PBT tumor cells lines are derived from patients with brain tumors (IRB07074) dissociated and grown in DMEM/F12 medium, supplemented with heparin, hepes, glutamax, and B27. Epidermal growth factor (EFG) and fibroblast growth factor (FGF) are added at the time of culture, as described previously (34). ICG-001 was provided by M. Kahn’s laboratory. Co-culture assay of PBT cell lines were grown as described above and seeded to 100,000 cells/2 ml in 6-well plates. ICG-001 was added to cell cultures in concentrations 0, 5, and 10 µM for 24–72h, as described previously (35).

RNA was extracted using the Total RNA Kit according to the manufacturer’s instructions (Qiagen, RNeasy PowerSoil). cDNA was generated with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Quantitative RT-PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems). Amplification of human Kif20A was performed on RNA samples isolated from PBT147 and PBT030 cells treated with ICG-001 at concentrations of 0, 5, and 10 µM. All data was normalized to the PBT147-24H-0 drug control sample. Human Kif20A RT-PCR primers used for RT-PCR analysis were: Forward: TGGTACGCAAGAACCTGC; Reverse: GATCAGGGTTGTGTCCGT. Human GAPDH primers were used as controls: Forward: GGATTTGGTCGTATTGGG; Reverse: GGAAGATGGTGATGGGATT.

The normalized distributions of protein levels were heavily skewed to the right, so we performed a logarithmic transformation of the data, which resulted in an approximate normal distribution of protein levels. The densities of the log-transformed normalized protein level for each cell line/dose are shown in Supplementary Figure 3 . To examine the effect of ICG-001, we calculated the difference in log normalized levels for each protein between the four treated samples and the appropriate untreated sample (PBT147 and PBT030 cell lines were treated with ICG001 at 0, 5, and 10 µM, and cells were collected after 24 or 72 h for protein analysis), which corresponds to examining the log of the fold changes in each protein after treatment. For each of the two cell lines (PBT147 and PBT030), we plotted the log fold change for each protein at 5µM against the log fold change at 10µM. These plots are shown in Supplementary Figure 3A , with the points colored-coded by cell line.

Anesthesia: For tumor models, mice were anesthetized by intraperitoneal (i.p.) injection of Ketamine/Xylazine and gaseous Isoflurane prior to tumor injection. Kluc [0.1×10⁵ were prepared in PBS−/− (2 μL per mouse)] and injected orthotopically in the brain parenchyma of female NSG mice via stereotactic injection. Tumor growth was monitored at least once a week via optical imaging (Spectral Instruments Imaging, LagoX) and flux signals were analyzed with Aura Imaging software (LagoX). For imaging, mice were injected intraperitoneally with 150 μL d-luciferin potassium salt (Perkin Elmer) suspended in PBS at 4.29 mg/mouse. Once flux signals reached desired levels, CAR T cells were prepared in PBS and mice were treated either by intratumoral/intracranial injection in 3 μL final volume. At desired time points or at moribund status, mice were euthanized, and tissues were processed for IHC as described below. Syngeneic mice (C57BL/6) of 8–12 weeks of age were implanted with subcutaneous K-Luc tumors (n=8). 7 days later, when tumors became palpable and after confirmation of tumor presence with BLU imaging, mice were implanted with Alzet minipumps. Pumps continuously provided a daily dose of ICG-001 (50 mg/kg/day). Tumor tissues were harvested on days 7, 14, and 21 post pump implantations, and tumors were prepared for IHC (paraffin sections) and NanoString analysis. Control mice were not treated with ICG-001 pumps, and tumors from control mice were harvested on day 7, 14 and 21 to match ICG-001-treated tumors.

