-secreting MCIPs significantly extended survival of mice injected with high-grade glioma. Therefore, we conclude that the use of MCIPs as a delivery vector for therapeutic agents could revolutionize the way we treat glioblastoma, as they allow for the local delivery of therapeutic agents in high concentrations, bypassing the need for these agents to cross the blood brain barrier while reducing the risk for systemic toxicities outside of the brain. ### Competing Interest Statement The authors have declared no competing interest.

Glioblastoma has a poor prognosis with limited therapeutic options, in part due to the presence of the blood brain barrier, which often prevents the delivery of therapeutic agents, and the immune suppressive tumor environment (TME), which is a challenge for T-cell based immunotherapies. We developed a cellular delivery vector where implanted modified post-mitotic Migratory Inhibitory Interneuron Precursors (MIPs) are chemoattracted to high-grade glioma (HGG), independent of expressed antigens and unhindered by the TME, to deliver therapeutic proteins of choice.

In vitro drug sensitivity screens are important tools in the discovery of anti-cancer drug combination therapies. Typically, these in vitro drug screens are performed on cells grown in a monolayer. However, these two-dimensional (2D) models are considered less accurate compared to three-dimensional (3D) spheroid cell models; this is especially true for glioma stem cell lines. Cells grown in spheres activate different signaling pathways and are considered more representative of in vivo models than monolayer cell lines. This protocol describes a method for in vitro drug screening of spheroid lines; mouse and human glioma stem cell lines are used as an example. This protocol describes a 3D spheroid drug sensitivity and synergy assay that can be used to determine if a drug or drug combination induces cell death and if two drugs synergize. Glioma stem cell lines are modified to express RFP. Cells are plated in low attachment round well bottom 96 plates, and spheres are allowed to form overnight. Drugs are added, and the growth is monitored by measuring the RFP signal over time using the Incucyte live imaging system, a fluorescence microscope embedded in the tissue culture incubator. Half maximal inhibitory concentration (IC50), median lethal dose (LD50), and synergy score are subsequently calculated to evaluate sensitivities to drugs alone or in combination. The three-dimensional nature of this assay provides a more accurate reflection of tumor growth, behavior, and drug sensitivities in vivo, thus forming the basis for further preclinical investigation.

Abstract BACKGROUND GD2-directed Chimeric Antigen Receptor (CAR) T cell therapy is a promising immunotherapeutic modality for diffuse midline glioma (DMG). We developed mRNA CAR T cells as a safe alternative to virally transduced CAR T cells and have shown that intra-tumoral (IT) infusions of mRNA GD2-directed CAR T cells result in significant tumor regression with improved toxicity in murine DMG models. For clinical translation, we sought to determine the optimal route and trafficking patterns of CAR T cells administered into the cerebrospinal fluid (CSF). METHODS NSG mice engrafted with DMG xenograft SU-DIPG13P* in the pons were treated with 5x106 mRNA CAR T cells either into the lateral ventricle (LV) or into the cisterna magna (CM), with brains harvested after 24 hours for analysis using confocal microscopy. RESULTS Both LV and CM treated mice showed higher GD2-directed CAR T cell migration to tumors compared to CD19-directed controls, with higher accumulation of GD2-directed CAR T cells in the parenchyma for LV treated mice compared to CM (p<0.001). To evaluate efficacy of delivery in the CSF, NSG mice engrafted with DMG xenograft 7316-6349 in the pons were treated with 5x106 mRNA CAR T cells IT, LV, or CM twice a week for 3 weeks totaling 6 doses. Mice were imaged weekly to monitor bioluminescent tumor signal, showing that mice treated with GD2-directed CAR T cells IT had significantly decreased tumor burdens compared to LV and CM GD2 CAR treated groups (p < 0.05 and p < 0.01 respectively), as well as compared to CD19 controls (p < 0.01). CONCLUSIONS Despite local infiltration, mRNA CAR T cells delivered into the CSF did not fully reduce tumor burden, and thus, future work is aimed at evaluating potential routes of priming tumors to attract subsequent cellular therapy doses delivered intra-ventricularly.

