Abstract Background: Medulloblastoma (MB) is the most common malignant pediatric brain tumor derived from transformed neuroepithelial stem (NES) cells, and is comprised of four main subgroups: WNT, SHH, Group 3, and Group 4. SHH MB tumors with amplification of MYCN and GLI2 and mutation of TP53 (MGT) are resistant to inhibitors of the SHH activator, SMO, and exhibit extremely poor prognosis. Given the poor druggability of MYCN and GLI2, we sought to identify new therapeutic targets by performing a drug screen. However, there are no reliable MGT MB cell lines that grow in culture. To circumvent this issue, we utilized a human induced pluripotent stem cell (iPSC) model of MB to generate MGT MB tumors in vivo which can subsequently be cultured in vitro to perform therapeutic screens along with an isogenic NES cell control line. Method: iPSCs from a healthy adult were differentiated towards NES cells, transduced with doxycycline-inducible MYCN and GLI2 and constitutive expression of TP53R248Q and implanted orthotopically into immunocompromised (NSG) mice. Resulting MGT tumors and control (empty vector) NES cells were analyzed by RNAseq and proteomics. MGT tumors were cultured in vitro and, along with control NES cells, were subjected to a drug screen (360 unique compounds from the SBP Oncology Dose Library (ODL3), and CTD2 Informer Set). MGT MB PDX tumors were maintained in NSG mice and analyzed for cell viability in response to drug treatments in 384 well plates with CellTiter-Glo cell viability assay. Results: MGT NES cells were fully penetrant within 35 days post injection in NSG mice. MGT tumor lines had transcriptomes similar to MB SHH subgroup and were resistant to in vitro treatment with chemotherapy drugs cisplatin and cyclophosphamide. The drug screen showed MGT tumors were more sensitive than control NES cells to homoharringtonine (HHT, inhibitor of translation elongation), MLN4924 (inhibitor of NEDD8 Activating Enzyme), and PI-103 (inhibitor of PI3K & mTOR). With three separate MGT MB PDX tumors, MLN4924 and PI-103 failed to reduce viability. In contrast, HHT was effective at reducing viability in all three MGT PDX tumors in a dose dependent manner and IC50s are in the nanomolar range (29.61nM/31.10nM/7.319nM)). Proteomic analysis of MGT NES cell tumors and control NES cells untreated or treated with HHT reveal that two of the 10 proteins most downregulated by HHT in MGT tumors but not control cells, are MYCN and GLI2. In support of this, HHT (1.52nM) induced lower viability in tumors with MYCN and GLI2 (+dox, 50%)) than tumors without (-dox, 93%). Conclusion: Here, we show for the first time that the human stem cell model of MB can be used to identify therapeutic targets that are effective against MB PDX tumors. Specifically, targeting translation elongation is a potential therapeutic target to treat MGT MB tumors. Citation Format: Bo Cheng, Jamie Vu, Darren Finlay, Shujun Cheng, Theophilos Tzaridis, Maria Poblete, Soumik Saha, Till Milde, Qianqian Li, Martine F. Roussel, Kristiina Vuori, John R. Prensner, William A. Weiss, Robert J. Wechsler-Reya, Miller Huang. Amplified MYCN, GLI2 and mutant TP53 medulloblastoma are vulnerable to the translation inhibitor homoharringtonine [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB495.

SWI/SNF chromatin remodeling complexes are perturbed in 20% of all cancers and in several developmental disorders, yet the mechanisms by which these mutations dysregulate transcription and drive disease are poorly understood. To both elucidate these mechanisms and identify vulnerabilities caused by these mutations, we leverage genome-wide CRISPR-Cas9 screening in hundreds of cancer cell lines and identify the chromatin reader protein PHIP as a specific dependency in cancers with broadly disrupted SWI/SNF function. Mechanistically, we reveal that PHIP cooperates with SWI/SNF to facilitate transcriptional activation by ubiquitinating and suppressing subunits of the repressive Nucleosome Remodeling and Deacetylase (NuRD) complex. We demonstrate that loss of SWI/SNF results in NuRD complexes accumulating at promoters where they would otherwise cause widespread transcriptional silencing if not antagonized by PHIP. Collectively, we identify PHIP as a regulator of the interplay between distinct chromatin regulators that function in development and disease and as a targetable vulnerability in cancers with broad SWI/SNF inactivation.

