BET bromodomain inhibition potentiates radiosensitivity in models of H3K27-altered diffuse midline glioma

Diffuse midline glioma (DMG) H3K27-altered is one of the most malignant childhood cancers. Radiation therapy remains the only effective treatment yet provides a 5-year survival rate of only 1%. Several clinical trials have attempted to enhance radiation antitumor activity using radiosensitizing agents, although none have been successful. Given this, there is a critical need for identifying effective therapeutics to enhance radiation sensitivity for the treatment of DMG. Using high-throughput radiosensitivity screening, we identified bromo- and extraterminal domain (BET) protein inhibitors as potent radiosensitizers in DMG cells. Genetic and pharmacologic inhibition of BET bromodomain activity reduced DMG cell proliferation and enhanced radiation-induced DNA damage by inhibiting DNA repair pathways. RNA-Seq and the CUT&RUN (cleavage under targets and release using nuclease) analysis showed that BET bromodomain inhibitors regulated the expression of DNA repair genes mediated by H3K27 acetylation at enhancers. BET bromodomain inhibitors enhanced DMG radiation response in patient-derived xenografts as well as genetically engineered mouse models. Together, our results highlight BET bromodomain inhibitors as potential radiosensitizer and provide a rationale for developing combination therapy with radiation for the treatment of DMG.


Introduction
Diffuse midline gliomas (DMGs) with H3K27M mutation (histone H3 lysine 27 replaced with methionine) are diffusely infiltrating glial neoplasms affecting midline structures of the CNS (1).DMG is one of the most malignant childhood tumors, with a median survival of 9-12 months from diagnosis (2).Factors that contribute to the dismal prognosis include the infiltrative nature and anatomic location of the tumor within the pons, which precludes surgical resection.The identification of effective therapies has been extremely challenging, with more than 250 clinical trials involving different combinations of chemotherapeutic agents commonly used in adult glioma proving ineffective in treating DMG (3).Fractionated focal radiation to a total dose of 54-60 Gy over a 6-week period remains the only standard treatment modality that can provide transient symptom relief and a delay in tumor progression in about 70%-80% of patients.However, radiation-treated children with DMG show evidence of disease progression within the first year of completing radiation therapy (4,5).Given this reality, the identification of efficacious therapeutic agents that enhance the antitumor effects of radiation is urgently needed for improving treatment outcomes for this patient population.
In contrast to adult gliomas, DMG is uniquely dependent on the H3K27M mutation for its initiation and maintenance (6)(7)(8)(9).H3K27M mutation occurs in H3F3A and HIST1H3B/C genes, encoding histone H3 variants H3.3 and H3.1, respectively, in as much as 80% of DMGs and is associated with shorter survival among patients with DMG (6,7,10).We and others have identified a key functional consequence of H3K27M mutation: mutant protein sequestration of the polycomb repressive complex 2 (PRC2) methyltransferase resulting in functional inactivation of PRC2 (8,9,11,12).This inactivation leads to a global reduction of H3K27 di-methylation (K27me2) and tri-methylation (K27me3), which, in turn, leads to extensive transcriptional reprogramming of mutant cells and promotes a stem cell-like, therapy-resistant phenotype.
Diffuse midline glioma (DMG) H3K27-altered is one of the most malignant childhood cancers.Radiation therapy remains the only effective treatment yet provides a 5-year survival rate of only 1%.Several clinical trials have attempted to enhance radiation antitumor activity using radiosensitizing agents, although none have been successful.Given this, there is a critical need for identifying effective therapeutics to enhance radiation sensitivity for the treatment of DMG.Using high-throughput radiosensitivity screening, we identified bromo-and extraterminal domain (BET) protein inhibitors as potent radiosensitizers in DMG cells.Genetic and pharmacologic inhibition of BET bromodomain activity reduced DMG cell proliferation and enhanced radiation-induced DNA damage by inhibiting DNA repair pathways.RNA-Seq and the CUT&RUN (cleavage under targets and release using nuclease) analysis showed that BET bromodomain inhibitors regulated the expression of DNA repair genes mediated by H3K27 acetylation at enhancers.BET bromodomain inhibitors enhanced DMG radiation response in patient-derived xenografts as well as genetically engineered mouse models.Together, our results highlight BET bromodomain inhibitors as potential radiosensitizer and provide a rationale for developing combination therapy with radiation for the treatment of DMG.
BET bromodomain inhibition potentiates radiosensitivity in models of H3K27-altered diffuse midline glioma bromodomain inhibitors were subsequently validated for their antiproliferative effects in DMG cells.AZD5153 and molibresib (I-BET762) in combination with radiation showed strong additive cytotoxic effects relative to each monotherapy (white dotted line, Figure 1C).However, methotrexate and temozolomide, which have been using in combination with radiation in adult glioblastoma (GBM), did not show additive radiosensitizing effects or monotherapy cytotoxic effects in DMG cells.Our results are consistent with the results from clinical trials, which show that DMG transiently responds to the combination of temozolomide and radiation, but with no survival benefit from the combination therapy (21).