High-grade gliomas (HGG) represent the leading cause of cancer-related mortality in children due to a lack of curative therapies. Challenges of cellular therapies in brain tumors include the lack of unique tumor antigens, downregulation of targeted antigens, and immune suppressive tumor microenvironment. To circumvent these obstacles, we devised a cellular delivery system using post-mitotic migratory cortical inhibitory interneuron precursors (MCIPs) to deliver therapeutic payloads in an antigen-independent manner. Notably, HGGs secrete the same chemoattractants stimulating long-distance MCIP migration as during normal fetal development, leading to the hypothesis that MCIPs would preferentially migrate to HGG cells. MCIP to HGG cell migration was assessed in vitro using transwell migration assays, and in vivo via stereotactic injection of MCIP and glioma xenografts in nude mice. As proof of principle, MCIPs were equipped with an EGFR or CD19 control bispecific T-cell engager (BiTE) which activates T-cells upon their linkage to the tumor cells resulting in glioma elimination. Equipped MCIPs were co-cultured with HGG cells and CD8 T-cells at an effector to target ratio of 4:1. Results were validated in vivo via xenografts of HGG, MCIPs, and either a single dose of CD8 T-cells or repeated T-cell dosing via intraventricular cannula. Survival was evaluated via Kaplan-Meier curves. MCIPs migrated preferentially to the majority of HGG cell lines in vitro and in vivo. This migration was mediated, at least in part, via CXCR4 ligands. MCIPs secreting an EGFR-BiTE potently killed HGG cells in vitro versus CD19-BiTE controls. Furthermore, nude mice xenografted with HGG and EGFR-expressing MCIPs experienced prolonged survival versus controls. These data suggest that MCIPs form a viable cellular delivery system and can be equipped with various agents targeting HGG. Modified MCIPs could be differentiated from induced pluripotent stem cells and injected into the resection cavity margins to eliminate residual cells.

Supplementary Figure 2. GPC2 CAR expression, activation, and polyfunctionality (see related main Fig. 3).

Supplementary Figure 4. Mouse body weights and GPC2 expression in the brains of study endpoint mice included in the CNS efficacy study (see related main Fig. 5).

Purpose: Retinoblastoma is the most common intraocular malignancy in children. Although new chemotherapeutic approaches have improved ocular salvage rates, novel therapies are required for patients with refractory intraocular and metastatic disease. Chimeric antigen receptor (CAR) T cells targeting glypican-2 (GPC2) are a potential new therapeutic strategy. Experimental Design: GPC2 expression and its regulation by the E2F1 transcription factor were studied in retinoblastoma patient samples and cellular models. In vitro, we performed functional studies comparing GPC2 CAR T cells with different costimulatory domains (4-1BB and CD28). In vivo, the efficacy of local and systemic administration of GPC2 CAR T cells was evaluated in intraocular and leptomeningeal human retinoblastoma xenograft models. Results: Retinoblastoma tumors, but not healthy retinal tissues, expressed cell surface GPC2, and this tumor-specific expression was driven by E2F1. GPC2-directed CARs with 4-1BB costimulation (GPC2.BBz) were superior to CARs with CD28 stimulatory domains (GPC2.28z), efficiently inducing retinoblastoma cell cytotoxicity and enhancing T-cell proliferation and polyfunctionality. In vivo, GPC2.BBz CARs had enhanced persistence, which led to significant tumor regression compared with either control CD19 or GPC2.28z CARs. In intraocular models, GPC2.BBz CAR T cells efficiently trafficked to tumor-bearing eyes after intravitreal or systemic infusions, significantly prolonging ocular survival. In central nervous system (CNS) retinoblastoma models, intraventricular or systemically administered GPC2.BBz CAR T cells were activated in retinoblastoma-involved CNS tissues, resulting in robust tumor regression with substantially extended overall mouse survival. Conclusions: GPC2-directed CAR T cells are effective against intraocular and CNS metastatic retinoblastomas.