Abstract The MYC oncoprotein promotes immune evasion of pancreatic ductal adenocarcinoma (PDAC), but the underlying molecular mechanisms are not fully understood. Here we show that MYC protects PDAC tumors from CD4 + T cell-dependent elimination. Single cell sequencing shows that MYC suppression in tumor cells increases amino acid availability and broadly activates amino acid-responsive gene expression programs in immune cell populations. This occurs because MYC-driven uptake depletes free amino acids from tumor interstitial fluid and plasma, while MYC compromises macropinocytosis and autophagy, both of which depend on lysosomal protein degradation. MYC engages the POZ/BTB transcription factor MIZ1 to suppress lysosomal genes regulated by the TFE3/TFEB/MITF network or by free MIZ1, thereby inhibiting lysosomal protein degradation. An orthogonal genetic model enabling transient, selective inhibition of amino acid uptake in tumor cells recapitulates the effects of MYC depletion on amino acid levels in the tumor microenvironment and induces complete, CD4 + T cell-dependent tumor eradication with long-term survival. We propose that MYC-mediated, cell-autonomous disruption of lysosome function coupled to non-cell-autonomous protection from immune clearance allows MYC-low cells to benefit from MYC-high neighbors, such that intratumoral heterogeneity in MYC expression confers a selective advantage to the entire tumor. Abstract Figure

PVT1 promoter and MYC transcription in ecDNA-amplified G3-MB.

Genome-wide Hi-C sequencing displays ecMYC E1 looping to the MYC promoter.

MYC is amplified on extrachromosomal DNA (ecDNA) or homogeneously staining regions (HSR) in group 3 medulloblastoma (G3-MB), conferring a poor prognosis. A better understanding of the mechanisms underlying MYC expression in ecDNA and HSRs could be leveraged to develop improved treatments for G3-MB. Using a structure-function approach, we identified and characterized an enhancer (ecMYC E1) that drives MYC activation specifically in G3-MB with MYC-amplified ecDNA or HSRs. The ecMYC E1 locus exhibited enhancer hallmarks exclusively in MYC-amplified G3-MB but not in other MYC-dependent cancer cell lines, including those with MYC amplification. Silencing of the ecMYC E1 enhancer significantly reduced MYC transcription, which was compensated for by increases in ecDNA copy number. Neuronal differentiation 1 and bromodomain-containing protein 4 interacted with each other and bound to ecMYC E1, looping the enhancer to the MYC promoter. Together, these findings define a mechanism that regulates amplified MYC gene expression within ecDNA or HSRs specifically in G3-MB.Significance: Combined structural and functional approaches identified a conserved enhancer activating MYC transcription exclusively in MYC-amplified ecDNA- and HSR-positive group 3 medulloblastoma, providing a potential therapeutic target for suppressing MYC.

Extensive molecular analyses by many groups have revealed a heterogenous landscape of embryonal brain tumors, but patient-derived tumor organoid (TO) model development has remained limited. Here, we describe the establishment of TO and TO xenografts (TOX) from patient-derived orthotopic xenografts (PDOXs) of medulloblastoma, embryonal tumor with multilayer rosettes, and atypical teratoid rhabdoid tumors. DNA methylation, bulk- and single-cell RNA sequencing, and whole-genome sequencing demonstrated that TOs and TOXs faithfully recapitulate the epigenetic, transcriptomic, and genetic landscape of PDOXs, as well as replicate the intratumor cellular heterogeneity of PDOXs that are often lost in established cell lines. We show that TOs and PDOXs have similar drug responses. The development of embryonal brain TOs will facilitate in vitro functional assays, including high-throughput drug or CRISPR screens, without the need for fresh tumors from tumor-bearing mice. This will accelerate the identification and validation of vulnerabilities and therapeutic strategies for preclinical testing toward clinical trials.

