Craig Michael Horbinski, MD, University of Rochester, PhD, University of Rochester - US grants

DESCRIPTION (provided by applicant): The goal of this research proposal is to determine the mechanism(s) by which mutant isocitrate dehydrogenase 1 (IDH1) causes brain tumors to be less aggressive. The most common type of brain tumor is the diffusely infiltrative glioma; these tumors cannot be completely excised surgically, and are difficult to treat with radiation and chemotherapy. Thus, infiltrative gliomas are incurable. A specific point mutation in IDH1 (and a less common analogous mutation in IDH2) has been found to be quite frequent in these gliomas. When present, it is a powerful favorable prognostic factor, being strongly associated with longer patient survival. Mutant IDH1 has recently been shown to produce a novel compound, 2-hydroxyglutarate (2-HG). However, the effects of mutant IDH1 and 2-HG on glioma cells are unknown. Other work showed that 2-HG causes oxidative stress in nonneoplastic tissue models, and our preliminary data indicate that 2-HG is toxic to glioma cells and induces autophagy, ERK activation, and reactive oxygen species production. We therefore hypothesize that the improved survival imparted by mutant IDH1 in diffuse gliomas is due to 2-HG-induced production of reactive oxygen species, leading to oxidative damage and cell death. We also hypothesize that the cell death is primarily by autophagy, a form of programmed cell death involving lysosomes that has been shown to be prominent in many gliomas. To test these hypotheses, glioma cells will be treated with 2-HG or transfected with mutant IDH1, and multiple well-described markers of autophagy and reactive oxygen species will be measured. Response of glioma cells to autophagy and reactive oxygen species modulation will be assessed. For patient-derived tumor biopsies and human-mouse xenografts, immunohistochemical markers of autophagy and oxidative stress will be quantified and correlated with IDH mutation status. Success in this project would determine whether mutant IDH1 causes increased oxidative stress and autophagy in gliomas, thereby producing a less aggressive glioma compared to tumors that are wild type for IDH1. This knowledge could then be exploited to develop novel ways of treating gliomas. I am fortunate to have been mentored by exceptional scientists and physicians, thus instilling in me a desire to pursue a career that synthesizes what I have learned as a scientist and neuropathologist.My graduate and postdoctoral work in neuroscience and my work in neuro-oncology have given me a diverse array of techniques and approaches that will be used in this project. My current position as an Assistant Professor in the Department of Pathology in the University of Kentucky offers the ideal opportunity to pursue my goal of being an independent investigator. I have nine person-months (75%) of guaranteed protected time for research, separate laboratory space that has been fully equipped, a full-time technician to increase output, and sufficient funds to conduct experiments for the next four years. All this has been put in place by my chairman, Dr. Paul Bachner, independent of my success in securing extramural funding. I also benefit from a collaborative environment, superb mentorship by Drs. Natasha Kyprianou, Arie Perry, and Jeremy Rich, and excellent technical resources. Now that I have completed my clinical training, the funding provided by this K08 award would allow me to develop as an independent investigator. Furthermore, the project described in this proposal provides a superb opportunity to discover why mutant IDH1 imparts a more favorable survival in patients afflicted with gliomas, in turn helping to identify pathways and targets for effective therapeutic interventions.

PROJECT SUMMARY/ABSTRACT Infiltrative glioma is the most common type of primary brain tumor in adults, causing over 17,000 deaths in the United States every year. Approximately 20-30% of infiltrative gliomas contain mutations in isocitrate dehydrogenase 1 (IDH1mut). IDH1mut causes global DNA hypermethylation, which contributes to gliomagenesis. Yet, IDH1mut gliomas are significantly less aggressive than gliomas lacking this mutation. It has remained unclear how DNA hypermethylation leads to this unique phenotype. Gliomas often produce blood clots (thrombi) within the tumor and throughout the bloodstream. These thrombi have long been known to be predictors of poor outcome. We recently reported that IDH1mut gliomas produce far fewer thrombi; our data strongly indicate that methylation-induced suppression of F3, the gene encoding Tissue Factor (TF), is the reason why. TF is a powerful procoagulant that, when produced and released by cancers, causes venous thromboemboli (VTE). This debilitating phenomenon occurs in ~25% of glioma patients, but never when IDH1mut is present. In addition to triggering thrombosis, TF binds and activates protease-activated receptor 2 (PAR2), a transmembrane receptor expressed by cancer cells that signals through multiple intracellular pathways to promote tumor malignancy. Our data show that: (i) among all genes that directly participate in blood clotting, F3 mRNA levels have the strongest inverse relationship with IDH1mut; (ii) F3 methylation is significantly higher in IDH1mut gliomas than IDH1wt gliomas; (iii) TF protein levels are consistently lower in IDH1mut gliomas than IDH1wt gliomas; (iv) circulating TF is lower in patients with IDH1mut gliomas than IDH1wt gliomas; (v) high circulating TF correlates with increased VTE risk; (vi) patient-derived glioma cells with endogenous IDH1mut produce smaller and fewer venous thrombi than IDH1wt gliomas in xenograft mouse models; (vii) suppression of TF in IDH1wt gliomas greatly reduces their in vitro and in vivo malignancy; (viii) patients whose gliomas express low TF have more than double the median survival of patients whose gliomas express high TF, independent of IDH1mut. Thus, we hypothesize that methylation- induced suppression of TF is a critical determinant of the less thrombogenic, and less malignant, IDH1mut phenotype. In Aim 1, we will conclusively establish that F3 hypermethylation is the mechanism by which IDH1mut suppresses TF expression. In Aim 2, we will modulate the expression of TF in a series of patient-derived IDH1wt and IDH1mut glioma cells, observing the effects on tumor-induced thrombosis and malignancy in cell cultures and in engrafted mice. In Aim 3, we will use molecular and pharmacologic approaches to investigate the therapeutic potential of blocking TF-PAR2 signaling in gliomas. Further, we will prospectively evaluate the utility of determining circulating TF levels, along with other clinical, blood-based, and tissue-based biomarkers, to create the first predictive model of VTE risk in glioma patients. In total, this research will greatly advance our understanding of IDH1mut tumor biology, and it will inform regarding novel treatment and diagnostic strategies for improving glioma patient care.