Abstract BACKGROUND Pediatric diffuse hemispheric glioma (DHG) is a histone-mutated (H3.3G34R/V) high-grade glioma with poor prognosis. Clinical observation and previous reports have identified that a subset of patients present with tumor-associated hemorrhage. Here, we present our findings from radiological review of these patients and determine genomic risk factors predictive for hemorrhage risk in this cohort. METHODS Data was abstracted through the Children’s Brain Tumor Network (CBTN) and EGAS00001004301. Transcriptomic and genomic analyses were completed in R 4.3.1 using edgeR, msigdbr, and xCell. Presence of blood products was determined on preoperative MRI by a board certified pediatric neuroradiologist. RESULTS 48 samples were available across cohorts with transcriptomic or genomic data. 10 samples (21%) had pre-operative imaging and transcriptomic data available. Initial analysis determined that 6 samples (60%) had acute/chronic hemorrhage based on radiological review. Samples with hemorrhage had increased levels of VEGFA (LFC: 3.45, p=2E-07) and CA9 (LFC: 5.97, p=8E-05) expression on the transcriptomic level suggesting that patients with high expression of these markers (AngioHi) had increased risk of hemorrhage compared to patients with low expression of these markers (AngioLo). AngioHi patients had notably higher levels of IL8 (LFC: 3.56, p=0.01) and increased macrophage populations. Genomic information was available for 15 samples (31%, 7 AngioHi, 8 AngioLo). PTEN alterations (n=4, 57%), FBXW7 alterations (n= 2, 29%), or PDGFRa driver alterations (n= 3, 43%) were found in the AngioHi cohort while none of these alterations were identified in the AngioLo cohort. Notably, at least one of these mutations was seen in all cases in the AngioHi cohort. CONCLUSIONS These findings identify potential mutational alterations predictive for tumor-associated hemorrhage in pediatric DHG. Current work is being completed to further characterize how these mutations impact vascular remodeling for patients with DHG and may offer targeted therapeutic opportunities.

BACKGROUND Cell surface proteins are valuable for generation of targeted and immune-based cancer therapies. While large molecular datasets are available for multiple central nervous system (CNS) tumors, identifying such targets on a tumor-level scale to generate novel therapeutics for patients is limited by a lack of analytical pipelines. Here, we present CARTarT (CAR-T Target identification Tool), a no-code, web application that utilizes user-inputted bulk transcriptomic data to identify putative surface targets for pediatric tumors. METHODS CARTarT was built on R 4.3.1 and performs differential gene expression between tumors and normal tissue (GTEX) utilizing an empirical bayes method. Data is then filtered based on predicted-surfaceome expression, non-immune cell expression, and limited normal tissue expression to mitigate possible off-target toxicities. Targets are finally scored using a weighted scoring metric based on RNA-level log fold change, t-value, average tumoral expression, and prevalence in oncogene databases to prioritize putative targets. Targets with a score less than 0.20 were removed from our final table. A total of 919 samples from OpenPedCan were utilized to validate the tool. RESULTS Amongst diffuse midline glioma specimens (n=196), we identified 100 putative surface targets. 60 (60%) of surface targets were previously identified oncogenic or cancer-associated genes. 17 surface targets (15%) are shown to have previously described immune or targeted therapeutics across cancer types, most notably identifying B7-H3, IL13RA2, EGFR, and GPC2. In addition, we identified 45 targets with potential across multiple pediatric neuro-oncologic malignancies (high grade glioma, medulloblastoma, ependymoma, atypical teratoid/rhabdoid tumor) including CHRNA5, GLP1R, B7-H3 and GPC2. CONCLUSION These findings highlight the potential utility of CARTarT as a no-code solution for surface target identification and describe established and novel surface targets across numerous pediatric CNS tumors. We present CARTarT to the scientific community as an open-access opportunity to advance immune-based and targeted therapies for tumors.

Supplementary Figure 1. Expression of GPC2 and GD2 in retinoblastoma patient samples and cell lines (see related main Fig. 1).

