Background. Despite recent advances in the biology of IDH-wildtype glioblastoma, it remains a devastating disease with median survival of less than 2 years. However, the molecular underpinnings of the heterogeneous response to the current standard-of-care treatment regimen consisting of maximal safe resection, adjuvant radiation, and chemotherapy with temozolomide remain unknown. Methods. Comprehensive histopathologic, genomic, and epigenomic evaluation of paired initial and recurrent glioblastoma specimens from 106 patients was performed to investigate the molecular evolution and cellular phenotypes underlying differential treatment responses. Results. While TERT promoter mutation and CDKN2A homozygous deletion were early events during gliomagenesis shared by initial and recurrent tumors, most other recurrent genetic alterations (eg, EGFR, PTEN, and NF1) were commonly private to initial or recurrent tumors indicating acquisition later during clonal evolution. Furthermore, glioblastomas exhibited heterogeneous epigenomic evolution with subsets becoming more globally hypermethylated, hypomethylated, or remaining stable. Glioblastoma that underwent sarcomatous transformation had shorter interval to recurrence and were significantly enriched in NF1, TP53, and RB1 alterations and the mesenchymal epigenetic class. Patients who developed somatic hypermutation following temozolomide treatment had significantly longer interval to disease recurrence and prolonged overall survival, and increased methylation at 4 specific CpG sites in the promoter region of MGMT was significantly associated with this development of hypermutation. Finally, an epigenomic evolution signature incorporating change in DNA methylation levels across 347 critical CpG sites was developed that significantly correlated with clinical outcomes. Conclusions. Glioblastoma undergoes heterogeneous genetic, epigenetic, and cellular evolution that underlies prognostically different treatment responses.

2067 Background: Isocitrate dehydrogenase mutant (IDHmut) glioma affects over 20,000 adults per year. Treatment failure often results from intratumoral heterogeneity, in which genomic subclones undergo selection for treatment-resistant populations. Understanding this heterogeneity is critical to selecting therapeutic combinations that are efficacious across the entire tumor, but few genomic studies go beyond analyzing a single tumor sample per patient. Methods: 3D whole tumor sampling was used to obtain 324 spatially mapped samples from 32 IDHmut gliomas. Whole exome (n=323), RNA (318), ATAC sequencing (92), and Hi-C (14) were performed. PyClone reconstructed clonal evolution, weighted gene/peak correlation network analyses derived co-expression (RNA) and gene regulatory (ATAC) programs, and comparison of tumor Hi-C results to normal brain identified tumor-specific chromatin loops. Results: Regional tumor content (purity) inversely correlated with distance from tumor centroid ( R = -0.32, p = 2.8×10 -8 ) and inter-sample distance was associated with subclonal ( R = 0.24, p = 8.1x10 -15 ) and transcriptional ( R = 0.12, p = 1.0x10 -5 ) similarity. Beyond IDH1 in the combined 17 astrocytomas and 15 oligodendrogliomas, tumor-wide alterations were recurrently observed in TP53 (altered in 64%; tumor-wide in 58%), ATRX (55%; 39%), TERT promoter (45%; 45%) and FUBP1 (18%; 6%). Despite high cohort prevalence, glioma-associated drivers CIC (altered in 42%), PIK3CA (36%), and ARID1A (18%) were tumor-wide in only one patient each, while MUC4 (33%), NOTCH1 (27%), SETD2 (24%), and PIK3R1 (15%) were never observed in the tumor founding clone. Tumors with CIC alterations harbored a median of 5 unique mutations with a median combined tumor-wide cancer cell fraction of 45%. Hi-C identified 5,579 tumor-specific promoter-enhancer interactions across 4,144 loops. The genes associated with these promoters enriched for tumor-specific RNA programs associated with OPC signature ( p adj = 3x10 -14 ), NPC signature ( p adj = 3.87x10 -14 ), and neuronal mimicry (1.2×10 -13 ). We discovered adversely prognostic gene expression programs in astrocytoma independently validated in chromatin regulatory programs. Conclusions: Tumor-wide sampling of IDHmut glioma revealed evidence of radial tumor growth and 3D patterns of cellular organization, in sharp contrast to intermixed patterns in IDH wildtype glioma. Most inter-patient heterogeneity in driver alterations was subclonal. Our data also revealed regulatory loops that explain tumor cell-specific expression signatures. A significant, adverse prognostic marker in astrocytoma was discovered and cross-validated.

