All-trans retinoic acid induces durable tumor immunity in IDH-mutant gliomas by rescuing transcriptional repression of the CRBP1-retinoic acid axis

Diffuse gliomas are epigenetically dysregulated, immunologically cold, and fatal tumors characterized by mutations in isocitrate dehydrogenase (IDH). Although IDH mutations yield a uniquely immunosuppressive tumor microenvironment, the regulatory mechanisms that drive the immune landscape of IDH mutant (IDHm) gliomas remain unknown. Here, we reveal that transcriptional repression of retinoic acid (RA) pathway signaling impairs both innate and adaptive immune surveillance in IDHm glioma through epigenetic silencing of retinol binding protein 1 (RBP1) and induces a profound anti-inflammatory landscape marked by loss of inflammatory cell states and infiltration of suppressive myeloid phenotypes. Restorative retinoic acid therapy in murine glioma models promotes clonal CD4 + T cell expansion and induces tumor regression in IDHm, but not IDH wildtype (IDHwt), gliomas. Our findings provide a mechanistic rationale for RA immunotherapy in IDHm glioma and is the basis for an ongoing investigator-initiated, single-center clinical trial investigating all-trans retinoic acid (ATRA) in recurrent IDHm human subjects.

Machine Learning Identification of Immunotherapy Targets in Low-Grade Glioma using RNA Sequencing Expression Data

PURPOSE

Immunotherapy has revolutionized cancer treatment in the past decade, but significant hurdles remain. Human studies with immune checkpoint inhibitors (ICIs) targeting programmed cell death protein have demonstrated suboptimal efficacy in the setting of low-grade gliomas (LGG). Identification of mechanisms leading to inadequate anti-tumor immunity is paramount. The current study evaluates and validates barriers to immunotherapy using a novel machine learning algorithm.

METHODS

We utilized The Cancer Genome Atlas (TCGA) to generate expression levels of 28 immune genes related to known immunotherapeutic targets or lymphocyte cytolytic activity. We created training and testing groups and three machine learning models to determine the genes most highly correlated to cytolytic activity (CYT). The three models were multiple regression by exhaustive selection, LASSO, and random forest. We validated computational results by comparing expression of pertinent genes in patient-derived glioma samples.

RESULTS

Our models demonstrated linearity, a low mean-squared error, and consistent results with respect to the most important variables. Expression of ICOS, IDO1, and CD40 were the most important variables in all models and demonstrated positive correlation with CYT. Other variables included TIGIT and CD137. Genetic analysis from three IDH-mutants (IDHm) and three IDH-wild type (IDHwt) patient-derived glioma samples validated TCGA data and demonstrated lower levels of CYT in IDHm gliomas compared to IDHwt.

CONCLUSION

This novel methodology has elucidated three potential targets for immunotherapy development in LGG. We also demonstrated a novel method of analyzing data using advanced statistical techniques that can be further employed in developing treatments for other diseases as well.

FACT maintains nucleosomes during transcription and stem cell viability in adult mice

Preservation of nucleosomes during replication has been extensively studied, while the maintenance of nucleosomes during transcription has gotten less attention. The histone chaperone FACT has a role in transcription elongation, although whether it disassembles or assembles nucleosomes during this process is unclear. To elucidate the function of FACT in mammals, we deleted the Ssrp1 subunit of FACT in adult mice. FACT loss is lethal, possibly due to the loss of the earliest progenitors in bone marrow and intestine, while more differentiated cells are not affected. Using cells isolated from several tissues, we show that FACT loss reduces the viability of stem cells but not of cells differentiated in vitro. FACT depletion increases chromatin accessibility in a transcription-dependent manner in adipose mesenchymal stem cells, indicating that nucleosomes are lost in these cells during transcription in the absence of FACT. We also observe activation of interferon (IFN) signaling and the accumulation of immunocytes in organs sensitive to FACT loss. Our data indicate that FACT maintains chromatin integrity during transcription in mammalian adult stem cells, suggesting that chromatin transcription in stem cells and differentiated cells is different.

