Genomic analyses of early responses to radiation in...

来自 : 发布时间：2024-05-03

GEnomic analyses of early responses to radiation in glioblastoma reveal new alterations at transcription, splicing, and translation levels AbstractHigh-dose radiation is the main component of glioblastoma therapy. Unfortunately, radio-resistance is a common problem and聽a major contributor to tumor relapse. Understanding the molecular mechanisms driving response to radiation is critical for聽identifying regulatory routes that could be targeted to improve treatment response.聽We conducted an integrated analysis in the U251 and U343 glioblastoma cell lines to map early alterations in the expression of聽genes at three levels: transcription, splicing, and translation in response to ionizing radiation.聽Changes at聽the聽transcriptional level were the most prevalent response. Downregulated genes are strongly associated with聽cell cycle and DNA replication and linked to a coordinated module of expression. Alterations in this group are likely driven by聽decreased expression of the transcription factor FOXM1 and members of the E2F family. Genes involved in RNA regulatory聽mechanisms were affected at the mRNA, splicing, and translation levels, highlighting their importance in radiation-response.聽We identified a number of oncogenic factors, with an increased expression upon radiation exposure, including BCL6, RRM2B,聽IDO1, FTH1, APIP, and LRIG2 and lncRNAs NEAT1 and FTX. Several of these targets have been previously implicated in聽radio-resistance. Therefore, antagonizing their effects post-radiation could increase therapeutic efficacy.聽Our integrated analysis provides a comprehensive view of early response to radiation in glioblastoma. We identify new biological聽processes involved in altered expression of various oncogenic factors and suggest new target options to increase radiation聽sensitivity and prevent relapse. IntroductionGlioblastoma is the most common intracranial malignant brain tumor with an aggressive clinical course. Standard of care entails maximally safe resection followed by radiotherapy with concomitant and adjuvant temozolomide as described in the landmark European Organization for Research and Treatment of Cancer (EORTC) Brain Tumor and Radiotherapy Groups and the National Cancer Institute of Canada study1. Adjuvant radiation therapy is commonly delivered 4鈥? weeks after maximally safe resection.. Nonetheless, the median overall survival remains approximately 16 months1,2, and the recent addition of tumor-treating fields to the standard of care has only increased median overall survival to 20.5 months1. Recurrence occurs in part because glioblastoma uses sophisticated cellular mechanisms to repair DNA damage from double-stranded breaks caused by ionizing radiation, specifically homologous recombination and non-homologous end-joining. Thus, the repair machinery confers a mechanism for resistance to radiation therapy. Ionizing radiation can also cause base damage and single-strand breaks, which are repaired by base excision and single-strand break repair mechanisms, respectively3. A comprehensive analysis of molecular mechanisms driving resistance to chemotherapy and radiation is required to surpass major barriers and advance treatments for glioblastoma.The Cancer Genome Atlas (TCGA) was instrumental in improving the classification and identification of tumor drivers4, but its datasets provide limited opportunities to investigate radiation response. Thus, studies using cell and murine models are still the best alternatives to evaluate radiation response at the genomic level. The list of biomarkers associated with radiation resistance in glioblastoma is still relatively small. Among the most relevant are FOXM15,6, STAT36, L1CAM7, NOTCH18, RAD519, EZH210, CHK1/ATR11, COX-212, and XIAP13. Dissecting how gene expression is altered by ionizing radiation is critical to identify possible genes and pathways that could increase radio-sensitivity. A few genomic studies14,15,16 have explored this question, but these analyses were restricted to describing changes in transcription.Gene expression is regulated at multiple levels, and RNA-mediated mechanisms such as splicing and translation are particularly relevant in cancer biology. A growing number of inhibitors against regulators of splicing and translation are being identified17. Splicing alterations are a common feature across cancer types and affect all hallmarks of cancer18. Numerous splicing regulators display altered expression in glioblastoma (e.g. PTBP1, hnRNPH, and RBM14) and function as oncogenic factors19. Importantly, a genome-wide study using patient-derived models revealed that transformation-specific depended on RNA splicing machinery. The SF3b-complex protein PHF5A was required for glioblastoma cells to survive, but not neural stem cells (NSCs). Moreover, genome-wide splicing alterations after PHF5A loss appear only in glioblastoma cells20. Translation regulation also plays a critical role in glioblastoma development. Many translation regulators such as elF4E, eEF2, Musashi1, HuR, IGF2BP3, and CPEB1 promote oncogenic activation in glioblastoma, and pathways linked to translation regulation (e.g., mTOR) promote cancer phenotypes21.To elucidate expression responses to radiation, we conducted an integrated study in U251 and U343 glioblastoma cell lines covering transcription (mRNAs and lncRNAs), splicing, and translation. We determined that the downregulation of FOXM1 and members of the E2F family are likely the major drivers of observed alterations in cell cycle and DNA replication genes upon radiation exposure. Genes involved in RNA regulatory mechanisms were particularly affected at the transcription, splicing, and translation levels. In addition, we identified several oncogenic factors and genes associated with poor survival in glioblastoma that displayed increased expression upon radiation exposure. Importantly, many have been implicated in radio-resistance, and therefore, their inhibition in combination with radiation could increase therapy efficacy.ResultsWe collected samples from U251 and U343 glioblastoma cells at 0 (T0), 1鈥塰our (T1), and 24鈥塰ours (T24) post irradiation and identified all the alterations at mRNA, splicing, and translation levels (Supplementary Fig.聽S1A). A Principal Component Analysis (PCA) performed on RNA-Seq and Ribo-Seq read counts revealed that most variation can be explained by the cell type along the first principal component, while radiation time-related changes were captured along the second principal component (Supplementary Fig.聽S1B). The distribution of fragment lengths for ribosome footprints was enriched in the 28鈥?0 nucleotides range (Supplementary Fig.聽S2). while the Ribo-seq metagene profiles exhibit high periodicity within the coding domain sequence (CDS) as expected since ribosomes traverse three nucleotides at a time (Supplementary Fig.聽S3). All the replicates in each condition show high level of similarity (Supplementary Fig.聽S1B and S4A).Changes in global transcriptome profile in response to radiationWe first conducted an integrated analysis to evaluate the early impact of radiation on the expression profile of U251 and U343 GBM lines. Using this approach, we focus on genes undergoing up regulation or downregulation post irraditaion after adjusting for cell line differences (See Methods). A relatively small number of genes displayed altered expression at T1 (Supplementary Fig.