Meningiomas are the most common primary intracranial tumors and the only brain tumors that are more common in females compared with males1. Progestin hormonal therapies increase the risk of meningioma, and progestin-induced or pregnancy-associated meningiomas can regress as serum progestogen levels normalize2. The mechanisms that underlie sex differences and progestogen signaling in meningioma are unknown. Here we show that sex hormone interaction with PGRMC1, a transmembrane progesterone binding protein, regulates the activity of RNA processing proteins FXR1 and RBM39 to drive meningioma sex differences. The genomic architecture and stem cells underlying meningiomas are conserved across vertebrate species3-5 and, using mass spectrometry-based proteomics to analyze 703 meningioma and meningeal samples, we demonstrate that meningiomas are enriched in RNA processing proteins in humans and dogs. Interactions between PGRMC1, FXR1, and RBM39 are inhibited by progestogens and stabilized by testosterone. After release from PGRMC1, FXR1 and RBM39 bind and stabilize progesterone receptor (PR) transcript to enable expression of PR protein, which induces cell cycle, membrane, and cytoskeleton remodeling genes that drive tumor growth. These findings reveal therapeutic strategies and a PR target gene biomarker that may improve outcomes for patients. More broadly, we elucidate an estrogen receptor-independent mechanism of PR expression that underlies sex differences in cancer.

Abstract Background: The low mutational burden and high heterogeneity of gliomas limit the efficacy of immunotherapy. Aberrant RNA splicing can generate targetable neojunctions (NJs) (Kwok et al., 2025), yet the impact of therapy and malignant transformation on this landscape remains unclear. We investigate these effects in IDH-mutant gliomas to identify conserved, upregulated, and potentially targetable splice-derived neoantigens. Methods: Paired bulk RNA-seq data from primary and recurrent IDH-mutant gliomas were obtained from the UCSF Brain Tumor Center patient data and AG-120-treated glioma cell line data (Wu et al., Science 2025). NJs and their subsequent peptide sequences were characterized using the SSNIP pipeline (Kwok et al., Nature 2025). High confidence n-mer sequences were prioritized using HLAthena, MHCflurry, NetMHCpan, MUNIS, and HLApollo across five HLA-I alleles (A0101, A0201, A0301, A1101, A2402). Peptides scoring within the top first percentile on ≥ 3 platforms were filtered by FPKM × splice read frequency. DESeq2 and GSEA were used for differential gene expression analyses. Publicly available eCLIP sequencing near NJ sites identified potential cis-binding RNA-binding regulators. Results: In patients treated with standard chemoradiation (n = 90), 1,806 of 57,400 NJs (3.1%) were significantly upregulated (log2[JPM] > 1; p < 0.05). In AG120-treated IDH-mutant glioma cell lines, 90 of 5,520 NJs (1.6%) were significantly upregulated versus non-treated controls. Across paired samples, NJ expression increased at recurrence (p = 0.033), a trend absent with AG120-treated cell lines (p = 0.7). NJ elevation was primarily associated with grade progression and differential gene expression analysis showed enrichment of intron-recognition/exclusion pathways. Further analysis of the chemoradiation cohort identified BCAN, PTPRZ1 and EEF1A1 NJ-derived targets as top HLA-A*11:01, HLA-A*24:02, HLA-A*11:01 candidates, respectively. Notably, the PTPRZ1 and EEF1A1 NJs were intratumorally conserved, and were more abundant than the immunogenic GNAS NJ reported by Kwok et al. Two upregulated AG120 NJs overlapped with those from the chemoradiation cohort and generated proteins. These two NJs generated distinct cancer 9-mers from S100A13 and RANBP2, which were recurrently expressed and strongly predicted to bind HLA-A*02:01. Leading-edge analysis showed IGF2BP3 upregulation, and eCLIP confirmed binding sites near the S100A13 and RANBP2 NJ sites. IGF2BP3 knockdown in HepG2 cells caused distinct splicing alterations without global loss, while S100A13 NJ expression decreased significantly post-KD. Conclusion: We show that therapy and malignant progression remodel the splicing landscape of IDH-mutant gliomas via altered splicing regulator expression. Moreover, this remodeling produces conserved, tumor-wide, and putatively immunogenic NJs. Citation Format: Tim Wu, Darwin Kwok, Joseph Costello, Hideho Okada. Therapy and malignant progression reshape the splicing landscape to generate shared, tumor-wide neoantigens in IDH-mutant gliomas [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 462.