Three immune cell markers (CD3, CD8, CD31, Abcam) were evaluated with IHC on whole tumor sections. IHC Staining was performed in the City of Hope Pathology Core according to the manufacturer’s instructions (n=6 mice). After IHC staining, slides were scanned using the NanoZoomer 2.0-HT slide scanner at 10× magnification (Hamamatsu Photonics). Scanned slides were then imported into Qupath as Brightfield H-DAB images for analysis. Tumor section annotations were manually outlined with the brush and wand tools. Once all tumor areas were selected, total cell count, positive cell count per area, and total area were counted using the Positive Cell Detection tool with the settings optimized for each CD3, CD8 and CD31 staining ( Figure 1 ). All data were extracted from QuPath and further calculations and quantifications were done with Microsoft Excel and Prism.

After IHC staining, the slides were scanned using the NanoZoomer 2.0-HT slide scanner (Hamamatsu Photonics). Scanned slides were then imported into Qupath as Brightfield H-DAB images for analysis. Tumor section annotations were manually outlined with the brush and wand tool. Once all tumor areas were selected, total cell count, positive cell detection, positive %, and total area were counted using the Positive Cell Detection tool with the auto settings ( Figures 1D, H, L ). All data were then extracted from Qupath and further calculations and quantifications were done with Microsoft Excel and Prism.

Raw gene count was averaged across three brain tissue samples from CAR T treated mice and across two brain tissue samples from CAR T + ICG001 treated groups. Brain tissue was isolated during euthanasia, and tumors were dissected and snap frozen in liquid nitrogen. Total RNA was isolated from K-Luc tumor tissue and analyzed by NanoString (NanoString nCount mouse PanCancer Immune profiling paneled assays, https://nanostring.com/products/ncounter-assays-panels/oncology/pancancer-immune-profiling). Multiplex gene expression analysis was performed in mice for 770 genes from different immune cell types, common checkpoint inhibitors, CT antigens, and genes covering both the adaptive and innate immune response. The panel measures many features of the immune response to facilitate rapid development of clinical actionable gene expression profiles in the context of cancer immunotherapy. Comparisons between brain tissue and ICG001-treated tissue were determined by the nonparametric U-test, and log₂ |fold change| ≥ 1.5, -Log₁₀ P-value < 0.05 using the Benjamini-Hochberg method to be statistically significant. Log2 normalized counts and expression ratios were generated using nSolver 4.0 and advanced analysis 2.0 (NanoString Technologies, Inc.) as well as ROSALIND analysis platform v3.38.0.1.

Data was analyzed by ROSALIND® version 3.38.0.1 (https://www.rosalind.bio/), with a HyperScale architecture developed by ROSALIND, Inc. (San Diego, CA). Read Distribution percentages, violin plots, identity heatmaps, and sample MDS plots were generated as part of the QC step. Normalization, fold changes and p-values were calculated using criteria provided by NanoString. ROSALIND® follows the nCounter® Advanced Analysis protocol of dividing counts within a lane by the geometric mean of the normalizer probes from the same lane. Housekeeping probes for normalization were selected based on the geNorm algorithm as implemented in the NormqPCR R library1. Abundance of various cell populations was calculated on ROSALIND using the NanoString Cell Type Profiling Module. ROSALIND was used to perform a filtering of Cell Type Profiling results to include results that have significant scores (p ≤ 0.05). Fold changes and p-values were calculated using the fast method, as described in the nCounter® Advanced Analysis 2.0 User Manual. P-value adjustment was performed using the Benjamini-Hochberg method of estimating false discovery rates (FDR). Clustering of genes for the final heatmap of differentially expressed genes was performed using the Partitioning Around Medoids (PAM) method using the fpc R library2 that takes into consideration the direction and type of all signals on a pathway (the position, role and type of every gene, etc.). Hypergeometric distribution was used to analyze the enrichment of pathways, gene ontology, domain structure, and other ontologies. The topGO R library3, was used to determine local similarities and dependencies between GO terms in order to perform Elim pruning correction. Several database sources were referenced for enrichment analysis, including Interpro4, NCBI5, MSigDB6,7, REACTOME8, and WikiPathways9. Enrichment was calculated relative to a set of background genes relevant for the experiment (ACTB, GAPDH).