聽S4B,C). Downregulated genes are mainly involved in transcription regulation and include 18 zinc finger transcription factors displaying high expression correlation in glioblastoma samples from TCGA (Supplementary Table聽S1). Upregulated sets contain genes implicated in cell cycle arrest, apoptosis, and stress such as ZFP36, FBXW7, SMAD7, BTG2, and PLK3 (Supplementary Table聽S1).Since many alterations were observed when comparing the T1 vs. T24 time points (Supplementary Table聽S1 and Supplementary Fig.聽S4C), we opted to focus on genes showing the most marked changes (log2 fold-change 1.0 or 鈭?.0 and adjusted p-value 0.05) to identify biological processes and pathways most affected at the T24 time point. Top enriched GO terms and pathways among downregulated genes include chromatin remodeling, cell cycle, DNA replication, and repair (Fig.聽1A). Additionally, we identified several GO terms associated with mRNA metabolism, decay, translation, and ncRNA processing, suggesting active participation of RNA-mediated processes in radio-response (Fig.聽1B). Network analysis indicated the set of genes in these categories is highly interconnected (Fig.聽1C and Supplementary Table聽S2).Figure 1Characteristics of downregulated genes at 24鈥塰ours (T24) after radiation exposure in glioblastoma cell lines. (A) Enriched gene ontology related to cell cycle, DNA replication, and repair among downregulated genes. (B) RNA-related Gene Ontology (GO) terms enriched among downregulated genes summarized using REVIGO97. (C) Protein-protein interaction network, according to STRING98 showing downregulated genes associated with RNA-related functions. Gene clusters based on the strength of connection and gene function are identified by color. Lines colors indicate the type of association: light green indicates an association based on literature findings; blue indicates gene co-occurrence; magenta indicates experimental evidence.Full size imageTo expand the expression analysis, we employed Weighted Gene Co-expression Network Analysis (WGCNA)22 to identify gene modules with significant co-expression variation in response to radiation. All identified modules, along with the complete list of genes in each module, are shown in Supplementary Fig.聽S5A,B and Supplementary Table聽S3. Seven modules were identified (({Z}_{text{summary}}, ,5)) as tightly regulated, independent of the cell line (Supplementary Fig.聽S5B). Among modules with the highest significant correlation (0.8, p-value 1e鈭?), module 2 contains genes downregulated in T24, with many involved in cell cycle, metabolism mRNA metabolism, processing, splicing, and transport (Supplementary Table聽S3), corroborating results described above.Next, we investigated downregulated genes with the gene set enrichment analysis (GSEA) tool Enrichr23 and conducted expression correlation analysis with Gliovis24. Based on their genomic binding profiles and effect of gene expression, FOXM1 and the E2F family of transcription factors emerged as potential regulators of a large group of cell cycle/DNA replication-related genes in the affected set (Fig.聽2A, Supplementary Table聽S4). In agreement, E2F1, E2F2, E2F8, and FOXM1 displayed a significant decrease upon radiation. FOXM1 and E2F factors have been previously implicated in chromatin remodeling, cell cycle regulation, DNA repair, and radio-resistance25,26. All four factors are highly expressed in glioblastoma with respect to low-grade glioma. Importantly, they display high expression correlation with a large set of downregulated genes implicated in cell cycle and DNA replication and among themselves in glioblastoma samples in TCGA (Fig.聽2B,C). We corroborated the changes in expression of these transcription factors and some of their potential targets by qRT-PCR in U343 cells and also observed similar changes in the glioblastoma line A172 and the glioma stem cell line 3565 (Supplementary Fig.聽S6).Figure 2E2Fs and FOXM1 in glioblastoma. (A) Correlation of E2F1, E2F2, E2F8, and FOXM1 with target genes involved in cell cycle. (B) Expression levels of E2F1, E2F2, E2F8, and FOXM1 in gliomas grades II, III, and IV in TCGA samples. (C) E2F1, E2F2, E2F8, and FOXM1 expression correlation in glioblastoma (TCGA samples) using Gliovis24. ***p-value 0.0001.Full size imageUpregulated genes at T24 are preferentially associated with the extracellular matrix receptor interaction pathway, extracellular matrix organization, axonogenesis, and response to type I interferon (Fig.聽3A, and Supplementary Table聽S2). With respect to the extracellular matrix, we observed changes in the expression levels of several collagens (types II, IV, V, and XI), glycoproteins of the laminin family (subunits (alpha ,beta ,,text{and},gamma )), and also integrins (subunits (alpha ), and (beta )) (Fig.聽3B, and Supplementary Table聽S1). Collagen type IV is highly expressed in glioblastoma and implicated in tumor progression27. In addition, it has been observed that the activation of two integrins, ITGB3 and ITGB5, contributes to radio-resistance28.Figure 3Global view of upregulated genes at T24 post-radiation in glioblastoma cells. (A) Gene ontology analysis of upregulated genes (B) Protein-protein interaction networks according to STRING99 showing genes associated with extracellular matrix organization and response to interferon. Gene clusters based on the strength of connection and gene function are identified by color. Lines colors indicate type of association: light green, association based on literature findings; blue indicates gene co-occurrence; magenta indicates experimental evidence.Full size imageRadiation treatment also induced the expression of genes involved in neuronal differentiation and axonogenesis. Some key genes in these categories include SRC, VEGFA, EPHA4, DLG4, MAPK3, BMP4, and several semaphorins. These genes can have very different effects on glioblastoma development, with some factors activating oncogenic programs and others behaving as tumor suppressors. Similarly, type I interferon鈥檚 effects on treatment are varied. For instance, interferon inhibited proliferation of glioma stem cells and their sphere-forming capacity and induced STAT3 activation29. On the other hand, chronic activation of type I IFN signaling has been linked to adaptive resistance to therapy in many tumor types30.Activation of oncogenic signals post-radiation could counteract treatment effects and later contribute to relapse. We searched the set of highly up-regulated genes post-radiation for previously identified radio-resistance genes in glioblastoma, oncogenic factors and genes whose high expression is associated with poor prognosis (Supplementary Table聽S5). In Table聽1, we list these genes according to their molecular function. Since several of these genes have never been characterized in the context of glioblastoma, our results open new opportunities to prevent radio-resistance and increase treatment efficiency. Importantly, there are inhibitors available against several of these proteins (Supplementary Table聽S4).Table 1 List of oncogenic factors, genes whose high expression is associated with poor survival and genes previously associated with radio-resistance in GBM that showed increased expression upon radiation. Genes are listed according to molecular function.Full size tableChanges in lncRNA profile in response to radiationlncRNAs have been implicated in the progression of glioblastoma31, but their role in response to ionizing radiation is still poorly understood. We identified 161 lncRNAs with expression alterations in T1 vs. T24 comparisons. Analysis of this set with LnC2Cancer32 identified several lncRNAs aberrantly expressed in cancer and with relevance to prognosis (Supplementary Table聽S1). We also detected significant downregulation of MIR155HG, whose