Targeted inhibition of BET bromodomain activity reduces cell proliferation and induces apoptosis in K27M-mutant DMG cells.To address whether BET bromodomain activity is required for K27M-mutant DMG cell growth, we studied the effects of depletion of BRDs (BRD2, -3, -4) on DMG cell proliferation using CRISPR/ Cas9 KO (Supplemental Figure 2).KO effects were confirmed at the protein level (Supplemental Figure 2A), and the effects of BRD depletion on cell proliferation were analyzed by use of the MTS assay (Supplemental Figure 2B).BRD4 depletion reduced H3K27ac and reciprocally increased H3K27me3 protein expression, whereas depletion of BRD2 and BRD3 did not affect the expression levels of H3K27ac and K27me3 (Supplemental Figure 2A).In addition, only BRD4 depletion suppressed the growth of DMG cells (Supplemental Figure 2B).We further analyzed the effects of BRD4 depletion on DMG cell growth (Figure 2 and Supplemental Figure 3).BRD4 KO and shRNA knockdown (KD) were confirmed at the protein level (Figure 2A and Supplemental Figure 3A), and the effects of BRD4 depletion on cell proliferation were analyzed by the MTS assay (Figure 2B and Supplemental Figure 3B), colony formation assays (Figure 2C and Supplemental Figure 3C), and BrdU incorporation assay (Figure 2D) in 2 K27M-mutant DMG cell lines (SF8628 and DIPG007).BRD4 depletion significantly reduced DMG cell growth relative to scramble control (Figure 2B and Supplemental Figure 3B).BRD4 depletion also suppressed colony formation in DMG cells (Figure 2C and Supplemental Figure 3C).The BrdU-positive cell population in S phase was decreased by BRD4 depletion in DMG cells (Figure 2D).These results indicate that BRD4 activity is required for DMG cell proliferation and suggest the possibility of a rational therapeutic target in DMG.
Through an unbiased high-throughput radiosensitivity screen, we found the BET bromodomain inhibitors to be potent radiosensitizers of H3K27M-mutant DMG cells.The depletion of BRD using shRNAs and sgRNAs, and BET bromodomain inhibition using small molecule inhibitors, reduced DMG cell proliferation and enhanced radiation-induced DNA damage by inhibiting DNA repair pathways.Moreover, BET bromodomain inhibition downregulated expression of DNA repair genes associated with H3K27 acyetylation (H3K27ac) occupancy and enhanced DMG radiation response in vitro and in vivo.Together, these results highlight BET bromodomain inhibition as a potential radiosensitizer and provide a rationale for developing combination therapy with radiation for the treatment of this deadly pediatric brain cancer.

Results
BET bromodomain inhibitors are identified as potent radiosensitizers by high-throughput drug screening.We first performed an unbiased high-throughput radiosensitivity screen in H3.3 WT and K27M-mutant DMG neurosphere cells using a total of 2,880 compounds, including 1,280 FDA-approved drugs and 1,600 clinical compounds (mainly small molecule inhibitors of epigenetic processes) in the presence or absence of 10 Gy irradiation.Radiosensitizing effects were quantified by cell death number using confocal image analysis combined with Hoechst nuclear staining and propidium iodide (PI) DNA staining (Figure 1A).We identified several clinical-grade BET bromodomain inhibitors as potent radiosensitizers in the screen (Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/JCI174794DS1),which increased cell death in combination with radiation (Figure 1B and Supplemental Figure 1).H3.3 K27M-mutant DMG neurosphere cells were more sensitive to BET bromodomain inhibitors in combination with radiation than H3.3 WT DMG neurosphere cells (Supplemental Figure 1).These BET vitro and in vivo.AZD5153 treatments induced dose-dependent inhibition of cell growth in 5 K27M-mutant DMG cell lines as well as human astrocytes expressing the K27M H3F3A transgene (Astro-KM cells), with IC 50 values of 0.41 μM (SF8628), 0.053 μM (DIPG007), 0.022 μM (SU-DIPG36), 0.063 μM (SU-DIPG4), 0.013 μM (genetically engineered mouse model-DMG [GEMM-DMG]), and 0.020 μM (Astro-KM) (Figure 3B).Normal human astrocytes (NHAs) and Astro-WT cells showed less sensitivity to NCT04817007, NCT02419417, NCT05372354).However, the preclinical efficacy of BMS-986158 was disappointing in orthotopic (brainstem) DMG PDX models (Supplemental Figure 3D) due to its poor penetrance across the blood-brain barrier (BBB), with a brain penetration ratio of 5.02% ± 1.32 % (Supplemental Table 2).AZD5153 showed a brain penetration ratio of 12.9% ± 1.25 %, higher than that of BMS-986158.Thus, we tested the efficacy of AZD5153 for cytotoxicity and radiosensitivity in DMG in Figure 1.High-throughput drug screening with radiation identified BET bromodomain inhibitors as radiosensitizers in DMG cells.Tumor cells isolated from GEMM-DMG (Ntv-a; p53fl/ fl; PDGFB; H3.3K27M; Cre) mice were cultured ex vivo as neurospheres and used to drug screen for radiosensitizers.