Ionizing radiation (IR) is part of standard care for group 3 (G3) medulloblastomas (MBs), the most aggressive molecular subtype of the most common malignant pediatric brain cancer. However, IR is often not curative and causes debilitating neurological side effects. There is a critical need for adjuvants to make IR more effective. We screened in-house (N=57) and public (N=375) single-cell and bulk RNA and epigenetic datasets for correlates of G3 MB progenitor cell types. We identified a telomere maintenance signature driven by zinc finger and scan domain 4 (ZSCAN4) that is upregulated in cycling progenitors vs. neuronal-like G3 MB cells. ZSCAN4 positively regulates telomere elongation and genomic stability in early embryos. Our data indicate that ZSCAN4 plays a similar role in G3 MBs. In particular, we found that ZSCAN4 and its targets are upregulated in embryonal brain tumors vs. diffuse gliomas, and in recurrent G3 MBs vs. matched primary cases. ZSCAN4 positively regulated telomere lengths in G3 cell lines. CRISPR inhibition of ZSCAN4 led to significant DNA damage, telomeric DNA damage, decreased tumor proliferation, and significantly increased survival and sensitivity to IR in vivo. ZSCAN4 expression was significantly reduced, and the survival of orthotopic xenografts significantly increased, by repurposing oral FDA-approved brain-penetrant drugs, with no toxicities. Our studies indicate that an embryonal mechanism of telomere maintenance and genomic stability mediated by ZSCAN4 is utilized by embryonal brain tumors and can be targeted for therapeutic benefit.

Astrocytomas and oligodendrogliomas are slow-growing and treatment-sensitive IDH-mutant gliomas diagnosed at ages 30-50. Local tumor regrowth and treatment resistance is inevitable resulting in 3-10 year astrocytoma and up to >20 years oligodendroglioma survival. We sought to identify genetic changes associated with tumor evolution in response to therapy through multi-timepoint whole-genome/whole-exome sequencing of 206 IDH-mutant glioma patient samples collected through the Glioma Longitudinal Analysis (GLASS) Consortium. We validated known genomic markers of tumor progression, including hypermutation and CDKN2A homozygous deletion, and discovered novel genetic alterations that distinguish the response to treatment in astrocytomas compared to oligodendrogliomas. Point mutations in PIK3CA , PIK3R1 , and NOTCH1 were newly acquired in recurrent oligodendrogliomas and associated with increased mutation rates. Focal oncogene amplifications, together with CDKN2A homozygous deletions, were associated with an increase in recurrence-specific chromosomal imbalances in astrocytomas. Mutational signature analysis revealed additional differences and detected enrichment for the SBS11, and SBS119 mutational signatures after temozolomide treatment in both IDH-glioma subtypes, whereas astrocytomas showed increased ID8 signatures after radiotherapy. These signatures suggest that the genomes of oligodendroglioma and astrocytoma adapt to the selective pressures of tumor progression and treatment in different ways. However, in both IDH-mutant glioma subtypes we observed a convergence of acquired driver gene alterations with genome-wide changes and worse patient outcomes, signaling selection of treatment-refractory clones. By identifying new prognostic markers and delineating the genomic divergence of oligodendrogliomas and astrocytomas after diagnosis, our results suggest that different DNA damage response mechanisms are engaged following chemo- and radiation therapy.