high expression is associated with glioma progression and poor survival33. Another downregulated lncRNA with relevance to prognosis is linc000152, whose increased expression has been observed in multiple tumor types34. On the other hand, we observed a significant upregulation of two 鈥渙ncogenic鈥?lncRNAs, NEAT1 and FTX. NEAT1 is associated with tumor growth, grade, and recurrence rate in gliomas35, while FTX promotes cell proliferation and invasion through negatively regulating miR-342鈥?p36. Thus, if further studies corroborate NEAT1 and FTX as players in radio-resistance, targeting these lncRNAs should be considered to improve treatment response.Effect of radiation on splicingAlternative splicing impacts genes implicated in all hallmarks of cancer37 and is an important component of changes in expression triggered by ionizing radiation38. All types of splicing events (exon skipping, alternative donor, and acceptor splice sites, multiple exclusive exons, and intron retention) were affected similarly upon exposure to radiation (Supplementary Table聽S6). At T24, we observed that transcripts associated with RNA-related functions (especially translation), showed the most splicing alterations. Affected transcripts encode ribosomal proteins, translation initiation factors, regulators of translation, and genes involved in tRNA processing and endoplasmic reticulum. Other enriched GO terms include mRNA and ncRNA processing, mRNA degradation, and modification. Catabolism is another process associated with several enriched terms, suggesting that splicing alterations in genes involved in catabolic routes could ultimately contribute to apoptosis (Fig.聽4A,B, and Supplementary Table聽S7). Changes in the splicing profile are likely driven by an alteration in the expression of splicing regulators. In Table聽2, we show a list of splicing factors displaying strong expression alterations. Among those previously connected to glioblastoma development, LGALS3 is the most extensively characterized. LGALS3 is a galactosidase-binding lectin and non-classic RNA binding protein implicated in pre-mRNA splicing and regulation of proliferation, adhesion, and apoptosis; LGALS3 also is a marker of the early stage of glioma39.Figure 4Impact of radiation on the splicing profile of glioblastoma cells. (A) GO-enriched terms among genes showing changes in splicing profiles at T24. GO-enriched terms are summarized using REVIGO97. (B) Protein-protein interaction networks according to STRING98 showing genes associated with RNA-related functions whose splicing profiles displayed alterations at T24. Gene clusters based on the strength of connection and gene function are identified by color. Lines color indicate type of association: light green, an association based on literature findings; blue indicates gene co-occurrence; magenta indicates experimental evidence.Full size imageTable 2 Splicing regulators showing changes in expression 24鈥塰ours post-radiation. Factors showing an increase in the expression are shown in bold, while factors showing a decrease in the expression are represented in italic.Full size tableDifferential translational efficiencyWe used Ribo-seq40 to identify changes in translation efficiency triggered by radiation (Supplementary Table聽S8). Translation, protein localization, and metabolism appear as top enriched terms among downregulated genes in T1 vs. T24 comparisons (Supplementary Table聽S9). In particular, several ribosomal proteins, along with translation initiation factors and mTOR, showed a significant decrease in translation efficiency (Fig.聽5A,B). Overall, these results indicate repression of the translation machinery post-radiation exposure and its strong auto-regulation. Since changes in components of the translation machinery are occurring at all levels (transcription, splicing, and translation) at T24, we expect that major translational alterations take place in later stages of post-radiation.Figure 5Impact of radiation on the translation profile of glioblastoma cells. (A) GO-enriched terms among genes showing changes in translation efficiency at T24. GO-enriched terms are summarized using REVIGO97. (B) Protein-protein interaction network, according to STRING98 showing genes whose translation efficiency decreased at T24. Gene clusters based on the strength of connection and gene function are identified by color. Line colors indicate the type of association: light green, an association based on literature findings; blue indicates gene co-occurrence; magenta indicates experimental evidence.Full size imageIn the upregulated set, we highlight three genes FTH1, APIP, and LRIG2 that could potentially counteract the impact of radiation (Supplementary Table聽S10). FTH1 encodes the heavy subunit of ferritin, an essential component of iron homeostasis41. Pang et al.42, reported that H-ferritin plays an important role in radio-resistance in glioblastoma by reducing oxidative stress and activating DNA repair mechanisms. The depletion of ferritin causes down-regulation of ATM, leading to increased DNA sensitivity towards radiation. APIP is involved in the methionine salvage pathway and has a key role in various cell death processes. It can inhibit mitochondria-mediated apoptosis by directly binding to APAF-143. LRIG2 is a member of the leucine-rich and immunoglobulin-like domain family44, and its expression levels are positively correlated with the glioma grade and poor survival. LRIG2 promotes proliferation and inhibits apoptosis of glioblastoma cells through activation of EGFR and PI3K/Akt pathway45.Crosstalk between regulatory processesParallel analyses of transcription, splicing, and translation alterations in the early response to radiation provided an opportunity to identify crosstalk between different regulatory processes. The datasets showed little overlap, with just a few genes showing alterations in two different regulatory processes. However, we identified several shared GO terms when comparing the results of alternative splicing, mRNA levels (transcription), and translation efficiency (Supplementary Table聽S10). These terms show two main groups of biological processes. The first group indicates that the expression of genes involved in DNA and RNA synthesis and metabolism is particularly compromised. The second group is related to translation initiation. Ribosomal proteins were particularly affected (Figs.聽4 and 5). There is growing support for the concept of specialized ribosomes. According to this model, variations in the composition of the ribosome due to the presence or absence of certain ribosomal proteins or alternative isoforms could ultimately dictate which mRNAs get preferentially translated46. Therefore, these alterations could later lead to translation changes of a specific set of genes.DiscussionWe performed the first integrated analysis to define global changes associated with the early response to radiation in glioblastoma. Our approach allowed the identification of 鈥渃onserved鈥?alterations at the transcription, splicing and translation levels and defined possible crosstalk between different regulatory processes. Alterations at the level of transcription were dominant, but changes affecting genes implicated in RNA mediated regulation were ubiquitous; they indicate that these processes are important components in radio-response and suggest that more robust changes in splicing and translation might take place later.E2F1, E2F2, E2F8, and FOXM1 as major drivers of transcriptional responses upon radiationWe observed marked changes in the mRNA levels of genes implicated in cell cycle, DNA replication, and