In vivo efficacy of decitabine as a natural killer cell-mediated immunotherapy against isocitrate dehydrogenase mutant gliomas

OBJECTIVE

Isocitrate dehydrogenase (IDH) mutations are found in more than 80% of low-grade gliomas and in the majority of secondary glioblastomas. IDH mutation (IDHmut) leads to aberrant production of an oncogenic metabolite that promotes epigenetic dysregulation by inducing hypermethylation to suppress transcription of various tumor suppressor genes. Hypermethylation in IDHmut gliomas leads to transcriptional repression of NKG2D ligands, especially UL16-binding protein (ULBP)-1 and ULBP-3, and subsequent evasion of natural killer (NK) cell-mediated lysis. The demethylating agent 5-aza-2'deoxycytodine (decitabine [DAC]) is a DNA methyltransferase 1 inhibitor that prevents hypermethylation and is capable of restoring NKG2D ligand expression in IDHmut gliomas to resensitize them to NK cells. Given its capacity for sustained epigenetic reprogramming, the authors hypothesized that DCA would be an effective immunotherapeutic agent in treating IDHmut gliomas in an NK cell-dependent manner by upregulating epigenetically repressed activating NKG2D ligands in IDHmut tumors. In this study, the authors sought to use a glioma stem cell, preclinical animal model to determine the efficacy of DAC in IDHmut and IDH wild-type (IDHwt) tumors, and to characterize whether the activity of DAC in gliomas is dependent on NK cell function.

METHODS

Xenograft models of IDHwt and IDHmut gliomas were established in athymic-nude mice. When tumors were grossly visible and palpable, mice were treated with either DCA or dimethylsulfoxide intraperitoneally every 7 days. Tumor sizes were measured every 2 to 3 days. After the animals were euthanized, xenografts were harvested and analyzed for the following: tumor expression of NKG2D ligands, tumor susceptibility to human and murine NK cells, immunohistochemistry for NK infiltration, and tumor-infiltrating lymphocyte characterization.

RESULTS

DAC significantly inhibited the growth of IDHmut xenografts in the athymic nude mice. This effect was abrogated with NK cell depletion. Ex vivo analysis of tumor cells from harvested xenografts confirmed that DAC increased NKG2D ligand ULBP-1 and ULBP-3 expressions, and enhanced susceptibility to lysis of both human and murine IDHmut glial cells with corresponding NK cells. Immunohistochemical analysis of the xenografts indicated that DCA-treated IDHmut gliomas had a greater level of NK infiltration into the tumor compared with the negative control. Finally, DCA radically altered the tumor-infiltrating lymphocyte landscape of IDHmut glioma xenografts by increasing NK cells, dendritic cells, and M1 macrophages, while decreasing suppressive monocyte infiltration.

CONCLUSIONS

DCA displayed novel immunotherapeutic functions in IDHmut gliomas. This effect was critically dependent on NK cells. Additionally, DCA significantly altered the tumor immune landscape in IDHmut gliomas from suppressive to proinflammatory.

History of atopy confers improved outcomes in IDH mutant and wildtype lower grade gliomas

PURPOSE

A history of atopy or allergy has been shown to be protective against the development of glioma, however the effect of atopy on patient outcomes, especially in conjunction with the survival benefit associated with IDH mutation, has not yet been investigated, and is the focus of the study we present here.

METHODS

Low grade glioma (LGG) data from the TCGA was downloaded, along with IDH, TERT, 1p/19q and ATRX mutational status and genetic alterations. History of asthma, eczema, hay fever, animal, or food allergies, as documented in TCGA, was used to determine patient atopy status. Patients with missing variables were excluded from the study.

RESULTS

374 LGG studies were included. Patients with a history of atopy demonstrated longer overall survival (OS) compared to those without (145.3 vs. 81.5 months, p = 00.0195). IDH mutant patients with atopy had longer OS compared those without atopy (158.8 vs. 85 months, p = 0.035). Multivariate cox regression analysis demonstrated that the effects of atopy on survival were independent of IDH and histological grade, (p = 0.002, HR 0.257, 95% 0.109-0.604), (p =  < 0.001, HR 0.217, 95% 0.107-0.444), and (p = 0.004, HR 2.72, 95% 1.373-5.397), respectively. In terms of treatment outcomes, patients with atopy did not differ in treatment response compared to their counterpart. Pathway analysis demonstrated an upstream activation of the BDNF pathway (p = 0.00027).

CONCLUSION

A history of atopy confers a survival benefit in patients with diffuse low-grade glioma. Activation of the BDNF pathway may drive the observed differences.