We initially tested the effects of the specific small molecule CBP/β-catenin antagonist ICG-001 in vitro as a single agent on 5 human PBT and 3 mouse glioma lines. Treatment with ICG-001 (0–10 µM) showed a concentration-dependent cytostatic effect in all tested human patient-derived glioma cell lines (PBT017, PBT030, PBT135, PBT144, and PBT147) ( Figures 1A, D–F ) and all murine-derived glioma cell lines (K-luc, GL261-parental, and GL261.IL13Rα2 engineered) ( Figure 1B ). We found that the KIF20A gene was downregulated in the PBT030 and PBT147 lines upon treatment of glioma cells with ICG-001 ( Figure 1C ). KIF20A, a mitotic kinesin, plays an important role in controlling the mode of division of neural progenitor cells (NPC) in both normal brain cells and brain tumor cells, and KIF20A knockdown induces a transition from proliferative to differentiative divisions in NPC (36, 37). Consistent with ICG-001-induced differentiation of PBT147 and PBT030 lines, ICG-001 treatment led to a significant reduction in the expression of KIF20A, as measured by qPCR ( Figure 1C ) (36), further confirming that the differentiating effects of ICG-001 on glioma lines are not cell line dependent. Additionally, all glioma lines exhibited a more differentiated phenotype based on their elongated cell morphology, with decreased proliferation and loss of clonal expansion, as shown by the PBT017.RFP human glioma cell line, imaged and quantified by Incucyte ( Figures 1D–G ). We also correlated human Wnt gene family with CD3 mRNAseq expression using public TCGA GBM datasets and RNA-seq data [downloaded from the GDC portal (https://portal.gdc.cancer.gov/)] ( Supplementary Figures 1A, B ). Upregulation of the Wnt pathway in glioma was inversely correlated with CD3 cell infiltration ( Supplementary Figure 1B ). A strong negative correlation was found among Wnt genes-APC, AXIN1, AXIN2, GSK3BTCF4 and CD3 score and a positive correlation was found for LEF1, CDC42 and CD3.

Decreased expression of the Wnt/CBP/β-catenin target gene BIRC5/Survivin, with concomitant upregulation of the Wnt/p300/β-catenin target gene EphB2, is a hallmark of specific CBP/β-catenin inhibition ( Figure 2A ) (38). In our studies, we used EphB2 and BIRC5/Survivin as a biomarker of the response to ICG-001. BIRC5/Survivin expression was also previously used as a biomarker for pharmacokinetic/pharmacodynamic (PK/PD) activity of the Wnt pathway in circulating tumor cells in the first-in-human clinical trial of the second-generation CBP/β-catenin specific antagonist PRI-724 (39). We demonstrated that BIRC5/Survivin gene expression was selectively downregulated and EphB2 was upregulated in our experiments in both human and mouse glioma cell lines post ICG-001 treatment, as shown by qPCR, which was further confirmed using a Survivin-luciferase reporter assay for PBT017 and PBT030 cell lines in vitro ( Figure 2B ). ICG-001’s selective effects on the expression of BIRC5/Survivin and EphB2 serve as an indicator of on target activity in vitro in the human and mouse G261, GL261.IL13Ra2, and K-luc cell lines ( Figures 2A, C ).