(A) Representative image of neurospheres in transmitted light, Hoechst staining (nuclear), PI staining (dead cell), and Hoechst/PI overlay.Evaluation of the number, area, and dead cell intensity of neurospheres.4x magnification, with each pixel being 3.4156 mm in size, for a raw image of 3.5 mm.Images were subsequently enlarged to enhance visibility.(B) A library of 1,280 FDA-approved drugs and 1,600 clinical candidates was screened in the presence or absence of 10 Gy radiation.Left: Compounds to the right of the diagonal light green line are identified as having an additive effect on neurospheres in the presence 10 Gy radiation Compounds to the right of the dark green diagonal line are identified as those that radiosensitized neurospheres beyond the additive drug effect when combined with 10 Gy radiation (at ≥3σ).Right: Representative images of neurospheres treated with BET bromodomain inhibitor (BRDi) in the presence or absence of 10 Gy radiation.norm., normalized.4× magnification, with each pixel being 3.4156 mm in size, for a raw image of 3.5 mm.Images were subsequently enlarged to enhance visibility.(C) Radiosensitizing effect (orange) and cytotoxic effect (gray) with methotrexate, AZD5153, temozolomide, and molibresib (I-BET762).AZD5153, with IC 50 values of 1.35 μM and 1.47 μM, respectively (Figure 3B).AZD5153 IC 50 also inhibited DMG cell growth in a time-dependent manner (Figure 3C) and reduced colony formation in DMG cells (Figure 3D).
BET bromodomain inhibition sensitizes DMG cells to radiation.To verify the radiosensitizing effect of BET bromodomain inhibition in K27M-mutant DMG cells, we conducted a clonogenic survival assay in 3 DMG cell lines (SF8628, DIPG007, GEMM-DMG).Cells were treated with AZD5153 (Figure 4A) and JQ1 (Supplemental Figure 4A) and depleted of BRD4 with shBRD4 and sgBRD4 (Supplemental Figure 4B), concurrently with being subjected to ionizing radiation (IR).To quantify the radiosensitizing effect, we calculated the dose enhancement factor (DEF), which represents the ratio of the dose with IR alone to the dose with combination of BRD4 inhibitor plus IR and BRD4 inhibition at 10% survival.If the DEF is greater than 1, the BRD4 inhibition will be functioning as a radiosensitizer.AZD5153 treatment showed a radiationenhancing effect, with DEFs of 1.22 (SF8628), 1.32 (DIPG007), and 1.10 (GEMM-DMG) (Figure 4A).JQ1 had similar effects on radiation response in DMG cells, with DEFs of 1.25 (SF8628), 1.40 (DIPG007), and 1.14 (GEMM-DMG) (Supplemental Figure 4A).shBRD4 KD and sgBRD4 KO also increased the radiation response of DMG cells, with DEFs of 1.22 (shBRD4-484 and -487, SF8628), 1.34 (shBRD4-484 and -487, DIPG007), and 1.18 (sgBRD4-1, SF8628), 1.20 (sgBRD4-2, SF8628), 1.28 (sgBRD4-1 and -2, DIPG007) (Supplemental Figure 4B).Furthermore, we conducted BrdU incorporation (Figure 4B), apoptosis (Figure 4C), senescence (Supplemental Figure 5, A and B), and sphere formation (Supplemental Figure 5, C and D) assays.AZD5153 treatment resulted in a decreased BrdU-positive S-phase cell population relative to control (Figure 4B).Combination treatment with AZD5153 plus IR further decreased the S-phase cell population when compared with AZD5153 alone (Figure 4B).The annexin V apoptosis assay showed that either AZD5153 or IR monotherapy increased the level of annexin-positive cells compared with control (Figure 4C).Combination treatment of AZD5153 plus IR increased the level of annexin V-positive cells, outperforming each monotherapy.The β-galactosidase assay revealed increasing senescence-associated β-galactosidase staining in DMG cells treated with either AZD5153 or IR monotherapy (Supplemental Figure 5A).Combination treatment with AZD5153 and IR further increased β-galactosidase-positive DMG cells.Cell size is known to be associated with senescence.To quantify cell size, we gated DMG cells for G 1 DNA content and sorted them with the side scatter parameter (SCC) using flow cytometry (Supplemental Figure 5B).Similar to the results with β-galactosidase staining, combination treatment with AZD5153 and IR further increased cell size relative to either monotherapy.Combination treatment also reduced self-renewal activity (Supplemental Figure 5C) and neurosphere formation compared with either monotherapy (Supplemental Figure 5D).These results suggest that when compared with monotherapy, combination treatment with AZD5153 plus IR further increased radiosensitivity in DMG cells by decreasing the population of radioresistant S-phase cells and stemness and increasing apoptosis and senescence.