repair 24鈥塰ours (T24) after radiation. Downregulation of several transcription factors, most of them members of the zinc finger family, was observed at one hour post-radiation. This group displays high correlations in expression within glioblastoma samples from TCGA, suggesting that they might work together to regulate gene expression. Unfortunately, most are poorly characterized, and the lack of information has prevented establishing further connections to changes in the cell cycle and DNA replication that we observed at T24.GSEA and expression correlation analysis suggested that the downregulation of members of the E2F family is likely responsible for several of the expression changes we observed at T24. E2Fs have been defined as major transcriptional regulators of the cell cycle. The family has eight members that could act as activators or repressors depending on the context, and are known to regulate one another. They are upregulated in many tumors due to overexpression of cyclin-dependent kinases (CDKs), inactivation of CDK inhibitors, or RB Transcriptional Corepressor 1 (RB1) and are linked to poor prognosis. Alterations in E2F genes can induce cancer in mice47,48. Specifically, we found that three E2F members showed decreased expression upon radiation: E2F1, E2F2, and E2F8, all of which have been previously implicated in glioblastoma development.E2F1 is probably the best-characterized member of the E2F family. Besides its known effect on cell cycle regulation and DNA replication, it is also a positive regulator of telomerase activity, binding the TERT promoter49. Recent studies show that lncRNAs and miRNAs function in an antagonistic fashion to regulate E2F1 expression, ultimately affecting cell proliferation, glioblastoma growth, and response to therapy50,51,52. E2F2 has been linked to the maintenance of glioma stem cell phenotypes and cell transformation53,54. Several tumor suppressor miRNAs (let7b, miR-125b, miR-218, and miR-138) decrease the proliferation and growth of glioblastoma cells by targeting E2F253,55,56,57. Although still poorly characterized in the context of glioblastoma, E2F8 drives an oncogenic phenotype in glioblastoma. Its expression is modulated by HOXD-AS1, which serves as a sponge and prevents the binding of miR-130a to E2F8 transcripts58. FOXM1 is another potential regulator of the group of cell cycle and DNA replication genes affected by radiation. FOXM1 is established as an important player in chemo- and radio-resistance and a contributor to glioma stem cell phenotypes5,6,59,60,61,62,63,64. FOXM1 and E2F protein have a close relationship and share target genes65. Additionally, FOXM1- and E2F2-mediated cell cycle transitions are implicated in the malignant progression of IDH1 mutant glioma66.E2F and FOXM1 targeting could be considered as an option to increase radio-sensitivity. Since the development of transcription factor inhibitors is very challenging, an alternative to be considered is the use of BET (bromodomain and external) inhibitors. BET is a family of proteins that function as readers for histone acetylation and modulates the transcription of oncogenic programs67. Recent studies in glioblastoma with a new BET inhibitor, dBET6, showed promising results and established that its effect on cancer phenotypes comes via disruption of the transcriptional program regulated by E2F168.RNA processing and regulation as novel categories in radio-responseBesides the expected changes in expression of cell cycle, DNA replication and repair genes, radiation affected preferentially the expression of genes implicated in RNA processing and regulation. Additionally, we identified a co-expression module containing multiple genes associated with translation initiation, rRNA and snoRNA processing, RNA localization, and ribonucleoprotein complex biogenesis.Many regulators of RNA processing are implicated in glioblastoma development, and splicing alterations affect all hallmarks of cancer69. Radiation-induced changes in the splicing patterns of oncogenic factors and tumor suppressors such as CDH11, CHN1, CIC, EIF4A2, FGFR1, HNRNPA2B1, MDM2, NCOA1, NUMA1, RPL22, SRSF3, TPM3, APC, CBLB, FAS, PTCH1, and SETD2. We also observed changes in expression of four RNA processing regulators previously identified in genomic/functional screening for RNA binding proteins contributing to glioblastoma phenotypes: MAGOH, PPIH, ALYREF, and SNRPE70.Potential new targets to increase radio-sensitivity and prevent relapseActivation of oncogenic signals is an undesirable effect of radiation that could influence treatment response and contribute to relapse. We observed increased expression or translation and splicing alterations of a number of pro-oncogenic factors, genes whose high expression is associated with poor survival and genes previously implicated in radio-resistance.Among genes with the most marked increase in expression upon radiation, we identified members of the Notch pathway (HES2, NOTCH3, MFNG, and JAG2). Notch activation has been linked to radio-resistance in glioblastoma, and Notch targeting improves the results of radiation treatment71,72. We also identified several genes associated with the PI3K-Akt, Ras, and Rap1 signaling pathways that increased expression levels upon radiation exposure. Targeting these pathways has been explored as a therapeutic option in glioblastoma71,73. Other oncogenic factors relevant to glioblastoma that had increased expression after radiation exposure include SRC, MUC1, LMO2, PML, PDGFR (beta ), BCL3, and BCL6.Anti-apoptotic genes (BCL6, RRM2B, and IDO1) also showed increased expression upon radiation. BCL6 is a member of the ZBTB family of transcription factors, which functions as a p53 pathway repressor. The blockage of the interaction between BCL6 and its cofactors has been established as a novel therapeutic route to treat glioblastoma74. RRM2B is an enzyme essential for DNA synthesis and participates in DNA repair, cell cycle arrest, and mitochondrial homeostasis. The depletion of RRM2B resulted in ADR-induced apoptosis, growth inhibition, and enhanced sensitivity to chemo- and radiotherapy75. IDO1 is a rate-limiting metabolic enzyme involved in tryptophan metabolism that is highly expressed in numerous tumor types76. The combination of radiation therapy and IDO1 inhibition enhanced therapeutic response77.Among genes whose high expression correlates with decreased survival in glioblastoma, we identified several components of the 鈥渕atrisome鈥?and associated factors (FAM20C, SEMA3F, ADAMTSL4, ADAMTS14, SERPINA5, and CRELD1). The core of the 鈥渕atrisome鈥?contains ECM proteins, while associated proteins include ECM-modifying enzymes and ECM-binding growth factors. This complex of proteins assembles and modifies extracellular matrices, contributing to cell survival, proliferation, differentiation, morphology, and migration78. In addition, several genes of the proteinase inhibitor SERPIN family (SERPINA3, SERPINA12, SERPINA5, and SERPINI1) implicated in ECM regulation79 were among those with high levels of expression upon radiation.In conclusion, our results generated a list of candidates for combination therapy. Contracting the effect of oncogenic factors and genes linked to poor survival could increase radio-sensitivity and treatment efficiency. Importantly, there are known inhibitors against several of these proteins (Supplementary Table聽S5). Moreover, RNA processing and translation were determined to be important components of radio-response. These additional vulnerable points could be explored in therapy, as many inhibitors against components of the RNA processing and translation machinery have been identified80,81.MethodsCell culture and radiation treatmentA172 was obtained from ATCC. U251 and U343 cells were obtained from the University of Uppsala (Sweden) and maintained in Dulbecco鈥檚 Modified Eagle Medium (DMEM, Hyclone) supplemented with 10% fetal bovine serum, 1% Penicillin/Streptomycin at 37鈥壜癈 in 5% CO2-humidified incubators and were sub-cultured twice a week. Cells were plated after appropriate dilution, and ionizing radiation treatment was performed on the next day at a dose of 5 Gray (Gy). 