ARF suppression by MYC but not MYCN confers increased malignancy of aggressive pediatric brain tumors

Medulloblastoma (MB) is the most common malignant pediatric brain tumor and is stratified into WNT, SHH, Group 3, and Group 4 subgroups with distinct biology and outcomes. Group 3 MBs frequently show MYC amplification/overexpression, photoreceptor transcriptional activity, and poor survival. Although MYC and MYCN are homologous and can substitute during development, their amplifications are largely mutually exclusive across MB subgroups and may drive distinct tumor phenotypes. The ARF/p53 failsafe pathway constrains MYC-driven oncogenesis; p53/CDKN2A loss is common in high-grade gliomas but rare in MB (outside SHH). Prior work suggests MYC can suppress ARF via methylation or transcriptional repression, whereas the impact of MYCN on ARF in brain tumorigenesis is less clear. Existing MB models often use viral methods; transgenic models allow tumor initiation and evolution in an immunocompetent brain. This study develops a regulatable transgenic MYC model (GMYC) to define how MYC versus MYCN modulate ARF, how ARF status shapes tumor phenotype and malignancy, and to identify therapies exploiting MYC-driven tumors with suppressed but functional ARF.

Prior models of MYC-driven MB predominantly used orthotopic transplantation or in utero electroporation to overexpress MYC in candidate cerebellar cells of origin. A regulatable MYCN-driven model (GTML) driven by the Glt1 promoter recapitulates aggressive Group 3 biology. MYC elevation triggers apoptosis via ARF/p53 failsafe; CDKN2A (ARF/INK4A) or TP53 loss is common in high-grade glioma but rare in MB. MYC can induce ARF yet cancers often silence ARF; MYC-driven transformation can select for ARF suppression and MYC can repress cell-cycle inhibitors through MIZ1 complexes. MYCN directly inhibits p53 but its regulation of ARF during brain tumor development is unclear. Distinct MYC/MYCN chromatin interactions (e.g., MIZ1 binding) and subgroup associations have been reported in transplantation models, prompting evaluation in transgenic, immunocompetent systems.

Animal models and genetics: A Tet-OFF transgenic mouse model overexpressing human MYC (TRE-MYC) under Glt1-tTA (GMYC) was generated; crosses included TRE-CRE-LC1 (luciferase) for in vivo bioluminescence imaging and PTRE-H2BGFP for fluorescent tracking. Doxycycline (dox) controlled MYC expression in vivo. GMYC mice were further crossed with Arf floxed mice (Arffl/fl; Ink4a intact) to create heterozygous and homozygous Arf knockouts in both GMYC and GTML backgrounds. For clonality, GMYC mice were crossed with R26-Confetti, enabling lineage tracing with four fluorescent reporters upon CRE activation. Tumor monitoring and interventions: IVIS imaging following D-luciferin administration tracked tumor burden and response to dox. Dox regimens: 30-day dox diet for established tumors; embryonic/postnatal exposure cohorts (E0 to birth; P0–P30) to suppress initiation. Histology and IHC characterized tumors (markers: Ki67, cleaved caspase-3, GFAP, OLIG2, Synaptophysin, TUJ1, NPR3, OTX2, MYC). In vitro models: Primary GMYC and GTML tumor cells were cultured as neurospheres; dox was used to suppress MYC/MYCN in vitro. Lentiviral constructs introduced exogenous MYC into GTML2 cells (GTML2/+MYC) and a MYC MIZ1-binding mutant (MYCV394D) into GMYC/GTML cells. Cell viability (Alamar Blue) and apoptosis markers were assessed. Molecular profiling: RNA-seq (GMYC n=6; GTML n=7–8) with cross-species projection to human MB datasets (e.g., GSE85217) enabled subgroup assignment and GSEA/ssGSEA analyses. Methylation profiling used MM285 Infinium Mouse Methylation BeadChip; human DNA methylation (GSE85212) was analyzed for CDKN2A CpG differential methylation and copy number intensities for MYC/MYCN amplification inference. Whole-exome sequencing assessed Trp53 mutations in tumors. Differential gene expression was performed with limma/voom; cross-species metagene mapping and t-SNE/PCA/hierarchical clustering assessed similarity to human MB and HGG subclasses. Drug prediction and testing: Drug gene set enrichment from CGP signatures prioritized compounds predicted to be effective in GMYC and Group 3γ. HSP90 inhibitor Onalespib and DNMT inhibitor 5-Azacytidine were tested in vitro (dose-response, EC50) and in vivo. In vivo treatments: Onalespib 20 mg/kg IP daily for 4 days; 5-Azacytidine 0.2 mg/kg IP daily; focal cranial irradiation 2 Gy/day for 5 days. Combination therapy in vitro tested Onalespib with cisplatin (1 μM) and 5-Azacytidine with cisplatin. Human cell lines and PDX: MYC-driven Group 3 human lines (D283, sD425) and SHH/MYCN line (DAOY) were used for Onalespib testing and heat shock response validation; MB PDX expression and metastasis data were examined for CDKN2A correlations. Statistics: Kaplan–Meier survival with log-rank tests; Welch’s t-test, Wilcoxon rank-sum, ANOVA where appropriate; GSEA/ssGSEA for pathway enrichment; combination index (CompuSyn) for synergy.