The N-termini of the two human Kat3 coactivators, CBP and p300, provide a highly evolutionarily conserved hub to integrate multiple signaling cascades that coordinate cellular metabolism with the regulation of symmetric versus asymmetric division of somatic stem cells and cancer stem cells, cellular proliferation, and differentiation status and function (40). More specifically, small molecule inhibition of the N-terminal region of CBP enhances p300/β-catenin mediated transcription, which is a prerequisite for increased mitochondrial oxidative metabolism during the initiation of cellular differentiation (40, 41). To explore ICG-001’s impact on metabolic changes associated with glioma differentiation after disrupting the CBP/β-Catenin interaction, we treated patient-derived GBM cell lines PBT147 and PBT030 with ICG-001 (0, 5, or 10 µM) for 24 and 72 h ( Figure 3 ). RNA isolated from the treated PBT cells was analyzed using the NanoString nCounter metabolic panel (NanoString Technologies) (42). We used Rosalind software v.3.38.0.1 to analyze differential gene expression between PBT030 and PBT147 cell lines ( Figures 3A–D ). Furthermore, we analyzed gene and pathway expression in ICG-001 treated PBT samples as compared with untreated controls ( Figures 3E, F ). Statistically significant genes and pathways were analyzed using log₂ Fold Change ≥ 1.5, < -1.5 and -Log₁₀ p-Adj <0.05 filter. First, we demonstrated differences between PBT030 and PBT147 patient-derived lines (X-axis shift on MDS plot) and that ICG-001 treatment caused a similar shift in gene expression on the Y-axis among untreated and ICG-001-treated cell lines ( Figure 3A ). We detected genes upregulated in PBT030 cells, such as SOX2, MYCN, NOS2, RUNX2, and VEGFA, while PBT147 cells upregulated level of PTEN, STAT6, CA9 and TLR2, demonstrating differences among PBT cell lines, which can be explained by various driving mutations and heterogeneity in glioma lines ( Figures 3A–D and Supplementary Figures 2A, B ).

Next, we tested if ICG-001 drives a similar metabolic shift in both lines. Fully consistent with our previous studies (40), we found similar changes in several genes and metabolic pathways after treatment of PBT030 and PBT147 with ICG-001–an increase in amino acid transporters, hypoxia, glutamine metabolism, MAPK, autophagy, mitochondrial respiration, cytokine and chemokine signaling, and arginine metabolism in both lines ( Supplementary Figure 7 ). The PBT030 line demonstrated a decrease of tryptophan/kynurenine, fatty acid synthesis and oxidation, glycolysis, and the pentose phosphate pathway ( Supplementary Figure 7 ). Similarly, in PBT147 lines treated with ICG-001, an increase in glutamine metabolism, autophagy, amino acid transport and synthesis, and MAPK, and strong down regulation in cell cycle, pentose phosphate, fatty acid synthesis, and oxidation was observed. ( Supplementary Figure 7 ). This change in tumor cell metabolism may provide CD8 effector cells with the ability to effectively compete metabolically within the TME (43), which increases the likelihood of immune cell infiltration into the otherwise hostile TME (44). Consistent with our in vitro data ( Figures 1 and 2 ), we also observed downregulation of cell cycle-related genes post ICG-001 treatment in PBT030 (PRIM2, NPM1) and PBT147 cell lines (including KIAA0101-PCNA, MKI67, RRM2, TYMS, CLSPN, CDCA8, UBE2C, EXO1, BRCA1, BRCA2, BRIP1, and PRIM1), which is associated with cytostasis and metabolic changes required for the switch from proliferation to differentiation ( Figures 3B, C ) (40). Fatty acid oxidation–the metabolic pathway preferred by quiescent stem-like cells (45)—was decreased in all treated cell lines at all timepoints and all doses. Furthermore, ALDH2 (PBT147), ACAT2, and HADH were downregulated in PBT030 cells, indicating that CBP/β-Catenin inhibition directs GSC activation and differentiation via metabolic reprogramming. These metabolic changes have been shown to be critical for the transition of quiescent to activated CSC (46) and their subsequent differentiation to bulk tumor cells (36). Upregulation of the p53 tumor suppressive pathway and mitochondrial respiration (up-PBT030-PPARGC1A, MCP1, PDP1, ME2, IDH3A, UQCR11, FAHD1, SOD2, and ATP6V1F), and glutamine metabolism (up-PBT030-PSAT1, GOT1, SERINC1; PBT147-ASNS, PYCR1, and PHGDH)—hallmarks of cell differentiation–were also observed at both time points, at both the 5- and 10-µM doses and in both cell lines ( Supplementary Figure 2 ). A full list of up and downregulated genes with log₂ Fold Change ≥ 1.5, < -1.5 and -Log₁₀ p-Adj 0.05 filter is shown in Supplementary Figure 2 . Selected data is shown for PBT147 and PBT030 in Figures 4B, C, E, F , and Supplementary Figure 7 .