H3K4me1 ChIP-seq in mouse ESCs and NPCs at the MYC locus.

BRD4 and NeuroD1 bind ecMYC E1 to activate MYC transcription. A–C, RT-qPCR for MYC mRNA transcription following 100 nmol/L SJ432 treatment in the (A) D425 and (B) D458 cell lines and the (C) SJMB016880 tumor organoid. D, CUT&RUN targeting BRD4 at the ecMYC E1 enhancer locus in the D425 (blue), D458 (purple), SJMB016880 (red), and HDMB03 (orange) G3-MB models. E, RT-qPCR for MYC mRNA transcription following NeuroD1 silencing in the D458 cell line. F and G, ChIP-seq targeting NeuroD1 at the (F) ecMYC E1 enhancer locus or (G) MYC promoter in the D425 (top row), D458 (second row), SJMB016880 (third row), and HDMB03 (bottom row) G3-MB cell models. H, ChIP-seq targeting NeuroD1 at the ecMYC E1 enhancer locus in the D458 cell line at DT3 after silencing (sg101 or sg102) or by nontargeting control (AAVS1). I, Log2 fold change (FC) in NeuroD1 occupancy at the ecMYC E1 enhancer at DT3 following silencing compared with the AAVS1 nontargeting control. J and K, Co-IP pulldown for (J) BRD4 or (K) NeuroD1. Lamin A/C: noninteracting control. L, Proposed mechanism for ecMYC E1 activation of MYC transcription through the binding of BRD4 and NeuroD1 at the enhancer and looping to the MYC promoter in MYC-amplified G3-MB. *, P <0.05; **, P <0.005; ***, P <0.0005; ****, P <0.00005. Created in BioRender. Friske, J. (2026) https://BioRender.com/9hm4nmn.

List of sgRNA sequences for the CRISPR inhibition screen of D425.

B7‑H3 is a promising immunotherapeutic target for pediatric brain tumors, yet the durability of B7‑H3-CAR T cell responses remains limited. Here, for the first time, we systematically evaluated whether knockout (KO) of TET2, DNMT3A, or Regnase‑1 (REG-1) enhances the antitumor activity of B7‑H3–CAR T cells. Across multiple G3MB orthotopic xenograft models spanning low to high antigen density, second‑generation B7‑H3–CAR T cells improved survival but failed to achieve curative responses, with efficacy influenced by donor‑specific variability and antigen density. While DNMT3A-KO enhanced CAR T cell persistence in vitro, neither TET2 nor DNMT3A-KO improved tumor control in vivo. In contrast, REG‑1-KO significantly augmented antitumor activity, inducing complete tumor eradication in a subset of mice and extending survival across both potent and poor‑performing donors. Single-cell transcriptomic and functional analyses show increased cytokine production, coordinated CD4–CD8 communication, and reduced exhaustion-associated programs, supporting a model in which effective responses depend on CD4-mediated immune orchestration rather than persistence or cytotoxicity alone. We also demonstrate CAR specificity: while DNMT3A and Regnase-1 disruption both enhance EphA2-targeted CAR T cells, only Regnase-1 disruption improves B7-H3 CAR T-cell efficacy, underscoring the need for context-specific engineering. Our findings challenge the prevailing assumption that persistence-enhancing strategies are central to successful CAR T cell therapies. Instead, they demonstrate that distinct engineering approaches should be considered for the specific disease and CAR target.

List of primary and secondary antibodies.

