© 2019 Jan C. Lumibao

CHCHD2 AND THE TUMOR MICROENVIRONMENT IN GLIOBLASTOMA

JAN C. LUMIBAO

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Nutritional Sciences in the Graduate College of the University of Illinois at Urbana-Champaign, 2019

Urbana, Illinois

Doctoral Committee:

Professor Brendan A. Harley, Chair Professor H. Rex Gaskins, Director of Research Assistant Professor Andrew J. Steelman Professor Rodney W. Johnson Professor Emeritus John W. Erdman

ABSTRACT

Glioblastoma (GBM) is the most common, aggressive, and deadly form of primary brain tumor in adults, with a median survival time of only 14.6 months. GBM tumors present with chemo- and radio-resistance and rapid, diffuse invasion, making complete surgical resection impossible and resulting in nearly universal recurrence. While investigating the genomic landscape of GBM tumors has expanded understanding of brain tumor biology, targeted therapies against cellular pathways affected by the most common genetic aberrations have been largely ineffective at producing robust survival benefits. Currently, a major obstacle to more effective therapies is the impact of the surrounding tumor microenvironment on intracellular signaling, which has the potential to undermine targeted treatments and advance tumor malignancy, progression, and resistance to therapy. Additionally, mitochondria, generally regarded as putative energy sensors within cells, also play a central role as signaling organelles. Retrograde signaling occurring from mitochondria to the nucleus allows mitochondria to sense intracellular metabolic stress imposed by the extracellular tumor microenvironment and respond appropriately by relaying their status to the nucleus, thus inducing changes in gene expression and facilitating cellular adaptation.

The focus of this work is on the mitochondrial protein coiled-coil-helix-coiled-coil-helix domain-containing protein 2 (CHCHD2) and its capacity to act in a signaling axis between mitochondria and the nucleus to facilitate cellular adaptation and plasticity as GBM cells navigate gradients in the tumor microenvironment. Chapters one, two, and three provide a general introduction of GBM, an analysis of mitochondrial biology in cancer and

ii retrograde signaling, and a review on the current state of the literature surrounding

CHCHD2. Chapter four focuses on characterizing the functional capacity of CHCHD2 in

GBM cells using CRISPR-Cas9 genome editing techniques. We hypothesized that

CHCHD2 knockout would abrogate malignant characteristics such as cell proliferation, sensitivity to therapeutics, and mitochondrial respiration. CHCHD2 gene amplification was found to occur in 9% of GBM tumors, nearly always with EGFR, and was associated with decreased overall survival and progression-free survival. CHCHD2 mRNA levels were increased in grade IV glioma, IDH-wt GBM tumors, and in tumor versus non-tumor tissue. CRISPR-Cas9 derived GBM U87vIII CHCHD2 KO cells displayed decreased OCR and spare respiratory capacity compared to U87vIII CHCHD2 WT cells. The redox state of the glutathione (GSH) pool was more reduced within the mitochondrial matrix of

CHCHD2 KO cells compared to WT cells, an effect independent of GSH synthesis, cystine import via xCT, or GPx-1, GPx-2, or GPx-4 levels. The cytosolic GSH pool was not perturbed by CHCHD2 KO. Furthermore, CHCHD2 KO abrogated the increased proliferation and invasion of U87vIII cells in hypoxia (1% O2) compared to standard normoxic culture conditions (20% O2). U87vIII CHCHD2 KO cells displayed increased sensitivity to sulfasalazine, erlotinib, and temozolomide, but not the procaspase activator

Pac-1. CHCHD2 exhibited a heterogeneous protein expression pattern among patient- derived cells investigated using western blot, with greater levels observed in more invasive samples. These results indicate that CHCHD2 mediates a variety of GBM cell hallmark characteristics, including cell proliferation, cell invasion in response to hypoxia, and cellular resistance to cytotoxic agents.

iii

The C-terminus of CHCHD2 contains redox-sensitive cysteines in twin CX9C motifs that are amenable to thiol/disulfide formation, which in turn determines protein folding and import into mitochondria. Chapter five focuses on defining the relationship between hypoxia and redox stress, CHCHD2 protein folding, and subcellular redistribution of

CHCHD2, in order to examine a mechanism by which CHCHD2 signals mitochondrial and cellular status back to the nucleus. We hypothesized that reduction of the C-terminal