Abstract Novel therapies are needed to for the treatment of atypical teratoid/rhabdoid tumor (ATRT), an aggressive brain tumor that predominantly affects young children and has an average 5-year survival under 50%. Claudin-6 (CLDN6) is a tight junction protein present during development and is expressed in up to 70% of ATRT specimens but not in normal tissue, making it a promising immunotherapeutic target. CLDN6-targeted chimeric antigen receptor (CAR) T cells in combination with a CAR T cell-amplifying mRNA vaccine have demonstrated antitumor activity against other CLDN6-expressing cancers in pre-clinical and a phase I adult trial (NCT04503278; Haanen J et al AACR, 2022). To assess the effectiveness of CLDN6-targeted CAR T cells against ATRT, we utilized a second-generation mRNA CAR with a 4-1BB costimulatory domain and single-chain variable fragment against CLDN6 (Reinhard et al, 2020). CLDN6 expression in patient-derived ATRT specimens was profiled by RNAseq (mean FPKM 11.4) and immunohistochemistry (positive staining in 53% of specimens). Tumor-derived cell lines were assessed for CLDN6 expression by flow cytometry. Co-culture of CLDN6-directed mRNA CAR T cells resulted in tumor-specific cytotoxicity compared to CD19-directed control CAR T cells in CLDN6-positive ATRT cell lines 7316-2187 (92% versus 15% at 10:1, p < 0.0001; 86% versus 0% at 5:1, p<0.0001) and 7316-2141 (75% versus 7% at 10:1, p<0.0001; 53% versus 0% at 5:1, p< 0.0001). Patient-derived, orthotopic xenograft ATRT models were created through intracranial injection of tumor cells into the cerebellum of NSG mice. Following engraftment with 7316-2141, repeated intratumoral administration of mRNA CLDN6 CAR T cells resulted in significant tumor regression (-1.9x108 vs. +2.4x109 p/sec/cm2/sr, p<0.001) and improved survival compared to mRNA CD19 CAR T cells (median OS not reached vs. 13 days, p=0.002). This work highlights the potential for targeting CLDN6 via CAR T cell therapy in patients with ATRT as a novel therapeutic strategy.

Abstract BACKGROUND GD2-directed chimeric antigen receptor (CAR) T cells have shown promise as a potential therapeutic for diffuse midline glioma. We have previously shown intra-tumoral infusions of mRNA CAR T cells result in tumor regression with improved safety and toxicity profile in murine models. For human trials, the optimal route of CAR T cell administration into the central nervous system should maximize in vivo potency while minimizing procedural morbidity. Prior mRNA CAR T literature has reported decreased solid tumor infiltration with systemic delivery, so we sought to determine mRNA CAR T cell trafficking into pontine tumors from the cerebral spinal fluid to inform clinical translation of this therapy. METHODS Mice engrafted with SU-DIPG13P* tumor cells in the pons were treated with 5x106 mRNA CAR T cells into the lateral ventricle (LV) and brains were harvested 24 hours later. Tumor and T cells were analyzed using confocal microscopy to determine the migration of CAR T cells. RESULTS LV injection of GD2-directed CAR T cells resulted in T cell trafficking to the subarachnoid spaces, tumor site, and surrounding parenchyma. The highest concentration of T cells was in the subarachnoid space directly adjacent to the tumor (p< 0.0001) and decreased on a gradient to subcortical tumor. Even at 24 hours after injection, T cells were identified within the center of the tumor mass. CD19-directed control CAR T cells showed similar distribution within the subarachnoid with significantly decreased T cells at the tumor site. CONCLUSION Ventricular infusions of mRNA CAR T cells showed effective migration to pontine tumors within 24 hours. While CAR T cells did not fully penetrate the parenchyma within that timeframe, GD2-directed mRNA CAR T cells were detected within the center of the tumor mass, indicating ventricular administration may be sufficient for clinical delivery. Additional routes are being explored in ongoing work.