BET bromodomain inhibition downregulates the genes involved in DNA repair and cell cycle in K27M-mutant DMG cells.We have previously shown that JQ1 treatment causes a change in the expression of genes that promote tumor growth in K27M-mutant DMG (15).In our current RNA-Seq analysis, we performed unsupervised principal component analysis of SF8628 DMG cells treated with DMSO and BET bromodomain inhibitors (AZD5153, JQ1) for 24 and 48 hours.We found a global gene expression shift in AZD5153-treated DMG cells compared with DMSO-treated samples (Figure 5A).We compared the RNA-Seq data from the samples treated with DMSO or AZD5153, and the previous RNA-Seq data from the samples treated with JQ1.The differentially expressed genes were highly correlated between the samples treated with JQ1 and AZD5153 (Figure 5B), including 3,301 genes upregulated and 3,591 downregulated in response to the BET bromodomain inhibitor.Gene set enrichment analysis (GSEA) (Figure 5, C and D, and Supplemental Figure 6, A and B) and gene ontology (GO) pathway analysis (Figure 6, A and B, and Supplemental Figure 6, C and D) showed that cell cycle (e.g., CDK6, CDCA7, and UHRF1) and DNA double-strand breaks (DSB) repair pathways (e.g., BRCA1, RAD51, XRCC1, XRCC4, and POLQ1) were among the most significantly downregulated with the BET bromodomain inhibitor treatment.AZD5153 and JQ1 treatments also upregulated gene pathways involved in autophagy (e.g., ATGA4, MAP1LC3B) and catabolism pathways including glycolysis and protein/macromolecule catabolic pathways (e.g., SIRT1, MTOR) (Figure 5, C and D, and Supplemental Figure 6, A and B).The senescence-associated genes CDKN1A and HMGA1 were upregulated by AZD5153 treatment (Supplemental Figure 7, A and B).However, CDKN2A was downregulated by AZD5153 treatment.This could be due to increased H3K27me3, which repressed the PRC2 targets, including CDKN2A (Supplemental Figure 7, A and B).
BET bromodomain inhibition is known to suppress gene expression by dissociating BRD from the active chromatin mark histone H3K27ac (26).We have shown that genomic occupancy of H3K27ac and BRD is required for enhancer activity and gene expression in DMG cells (15).To determine the effects of BET bromodomain inhibition on gene expression associated with H3K27ac occupancy, we performed CUT&RUN followed by next-generation sequencing in DMG cells treated with AZD5153 (Figure 7).The CUT&RUN data showed that the majority of H3K27ac enrichments occurred in introns (first intron: 13.47%; other intron: 28.97%) and intergenic regions (38.8%) (Figure 7A).Metaplots and heatmaps showed the enrichments of H3K27ac signal near the previously defined enhancer regions in DMG cells (Figure 7B).AZD5153 treatment dramatically reduced H3K27ac occupancy at enhancer regions.To investigate the enrichment of transcription factors among the H3K27ac DNA-binding sites, we used the DiffBind R package to determine differential peaks between DMSO-and AZD5153-treated samples (FDR < 0.05).We performed simple enrichment motif analysis in SF8628 DMG cells and found a significant enrichment of DNA sequencing motifs involving neuronal developmental transcriptional factors such as LHX1-3 and HOX13 (e-value < 0.05; Supplemental Table 3).Interestingly, the H3K27ac peaks of 2 representative DNA repair genes, BRCA1 and RAD51, were diminished at enhancer regions in the SF8628 cells treated with AZD5153 (Figure 7C).Expression of these DNA repair genes was significantly downregulated in AZD5153-treated samples in the RNA-Seq analysis (Figure 5, B  and C).Senescence-associated gene expression was not correlat-ing radiation (Figure 9B and Supplemental Figure 4D).Expression of BRCA1, RAD51, and RAD50 was also induced by radiation and peaked at 3-6 hours following radiation (Figure 9B).BET bromodomain inhibitors extended radiation-induced γH2X expression over 6 hours following radiation (Figure 9B and Supplemental Figure 4D).In contrast, expression of BRCA1, RAD51, and RAD50 was decreased by BET bromodomain inhibitors over the time of radiation.These results suggest that BET bromodomain inhibition extends radiation-induced DNA damage signaling by suppressing the DNA repair pathway.
DNA damage is repaired by 2 major pathways: homologous recombination (HR) and nonhomologous end-joining (NHEJ) repair (27,28).HR repairs DNA DSB during S and G 2 phases and provides a template for error-free repair.In contrast, NHEJ is active throughout the cell cycle and directly involves ligation of DNA ends without homology.To analyze the DNA damage repair pathways in DMG cells, we transfected GFP reconstitution reporter cassettes for HR and NHEJ (29) into SF8628 DMG cells in the presence or absence of AZD5153.AZD5153 treatment reduced DNA repair ability through both the HR and NHEJ DNA repair pathways (Figure 9C), which is consistent with the RNA-Seq results showing that AZD5153 downregulated the genes involved in both HR and NHEJ repair pathways (Figure 5).Collectively, our results indicate that BET bromodomain inhibition increased radiation-induced DNA damage by inhibiting the HR and/or NHEJ DNA repair pathway in K27M-mutated DMG cells.