Glioblastoma (GBM) is the most prevalent malignant brain tumor with poor prognosis. Although chromatin intratumoral heterogeneity is a characteristic feature of GBM, most current studies are conducted at a single tumor site. To investigate the GBM-specific 3D genome organization and its heterogeneity, we conducted Hi-C experiments in 21 GBM samples from nine patients, along with three normal brain samples. We identified genome subcompartmentalization and chromatin interactions specific to GBM, as well as extensive intertumoral and intratumoral heterogeneity at these levels. We identified copy number variants (CNVs) and structural variations (SVs) and demonstrated how they disrupted 3D genome structures. SVs could not only induce enhancer hijacking but also cause the loss of enhancers to the same gene, both of which contributed to gene dysregulation. Our findings provide insights into the GBM-specific 3D genome organization and the intratumoral heterogeneity of this organization and open avenues for understanding this devastating disease.

T cell-based immunotherapies hold promise in treating cancer by leveraging the immune system's recognition of cancer-specific antigens1. However, their efficacy is limited in tumours with few somatic mutations and substantial intratumoural heterogeneity2-4. Here we introduce a previously uncharacterized class of tumour-wide public neoantigens originating from RNA splicing aberrations in diverse cancer types. We identified T cell receptor clones capable of recognizing and targeting neoantigens derived from aberrant splicing in GNAS and RPL22. In cases with multi-site biopsies, we detected the tumour-wide expression of the GNAS neojunction in glioma, mesothelioma, prostate cancer and liver cancer. These neoantigens are endogenously generated and presented by tumour cells under physiologic conditions and are sufficient to trigger cancer cell eradication by neoantigen-specific CD8+ T cells. Moreover, our study highlights a role for dysregulated splicing factor expression in specific cancer types, leading to recurrent patterns of neojunction upregulation. These findings establish a molecular basis for T cell-based immunotherapies addressing the challenges of intratumoural heterogeneity.

2519 Background: Tumor heterogeneity and low mutational burden limits the availability of effective immunotherapy targets. Aberrant RNA-splicing (neojunctions) represents an underexplored yet promising source of neoantigens. To address this, we developed a neoantigen discovery platform (SNIPP) that characterizes a novel class of clonally-expressed, splicing-derived neoantigens. Furthermore, we validated the immunogenicity of these neoantigens by identifying specific TCRs that drive CD8+ T-cell-mediated tumor killing. Methods: SNIPP identified public neojunctions by analyzing TCGA RNA-seq data, selecting neojunctions with a positive sample rate (PSR) > 10% and filtering out those found in GTEx normal tissue RNA-seq data (PSR < 1%) across 12 cancer types. To characterize intratumorally conserved neojunctions, we performed maximally-distanced multi-site biopsies ( n = 535) within glioma patients ( n = 56) and generated RNA-seq data for each intratumoral site. Two independent algorithms were utilized to predict peptide processing likelihood and HLA-binding affinity of splicing-derived neoantigen candidates. Neoantigen-specific TCR sequences were identified via in vitro sensitization of PBMCs and subsequent 10x V(D)J scRNA-seq. These TCRs were transduced into CD8+ T-cells, which were tested downstream for immunogenicity and cytotoxicity against glioma cell lines. Results: Our pipeline identified 789 public neojunctions, including 32 neojunctions concurrently detected in transcriptomic and proteomic glioma datasets and confidently predicted to be presented by HLA-A*02:01. IVS and subsequent 10x V(D)J scRNA-seq identified TCR clonotypes reactive against neojunctions in RPL22 ( n = 7) and GNAS ( n = 1), with the latter exhibiting high intratumoral conservation (detected in > 90% of spatially-mapped biopsies across 17/56 patients (26.78%)). TCR-transduced CD8+ T-cells recognized and were immunogenically activated and demonstrated cytotoxicity against endogenously processed and presented neoantigens in GBM and melanoma lines. Additionally, IDH1-mutant oligodendrogliomas exhibited significantly higher neojunction expression compared to IDH1-mutant astrocytomas and IDH1wt subtypes. Differential gene expression analysis (DESeq2) revealed reduced expression of splicing factors in oligodendrogliomas, attributed to their specific co-deletion of chromosomes 1p and 19q. CRISPRi-mediated knockdown of these splicing factors (e.g. SF3A3, SNRPD2) in IDH1wt glioma cells resulted in significantly increased expression of corresponding neojunctions. Conclusions: Our study highlights a novel class of neoantigens derived from tumor-wide aberrant RNA splicing. The SNIPP platform effectively identifies public intratumorally-conserved neojunctions with strong therapeutic potential. Furthermore, elevated neojunction expression in oligodendroglioma underscores the mechanistic link between dysregulated splicing factor expression and RNA splicing abnormalities.