Characterization of systemic immunosuppression by IDH mutant glioma small extracellular vesicles

BACKGROUND

Gliomas are the most common primary brain tumors and are universally fatal. Mutations in the isocitrate dehydrogenase genes (IDH1 and IDH2) define a distinct glioma subtype associated with an immunosuppressive tumor microenvironment. Mechanisms underlying systemic immunosuppression in IDH mutant (mutIDH) gliomas are largely unknown. Here, we define genotype-specific local and systemic tumor immunomodulatory functions of tumor-derived glioma exosomes (TEX).

METHODS

TEX produced by human and murine wildtype and mutant IDH glioma cells (wtIDH and mutIDH, respectively) were isolated by size exclusion chromatography (SEC). TEX morphology, size, quantity, molecular profiles and biodistribution were characterized. TEX were injected into naive and tumor-bearing mice, and the local and systemic immune microenvironment composition was characterized.

RESULTS

Using in vitro and in vivo glioma models, we show that mutIDH TEX are more numerous, possess distinct morphological features and are more immunosuppressive than wtIDH TEX. mutIDH TEX cargo mimics their parental cells, and induces systemic immune suppression in naive and tumor-bearing mice. TEX derived from mutIDH gliomas and injected into wtIDH tumor-bearing mice reduce tumor-infiltrating effector lymphocytes, dendritic cells and macrophages, and increase circulating monocytes. Astonishingly, mutIDH TEX injected into brain tumor-bearing syngeneic mice accelerate tumor growth and increase mortality compared with wtIDH TEX.

CONCLUSIONS

Targeting of mutIDH TEX represents a novel therapeutic approach in gliomas.

TIGIT and PD-1 Immune Checkpoint Pathways Are Associated With Patient Outcome and Anti-Tumor Immunity in Glioblastoma

Glioblastoma (GBM) remains an aggressive brain tumor with a high rate of mortality. Immune checkpoint (IC) molecules are expressed on tumor infiltrating lymphocytes (TILs) and promote T cell exhaustion upon binding to IC ligands expressed by the tumor cells. Interfering with IC pathways with immunotherapy has promoted reactivation of anti-tumor immunity and led to success in several malignancies. However, IC inhibitors have achieved limited success in GBM patients, suggesting that other checkpoint molecules may be involved with suppressing TIL responses. Numerous IC pathways have been described, with current testing of inhibitors underway in multiple clinical trials. Identification of the most promising checkpoint pathways may be useful to guide the future trials for GBM. Here, we analyzed the The Cancer Genome Atlas (TCGA) transcriptomic database and identified PD1 and TIGIT as top putative targets for GBM immunotherapy. Additionally, dual blockade of PD1 and TIGIT improved survival and augmented CD8⁺ TIL accumulation and functions in a murine GBM model compared with either single agent alone. Furthermore, we demonstrated that this combination immunotherapy affected granulocytic/polymorphonuclear (PMN) myeloid derived suppressor cells (MDSCs) but not monocytic (Mo) MDSCs in in our murine gliomas. Importantly, we showed that suppressive myeloid cells express PD1, PD-L1, and TIGIT-ligands in human GBM tissue, and demonstrated that antigen specific T cell proliferation that is inhibited by immunosuppressive myeloid cells can be restored by TIGIT/PD1 blockade. Our data provide new insights into mechanisms of GBM αPD1/αTIGIT immunotherapy.

Epigenomic tensor predicts disease subtypes and reveals constrained tumor evolution

Understanding the epigenomic evolution and specificity of disease subtypes from complex patient data remains a major biomedical problem. We here present DeCET (decomposition and classification of epigenomic tensors), an integrative computational approach for simultaneously analyzing hierarchical heterogeneous data, to identify robust epigenomic differences among tissue types, differentiation states, and disease subtypes. Applying DeCET to our own data from 21 uterine benign tumor (leiomyoma) patients identifies distinct epigenomic features discriminating normal myometrium and leiomyoma subtypes. Leiomyomas possess preponderant alterations in distal enhancers and long-range histone modifications confined to chromatin contact domains that constrain the evolution of pathological epigenomes. Moreover, we demonstrate the power and advantage of DeCET on multiple publicly available epigenomic datasets representing different cancers and cellular states. Epigenomic features extracted by DeCET can thus help improve our understanding of disease states, cellular development, and differentiation, thereby facilitating future therapeutic, diagnostic, and prognostic strategies.