BET bromodomain inhibitors enhance radiation-mediated antitumor effects in patient-derived and genetically engineered DMG animal models.Based on the radiosensitizing effect of BET bromodomain inhibition on the growth of K27M-mutant DMG cells, we hypothesized that BET bromodomain inhibition increases radiation-mediated antitumor activity and survival benefit in DMG mouse models.To address this, we implanted SF8628 or GEMM-DMG cells into mouse pons and treated them with AZD5153 (50 mg/kg) or JQ1 (30 mg/kg) for 2 weeks in the presence or absence of radiation at total dose of 9 Gy (1.5 Gy per day for 3 days a week for 2 weeks) (Figure 10A).BET bromodomain inhibitor monotherapy inhibited tumor growth and extended the survival of mice with SF8628 DMG PDX (Figure 10, B and C) as well as GEMM-DMG models (Supplemental Figure 8).Similarly, radiation monotherapy provided a significant therapeutic benefit (Figure 10, B  and C, and Supplemental Figure 8).We found that combination treatment with AZD5153 and radiation therapy significantly prolonged animal survival (Figure 10B and Supplemental Figure 8).Similarly, the combination treatment with JQ1 and radiation showed a significant survival benefit (Figure 10C).These in vivo efficacy studies included euthanizing the mice at the end of treatment to obtain brainstem tumor samples for analysis of tumor cell proliferation (Ki-67; Figure 10D), apoptosis (TUNEL, Figure 10D), senescence (p21, p16; Supplemental Figure 9), and migration (normal human nuclear antigen [NHNA]; Supplemental Figure 9, bottom).Analysis of intratumor Ki-67 staining showed that all therapies significantly reduced SF8628 DMG cell proliferation relative to the control group (Figure 10D).There were significantly fewer Ki-67-positive cells in the samples treated with combination therapy compared with either monotherapy.TUNEL staining results showed that the proportion of positive cells was highest in tumors ed with H3K27ac occupation (Supplemental Figure 7C).Taken together, our results suggest that BET bromodomain inhibition promotes a transcriptionally silent chromatin state by reducing H3K27ac occupancy and represses the expression of the genes involving DSB repair in K27M-mutant DMG cells.
BET bromodomain inhibition enhances radiation-induced DNA damage.We next analyzed the effects of BET bromodomain inhibition on radiation-induced DNA damage and repair pathways in DMG cells.We examined fluorescence immunocytochemistry of the DNA DSB marker γH2AX and repair marker 53BP1 to quantify the extent of DNA damage and repair in irradiated SF8628 DMG cells in either the presence or absence of the BET bromodomain inhibitors AZD5153 (Figure 8A) and JQ1 (Supplemental Figure 4C).The number of γH2AX and 53BP1 foci increased 1 hour following IR, indicating increased DNA DSB damage and repair by IR.At 24 hours after IR, γH2AX and 53BP1 foci were largely reduced in those cells due to successful repair upon DNA damage.However, irradiated DMG cells treated with BET bromodomain inhibitors sustained high levels of γH2AX at 24 hours compared with cells treated with IR alone, while 53BPI foci were decreased (Figure 8A and Supplemental Figure 4C).Similarly, comet assays showed that IR increased comet tail formation in SF8628 and DIPG007 DMG cell lines, indicating increased unrepaired DNA damage (Figure 8B).DNA damage further increased comet tail formation in irradiated DMG cells treated with AZD5153 compared with cells treated with IR alone.These results suggest that BET bromodomain inhibition may contribute to the DNA repair process to enhance radiationinduced DNA damage.Western blotting showed that AZD5153 treatment decreased expression of the DNA repair markers BRCA1, RAD51, and XRCC1 in DMG cell lines (Figure 9A).H3K27ac was also decreased by AZD5153 in a dose-dependent manner (Figure 9A).Radiation-induced γH2X expression peaked at 1 hour follow-    successfully integrated the HTS with neurosphere-based assays using automated fluorescence live-cell imaging in the presence or absence of radiation and identified several clinical-grade BET bromodomain inhibitors as top candidates for radiosensitization (Figure 1, Supplemental Figure 1, and Supplemental Table 1).In this assay, we used PI to detect the dead cell population.However, PI staining may not capture the long-term mechanism of radiation-induced cell death.Further evaluation of proliferative cell death caused by radiation, such as mitotic catastrophe (31), would be needed to understand the mechanism of radiation-induced cell death.