5鈥塆y was chosen since in current hypofractionated regimen, radiation therapy is delivered over the course of 5 fractions at 5鈥塆y per fraction. A cabinet X-ray system (CP-160 Cabinet X-Radiator; Faxitron X-Ray Corp., Tucson, AZ) was used. Glioma stem line 3565 was established in Dr. Jeremy Rich;s lab; cells were cultured in Neurosphere Media consisting of Neurobasal-A supplemented with B-27, 50鈥塶g/ml EGF, and 50鈥塶g/ml hFGF.qRT-PCR analysisAfter exposure to ionizing radiation, cells were cultured for 14, 24, and 48鈥塰ours (hrs). Total RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer鈥檚 instructions. Reverse transcription was performed using High-Capacity cDNA reverse transcription kit (Applied Biosystems). Quantitative PCR was performed using TaqMan Universal PCR Master Mix (Applied Biosystems) or PowerUp SYBR Green Master Mix (Applied Biosystems) and reactions were performed on ({text{ViiA}}^{text{TM}}) 7 Real-Time PCR System (Applied Biosystems). Data were acquired using ViiA 7 RUO software (Applied Biosystems) and analyzed using the 2-螖螖 CT method with GAPDH as an endogenous control. Probes and primers used in qRT-PCR are listed in Supplementary Table聽S11.RNA preparation, RNA-seq and ribosome profiling (Ribo-seq)RNA was purified using a GeneJet RNA kit from Thermo Scientific. The TruSeq Ribo Profile (Mammalian) kit from Illumina was used to prepare material for ribosome profiling (Ribo-seq). RNA-seq and Ribo-seq samples were prepared according to Illumina protocols and sequenced at UTHSCSA Genome Sequencing Facility.Overall strategy to identify gene expression alterations upon radiationTo identify the most relevant expression alterations in the early response to radiation, we analyzed samples from U251 and U343 cells collected at 0 (T0), 1 (T1), and 24 (T24) hours post-radiation. The time of one hour and 24鈥塰ours were chosen as this reflects the immediacy of the DNA damage response and resolution of double stranded DNA breaks. The production of double stranded DNA breaks occurs nearly immediately with the subsequent resolution of the breaks nearly immediately (within 15鈥?0鈥塵inutes after exposure to ionizing radiation) with near resolution of all breaks by 24鈥塰ours. To capture the progressive dynamics of expression alterations, we compared T0 to T1 samples and T1 to T24 samples. Our strategy to identify the most relevant alterations in expression with maximal statistical power was to combine all samples and use a design matrix with cell type defined as a covariate with time points (Supplementary Fig.聽S1).Sequence data pre-processing and mappingThe quality of raw sequences reads from RNA-Seq and Ribo-Seq datasets were assessed using FastQC82. Adaptor sequences and low-quality score (phred quality score 5) bases were trimmed from RNA-Seq and Ribo-Seq datasets with TrimGalore (v0.4.3)83. The trimmed reads were then aligned to the human reference genome sequence (Ensembl GRCh38.p7) using STAR aligner (v.2.5.2b)84 with GENCODE85 v25 as a guided reference annotation, allowing a mismatch of at most two positions. All the reads mapping to rRNA and tRNA sequences were filtered out before downstream analysis. Most reads in the Ribo-seq samples mapped to the CDS. The periodicity analysis was performed using ribotricer86. The number of reads assigned to annotated genes included in the reference genome was obtained by htseq-count87.Differential gene expression analysisFor differential expression analysis, we performed counting over exons for the RNA-seq samples. For translational efficiency analyses, counting was restricted to the CDS. Differential gene expression analysis was performed by employing the DESeq2 package88, with read counts from both U251 and U343 cell samples as inputs. Both the cell line and time were treated as covariates along with their interaction term. To find differentially expressed genes that show changes between two time points, we used time specific contrasts. The p-values were adjusted for controlling for the false discovery rate (adjusted p-value) using the Benjamini and Hochberg (BH) procedure89. Differentially expressed genes were defined with an adjusted p-value 0.05.Weighted gene co-expression network analysisWGCNA22 uses pairwise correlations on expression values to identify genes significantly co-expressed across samples. We used this approach to identify gene modules with significant co-expression variations as an effect of radiation. The entire set of expressed genes, defined here as those with one or higher transcripts per million higher (TPM), followed by variance stabilization) from U251 and U343 samples were clustered separately using the signed network strategy. We used the ({Z}_{text{summary}})90 statistic as a measure of calculating the degree of module preservation between U251 and U343 cells. ({Z}_{text{summary}}) is a composite statistic defined as the average of the density and connectivity based statistic. Thus, both density and connectivity are considered for defining the preservation of a module. Modules with ({Z}_{text{summary}}, ,5) were considered as significantly preserved. The expression profile of all genes in each co-expression module can be summarized as one 鈥渆igengene鈥? We used the eigengene-based connectivity (kME) defined as the correlation of a gene with the corresponding module eigengene to assess the connectivity of genes in a module. The intramodular hub genes were then defined as genes with the highest module membership values (kME鈥?gt;鈥?.9). All analysis was performed using the R package WGCNA. The protein-coding hub genes were then selected for gene ontology enrichment analysis.Translational efficiency analysisWe used Riborex91 to perform differential translational efficiency analysis. The underlying engine selected was DESeq288. DESeq2 estimates a single dispersion parameter per gene. However, RNA-Seq and Ribo-Seq libraries can have different dispersion parameters owing to different protocols. We estimated the dispersion parameters for RNA-Seq and Ribo-Seq samples separately and found them to be significantly different (mean difference = 0.04, p-value ( ,2.2e-16)). This leads to a skew in translational-efficiency p-value distribution since the estimated null model variance for the Wald test is underestimated. To address this issue, we performed a p-value correction using fdrtool92 that re-estimates the variance using an empirical bayes approach. Normalized read counts for both Ribo-seq and RNA-seq samples are provided in Supplementary Table聽S12.Alternative splicing analysisAlternative splicing analysis was performed using rMATS93. All reads were trimmed using cutadapt94 with parameters (-u -13 -U -13) to ensure trimmed reads had equal lengths (138鈥塨p). rMATs was run with default parameters in paired end mode (-t paired) and read length set to 138鈥塨p (-len 138) using GENCODE GTF (v25) and STAR index for GRCh38.Gene ontology (GO) and pathway enrichment analysisTo classify the functions of differentially enriched genes, we performed GO enrichment, and the Reactome pathway95 analysis using Panther96. For