Abstract MYC is amplified on extrachromosomal DNA (ecDNA) or homogenously staining regions (HSRs) in Group 3 medulloblastoma (G3-MB), conferring a poor prognosis, but the underlying mechanisms controlling MYC expression within ecDNA and HSRs are poorly understood. Using a structure-function approach, we identified and characterized a novel enhancer (ecMYC E1) that drives MYC activation specifically in G3-MB with MYC-amplified ecDNA or HSRs. The ecMYC E1 locus exhibits enhancer hallmarks exclusively in MYC-amplified G3-MB but not in other MYC-dependent cancer cell lines, including those with MYC amplification. Silencing of the ecMYC E1 enhancer significantly reduced MYC transcription, which was compensated by increases in ecDNA copy number, but not in HSR-driven G3-MB tumor. NeuroD1 and BRD4 interact with each other and bind to ecMYC E1, looping to the enhancer to the MYC promoter, and defining a novel mechanism that regulates amplified MYC gene expression within ecDNA or HSRs specifically in G3-MB. Citation Format: Jake D. Friske, Flore Cuisin, Paloma Guernalec, Hayden Malone, Stephanie Nance, Declan Bennett, Steven Burden, Ti-Cheng Chang, Hao Shi, Justin S. Williams, Virginia Valentine, Barbara Passaia, Bensheng Ju, Modupeore Adetunji, Paul Geeleher, Brian J. Abraham, Gang Wu, Chunliang Li, Martine F. Roussel. A conserved enhancer locus in ecDNA and HSRs activates MYC transcription in group 3 medulloblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB493.

BACKGROUND:Ependymoma (EPN) is the third most common pediatric brain tumor with no targeted therapies available to patients. In supratentorial ependymoma, the most frequent driver alteration is a gene fusion between ZFTA and RELA (denoted ZR), leads to constitutive localization of ZR in the nucleus. Because ZR is not currently druggable, we tested whether ZR expression leads to aberrant protein interactions that could represent therapeutic vulnerabilities. METHODS:Using CRISPR-Cas9 pooled screening, we identified many novel druggable ZR interacting proteins including XPO1, CARM1, SMARCA4, and CDK1. We focused on the nuclear export protein (XPO1), given the ability of most XPO1 inhibitors (i.e. Selinexor) to cross the blood brain barrier, FDA approval, and documented safety profiles in children. RESULTS:We found that specific nuclear ZR levels are needed for cell proliferation and are regulated by XPO1. Increased ZR accumulation in the nucleus does not increase oncogenic gene expression but drives tumor cells out of cell cycle, as compared to a defective ZR DNA binding mutant. Treatment of ZR driven patient-derived mouse models with Selinexor impairs cell growth and extends survival of animals in vivo. The combination of Selinexor treatment with Gemcitabine and Ribociclib (used in a clinical trial for relapsed EPN at St Jude Children's Research Hospital (SJDAWN)) further extends mouse survival. CONCLUSION:Our findings demonstrate that ZR interacting proteins constitute therapeutic leads, and that XPO1 is critical for titrating 'goldilocks' levels of ZR nuclear expression. We identify a novel combination therapy of Selinexor, Gemcitabine, and Ribociclib that may be immediately translated into clinical trials for EPN patients that are currently without targeted treatments.

Abstract Rampant genomic instability of osteosarcoma (OS) and associated inter- and intratumoral heterogeneity convey a high risk of metastasis despite contemporary multimodal therapy. A mouse primary OS tumor model, originating from Myc-overexpressing Trp53 -null mesenchymal progenitor cells, closely mimics cardinal genetic features and gene expression patterns of human OS samples. Subcutaneous injection of OS cells into immunocompromised NSG mice produced extensive metastases, whereas many fewer metastases occurred in T cell-deficient nude mice, suggesting a principal role for innate immunity in controlling OS dissemination. Depletion of natural killer (NK) cells in nude mice facilitated OS metastasis. OS cells released a suite of chemokines, with CCL2 the most prominent. A genome-wide CRISPR/Cas9 screen in OS cells identified 11 genes, including Ccl2 , whose loss facilitated pulmonary metastasis in nude mice. Disruption of CCL2 in OS cells partially phenocopied the effects of antibody-dependent NK cell depletion, underscoring a plausible OS-NK signaling pathway that limits OS metastasis. Significance Osteosarcoma exhibits complex genomic instability and a propensity for pulmonary metastasis that limit chemotherapeutic response and patient survival. Despite the plethora of heterogeneous genetic alterations that connote poor prognosis, a novel preclinical in vivo model for studying metastasis highlights potentially targetable signaling between osteosarcoma cell-derived chemokines and pulmonary NK cells.