Activating mutations in the telomerase reverse transcriptase (TERT) promoter are prevalent in cancer and enable limitless cell division characteristic of immortal cells. Solving the immortality mechanism represents a major step toward selective reversal in cancer cells. TERT promoter (TERTp) mutations create a de novo E26 transformation-specific (ETS) transcription factor binding motif. Here, we analyzed 53 cell lines representing 16 cancer types and 6 recurrent TERTp mutations and found that the GA-binding protein (GABP) tetramer is responsible for promoter activation in all cases. Surprisingly, TERT expression is maintained after tetramer depletion. Further investigation revealed an underlying network of auto-suppression among the GABP subunits. Release from it drives TERT maintenance via upregulated GABP dimers or a paralogous tetramer. The GABPB1L tetramer is therefore a pan-cancer, pan-mutation activator of the mutant TERT promoter, but it is replaceable. Domains shared by the three GABP complexes, rather than solely the B1L tetramer, are mutation-specific vulnerabilities.

Supplementary Figures and Figure Legends

Advances in digital pathology and machine learning have the potential to revolutionize diagnostic neuropathology. Current brain tumor models are typically trained and validated using morphologic features from a single hematoxylin and eosin (H&E)-stained slide per patient. Yet, brain tumors such as diffuse glioma are known for their epigenetic, genetic, and transcriptional heterogeneity within an individual patient. The impact of this heterogeneity on model accuracy and development is unknown. To quantitatively investigate morphologic intratumoral heterogeneity in glioblastoma (GBM), we acquired 92 regionally distinct samples representing maximal tumor sampling across 10 patients with isocitrate dehydrogenase-wildtype GBM and quantified cell density, nucleus area, and nucleus circularity from whole-slide scanned images of H&E-stained slides. All 3 parameters exhibited significant morphologic variation between tumors from different patients and within a given tumor. To identify potential drivers of this variation, tumor-level and sample-level mutation profiling was performed. Mutations in tumor protein 53 both at the tumor level and the sample level had larger nuclear area and decreased nuclear circularity. Morphological features were not associated with regional location within the tumor. Accurate and robust H&E-based models to improve diagnosis and disease prognostication may require training sets that incorporate multiple spatially distinct samples per patient.