both analyses, we considered terms to be significant if BH adjusted p-values weree 0.05, and fold enrichment is 2.0. Further, we used REVIGO97 to reduce redundancy of the enriched GO terms and visualize the semantic clustering of the identified top-scoring terms. We used STRING database (v10)98 to construct protein-protein interaction networks and determine associations among genes in a given dataset. The interactions are based on experimental evidence procured from high-throughput experiments text-mining, and co-occurrence. Only high-confidence (0.70) nodes were retained.Expression correlation analysisGene expression correlation analysis was done using Gliovis24 using glioblastoma samples (RNAseq) from the TCGA. To select correlated genes, we used Pearson correlation, (R, ,0.3), and p-value 0.05. A list of genes affecting survival in glioblastoma was downloaded from GEPIA99. A list of long non-coding RNAs (lncRNAs) implicated in glioma development was obtained from Lnc2Cancer32. Drug-gene interactions were identified using the Drug-Gene Interaction Database100.Accession codesThe processed data of read abundance matrices are available through GEO accession GSE141013. Code for differential expression analysis and translational efficiency analysis is available at https://github.com/saketkc/2019_radiation_gbm. Change history04 August 2020An amendment to this paper has been published and can be accessed via a link at the top of the paper.References1.Stupp, R. et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. New Engl. J. Medicine 352, 987鈥?96 (2005).CAS聽 Article聽Google Scholar聽 2.Gilbert, M. R. et al. A Randomized Trial of Bevacizumab for Newly Diagnosed Glioblastoma. New Engl. J. Medicine 370, 699鈥?08 (2014).CAS聽 Article聽Google Scholar聽 3.Hegi, M. E. et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. New Engl. J. Medicine 352, 997鈥?003 (2005).CAS聽 Article聽Google Scholar聽 4.Weinstein, J. N. et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat. Genet. 45, 1113 (2013).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 5.Lee, Y. et al. FoxM1 Promotes Stemness and Radio-Resistance of Glioblastoma by Regulating the Master Stem CellRegulator Sox2. PLoS ONE 10, e0137703 (2015).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 6.Maachani, U. B. et al. FOXM1 and STAT3 interaction confers radioresistance in glioblastoma cells.Oncotarget 7, 77365 (2016).7.Cheng, L. et al. L1CAM regulates DNA damage checkpoint response of glioblastoma stem cells through NBS1. TheEMBO J. 30, 800鈥?13 (2011).CAS聽Google Scholar聽 8.Han, X. et al. Notch Represses Transcription by PRC2 Recruitment to the Ternary Complex. Mol. Cancer Res. 15, 1173鈥?183 (2017).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 9.Balbous, A. et al. A radiosensitizing effect of rad51 inhibition in glioblastoma stem-like cells. BMC Cancer 16, 604 (2016).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 10.Kim, S.-H. et al. EZH2 protects glioma stem cells from radiation-induced cell death in a MELK/FOXM1-dependentmanner. Stem Cell Reports 4, 226鈥?38 (2015).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 11.Ahmed, S. U. et al. Selective inhibition of parallel dna damage response pathways optimizes radiosensitization ofglioblastoma stem-like cells. Cancer Res. 75, 4416鈥?428 (2015).CAS聽 PubMed聽 Article聽Google Scholar聽 12.Karim, A. et al. Differential cyclooxygenase-2 enzyme expression in radiosensitive versus radioresistant glioblastomamultiforme cell lines. Anticancer. Res. 25, 675鈥?79 (2005).CAS聽 PubMed聽Google Scholar聽 13.Vellanki, S. H. K. et al. Small-molecule xiap inhibitors enhance纬-irradiation-induced apoptosis in glioblastoma. Neoplasia 11, 743鈥揥9 (2009).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 14.Ma, H. et al. Transcriptome analysis of glioma cells for the dynamic response to纬-irradiation and dual regulation ofapoptosis genes: a new insight into radiotherapy for glioblastomas. Cell Death Dis. 4, e895 (2013).CAS聽 Article聽Google Scholar聽 15.Godoy, P. et al. Ionizing radiation-induced gene expression changes in TP53 proficient and deficient glioblastoma celllines. Mutat. Res. Toxicol. Environ. Mutagen. 756, 46鈥?5 (2013).CAS聽 Article聽Google Scholar聽 16.Bhat, K. P. et al. Mesenchymal differentiation mediated by NF-魏B promotes radiation resistance in glioblastoma. CancerCell 24, 331鈥?46 (2013).CAS聽Google Scholar聽 17.Effenberger, K. A., Urabe, V. K. Jurica, M. S. Modulating splicing with small molecular inhibitors of the spliceosome. Wiley Interdiscip. Rev. RNA 8, e1381 (2017).Article聽 CAS聽Google Scholar聽 18.Dvinge, H., Kim, E., Abdel-Wahab, O. Bradley, R. K. RNA splicing factors as oncoproteins and tumour suppressors. Nat. Rev. Cancer 16, 413 (2016).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 19.Meliso, F. M., Hubert, C. G., Galante, P. A. F. Penalva, L. O. Rna processing as an alternative route to attackglioblastoma. Hum. Genet. 136, 1129鈥?141 (2017).PubMed Central聽 Article聽 CAS聽Google Scholar聽 20.Hubert, C. G. et al. Genome-wide RNAi screens in human brain tumor isolates reveal a novel viability requirement forPHF5A. Genes Dev. 27, 1032鈥?045 (2013).CAS聽 Article聽Google Scholar聽 21.Grzmil, M. Hemmings, B. A. Translation regulation as a therapeutic target in cancer. Cancer research 72, 3891鈥?900 (2012).CAS聽 PubMed聽 Article聽Google Scholar聽 22.Langfelder, P. Horvath, S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinforma. 9, 559 (2008).Article聽 CAS聽Google Scholar聽 23.Kuleshov, M. V. et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 44, W90鈥揥97 (2016).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 24.Bowman, R. L., Wang, Q., Carro, A., Verhaak, R. G. Squatrito, M. GlioVis data portal for visualization and analysis ofbrain tumor expression datasets. Neuro-Oncology 19, 139鈥?41 (2016).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 25.Nandi, D., Cheema, P. S., Jaiswal, N. Nag, A. FoxM1: repurposing an oncogene as a biomarker. In Seminars in CancerBiology, vol. 52, 74鈥?4 (Elsevier, 2018).26.Chen, H.-Z., Tsai, S.-Y. Leone, G. Emerging roles of E2Fs in cancer: an exit from cell cycle control. Nat. Rev. Cancer 9, 785 (2009).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 27.Payne, L. S. Huang, P. H. The pathobiology of collagens in glioma. Mol. Cancer Res. 11, 1129鈥?140 (2013).CAS聽 PubMed聽 Article聽Google Scholar聽 28.Monferran, S. et al. 伪v尾3 and伪v尾5 integrins control glioma cell response to ionising radiation through ilk and rhob. Int. J. Cancer 123, 357鈥?64 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 29.Du, Z. et al. The effects of type I interferon on glioblastoma cancer stem cells.Biochem. Biophys. Res. Commun. 491, 343鈥?48 (2017).CAS聽 Article聽Google Scholar聽 30.Budhwani, M., Mazzieri, R. Dolcetti, R. Plasticity of type I interferon-mediated responses in cancer therapy: fromanti-tumor immunity to resistance. Frontiers in Oncology 8, 322 (2018).PubMed聽 PubMed Central聽 Article聽Google Scholar聽 31.Peng, Z., Liu, C. Wu, M. New insights into long noncoding rnas and their roles in glioma. Mol. Cancer 17, 61 (2018).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 32.Ning, S. et al. Lnc2Cancer: a manually curated database of experimentally supported lncRNAs associated with varioushuman cancers. Nucleic Acids Res. 44, D980鈥揇985 (2015).MathSciNet聽 PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 33.Wu, X. et al. Blocking mir155hg/mir-155 axis inhibits mesenchymal transition in glioma. Neuro-oncology 19, 1195鈥?205 (2017).