To see the effect of ICG-001 treatment on the protein level in glioma cell lines, PBT147 and PBT030 cells were treated with ICG-001 at 0, 5, and 10 µM for 72 h ( Figure 4 ). We collected cells and conducted global proteomic analyses after treatment ( Figure 4 ). Using 2-DICAL, we detected 1,553 common proteins expressed in both PBT030 and PBT147 lines (47). The resulting data was quantile normalized and differential protein expression was further analyzed by ANOVA. Differentially expressed proteins were subjected to pathway enrichment analysis utilizing the KEGG database. Pathways with statistically significant enrichment (p-value <0.0001) following 5 µM or 10 µM ICG-001 treatments are represented in Figure 4 . Both PBT147 and PBT030 cell lines demonstrated upregulation of pathways involved in neurogenesis, autophagy, regulation of the actin skeleton, and protein synthesis in the ER ( Figures 4A, B ). Consistent with the activation of quiescent stem-like cells, ICG-001 treatment of PBT147 and PBT030 induced metabolic reprogramming with increased ATP production, and an increase in protein translation and differentiation (48) ( Supplementary Figure 3 ). Specifically, we found increased protein expression of APOD (lipid metabolism and neuroprotection), AAAS (metabolic protein transport), MFGM (promotes phagocytosis of apoptotic cells, wound healing), PDPR (metabolism), PGM1 (metabolism-glycosylation), RRAGA (metabolism), RRAGB (regulator of TOR signaling), RTC1(metabolism), SERC, and SYMC, and downregulation of MCM6 (cell cycle, DNA replication), GSTM5 (oxidative stress, metabolism), SORCN (Proto-oncogene), FADS2 (fatty acid metabolism), and PSD7 (cytoskeletal remodeling) upon treatment of PBT147 and PBT030 with ICG-001. We postulate that these metabolic changes in tumor cells post ICG-001 treatment, leading towards a more differentiated glioma phenotype, modulate the TME and influence CAR T cell efficacy.

We anticipated that modulation of the cold TME by ICG-001 should enhance the recruitment of T cells into the TME, because upregulation of the Wnt pathway has been shown to be involved in immune evasion ( Figure 5 ) (49). To test this hypothesis, we used a subcutaneous model of syngeneic K-Luc glioma cells implanted into C57BL/6 mice. Mice received 1×10⁶ tumor cells and were either untreated or treated with ICG-001 on day 7 after tumor implantation (delivered by subcutaneous Alzet minipumps, 50 mg/kg/day) ( Figures 5A–C ). The tumors were harvested and evaluated for CD3, CD8, and CD31 by immunohistochemistry (IHC) on days 7, 14, and 21 post-Alzet minipump implantation (ICG-001 release by Alzet pump continues for 28 days post implantation) and compared to the tumors from untreated controls that were harvested on the same days ( Figures 5E–G, I–K ). We observed an increase in host immune cells, including CD3⁺ and CD8⁺ T cells, infiltrating tumors by day 14 post-ICG-001 treatment ( Figures 5D, H ). A time-dependent increase in CD31 expression was also observed, when compared with controls ( Figure 5L ).