BET bromodomain inhibitors disrupt the binding between acetylated histone and BRD proteins and inhibit active transcription, leading to enhancement of the radiation effect in DMG (Figure 11B) (14-16).K27M-mutant DMG cells are vulnerable to BET bromodomain inhibition due to transcriptional dysregulation resulting from the mutation (14,15,20,32).We demonstrated that BET bromodomain inhibition, by shRNA-or sgRNA-mediated BRD depletion (Figure 2, Supplemental Figure 2, and Supplemental Figure 3) and treatment with small molecule inhibitors (AZD5153 and JQ1) (Figure 3), suppressed the growth of human and mouse K27M-mutant DMG cells.Importantly, BET bromodomain inhibition in combination with radiation further increased radiosensitivity of DMG cells by reducing the radioresistant derived from mice receiving combination therapy of AZD5153 and radiation relative to those receiving either monotherapy (Figure 10D).No TUNEL positivity was evident in normal brain surrounding tumor in mice receiving any of the combination treatments.Senescence marker p21 staining revealed an increase in positive cells in tumors treated with AZD5153 and in combination with radiation (Supplemental Figure 9).However, p16-positive cells were decreased by the treatment.This could be due to downregulation of the expression of the CDKN2A gene, which codes p16 protein.NHNA staining revealed decreasing NHNA-positive cells in the tumors of mice treated with either AZD5153 or radiation (Supplemental Figure 9).Combination treatment further decreased NHNA positive cells relative to each monotherapy.

Discussion
DMG is one of the most malignant childhood cancers, with a limited response to radiation therapy, resulting in a dismal prognosis: median overall survival is less than 12 months.There is a critical need for new therapeutics that enhance the effect of radiation in the treatment of DMG.Here, we identified BET bromodomain inhibitors as potent radiosensitizers in DMG using unbiased high-throughput radiosensitivity library screening (Figure 11A).High-throughput screening (HTS) is a useful tool for identifying candidate compounds from a large chemical library (30).We JQ1) revealed significant decreases in transcripts from the genes involved in DNA repair pathways for both HR and NHEJ, including BRCA1, RAD51, XRCC1, and XRCC4 (Figures 5 and 6, and Supplemental Figure 6).To advance understanding of the transcriptional regulation in DNA repair gene pathways by BET bromodomain inhibition, we mapped genome-wide occupancy of H3K27ac in K27M-mutant DMG cells using cleavage under targets and release using nuclease (CUT&RUN; Figure 7).We have previously profiled the epigenome of K27M-mutant DMG cells and shown that K27M mutation associates with increased H3K27ac and that the heterotypic H3K27M-K27ac nucleosomes colocalize with BET BRD2 and BRD4 at the loci of actively transcribed genes (15).We analyzed S-phase cell population and stemness, and increasing apoptosis and senescence (Figure 4 and Supplemental Figure 5).
DNA damage is thought to be the most important consequence of the radiation effect, and the genetic alterations of DNA repair pathways are frequently detected in pediatric high-grade glioma, including DMG (6,10,33,34).We and others have demonstrated that the majority of DNA DSB caused by radiation are repaired within 24 hours of completing radiation (35)(36)(37).Thus, DNA repair is a key factor in radiosensitivity and can be a therapeutic target to enhance radiation-mediated antitumor activity in K27Mmutant DMG.Our gene expression profiling of K27M-mutant DMG cells treated with BET bromodomain inhibitors (AZD5153 and  (Figure 7).In fact, the BET bromodomain inhibitors AZD5153 and JQ1 inhibited HR and NHEJ DNA repair pathways and prolonged radiation-induced DNA damage in K27-mutant DMG cells (Figures 8 and 9, and Supplemental Figure 4).
The molecular mechanisms of BET bromodomain inhibition in transcription and chromatin machinery for DNA damage repair in DMG are not fully understood.Upon binding to the chromatin, BET BRDs are known to function in the assembly of complexes that facilitate chromatin accessibility to transcription factors, allowing for the recruitment of RNA polymerases II (RNAPII) (39)(40)(41)(42).In particular, BRD4 is required for subsequent progression of RNAPII through hyperacetylated nucleosomes during transcription elongation through interactions of its bromodomains with acetylated histones in order to prevent transcriptional stalling (40)(41)(42).Yaffe's group demonstrated that deregulated transcription following inhibition or loss of BRD4 in cancer cells leads to the accumulation of RNA:DNA hybrids (R-loops) and collisions with the replication machinery causing replication stress, DNA damage, and apoptotic cell death during S phase (43).We observed that BET bromodomain inhibition decreased the radioresistant S-phase cell population, diminished self-renewal activity, and resulted in increased apoptotic cell death and cellular senescence (Figure 4 and Supplemental Figure 5).In a study of H3K27me3-deficient medulloblastoma cells (44), JQ1 inhibition sensitized medulloblastoma cells to radiation by enhancing the apoptotic response through suppression of Bcl-xL and upregulation of Bim.Loss of H3K27me3 caused an epigenetic switch from H3K27me3 to H3K27ac at specific genomic loci, altering the transcriptional profile, which associated with a radioresistant phenotype in H3K27me3-deficient medulloblastoma.Stemness is a key characteristic of radioresistance in glioma (45).BET bromodomain inhibition may sensitize H3K27me3-deficient tumors to radiation by diminishing the radioresistance phenotype and enhancing apoptotic response and cellular senescence.We will further investigate the role of BET bromodomain inhibition in the transcription machinery associated with histone modification of H3K27me3 and H3K27ac to understand the radiation-induced DNA damage response in DMG.