Abstract Extrachromosomal DNA (ecDNA) is an emergent prognostic marker for poor disease outcome in a variety of cancer types, including pediatric medulloblastoma, where MYC is commonly amplified on ecDNA in Group 3 tumors (G3-MB). Tumors with ecDNA have higher recurrence and relapse rates with current therapies, highlighting a need for more targeted therapeutic intervention; however, the underlying mechanisms controlling MYC expression from ecDNA are still poorly understood. A structure-function investigation identified and characterized a conserved enhancer that drives MYC activation specifically in G3-MB with MYC amplified on ecDNA or homogenous staining regions (HSRs). This enhancer, termed ecMYC E1, exhibits enhancer hallmarks of open chromatin and H3K27ac enrichment exclusively in MYC-amplified G3-MB but not in other MYC-dependent cancer cell lines, including those with and without MYC amplification on ecDNA or HSRs. This provides a tumor-specific regulatory mechanism for oncogenic gene transcription. Silencing of the ecMYC E1 enhancer locus significantly reduced MYC transcription acutely, but this reduction in oncogene expression was compensated for by an increase in ecDNA copy number during prolonged enhancer silencing. This compensation was not observed in HSR-driven G3-MB. The increase in ecDNA copy number resulted in tumor cells reverting back to baseline levels of MYC transcription; therefore, ecDNA copy number could be used as a biomarker for therapy evasion or adaptation. Mechanistically, we showed that the transcription factor NeuroD1 and epigenetic regulator BRD4 interact with each other and bind to ecMYC E1 in G3-MB. This study defines a novel mechanism for ecDNA or HSR regulation of amplified MYC gene expression exclusively in G3-MB. Citation Format: Jake D. Friske, Flore Cuisin, Paloma Guernalec, Hayden A. Malone, Stephanie Nance, Declan Bennett, Steven Burden, Ti-Cheng Chang, Hao Shi, Justin S. Williams, Virginia Valentine, Barbara Passaia, Bensheng Ju, Modupeore Adetunji, Paul Geeleher, Brian J. Abraham, Gang Wu, Chunliang Li, Martine F. Roussel. A conserved enhancer locus in ecDNA and HSRs activates MYC transcription in Group 3 Medulloblastoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(6_Suppl):Abstract nr A047.

Abstract SWI/SNF chromatin remodeling complexes are perturbed in 20% of all cancers and in several developmental disorders, yet the mechanisms by which these mutations dysregulate transcription and drive disease are poorly understood. To both elucidate these mechanisms and identify vulnerabilities caused by these mutations, we leveraged genome-wide CRISPR-Cas9 screening in hundreds of cancer cell lines and identified the chromatin reader protein PHIP as a specific dependency in cancers with broadly disrupted SWI/SNF function. Mechanistically, we reveal that PHIP cooperates with SWI/SNF to facilitate transcriptional activation by ubiquitinating and suppressing subunits of the repressive Nucleosome Remodeling and Deacetylase (NuRD) complex. We demonstrate that loss of SWI/SNF results in NuRD complexes accumulating at promoters where they would otherwise cause widespread transcriptional silencing if not antagonized by PHIP. Collectively, we identify PHIP as a regulator of the interplay between distinct chromatin regulators that function in development and disease and as a targetable vulnerability in cancers with broad SWI/SNF inactivation. Citation Format: Hayden A. Malone, Jacqueline A. Myers, Emma G. Gruss, Marc A. Morgan, Jake D. Friske, Tabitha C. McCarty, John J. Navarro, Sarah Robinson, Rebecca L. Halliburton, Sandra J. Kietlinska, Francisca N. De Luna Vitorino, Baranda S. Hansen, Shondra M. Pruett-Miller, Benjamin A. Garcia, Martine F. Roussel, Janet F. Partridge, Charles W. Roberts. PHIP suppresses NuRD to enable the growth of SWI/SNF-mutant cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB093.