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 34.Miao, C. et al. Clinicopathological and prognostic role of long noncoding rna linc00152 in various human neoplasms:Evidence from meta-analysis. BioMed Res. Int. 2017, 1鈥?1 (2017). Google Scholar聽 35.Zhang, H. et al. Long noncoding rna neat1 regulate papillary thyroid cancer progression by modulating mir-129-5p/klk7expression. J. Cell. Physiol. 233, 6638鈥?648 (2018).ADS聽 CAS聽 PubMed聽 Article聽Google Scholar聽 36.Zhang, W. et al. Long noncoding rna ftx is upregulated in gliomas and promotes proliferation and invasion of gliomacells by negatively regulating mir-342-3p. Lab. Investig. 97, 447 (2017).CAS聽 PubMed聽 Article聽Google Scholar聽 37.Climente-Gonz谩lez, H., Porta-Pardo, E., Godzik, A. Eyras, E. The functional impact of alternative splicing in cancer. Cell Reports 20, 2215鈥?226 (2017).PubMed聽 Article聽 CAS聽Google Scholar聽 38.Macaeva, E. et al. Radiation-induced alternative transcription and splicing events and their applicability to practicalbiodosimetry. Sci. Reports 6, 19251 (2016).ADS聽 CAS聽Google Scholar聽 39.Binh, N. H. et al. Galectin-3 in preneoplastic lesions of glioma. J. Neuro-Oncology 111, 123鈥?32 (2013).CAS聽 Article聽Google Scholar聽 40.Ingolia, N. T. Ribosome profiling: new views of translation, from single codons to genome scale. Nat. Rev. Genet. 15, 205 (2014).CAS聽 PubMed聽 Article聽Google Scholar聽 41.Orino, K. Watanabe, K. Molecular, physiological and clinical aspects of the iron storage protein ferritin. The Vet. J. 178, 191鈥?01 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 42.Pang, M., Liu, X., Slagle-Webb, B., Madhankumar, A. Connor, J. Role of h-ferritin in radiosensitivity of human gliomacells. J Cancer Biol Treat 3, 1鈥?0 (2016). Google Scholar聽 43.Cho, D.-H. et al. Induced inhibition of ischemic/hypoxic injury by apip, a novel apaf-1-interacting protein. J. Biol. Chem. 279, 39942鈥?9950 (2004).CAS聽 PubMed聽 Article聽Google Scholar聽 44.Simion, C., Cedano-Prieto, M. E. Sweeney, C. The lrig family: enigmatic regulators of growth factor receptor signaling. Endocrine-related cancer 21, R431鈥揜443 (2014).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 45.Xiao, Q. et al. Soluble lrig2 ectodomain is released from glioblastoma cells and promotes the proliferation and inhibitsthe apoptosis of glioblastoma cells in vitro and in vivo in a similar manner to the full-length lrig2. PLoS ONE 9, e111419 (2014).ADS聽 PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 46.Dalla Venezia, N., Vincent, A., Marcel, V., Catez, F. Diaz, J.-J. Emerging Role of Eukaryote Ribosomes in TranslationalControl. Int. J. Mol. Sci. 20, 1226 (2019).CAS聽 PubMed Central聽 Article聽Google Scholar聽 47.Kent, L. N. Leone, G. The broken cycle: E2f dysfunction in cancer. Nat. Rev. Cancer 19, 326鈥?38 (2019).CAS聽 PubMed聽 Article聽Google Scholar聽 48.Thurlings, I. de Bruin, A. E2f transcription factors control the roller coaster ride of cell cycle gene expression. In Methods in Molecular Biology, 71鈥?8 (Springer New York, 2016).49.Alonso, M. M. et al. Expression of Transcription Factor E2F1 and Telomerase in Glioblastomas: Mechanistic Linkageand Prognostic Significance. J. Natl. Cancer Inst. 97, 1589鈥?600 (2005).CAS聽 PubMed聽 Article聽Google Scholar聽 50.Li, X., Zhang, H. Wu, X. Long noncoding RNA DLX6-AS1 accelerates the glioma carcinogenesis by competingendogenous sponging miR-197-5p to relieve e2f1. Gene 686, 1鈥? (2019).CAS聽 PubMed聽 Article聽Google Scholar聽 51.Xia, L., Nie, D., Wang, G., Sun, C. Chen, G. FER1l4/miR-372/e2f1 works as a ceRNA system to regulate theproliferation and cell cycle of glioma cells. J. Cell. Mol. Medicine 23, 3224鈥?233 (2019).CAS聽 Article聽Google Scholar聽 52.Yang, B., Meng, Q., Sun, Y., Gao, L. Yang, J. Long non-coding RNA SNHG16 contributes to glioma malignancy bycompetitively binding miR-20a-5p with E2F1. J. Biol. Regul. Homeost. Agents 32, 251鈥?61 (2018).CAS聽Google Scholar聽 53.Wu, N. et al. miR-125b regulates the proliferation of glioblastoma stem cells by targeting e2f2. FEBS Lett. 586, 3831鈥?839 (2012).CAS聽 PubMed聽 Article聽Google Scholar聽 54.Okamoto, O. K., Oba-Shinjo, S. M., Lopes, L. Marie, S. K. N. Expression of HOXC9 and E2F2 are up-regulated inCD133+ cells isolated from human astrocytomas and associate with transformation of human astrocytes.Biochimica etBiophys. Acta (BBA)-Gene Struct. Expr. 1769, 437鈥?42 (2007).CAS聽 Article聽Google Scholar聽 55.Song, H. et al. Let-7b inhibits the malignant behavior of glioma cells and glioma stem-like cells via downregulation ofE2F2. J. Physiol. Biochem. 72, 733鈥?44 (2016).CAS聽 PubMed聽 Article聽Google Scholar聽 56.Qiu, S. et al. Suppression of tumorigenicity by MicroRNA-138 through inhibition of EZH2-CDK4/6-pRb-e2f1 signalloop in glioblastoma multiforme. Biochimica et Biophys. Acta (BBA) - Mol. Basis Dis. 1832, 1697鈥?707 (2013).CAS聽 Article聽Google Scholar聽 57.Zhang, Y. et al. MiR-218 inhibited growth and metabolism of human glioblastoma cells by directly targeting e2f2. Cell. Mol. Neurobiol. 35, 1165鈥?173 (2015).CAS聽 PubMed聽 Article聽Google Scholar聽 58.Chen, Y. et al. HOXD-AS1/miR-130a sponge regulates glioma development by targeting e2f8. Int. J. Cancer 142, 2313鈥?322 (2018).CAS聽 PubMed聽 Article聽Google Scholar聽 59.Gouaz茅-Andersson, V. et al. FGFR1/FOXM1 pathway: a key regulator of glioblastoma stem cells radioresistance and aprognosis biomarker. Oncotarget 9, 31637 (2018).PubMed聽 PubMed Central聽 Article聽Google Scholar聽 60.Ma, Q., Liu, Y., Shang, L., Yu, J. Qu, Q. The FOXM1/BUB1b signaling pathway is essential for the tumorigenicity andradioresistance of glioblastoma. Oncol. Reports 38, 3367鈥?375, https://doi.org/10.3892/or.2017.6032 (2017).CAS聽 Article聽Google Scholar聽 61.Zhang, S. et al. m 6 a demethylase ALKBH5 maintains tumorigenicity of glioblastoma stem-like cells by sustainingFOXM1 expression and cell proliferation program. Cancer Cell 31, 591鈥?06.e6 (2017).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 62.Quan, J.-J., Song, J.-N. Qu, J.-Q. PARP3 interacts with FoxM1 to confer glioblastoma cell radioresistance. Tumor Biol. 36, 8617鈥?624 (2015).CAS聽 Article聽Google Scholar聽 63.Hua Gong, A. et al. FoxM1 drives a feed-forward STAT3-activation signaling loop that promotes the self-renewal andtumorigenicity of glioblastoma stem-like cells. Cancer Res. 75, 2337鈥?348 (2015).Article聽 CAS聽Google Scholar聽 64.Zhang, N. et al. FoxM1 inhibition sensitizes resistant glioblastoma cells to temozolomide by downregulating theexpression of DNA-repair gene rad51. Clin. Cancer Res. 18, 5961鈥?971 (2012).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 65.Fischer, M., Grossmann, P., Padi, M. DeCaprio, J. A. Integration of TP53, DREAM, MMB-FOXM1 and RB-e2f targetgene analyses identifies cell cycle gene regulatory networks. Nucleic Acids Res. 44, 6070鈥?086 (2016).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 66.Bai, H. et al. Integrated genomic characterization of IDH1-mutant glioma malignant progression. Nat. Genet. 48, 59鈥?6 (2015).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 67.Belkina, A. C. Denis, G. V. BET domain co-regulators in obesity, inflammation and cancer. Nat. Rev. Cancer 12, 465鈥?77 (2012).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 68.Xu, L. et al. Targetable BET proteins- and e2f1-dependent transcriptional program maintains the malignancy ofglioblastoma. Proc. Natl. Acad. Sci. 115, E5086鈥揈5095 (2018).CAS聽 PubMed聽 Article聽Google Scholar聽 69.Meliso, F. M., Hubert, C. G., Galante, P. A. F. Penalva, L. O. RNA processing as an alternative route to attackglioblastoma. Hum. Genet. 136, 1129鈥?141 (2017).PubMed Central聽 Article聽 CAS聽Google Scholar聽 70.Correa, B. R. et al. Functional genomics analyses of RNA-binding proteins reveal the splicing regulator SNRPB as anoncogenic candidate in glioblastoma. Genome Biol. 17, 125 (2016).