Consistent with in vitro experiments, our animal studies demonstrated that the combination therapy with BET bromodomain inhibitor and radiation inhibited tumor growth and increased survival in human and murine DMG mouse models compared with either therapy alone (Figure 10).The improvement resulting from the combination therapy proved to be modest.One limitation to the in vivo efficacy of BET bromodomain inhibitors is poor brain penetration (Supplemental Table 2).To increase drug concentrations in the brain, we would further investigate new drug delivery systems such as disrupting the BBB using focused ultrasound (46,47) or bypassing the BBB using convection-enhanced delivery (48) and intranasal delivery (49).Nevertheless, our findings support the possible use of BET bromodomain inhibitor to increase the radiation-mediated antitumor effect for the treatment of DMG.

Methods
Further information can be found in Supplemental Methods.
Sex as a biological variable.Our study examined 6-week-old female athymic mice (rnu/rnu genotype, BALB/c background).The animals the specific loci with H3K27M-K27ac occupation from the previous study (15) and found that BET bromodomain inhibition diminished a genome-wide distribution of H3K27ac at enhancer regions including 2 representative DNA repair genes, BRCA1 and RAD51 (Figure 7).DNA damage induces cellular senescence (38).We found that the senescence-associated genes, CDKN1A and HMGA1 were upregulated by BET bromodomain inhibition (Supplemental Figure 7, A and B).However, CDKN2A was downregulated by BET bromodomain inhibition.There was no association between H3K27ac occupation and senescence-associated gene expression (Supplemental Figure 7C).It is possible that senescenceassociated genes are controlled by different epigenetic regulation mechanisms, such as H3K27me3.Indeed, BET bromodomain inhibition reciprocally increased H3K27me3 (Supplemental Figure 2A), which resulted in silencing of PRC2 target genes such as CDKN2A.Our results indicated that BET bromodomain inhibitors downregulate the genes involved in DNA damage repair mediated by H3K27ac at enhancers, providing a basis for the possibility that BET bromodomain inhibitors act as radiation enhancers in DMG the identity of the cell lines.All cells were cultured in an incubator at 37°C in a humidified atmosphere containing 95% O 2 and 5% CO 2 and were mycoplasma-free at the time of testing with a Mycoplasma Detection Kit (InvivoGen).
shRNAs and sgRNA treatments.BRD4 and scrambled control shRNAs (BRD4 shRNAs: V3THS_378004, V3THS_326487, V3THS_326484, Control shRNA: RHS4346; Dharmacon IDs) were used to generate lentivirus and infected tumor cells according to the manufacturer's instructions.At 24 hours after lentiviral infection, cells were selected using 2 mg/mL puromycin for 5 days prior to in vitro assays.We also generated sgRNAs to knock out BRD2, BRD3, and BRD4 expression.sgRNA for the ROSA26 gene was used as control (see Supporting Data Values).The lentiCRISPRv2 vector (a gift from F. Zhang; Addgene plasmid 52961) was digested with BsmBI and inserted the sgRNAs into the vector (53).The ligation reactions were transfected into Stbl3 cells.Positive clones were confirmed with Sanger sequence.These plasmids were cotransfected into HEK293T cells with psPAX2, pMD2.G with PEI reagents (23966, Polysciences).Supernatants containing virus particles were collected at 48 and 72 hours and were used to infect DMG cells.After 48 hours of lentiviral infection, cells were selected with 2 μg/mL puromycin for 5 days prior to the in vitro assay.
Clonogenic survival assay.Six-well tissue culture plates were seeded with 400-10,000 cells and allowed to adhere for 12 hours.The modified cells with BRD4 shRNAs or sgRNAs, or unmodified cells treated with 50 nM AZD5153 or 50-100 nM JQ1 alone, were irradiated at doses of 0.5, 1, 2, 3, 4, 6, and 8 Gy.Radiation was delivered by a gamma irradiator.Cells were incubated at 37°C for 2 weeks, after which colonies were counted following staining with 0.05% crystal violet.Plating efficiencies were calculated as the ratio of the number of colonies formed to the number of cells seeded.Colonies of more than 50 cells were used to indicate surviving fractions.Surviving fractions were calculated as the plating efficiency of treated cells divided by the plating efficiency of control cells.DEFs were calculated as the ratio of the dose with radiation alone to the dose with radiation and BRD4 inhibition at 10% survival.