Leistico et al. apply tensor decomposition and classification methods to integrate information from hierarchical heterogenous epigenomic datasets and identify histone modification patterns that discriminate disease conditions, tissue types, and differentiation states. Leiomyomas are shown to possess alterations in distal enhancers and large-scale regions confined to chromatin contact domains.

Abstract 1747: Induction of nucleosome disassembly with small molecules as a novel anti-cancer approach

CBL0137 is a clinical-stage anti-cancer drug candidate that is active against multiple preclinical cancer models. It is a carbazole-based DNA intercalator, discovered by cell-based screening for p53 activation in cells lacking DNA damage. Binding of curaxin to DNA results in alterations in DNA length, charge, and flexibility, which together impair the ability of DNA to wrap around the histone core. Treatment of cells with CBL0137 leads to decondensation of chromatin accompanied by histone eviction. This causes the binding of histone chaperone FACT to epitopes normally hidden within nucleosomes and the activation of p53 by FACT-bound CK2. p53 activation occurs in the absence of detectable DNA damage. However, CBL0137 is also toxic to p53-deficient cells, suggesting that its toxicity is mediated by events upstream of p53 activation. Trying to understand the mechanisms of CBL0137 toxicity, we found that FACT is essential for the viability of tumor but not normal cells. FACT was discovered as a transcription elongation factor performing disassembly and reassembly of nucleosomes in cell-free conditions. However, no reduction of transcription is observed with FACT depletion in cells. Moreover, measurement of FACT chromatin binding and gene expression upon FACT loss suggest that FACT negatively regulates gene expression. mammalian FACT does not disassemble nucleosomes during transcription but binds “open” nucleosomes, with partially unwrapped DNA, appearing during transcription, and tethers nucleosome components together to prevent histone loss from chromatin. In cells treated with CBL0137 nucleosomes are mostly lost in regions enriched with tandem AC/TG dinucleotide repeats, known as microsatellites. These regions accumulate FACT in treated cells. To understand the effects of FACT loss from other chromatin regions, we measured nucleosome distribution in cells with genetically-depleted FACT. Regions normally enriched with FACT lose nucleosomes in a transcription-dependent manner. Thus, curaxin induces chromatin disassembly via a direct effect on nucleosomal DNA and inhibition of FACT-mediated nucleosome restoration at transcribed regions. To explain the reasons for the higher toxicity of curaxin in tumor versus normal cells, we hypothesized that chromatin is, in general, less stable in tumor than in normal cells. Thus, less drug is needed to cause unrepairable nucleosome disassembly in tumor cells. We compared chromatin organization in syngeneic normal, immortalized, and transformed cells and found that during tumor progression, there is loss of histones, reduction of condensed chromatin and increase of DNA accessibility to the transcriptional machinery. The same signs are observed in human tumors. In summary, curaxin kills cells via the destabilization of nucleosomes and disassembly of chromatin, a phenomenon that we have named “chromatin damage.” Tumor cells are more sensitive to chromatin damage due to their preexisting, less stable chromatin.

Citation Format: Alfiya Safina, Poorva Sandlesh, Jianmin Wang, Katerina V. Gurova. Induction of nucleosome disassembly with small molecules as a novel anti-cancer approach [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1747.

Prevention of Chromatin Destabilization by FACT Is Crucial for Malignant Transformation

Histone chaperone FACT is commonly expressed and essential for the viability of transformed but not normal cells, and its expression levels correlate with poor prognosis in patients with cancer. FACT binds several components of nucleosomes and has been viewed as a factor destabilizing nucleosomes to facilitate RNA polymerase passage. To connect FACT's role in transcription with the viability of tumor cells, we analyzed genome-wide FACT binding to chromatin in conjunction with transcription in mouse and human cells with different degrees of FACT dependence. Genomic distribution and density of FACT correlated with the intensity of transcription. However, FACT knockout or knockdown was unexpectedly accompanied by the elevation, rather than suppression, of transcription and with the destabilization of chromatin in transformed, but not normal cells. These data suggest that FACT stabilizes and reassembles nucleosomes disturbed by transcription. This function is vital for tumor cells because malignant transformation is accompanied by chromatin destabilization.