NanoString gene expression analysis of subcutaneous tumors from ICG-001-treated mice (daily release of ICG-001 by Alzet pump is 50 mg/kg/day) versus untreated (harvested on the same days 7, 14, 21) tumors identified genes that were significantly upregulated (p ≤ 0.05), such as C9, CD83, Gpr44, Cxcr4, and others ( Figure 5M and Supplementary Table 1 ) with a demonstrated fold-change increase (>=4) such as Dmbt1, Nsr1, Irf4, Klrg1, Gzmb, and others ( Figure 5M ). On day 7 and 14 post Azlet pump implantation, we observed upregulation of tumor suppressor gene DMBT1 and the chemotactic factor CCL24, which displays chemotactic activity on resting T lymphocytes, and both GZMA and GMZB, which are associated with cytotoxic T cell activation on day 7 and 14 post Alzet pump implantation ( Supplementary Table 1 ) was observed. These results further support our hypothesis that specific downregulation of Wnt/CBP/β-catenin signaling along with reprogramming of the TME may enhance the efficacy of tumor immunotherapy, increase recruitment of host T-cells, and improve immunotherapy to solid tumors (50).

To understand the effect of ICG-001 on tumor metabolism alone and in combination with IL13Rα2–CAR T cells, we established orthotopic immunocompetent mouse models of syngeneic glioma using K-Luc glioma cells and engineered murine IL13Rα2-targeted CAR T cells (51). The murine IL13Rα2- targeted CAR T cells (mIL13BBζ CAR T cells) were characterized by FACS and contained comparable numbers of CD4+ and CD8+ T-cell subsets, with a mixture of early memory (CD62L+) and effector (CD62L−) T-cell populations, as described previously (51). We developed intracranial tumors by administration of K-luc glioma cells (0.1×10⁶) into the right frontal lobe of C57BL/6 mice. The K-luc tumor line derived from a spontaneous glioma arising from Nf1, Trp53 mutant mice (KR158) is poorly immunogenic, as indicated by its unresponsiveness to anti–PD-1 checkpoint therapy (52). This line has been further engineered to express the murine IL13Rα2 and used to recapitulate invasive glioma in syngeneic mouse models. On day 14 post-tumor implantation, mice received intratumoral administration of CAR T cells at a sub-therapeutic dose (1×10⁶) (cF11240 Mouse T cell mIL13-mCD8h-mCD8tm3-m41BB-mZeta-T2A-mCD19t(CO)_MSCV) (n=14). Treatment groups were as follows: 1) Tumor only [GR1]; 2) Tumor + ICG-001 (D16-[MI-GR2]); 3) Tumor + CAR Ts (D14–[MC-GR3]); 4) Tumor + CAR Ts (D14) and ICG-001 (D16—[MCI-GR4]) (n=14); 5) Tumor + ICG-001 (D7)+ CAR Ts (D14–[MIC-GR5]). Tumor-bearing mice in GR5 were pretreated with ICG-001 (starting on Day 7 post tumor implantation for 28 days) 1 week prior to CAR T administration [MIC-GR5]. ICG-001 was given via Alzet subcutaneous minipump delivering a dose of 50 mg/kg/day over 28 days. Otherwise, pumps were implanted on day 16, 2 days post CAR T administration (D14) in the treatment group GR4. Mice were monitored until they developed tumors larger than 0.5 cm². Normalized RNA concentration per tumor tissue (1 mg) was used for NanoString analysis. NanoString pathway analysis revealed the top genes that were up- or downregulated in CAR T vs CAR T + ICG-001 groups using a mouse immune panel to detect metabolic changes in mouse K-luc tumors and to visualize endogenous immune cells infiltration upon inhibition of Wnt with ICG-001 (nCounter_Mouse_PanCancer_Immune_Profiling_Panel). Once tumors reached the predetermined criteria for euthanasia, they were harvested, isolated, and subjected to IHC and NanoString analysis to detect differentially expressed genes in ICG + CAR T[MIC-GR5] versus CAR T + ICG-001[MCI-GR4] ( Figure 6 ). Rosalind analysis (https://www.rosalind.bio) revealed that gene expression of all CAR T, ICG001, and CAR T+ ICG-001 treated tumors were clustered together on the MDS plot, whereas ICG-001 pretreated and CAR T treated tumors demonstrated very different gene expression profiles on the MDS plot ( Figure 6A ). The volcano plot shows difference in expression of selected genes between ICG-001 + CART versus CART+ ICG-001 with the x-axis of log2 fold change and y-axis of the negative log10 of p-value. We observed downregulation of Kit, CD27A, and CCl23A, and upregulation of CCL2, VEGFC, Erbb2, Cd274, CCL22, and CSF1, contributing to improved T cell trafficking ( Figure 6B ). The full list of differentially regulated genes is presented in Supplementary Table 1 . The top up and downregulated genes are displayed in Figures 6E, F and the top biological processes affected are displayed in Figure 6D . These changes demonstrate the reprogramming of tumors pre-treated with ICG-001 and then CAR T and changes in metabolic pathways and genes, shifting to a more differentiated, immune responsive tumor.