DNA repair assays.GFP reconstitution reporter cassettes for detection of HR and NHEJ have been previously reported (29,34).Plasmids containing HR or NHEJ reporter cassettes were linearized and transfected into cells to measure HR or NHEJ as a function of GFP expression.Transfections were performed using Lipofectamine 2000 (ThermoFisher Scientific 11668027).Cells with integrated reporter constructs were selected by adding 1 mg/mL geneticin (ThermoFisher Scientific 10131-035).HR or NHEJ cassette-expressing cells were treated with 1 μM AZD5153 for 72 hours, then transfected with a mixture of 5 μg ISceI-expressing plasmid and 2 μg pDsRed2-N1 (Clonetech 632406).Four days following transfection, cells were harvested, suspended in PBS, and placed on ice.Cells were then analyzed by FACS using LSRFortessa cell analyzer (BD Biosciences).Cells expressing GFP, pDsRed2-N1, or no fluorescent protein were used as calibration controls.Data were analyzed using FlowJo software.DNA repair efficiency was determined as a ratio of GFP + to DsRed + cells normalized to 100% of vehicle control (DMSO).
Comet assay.Cells were treated with 500 nM AZD5153 or 0.5% DMSO, followed by 4 Gy irradiation, and alkaline comet assays were performed (54).Briefly, 10,000 cells were resuspended in 75 μL of 0.5% (wt/vol) low-melting-point agarose and pipetted on slides pre-were purchased from Envigo and housed under aseptic conditions.The animals are well established and were used to develop DMG PDXs in our published studies (15,35,37,50,51).There are no reported sex differences among DMG patients.
Statistics.Survival plots were generated and analyzed using the Kaplan-Meier method and Graph-Pad Prism v9.5 software.Differences between survival plots were estimated using a log-rank test with Holm's adjustment.For other analyses, 1-way ANOVA was applied for multiple-group comparison with a post hoc Tukey's test and a 2-tailed unpaired t test for comparison in 2 groups using the Prism software.
Study approval.All animal protocols were approved by the Northwestern University Institutional Animal Care and Use Committee.
Histology and Phenotypic Core, and NUSeq Core facilities.The initial drug screen was supported by a developmental research project awarded to OJB, RH, and MRC through the NU Brain SPORE award mechanism (5P50CA221747).
Address correspondence to: Oren J. Becher, Department of Pediatrics, Icahn School of Medicine at Mount Sinai, 1468 Madison Avenue, 4th Floor, New York, New York 10029, USA.Phone: 212.241.7022;Email: oren.becher@mssm.edu.Or to: Rintaro Hashizume, Department of Pediatrics, University of Alabama at Birmingham, 1670 University Boulevard, G019A, Birmingham, Alabama 35294, USA.Phone: 205.975.0285;Email: rhashi23@uab.edu. of data.All authors commented on the manuscript and approved the included data.

Figure 7 .
Figure 7. BET bromodomain inhibition altered genome-wide H3K37ac occupancy and transcription in DMG cells.CUT&RUN was performed using H3K27ac antibody in SF8628 DMG cells treated with 1 μM AZD5153 or 0.5% DMSO for 48 hours.(A) Pie chart showing the distribution of H3K27ac across the DMG genome.(B) Heatmaps showing H3K27ac occupancy with DMSO versus AZD5153 treatment.Metaplots above indicate corresponding H3K27ac occupancy.Each plot is centered on the summit of the average occupancy and extended 5 kb upstream and downstream (-5 kb and +5 kb, respectively).Corresponding gene expression at the H3K27ac binding sites generated from RNA-Seq are shown to the right.(C) Gene annotation tracks showing H3K27ac occupancy and gene expression for the BRCA1 and RAD51 loci.The enhancer region is highlighted with a square for each gene.

Figure 9 .
Figure 9. BET bromodomain inhibition induced DNA damage and suppressed DNA repair in DMG cells.(A) Western blotting showing the effect of AZD5153 (0-10 μM) on expression of DNA repair marker: BRCA1, RAD51, and XRCC1, DNA damage marker: γH2AX, and H3K27ac.(B) Western blot showing effects of AZD5153 (5 μM) on expression change over time after 6 Gy IR in SF8628 DMG cells.(C) DNA repair assay showing effect of AZD5153 (500 μM) on HR and NHEJ pathways in SF8628 DMG cells.Flow plots represent fluorescent signals from HR and NHEJ reporter cassettes.Repair efficiency represents the ratio of GFP + to DsRed + cells normalized to 100% of vehicle control (0.5% DMSO).Values (mean ± SEM) shown are based on averages from quadruplicate samples.Unpaired t test values for comparisons between control and AZD5153 samples: ***P = 0.0004 (HR), ****P < 0.0001 (NHEJ).

Figure 11 .
Figure 11.Working model.(A) High-throughput drug screening.BET bromodomain inhibitors (BRDi) were identified as radiosensitizers using high-throughput drug screening in the DMG cells treated with radiation.BRDi decreased H3K37ac occupancy at enhancer regions, which led to suppressed transcription involving DNA repair in DMG cells.(B) Epigenetic inhibition of DNA repair genes.BRDi disrupts the interaction between acetylated histone (Ac) and BRDs, inhibiting active transcription for the genes involving radiation-induced DNA damage repair, which results in enhancement of the radiation effect in DMG.