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 71.Liu, Y., Shen, Y., Sun, T. Yang, W. Mechanisms regulating radiosensitivity of glioma stem cells. Neoplasma 64, 655鈥?65 (2017).CAS聽 PubMed聽 Article聽Google Scholar聽 72.Wang, J. et al. Notch Promotes Radioresistance of Glioma Stem Cells. Stem Cells 28, 17鈥?8 (2010).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 73.Hannen, R., Hauswald, M. Bartsch, J. W. A rationale for targeting extracellular regulated kinases ERK1 and ERK2 inglioblastoma. J. Neuropathol. Exp. Neurol. 76, 838鈥?47 (2017).CAS聽 Article聽Google Scholar聽 74.Xu, L. et al. Bcl6 promotes glioma and serves as a therapeutic target. Proc. Natl. Acad. Sci. 114, 3981鈥?986 (2017).CAS聽 PubMed聽 Article聽Google Scholar聽 75.Cho, E. Yen, Y. Novel regulators and molecular mechanisms of p53R2 and its disease relevance. Biochimie 123, 81鈥?4 (2016).CAS聽 PubMed聽 Article聽Google Scholar聽 76.Zhai, L. et al. Ido1 in cancer: a gemini of immune checkpoints. Cell. Mol. Immunol. 15, 447 (2018).CAS聽 Article聽Google Scholar聽 77.Kesarwani, P. et al. Tryptophan metabolism contributes to radiation-induced immune checkpoint reactivation in glioblas-toma. Clin. Cancer Res. 24, 3632鈥?643 (2018).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 78.Hynes, R. O. Naba, A. Overview of the matrisome鈥揳n inventory of extracellular matrix constituents and functions. Cold Spring Harb. Perspectives Biol. 4, a004903鈥揳004903 (2011). Google Scholar聽 79.Heit, C. et al. Update of the human and mouse SERPIN gene superfamily. Hum. Genomics 7, 22 (2013).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 80.Bonnal, S., Vigevani, L. Valc谩rcel, J. The spliceosome as a target of novel antitumour drugs. Nat. Rev. Drug Discov. 11, 847鈥?59 (2012).CAS聽 PubMed聽 Article聽Google Scholar聽 81.Pal, I., Safari, M., Jovanovic, M., Bates, S. E. Deng, C. Targeting translation of mRNA as a therapeutic strategy incancer. Curr. Hematol. Malig. Reports 1鈥? (2019).82.Andrews, S. et al. Fastqc: a quality control tool for high throughput sequence data, https://www.bioinformatics.babraham.ac.uk/projects/fastqc (2010).83.Krueger, F. TrimGalore! A wrapper around Cutadapt and FastQC to consistently apply adapter and quality trimming toFastQ files URL https://www.bioinformatics.babraham.ac.uk/projects/trim_galore (2012).84.Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15鈥?1 (2013).CAS聽 PubMed聽 Article聽Google Scholar聽 85.Harrow, J. et al. GENCODE: the reference human genome annotation for The ENCODE Project. Genome Res. 22, 1760鈥?774 (2012).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 86.Choudhary, S., Li, W. Smith, A. D. Accurate detection of short and long active ORFs using Ribo-seq data. Bioinformatics https://doi.org/10.1093/bioinformatics/btz878 (2019).87.Anders, S., Pyl, P. T. Huber, W. HTSeq - a Python framework to work with high-throughput sequencing data. Bioinformatics 31, 166鈥?69 (2015).CAS聽 Article聽Google Scholar聽 88.Love, M. I., Huber, W. Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 89.Benjamini, Y. Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. Royal Stat. Soc. Ser. B (Methodological) 57, 289鈥?00 (1995).MathSciNet聽 MATH聽Google Scholar聽 90.Langfelder, P., Luo, R., Oldham, M. C. Horvath, S. Is my network module preserved and reproducible? PLoS Comput.Biol. 7, e1001057 (2011).ADS聽 MathSciNet聽 CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 91.Li, W., Wang, W., Uren, P. J., Penalva, L. O. Smith, A. D. Riborex: fast and flexible identification of differentialtranslation from Ribo-seq data. Bioinformatics 33, 1735鈥?737 (2017).CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 92.Strimmer, K. fdrtool: a versatile R package for estimating local and tail area-based false discovery rates. Bioinformatics 24, 1461鈥?462 (2008).CAS聽 PubMed聽 Article聽Google Scholar聽 93.Shen, S. et al. rMATS: robust and flexible detection of differential alternative splicing from replicate RNA-Seq data. Proc.Natl. Acad. Sci. 111, E5593鈥揈5601 (2014).CAS聽 PubMed聽 Article聽Google Scholar聽 94.Martin, J. A. Wang, Z. Next-generation transcriptome assembly. Nat. Rev. Genet. 12, 671 (2011).CAS聽 PubMed聽 Article聽Google Scholar聽 95.Fabregat, A. et al. The reactome pathway knowledgebase. Nucleic Acids Res. 44, D481鈥揇487 (2015).PubMed聽 PubMed Central聽 Article聽 CAS聽Google Scholar聽 96.Mi, H. et al. The PANTHER database of protein families, subfamilies, functions and pathways. Nucleic Acids Res. 33, D284鈥揇288 (2005).CAS聽 PubMed聽 Article聽Google Scholar聽 97.Supek, F., Bo拧njak, M., 艩kunca, N. 艩muc, T. REVIGO summarizes and visualizes long lists of gene ontology terms. PLoS ONE 6, e21800 (2011).ADS聽 CAS聽 PubMed聽 PubMed Central聽 Article聽Google Scholar聽 98.Szklarczyk, D. et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic AcidsRes. 43, D447鈥揇452 (2014).Article聽 CAS聽Google Scholar聽 99.Tang, Z. et al. GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. NucleicAcids Research 45, W98鈥揥102 (2017).CAS聽 Article聽Google Scholar聽 100.Cotto, K. C. et al. DGIdb 3.0: a redesign and expansion of the drug鈥揼ene interaction database. Nucleic Acids Res. 46, D1068鈥揇1073 (2017).PubMed Central聽 Article聽 CAS聽Google Scholar聽 Download referencesAcknowledgementsThis research was supported by NIH grants R01HG006015 and R21CA205475-01.Author informationAuthor notesThese authors contributed equally: Saket Choudhary and Suzanne C. Burns.Suzanne C. Burns is deceased.AffiliationsComputational Biology and Bioinformatics, University of Southern California, California, USASaket Choudhary,聽Hoda Mirsafian,聽Wenzheng Li聽 聽Andrew D. SmithGreheey Children鈥檚 Research Institute, University of Texas Health Science Center at San Antonio, Texas, USASuzanne C. Burns,聽Mei Qiao,聽Xiufen Lei聽 聽Luiz O. PenalvaDepartment of Cell Systems and Anatomy, University of Texas Health Science Center at San Antonio, Texas, USALuiz O. PenalvaDepartment of Radiation Oncology, University of Texas Southwestern Medical Center, Texas, USADat T. VoAuthorsSaket ChoudharyView author publicationsYou can also search for this author in PubMed聽Google ScholarSuzanne C. BurnsView author publicationsYou can also search for this author in PubMed聽Google ScholarHoda MirsafianView author publicationsYou can also search for this author in PubMed聽Google ScholarWenzheng LiView author publicationsYou can also search for this author in PubMed聽Google ScholarDat T. VoView author publicationsYou can also search for this author in PubMed聽Google ScholarMei QiaoView author publicationsYou can also search for this author in PubMed聽Google ScholarXiufen LeiView author publicationsYou can also search for this author in PubMed聽Google ScholarAndrew D. SmithView author publicationsYou can also search for this author in PubMed聽Google ScholarLuiz O. PenalvaView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsS.C. Contributed to experimental design, conducted most of the data analysis, manuscript writing. S.C.B. Performed all experiments. H.M. Contributed to data analysis, manuscript writing. W.L.. Contributed to data analysis. D.V. Contributed to data interpretation and writing. M.Q. Contributed to the experimental part. X.L. Performed radiation experiments and qRT-PCR validation analysis. A.D.S. Design of analysis pipeline. L.O.F.P. Experimental design and concept, data analysis, manuscript writing. This research was supported by NIH grants R01HG006015 and R21CA205475-01.聽In memoriam聽Suzanne C. Burns.聽Corresponding authorCorrespondence to Luiz O. Penalva.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPublisher鈥檚 note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary information CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.