• A regulatable, immunocompetent MYC-driven transgenic model (GMYC) forms hindbrain tumors resembling Group 3 medulloblastoma (MB), with 60% penetrance and 90–150 day latency; tumors are MYC-addicted and fully regress with 30-day doxycycline (no relapses; n=8 treated).
• GMYC tumors cluster with human Group 3 MB and specifically align with Group 3γ transcriptional signatures, whereas MYCN-driven GTML aligns with Group 3α and shows photoreceptor/cilium gene enrichment.
• ARF (Cdkn2a/Arf transcript) expression is markedly lower in GMYC than GTML tumors; GTML shows predominant Arf over Ink4a usage. Human data show higher CDKN2A in MYCN-high vs MYC-high when pooling Group 3/4, and higher CDKN2A associates with longer survival in Group 3/4.
• Differential methylation at the ARF promoter is frequent in MYC-amplified Group 3 patients (hypermethylation) and hypomethylated in MYCN-amplified; mouse tumors show epigenetic silencing at Cdkn2a, including upstream of Arf TSS.
• Genetic modulation of Arf: • GMYC: Arf+/− had no effect; Arf−/− increased penetrance to 95% and shortened latency to ~100 days. • GTML: Arf+/− increased penetrance from 40% to 90% and shortened latency from ~200 to ~150 days; Arf−/− reached 100% penetrance with ~80-day latency.
• Complete Arf loss shifts tumor identity: many Arf−/− tumors in both models become high-grade glioma (HGG)-like (HGG-RTK or HGG-G34), lose Group 3 MB markers (NPR3, OTX2), reduce photoreceptor pathway enrichment, and show reduced MYC/MYCN pathway association.
• Metastasis: Baseline GMYC shows rare spinal leptomeningeal spread (5%); GMYC Arf−/− increases metastasis to 45% (5/11). GTML metastasis ~10% and does not increase with Arf loss (most Arf−/− GTML tumors are HGG-like, which seldom metastasize in spinal cord).
• MYC but not MYCN mediates de novo ARF suppression: In GTML2 cells with ectopic MYC, dox-mediated MYCN suppression increases MYC and reduces ARF by day 7; MYC’s ARF suppression is independent of MIZ1 binding (MYCV394D still suppresses ARF).
• DNMT inhibition: 5-Azacytidine restores ARF in GMYC1 but not GTML2; affects viability mainly at 1 μM; no significant synergy with cisplatin and no in vivo survival benefit versus control.
• HSP90 pathway targeting: HSP90AB1 correlates with MYC and poorer survival. HSF1-mediated heat-shock response is enriched specifically in MYC-high/CDKN2A-low Group 3 patients. Onalespib (HSP90 inhibitor) shows selective sensitivity in MYC-driven GMYC cells (EC50 ~0.05 μM) versus GTML2 (~2 μM), induces HSF1/HSP70, restores ARF, elevates p21 and cleaved caspase-3, and reduces viability. Onalespib plus cisplatin is synergistic in GMYC in vitro.
• In vivo, Onalespib prolongs survival of GMYC allograft-bearing mice (median 27 vs 16 days, p=0.013) versus vehicle; 5-Azacytidine shows no benefit; irradiation trend not significant. Onalespib fails to extend survival in ARF-deficient MYC tumors, indicating ARF dependence.
• Human validation: MYC-driven, CDKN2A-present MB lines (D283, sD425) are more sensitive to Onalespib than DAOY (SHH/MYCN); treatment activates HSF1/HSP70 without degrading HSP90, consistent with on-target heat-shock response.