BIOSPECIMEN CORE: SUMMARY The Biospecimen Core is a vital resource that supports all brain tumor research at Northwestern, and is especially important in seeing that all projects experience clinical translation within the period of SPORE funding. Regular interactions with the Biostatistics & Bioinformatics Core and Administrative Core will ensure that each project achieves its objectives, and has positive impact on the care of brain tumor patients. The Biospecimen Core will have extensive interactions with all projects and cores by acting as the central repository of patient tissues and renewable tumor resources, by providing the highest quality of tissue analytical services, and by providing neuropathological consultation. The Director of the Core, Dr. Horbinski, as well as the Co-Director, Dr. Daniel Brat, will supervise quality control testing of banked specimens, and will perform all microscopy-based analyses of patient tissues and tumor models for each SPORE project. Biospecimen Core activities will be performed to accomplish the following aims: Aim 1: Provide Northwestern brain tumor researchers with annotated biospecimens from brain tumor patients. Aim 2: Subject all biospecimens to rigorous quality control. Aim 3: Propagate, bank, and characterize surgical specimen derivatives, including PDX and cancer stem cell (CSC) models, to meet ongoing needs for renewable tumor cell sources. Aim 4: Support SPORE projects with biospecimens and neuropathologic analyses.

ABSTRACT ? MOUSE HISTOLOGY AND PHENOTYPING LABORATORY The Mouse Histology and Phenotyping Laboratory (MHPL) was originally established in 2009 as a developing core laboratory for the Lurie Cancer Center (LCC), and emerged as an established core after the last competing CCSG renewal in 2013. The core provides LCC members and the Northwestern University (NU) research community with histopathologic assessment of experimental research animal tissues by trained pathologists and expert histotechnologists. The mission of MHPL is to provide investigators with highly specialized tissue-based histology and phenotyping services to support their animal-based research work. Since the last competing renewal, the laboratory has expanded its services and now provides comprehensive histology services for a wide variety of research animal species. Phenotyping services include immunohistochemistry (IHC), immunofluorescence (IF), whole mouse necropsy with detailed phenotyping of organs and tissues, and tissue- based toxicology analyses. The laboratory also provides training opportunities for students, post-doctoral fellows, staff and research faculty to learn histology and phenotyping analysis techniques. The MHPL Director is a board- certified pathologist and is available to assist LLC investigators in developing the best strategies to elucidate specific phenotypes, gain mechanistic insights regarding the biologic actions of targeted molecules, and study the toxicity and/or therapeutic efficacy of exogenously administered substances in rodents. The comprehensive histopathology support services provided by the MHPL enhance the ability of LCC investigators to characterize viable and embryonic lethal mouse models, and to develop and analyze new in vivo model systems for studies of cancer biology, prevention, and therapy. The use of the MHPL by LCC members has increased substantially over the current funding period. During the last year, the core provided services to 74 LCC members, of which 69 were peer review funded and whose work resulted in multiple high impact publications.