•

FACT is essential for viability of the tumor, but not for normal cells

•

FACT level depends on transcription, but transcription does not depend on FACT

•

FACT preserves nucleosomes during transcription to maintain chromatin integrity

•

FACT maintains chromatin in destabilized state during malignant transformation

Abstract 5193: Conditional knockout of histone chaperone FACT in mice: Consequences for normal tissues

Histone chaperone FAcilitates Chromatin Transcription (FACT) is, a complex of SSRP1 and SPT16, involved in chromatin remodeling during transcription, replication, and DNA repair. FACT is selectively expressed during early stages of development and later exclusively present in the stem cell niches in adult normal tissues. It is expressed in undifferentiated aggressive tumors with poor prognosis.¹ FACT inhibition is not tolerated by most tumor, unlike normal cells. This vulnerability of tumor cells upon FACT inactivation suggests evaluating FACT inhibition as an antitumor therapeutic approach.¹ Since FACT knockout (KO) is embryonically lethal (E3.5),² study of FACT’s function in different non-tumor cells of mammalian organisms was not possible and was mostly studied in single cell organisms and in tumor cell lines. Therefore, to assess the safety of FACT inhibition as an anticancer approach, by understanding its role in normal physiology, we generated a conditional KO of Ssrp1 mouse model using CreERT2-LoxP system.³ We harvested stem cells of different lineages to better understand the role of FACT in normal mouse physiology. Using the hematopoietic stem cells (HSCs) from the bone marrow (BM), we performed BM colony formation assay. We found that FACT inactivation significantly decreased the number of colonies of HSCs. Furthermore, inhibiting FACT significantly increased caspase 3 expression in mouse adipose stem cells (ASCs), resulting in apoptosis and decreased survival. These results prompted us to comprehensively evaluate roles of FACT in stem cell development and survival. To this end, fibroblasts in a conditional knockout mouse were reprogrammed by retroviral vector encoding four pluripotency transcription factors (Oct3/4, Klf4, Sox2 and c-Myc). We have found successful generation of induced pluripotent stem (iPS) cells 14 days after reprogramming. In conclusion, FACT associates with development and/or survival of mouse HSCs and ASCs. Further evaluation using iPS cells is expected to delineate roles of FACT in various stages of stem cell development and thus normal physiology. 1. Garcia H, et al. Cell Rep, 2013 4(1) 2. Cao S, et al. Mol Cel Bio. 2003 23(15) 3. Sandlesh P et al. PLOS ONE. 2018 13(6)

Citation Format: Masaki A. Ito, Poorva Sandlesh, Imon Goswami, Laura A. Prendergast, Katerina Gurova. Conditional knockout of histone chaperone FACT in mice: Consequences for normal tissues [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 5193.

Structure and function of the histone chaperone FACT - Resolving FACTual issues

FAcilitates Chromatin Transcription (FACT) has been considered essential for transcription through chromatin mostly based on cell-free experiments. However, FACT inactivation in cells does not cause a significant reduction in transcription. Moreover, not all mammalian cells require FACT for viability. Here we synthesize information from different organisms to reveal the core function(s) of FACT and propose a model that reconciles the cell-free and cell-based observations. We describe FACT structure and nucleosomal interactions, and their roles in FACT-dependent transcription, replication and repair. The variable requirements for FACT among different tumor and non-tumor cells suggest that various FACT-dependent processes have significantly different levels of relative importance in different eukaryotic cells. We propose that the stability of chromatin, which might vary among different cell types, dictates these diverse requirements for FACT to support cell viability. Since tumor cells are among the most sensitive to FACT inhibition, this vulnerability could be exploited for cancer treatment.