This work demonstrates that MYC and MYCN differentially engage the ARF/p53 axis during brain tumorigenesis. In an immunocompetent, regulatable transgenic context, MYC-driven Group 3-like MBs exhibit pronounced ARF suppression, whereas MYCN-driven tumors retain higher ARF expression and photoreceptor signatures. Modulating ARF reveals its pivotal role in determining malignancy and lineage: partial or complete ARF loss accelerates tumorigenesis and, when complete, often redirects tumors toward HGG-like states with reduced MYC/MYCN pathway activity and loss of photoreceptor programs. These findings align with clinical observations that CDKN2A loss is common in pediatric HGG but rare in MB, and that higher CDKN2A expression correlates with improved survival in Group 3/4. Therapeutically, the study identifies HSP90 inhibition as a strategy that preferentially targets MYC-driven tumors with suppressed yet functional ARF. Onalespib restores ARF levels and triggers a robust HSF1/HSP70 stress response, p21 induction, apoptosis, and synergizes with cisplatin in MYC-driven settings. The dependency on ARF for Onalespib efficacy underscores a mechanistic link between HSP90 signaling and ARF stabilization. Conversely, broad DNMT inhibition can raise ARF in vitro but lacks specificity and clinical utility in this context. Overall, exploiting the MYC–ARF interplay enables rational selection of therapies for MYC-driven Group 3 MBs, with potential for combination regimens that include HSP90 inhibition.

A regulatable MYC transgenic model (GMYC) faithfully mirrors Group 3 medulloblastoma biology and reveals that MYC, unlike MYCN, suppresses ARF to promote aggressive disease. ARF status governs both malignancy and tumor identity: complete ARF loss accelerates disease and frequently yields HGG-like tumors with diminished photoreceptor and MYC pathway activity. Therapeutically, HSP90 inhibition with Onalespib selectively targets MYC-driven, ARF-competent tumors, restores ARF, induces apoptotic programs, synergizes with cisplatin, and extends survival in vivo. These insights nominate ARF status and MYC/MYCN context as biomarkers for therapy selection and support clinical evaluation of HSP90 inhibitors, potentially as radiosensitizers or in combination with standard chemotherapy, in MYC-driven Group 3 MB. Future work should validate ARF/MYC biomarkers prospectively in patient cohorts, optimize dosing and combinations (e.g., with radiation or DNA-damaging agents), dissect the molecular basis of HSP90–ARF stabilization, and explore strategies to reverse MYC-driven ARF silencing with greater specificity than global demethylation.

The baseline GMYC model exhibits low leptomeningeal metastasis relative to human Group 3 MB, potentially limiting modeling of dissemination; metastasis increases only upon complete Arf loss. Some analyses in human datasets are constrained by limited sample numbers when segregating MYC-high versus MYCN-high within Group 3 alone, and there is a lack of public data to distinguish ARF versus INK4A isoform-specific expression as compensation for promoter methylation. DNMT inhibition is non-specific and produced limited therapeutic benefit in vivo. Trp53 mutations occurred in a subset of tumors in both models, complicating interpretation of p53 pathway function. The in vivo Onalespib regimen was brief and tested in allografts; broader dosing schedules, pharmacokinetics, and combinations need evaluation. Findings derive from mouse models and selected cell lines/PDXs, which may not capture full patient heterogeneity.