CHCH domain would induce mitochondrial export of CHCHD2. Total CHCHD2 protein levels were not significantly altered by decreasing oxygen tensions in U87 cells. U87vIII cells exhibited constant levels of total CHCHD2 protein at 20%, 7%, and 4% O2, while incubation in 1% O2 led to a significant decrease in total CHCHD2. Half-life of CHCHD2 was determined to be approximately 3 h in both U87 and U87vIII cells. U87vIII cells constitutively displayed a greater amount of nuclear CHCHD2 compared to isogenic U87 cells at 20% O2. Metabolic stress induced by incubation in 1% O2 for 48 h or serum-free media for 30 min both resulted in increased CHCHD2 detected in nuclei of U87 and

U87vIII cells. Oxidative stress induced by the glutathione synthesis inhibitor BSO increased nuclear CHCHD2 in U87vIII cells, while the glutathione precursor NAC did not significantly affect nuclear CHCHD2 levels. Exposure to H2O2 in serum-free media abrogated the amount of nuclear CHCHD2 detected in nuclei of U87 and U87vIII cells.

Reductive unfolding of CHCHD2 with the disulfide reductant DTT consistently resulted in the greatest and greatest degree of nuclear accumulation of CHCHD2 in both cell lines within 15-30 min. Ectopic expression of exogenous 3X-FLAG-tagged CHCHD2 in U87vIII cells did not result in increased expression of endogenous CHCHD2. These findings

iv indicate that reductive unfolding of CHCHD2 is responsible for mitochondrial export and nuclear translocation in response to hypoxia, serum deprivation, and redox stress.

Finally, in addition to gradients in oxygen, nutrients, and biophysical cues, tumor- associated microglia and macrophages represent a substantial portion of GBM tumors and play a critical role in cultivating a microenvironment that facilitates tumor expansion.

Chapter six focuses on examination of the crosstalk between microglia and GBM cells that promotes angiogenesis, cell invasion, and remodeling of the tumor immune landscape. We hypothesized that GBM-microglia crosstalk would facilitate secretion of soluble factors that promote angiogenesis and extracellular matrix remodeling, as well as alter GBM cell glutathione redox balance. The expression of EGFRvIII on U87 cells resulted in an altered profile of secreted angiogenesis-related factors under basal culture conditions. Similarly, primary neonatal murine microglia, but not the human microglia cell line HMC3, exhibited differential angiogenesis-related secretomes when cultured in U87 or U87vIII conditioned media. Primary murine microglia additionally displayed increased cell proliferation when cultured in U87 and U87vIII conditioned media, while HMC3 cells did not. Conditioned media from patient-derived, EGFRvIII-expressing xenograft GBM6 cells led to a striking increase in HMC3 secretion of angiogenesis-related factors, inducing the greatest secretion of CCL2, IGFBP-2, TIMP-4, angiogenin, and IGFBP-3. HMC3

- conditioned media decreased expression of the system xc antiporter and increased glutathione oxidation in mitochondria of U87 cells, while xCT expression and mitochondrial glutathione redox status in U87vIII cells remained stable. These findings indicate that GBM cells stimulate the secretion of cytokines, chemokines, and soluble factors by microglia to cultivate an immunosuppressive microenvironment promoting

- angiogenesis and cell invasion. Additionally, system xc expression and, consequently, intracellular glutathione redox balance remained stable in U87vIII cells exposed to HMC3 conditioned media. Together, the studies demonstrate a dynamic crosstalk between GBM cells and microglia that facilitates microenvironment remodeling and has the potential to influence GBM cell glutathione metabolism.

ACKNOWLEDGEMENTS

From the start, my advisor Dr. H. Rex Gaskins has been a constant source of support, encouragement, and motivation. You had seen potential in me I had not known existed.

You consistently pushed me to explore new realms and not let my curiosity go unsatisfied.

Under your mentorship, I have gained not only research experience and skills as a meticulous writer, but also increasing amounts of confidence. I cannot begin to list the things that you have taught me, not just from a scientific perspective, but also in the areas of professionalism, leadership, empathy, education, and community involvement. Your resounding passion for all these areas has always been something I have admired and seek to emulate as my career moves onward. I am ever grateful and privileged to have had a mentor whom I can also confidently call a genuine friend.

I would also like to sincerely acknowledge my committee members, Drs. Brendan

Harley, Andrew Steelman, Rodney Johnson, and John Erdman. I deeply appreciate the time you have spent guiding me as I moved through my graduate career, and the trans- disciplinary mentorship you have provided. In particular, I would like to thank Dr. Harley for his constant support and guidance as a collaborator in the realm of GBM and willingness to share laboratory supplies and methodologies; Dr. Steelman for his patience, mentorship, and enthusiasm as we expanded our questions into microglia and into mice; and Drs. Steelman, Johnson, and Erdman for their genuine advice and support of my future career. My work as a scientist, this document, and my career moving onward have all been improved by your mentorship.