Next, we compared CD45, Macrophages, and NK cell infiltration within tumors treated with 2 different schedules of administration of ICG-001 [CAR T + ICG001[GR4] versus ICG-001 + CAR Ts[GR5]]. We compared groups treated with ICG-001 only, CAR T only, and CAR T [D14] + ICG-001[D16] in comparison with ICG-001[D7] only or CAR T only treated groups[D14]. We found increased infiltration of CD45+ macrophages and NK-CD56dim (natural killer cells responsible for cytolytic activity and target cell killing) cells in the ICG-001+CAR T group [GR5] ( Figures 7A–D ). Cell type profiler analysis identified clearly different immune cell infiltration in the GR5 [ICG-001 + CAR T] treated group when compared with controls (GR1[ICG001], GR2 [CAR T], and GR4 [CAR T + ICG001] treated groups). We also observed an increase in the expression of the gene PTPRC encoding CD45RA that plays role in lymphocyte function, and an increase in KLRG1- cells contributing to an increase in all memory T cell lineages, including peripheral memory and tissue-resident memory cells, which should enhance long-term protective immunity (53) ( Supplementary Figure 4A ). Upregulation of the SELL (aka E-selectin CD62L) gene T_scm CD45RA⁺, IL7R⁺, CCR7⁺, CD62L⁺, KLRG^- was found within a group treated with ICG-001 first and then CAR T [ICG001 + CAR T], which is a marker associated with the T cell memory phenotype ( Supplementary Figure 4B ). We also found upregulation of IL7R- Interleukin-7 receptor (IL7R), which provides the potential for long-term survival of both CD62L^high central memory T cells and Th1 effector cells ( Supplementary Figure 4C ) (54). CCR7 is important in the migration of memory CD8+ T cells and survival in their niches ( Supplementary Figures 4D, E ).

We next tested the efficacy of a titrated dose of CAR T cells ± ICG-001 to induce tumor cell killing in vitro ( Supplementary Figure 5 ). We evaluated the combination of HER2-engineered CAR T (HER2-CAR T) cells and PBT106 tumor cells (high expressing HER2) in combination with IGC-001 using an in vitro co-culture assay. ICG-001, in combination with HER2-CAR T cells, demonstrated synergistic cytotoxicity in several patient-derived glioma lines that expressed high levels of the tumor antigen HER2 ( Supplementary Figure 5 ). ICG-001 combined with HER2-CAR T cells demonstrated synergistic killing in a CAR T cell concentration-dependent fashion ( Supplementary Figure 5A ). Enhanced killing was observed with the combination of HER2-CAR T cells and ICG-001 at effector/target cell ratios of both 1:2 and 1:4 ( Supplementary Figures 5B, C ). Furthermore, CBP/β-catenin antagonism with ICG-001 was not toxic to T cells in vitro or in vivo, providing important insights for the advancement of CAR T cell therapy combined with CBP/β-catenin antagonism for the treatment for brain tumors ( Supplementary Figure 6 ).