Uncovering the fine print of the CreERT2-LoxP system while generating a conditional knockout mouse model of Ssrp1 gene

FAcilitates Chromatin Transcription (FACT) is a complex of SSRP1 and SPT16 that is involved in chromatin remodeling during transcription, replication, and DNA repair. FACT has been mostly studied in cell-free or single cell model systems because general FACT knockout (KO) is embryonically lethal (E3.5). FACT levels are limited to the early stages of development and stem cell niches of adult tissues. FACT is upregulated in poorly differentiated aggressive tumors. Importantly, FACT inhibition (RNAi) is lethal for tumors but not normal cells, making FACT a lucrative target for anticancer therapy. To develop a better understanding of FACT function in the context of the mammalian organism under normal physiological conditions and in disease, we aimed to generate a conditional FACT KO mouse model. Because SPT16 stability is dependent on the SSRP1-SPT16 association and the presence of SSRP1 mRNA, we targeted the Ssrp1 gene using a CreERT2- LoxP approach to generate the FACT KO model. Here, we highlight the limitations of the CreERT2-LoxP (Rosa26) system that we encountered during the generation of this model. In vitro studies showed an inefficient excision rate of ectopically expressed CreERT2 (retroviral CreERT2) in fibroblasts with homozygous floxed Ssrp1. In vitro and in vivo studies showed that the excision efficiency could only be increased with germline expression of two alleles of Rosa26CreERT2. The expression of one germline Rosa26CreERT2 allele led to the incomplete excision of Ssrp1. The limited efficiency of the CreERT2-LoxP system may be sufficient for studies involving the deletion of genes that interfere with cell growth or viability due to the positive selection of the phenotype. However, it may not be sufficient for studies that involve the deletion of genes supporting growth, or those crucial for development. Although CreERT2-LoxP is broadly used, it has limitations that have not been widely discussed. This paper aims to encourage such discussions.

TRAIN (Transcription of Repeats Activates INterferon) in response to chromatin destabilization induced by small molecules in mammalian cells

Cellular responses to the loss of genomic stability are well-established, while how mammalian cells respond to chromatin destabilization is largely unknown. We previously found that DNA demethylation on p53-deficient background leads to transcription of repetitive heterochromatin elements, followed by an interferon response, a phenomenon we named TRAIN (Transcription of Repeats Activates INterferon). Here, we report that curaxin, an anticancer small molecule, destabilizing nucleosomes via disruption of histone/DNA interactions, also induces TRAIN. Furthermore, curaxin inhibits oncogene-induced transformation and tumor growth in mice in an interferon-dependent manner, suggesting that anticancer activity of curaxin, previously attributed to p53-activation and NF-kappaB-inhibition, may also involve induction of interferon response to epigenetic derepression of the cellular 'repeatome'. Moreover, we observed that another type of drugs decondensing chromatin, HDAC inhibitor, also induces TRAIN. Thus, we proposed that TRAIN may be one of the mechanisms ensuring epigenetic integrity of mammalian cells via elimination of cells with desilenced chromatin.

Pharmacological Targeting of the Histone Chaperone Complex FACT Preferentially Eliminates Glioblastoma Stem Cells and Prolongs Survival in Preclinical Models

The nearly universal recurrence of glioblastoma (GBM) is driven in part by a treatment-resistant subpopulation of GBM stem cells (GSC). To identify improved therapeutic possibilities, we combined the EGFR/HER2 inhibitor lapatinib with a novel small molecule, CBL0137, which inhibits FACT (facilitates chromatin transcription), a histone chaperone complex predominantly expressed in undifferentiated cells. Lapatinib and CBL0137 synergistically inhibited the proliferation of patient-derived GBM cells. Compared with non-stem tumor cells (NSTC) enriched from the same specimens, the GSCs were extremely sensitive to CBL0137 monotherapy or FACT knockdown. FACT expression was elevated in GSCs compared with matched NSTCs and decreased in GSCs upon differentiation. Acute exposure of GSCs to CBL0137 increased asymmetric cell division, decreased GSC marker expression, and decreased the capacity of GSCs to form tumor spheres in vitro and to initiate tumors in vivo Oral administration of CBL0137 to mice bearing orthotopic GBM prolonged their survival. Knockdown of FACT reduced the expression of genes encoding several core stem cell transcription factors (SOX2, OCT4, NANOG, and OLIG2), and FACT occupied the promoters of these genes. FACT expression was elevated in GBM tumors compared with non-neoplastic brain tissues, portended a worse prognosis, and positively correlated with GSC markers and stem cell gene expression signatures. Preferential targeting of GSCs by CBL0137 and synergy with EGFR inhibitors support the development of clinical trials combining these two agents in GBM. Cancer Res; 76(8); 2432-42. ©2016 AACR.