PROJECT SUMMARY Glioblastoma (GBM) is the most common cancer arising in the adult brain, and is lethal in nearly all cases. A key contributing factor to poor outcomes for GBM patients is a subpopulation of cells, known as cancer stem cells (CSCs), that are highly resistant to routinely used genotoxic/cytotoxic therapies, and ultimately manifest as recurrent tumor. Inhibiting tumor recurrence from CSCs that survive therapy would therefore improve GBM patient outcomes. Our preliminary and recently published work show that CSC survival is dependent on Tissue Factor (TF), a conserved transmembrane and secreted protein involved in blood clotting. TF activates protease- activated receptor 2 (PAR2), a G-protein-coupled receptor on GBM cells, which promotes CSC maintenance and expansion, as indicated by analysis of marker expression, self-renewal capacity, and in vivo growth of CSCs. TF suppression greatly reduces CSC subpopulations, in some cases even leading to complete tumor eradication in vivo. Protective and proliferative effects of TF-PAR2 signaling on CSCs appears to be through activation of multiple classes of oncogenic receptor tyrosine kinases (RTKs) like EGFR, although the mechanism by which TF-PAR2-RTKs stimulate CSCs is unclear. Our preliminary data also show that TF positively correlates with expression of Junctional Adhesion Molecule-A (JAM-A), a protein that promotes cell-cell adhesion by stabilizing integrin ?1. CSCs depend on such cell-cell adhesion, and our data show that JAM-A is necessary for CSC behavior in GBM. Because TF can also signal through integrin ?1, our overarching hypothesis is that TF upregulates JAM-A expression, which stabilizes integrin ?1 and enhances the ability of TF to act on integrin ?1 and promote self-renewal in GBM. This has therapeutic relevance, because the pro-CSC effects of TF are independent of its role in hemostasis, and blocking JAM-A could potentially reduce the pro-tumor effects of TF, without causing bleeding that would result from targeting TF directly. In Specific Aim 1, we will test the hypothesis that TF drives JAM-A expression through PAR2-RTK signaling. We will identify components of the TF-PAR2 complex that are essential for TF-PAR2 pro-tumor activities, and will test the ability of TF to trigger JAM-A expression while inhibiting each major downstream pathway of RTK signaling. In Specific Aim 2, we will determine whether JAM-A requires serpin B3, a serine-protease inhibitor that we have shown binds to JAM-A, for its pro-CSC activities. In Specific Aim 3, we will use molecular knockouts to determine whether JAM-A is an effective therapeutic target against aggressive, high TF-expressing GBM. Completion of this project will advance our understanding of how CSC subpopulations are maintained in GBM, could well be applicable to a wide range of cancers, and would demonstrate a compelling new therapeutic target in treating numerous malignancies, including GBM.

PROJECT SUMMARY Diffusely infiltrative glioma is the most common primary brain tumor in adults. Most glioma patients experience at least one seizure during the course of their disease, and over 30% suffer from repeated seizures, known as tumor-associated epilepsy (TAE). Current front-line treatment for TAE is levetiracetam (LEV) (a.k.a. Keppra®), but this fails to control seizures in over 50% of patients. Such patients then require more powerful second-line antiepileptic drugs that often have greater side effects. TAE is more common in World Health Organization (WHO) grade II-III gliomas than in grade IV glioblastomas, but the reason for this is not clear. The vast majority of grade II-III gliomas contain mutations in isocitrate dehydrogenases 1 and 2 (collectively ?IDHmut?), which lead to the production and release of large amounts of D-2-hydroxyglutarate (D2HG). D2HG bears a great deal of structural similarity to glutamate, an excitatory neurotransmitter that binds to N-methyl-D-aspartate receptor (NMDAR) on neurons. Our data show that D2HG increases in vitro neuronal membrane depolarization and neuronal network activity, and that this can be completely blocked by an NMDA receptor (NMDAR) antagonist. We also found that IDHmut glioma increases seizure activity in engrafted mice compared to IDHwt glioma, and that this is greatly reduced by treatment with IDHmut enzyme inhibitor. Finally, we found that IDHmut gliomas are much more likely to cause seizures compared to IDHwt gliomas. This is the first direct evidence of a mechanistic link between IDHmut and seizures; therefore, our hypothesis is that D2HG contributes to an increased incidence of seizures in patients with IDHmut gliomas, and that new targeted therapeutic strategies can decrease seizures in these patients. In Aim 1, we will explore the mechanisms by which D2HG triggers neuronal depolarization and increased neuronal network activity. Our two main hypotheses are: (i) D2HG directly stimulates NMDA receptors; (ii) D2HG inhibits glutamate reuptake transporters that normally prevent the pathologic accumulation of glutamate in the synaptic cleft. We will use patch clamping and multi-electrode arrays to study the effects of D2HG on the electrical activity of cultured mouse cortical neurons, as well as on mouse brain slices. In Aim 2, we will explore the effects of IDHmut glioma on the surrounding nonneoplastic tissue in vivo, focusing on changes that are characteristic of epilepsy, including neuronal loss, NMDAR downregulation, oxidative stress, inflammation, hippocampal damage, and altered mouse behavior. Results will be validated in patient-derived IDHwt and IDHmut gliomas. In Aim 3, we will compare the anti-seizure effects of two next- generation IDHmut inhibitors, AG-120 and AG-881, as well as memantine, an NMDAR antagonist that is already used to treat Alzheimer?s Disease. Each of these drugs will be tested as monotherapy and in combination with LEV. Successful completion of these Aims will establish the D2HG product of IDHmut as an epileptogenic agent, will shed more light on how IDHmut alters the nonneoplastic neural tissues surrounding glioma, and will foster clinical trials to determine the efficacy of IDHmut inhibitors, and memantine, against seizures in these patients.