Low-grade gliomas (LGGs) exhibit low overall mutational burden and an immunosuppressive microenvironment, contributing to immunotherapy resistance. Intratumoral heterogeneity (ITH) further complicates the targeting of tumor-specific antigens (TSAs), yet its influence on the antigenic and immune landscapes in LGG remains understudied, particularly due to reliance on single-biopsy analyses. We performed exome and transcriptome sequencing on 70 spatially mapped biopsies from seven IDH-mutant Grade II astrocytoma patients. Tumor purity, immune deconvolution, and neoantigen prediction were conducted to assess regional immune variation and antigenic burden. In one patient (P375), neoantigen-specific CD8+ T cells were isolated using barcoded peptide-HLA multimers. Reactive T-cell receptor (TCR) clonotypes were identified via single-cell TCR sequencing and functionally validated in Jurkat76 and CD8+ T cells. Spatial profiling of whole tumors revealed most mutations were biopsy-specific, with a steep drop in mutations shared across multiple regions. RNA-seq–based unsupervised clustering with xCell and DESeq2 revealed two distinct immune microenvironments: immune-hot regions with elevated immune infiltration, and immune-cold regions with minimal immune presence. Notably, mutation-derived n-mer peptides from immune-hot regions showed enhanced predicted immunogenicity relative to those from immune-cold areas. In one patient (P375), neoantigen-specific CD8⁺ TCR clonotypes were identified targeting PRMT5 mutations found in immune-hot regions. A mutant PRMT5-specific TCR demonstrated robust, antigen-dependent activation and dose-dependent cytotoxicity in vitro. This TCR showed high specificity toward the mutant peptide with minimal cross-reactivity to the wild-type sequence. Furthermore, engineered CD8⁺ T cells effectively killed glioma cells endogenously expressing the full-length mutant PRMT5, supporting its candidacy as a viable immunotherapeutic target. Our findings reveal significant 3D spatial and immunologic heterogeneity in low-grade gliomas and demonstrate that neoantigens such as mutant PRMT5 can be selectively targeted by highly specific, cytotoxic TCR-engineered CD8⁺ T cells. These results support the potential of personalized TCR-based immunotherapy for low-grade glioma.

Table S1. RNA sequencing comparison data

Telomerase Reverse Transcriptase promoter (TERTp) mutations enable tumor cell immortality in millions of cancer patients annually. TERTp mutations are the most common non-coding mutations across all cancers, including glioblastoma, oligodendroglioma, medulloblastoma, and high-grade meningioma. A multitude of TERTp mutations, including the two hotspots, create de novo E26 transformation specific transcription factor binding sites. Among the 28 family members, we have shown that only the GA-binding protein (GABP, composed of DNA binding GABPA and transactivating GABPB subunits) activates the mutant TERTp. Prior therapies targeting the telomerase RNA component, TERC, lacked tumor selectivity and were poorly tolerated. The GABP-mediated reactivation of the mutated TERTp presents a unique therapeutic opportunity for tumor specific reversal of cellular immortality. Breakthroughs in the design of biological proteolysis-targeting chimera (bioPROTACs) offer a new approach to selectively degrade previously intractable targets such as transcription factors. We combined in silico protein-protein interaction modeling via AlphaFold and experimental validation to identify a minimal GABPB to bind GABPA and fused it with an E3 ubiquitin ligase. Introduction of this GABPA bioPROTAC into TERTp mutant glioblastoma cells depleted GABPA protein and eliminated GABPA binding to the mutant TERT promoter, reducing transcriptional activating (H3K4me3) and increasing suppressive (H3K27me3) histone marks. Reversion of mutant TERTp to an epigenetically silenced state reduced TERT expression by 73% to 95% and shortened tumor cell telomeres in a promoter mutation-specific manner. The GABP bioPROTAC reduced tumor growth and improved survival of mice bearing an orthotopic xenograft of TERTp mutated glioblastoma cells. The effects of the bioPROTAC were validated in vivo via magnetic resonance imaging of metabolic correlates of TERT expression. These data demonstrate that this GABP bioPROTAC potently degrades GABPA, leading to tumor-specific silencing of telomerase expression, and a reversal of glioblastoma tumor cell immortality. This artificial intelligence-guided approach may be applicable to other intractable cancer specific targets.

Isocitrate dehydrogenase (IDH)-mutant astrocytomas represent the most frequent primary intraparenchymal brain tumor in young adults, which typically arise as low-grade neoplasms that often progress and transform to higher grade despite current therapeutic approaches. However, the genetic alterations underlying high-grade transformation and disease progression of IDH-mutant astrocytomas remain inadequately defined. Genomic profiling was performed on 205 IDH-mutant astrocytomas from 172 patients from both initial treatment-naive and recurrent post-treatment tumor specimens. Molecular findings were integrated with clinical outcomes and pathologic features to define the associations of novel genetic alterations in the RAS-MAPK signaling pathway. Likely oncogenic alterations within the RAS-MAPK mitogenic signaling pathway were identified in 13% of IDH-mutant astrocytomas, which involved the KRAS, NRAS, BRAF, NF1, SPRED1, and LZTR1 genes. These included focal amplifications and known activating mutations in oncogenic components (e.g. KRAS, BRAF), as well as deletions and truncating mutations in negative regulatory components (e.g. NF1, SPRED1). These RAS-MAPK pathway alterations were enriched in recurrent tumors and occurred nearly always in high-grade tumors, often co-occurring with CDKN2A homozygous deletion. Patients whose IDH-mutant astrocytomas harbored these oncogenic RAS-MAPK pathway alterations had inferior survival compared to those with RAS-MAPK wild-type tumors. These findings highlight novel genetic perturbations in the RAS-MAPK pathway as a likely mechanism contributing to the high-grade transformation and treatment resistance of IDH-mutant astrocytomas that may be a potential therapeutic target for affected patients and used for future risk stratification.

Abstract Alterations in cis-regulatory elements, such as hypomethylation of the O-6-methylguanine-DNA methyltransferase (MGMT) promoter and activating mutations within the telomerase reverse transcriptase (TERT) promoter, are pervasive drivers of glioblastoma (GBM) tumorigenesis that are challenging to target. CRISPRoff is an engineered CRISPR/Cas9 system that can potently and heritably silence cis-regulatory elements through DNA methylation. Here, we develop an mRNA-based CRISPRoff platform to target the MGMT and TERT promoters in GBM, and we establish proof-of-principal in vivo targeting through lipid nanoparticles (LNP). Epigenetic editing in GBM cells and primary GBM organoids was performed using either electroporation or LNP encapsulation of CRISPRoff mRNA and sgRNAs. Target gene silencing and functional phenotypes were assessed by bisulfite sequencing, RT-qPCR, RNA-seq, western blot, cell viability/apoptosis, and telomere restriction fragment assays. In vivo efficacy was demonstrated using intracranial GBM xenografts, first through transplantation of cells with CRISPRoff delivered ex vivo, then with direct delivery of LNPs using convection enhanced delivery (CED) coupled with immunofluorescence/ immunohistochemistry. CRISPRoff targeting of MGMT promoter resulted in >99% reduction in MGMT expression in MGMT unmethylated primary GBM organoids, inducing up to 150-fold sensitization to temozolomide. Temozolomide sensitization was retained after intracrania transplantation, and silencing was durable in clonally isolated GBM cells continuously passaged for over 8 months after transient delivery of CRISPRoff. CRISPRoff targeting of the TERT promoter in primary GBM cultures harboring the G228A TERT promoter mutation silenced TERT expression by up to 99%, overcoming constitutive activation of TERT, and induced telomere shortening sufficient for complete replicative senescence. CED of mRNAs encapsulated by LNPs based on cKK-E12 or Lipid A9 resulted in tumor-selective uptake of mRNA transcripts, establishing a foundation for direct delivery of CRISPRoff in vivo bypassing the blood-brain-barrier. In summary, we establish a flexible and durable epigenetic editing system against multiple regulatory elements driving GBM, leveraging LNPs with potential for in vivo application.

Treatment failure for the lethal brain tumor glioblastoma (GBM) is attributed to intratumoral heterogeneity and tumor evolution. We utilized 3D neuronavigation during surgical resection to acquire samples representing the whole tumor mapped by 3D spatial coordinates. Integrative tissue and single-cell analysis revealed sources of genomic, epigenomic, and microenvironmental intratumoral heterogeneity and their spatial patterning. By distinguishing tumor-wide molecular features from those with regional specificity, we inferred GBM evolutionary trajectories from neurodevelopmental lineage origins and initiating events such as chromothripsis to emergence of genetic subclones and spatially restricted activation of differential tumor and microenvironmental programs in the core, periphery, and contrast-enhancing regions. Our work depicts GBM evolution and heterogeneity from a 3D whole-tumor perspective, highlights potential therapeutic targets that might circumvent heterogeneity-related failures, and establishes an interactive platform enabling 360° visualization and analysis of 3D spatial patterns for user-selected genes, programs, and other features across whole GBM tumors.