vii

I also would like to thank members of the Gaskins lab for their support, encouragement, and mentorship. In particular, I am deeply grateful for the mentorship I have received from Dr. Vladimir Kolossov. Your meticulous eye for detail in the lab and in experimental design is something that I am constantly seeking to emulate and improve in myself every day. I have learned and honed countless techniques and skills with your guidance, and you have always received all my questions with thoughtful, critical, and meaningful responses. You have undoubtedly shaped me into becoming a better bench scientist. Additionally, I would like to offer huge thanks to Dr. Nagendra Prabhu Ponnuraj, who taught me RT-qPCR; Drs. Patty Wolf and Matthew Leslie, who provided endless encouragement, support, and positivity during trying times; and our undergraduates, whose enthusiasm propelled our research onward and who have been a joy to mentor.

Also, I would like to sincerely acknowledge members of the Harley lab, and in particular

Emily Chen, for their willingness to answer my questions, share laboratory materials and techniques, and whom I have greatly enjoyed working with.

Lastly, I would like to extend my deepest acknowledgements to my friends and family.

I am astounded by how blessed and fortunate I am to be surrounded by such incredible, selfless people, who constantly have words and actions of encouragement and confidence. To Steve and Tom, may our friendship from the streets of Berwyn and Midway continue to hold fast. To my mother and father, the examples you lead in your lives have shaped me more than you will ever know. Your constant love and support, whether domestic or overseas, has always pushed me to new heights. My qualities and accomplishments are due entirely to virtues I have learned from you. It is without a doubt that this would not have been possible without you.

viii

TABLE OF CONTENTS

CHAPTER 1: A GENERAL INTRODUCTION TO GLIOBLASTOMA ...... 1

CHAPTER 2: RETROGRADE SIGNALING BETWEEN MITOCHONDRIA AND

NUCLEI ...... 21

CHAPTER 3: COILED-COIL-HELIX-COILED-COIL-HELIX DOMAIN-CONTAINING

PROTEIN 2 ...... 33

CHAPTER 4: CRISPR-CAS9-MEDIATED CHCHD2 KNOCKOUT COMPROMISES

MALIGNANT CHARACTERISTICS OF GLIOBLASTOMA CELLS EXPRESSING

EGFRvIII ...... 42

CHAPTER 5: DEFINING THE RELATIONSHIP BETWEEN HYPOXIA, REDOX

STATUS, AND SUBCELLULAR LOCALIZATION OF CHCHD2 IN GLIOBLASTOMA

CELLS ...... 69

CHAPTER 6: CHARACTERIZING MICROGLIA-GLIOBLASTOMA CROSSTALK ...... 93

CHAPTER 7: OVERALL CONCLUSIONS AND FUTURE DIRECTIONS ...... 116

REFERENCES ...... 120

APPENDIX A: THE INTERPLAY AMONG PSYCHOLOGICAL DISTRESS, THE

IMMUNE SYSTEM, AND BRAIN TUMOR PATIENT OUTCOMES ...... 144

CHAPTER 1: A GENERAL INTRODUCTION TO GLIOBLASTOMA

1.1. GLIOBLASTOMA

Tumors of the brain and central nervous system (CNS) account for approximately 1% of newly diagnosed cancers and 2% of cancer deaths in the United States (Barnholtz-Sloan,

Ostrom, & Cote, 2018). Unlike other cancer types, which are staged, brain and CNS tumors are broadly classified as nonmalignant (benign) and malignant, as determined by the World Health Organization (WHO) grading system, which ranges from grade I to grade IV (Barnholtz-Sloan et al., 2018). Grade I and II brain and CNS tumors are classified as low grade/nonmalignant, while grade III and IV are classified as high grade/malignant.

Nonmalignant meningiomas represent the most common benign brain and CNS tumor, while malignant gliomas are the most commonly diagnosed malignant tumor, accounting for approximately 80% of all malignant brain tumors (Omuro & DeAngelis, 2013). Of gliomas, glioblastoma occurs at the greatest frequency, accounting for greater than 50% of diagnosed gliomas (Barnholtz-Sloan et al., 2018).

Glioblastoma [GBM, WHO grade IV glioma] is the most common, aggressive, and deadly form of primary malignant brain tumor in adults (Omuro & DeAngelis, 2013; Stupp et al., 2005). In 2016, the WHO published an updated classification of tumors of the CNS, focusing on incorporation of both histological and molecular features to stratify diffuse gliomas (Louis et al., 2016). In this new classification system, GBMs are stratified into either: (1) GBM, isocitrate dehydrogenase 1/2 (IDH-1/2)-wild type; (2) GBM, IDH1/2- mutant; or (3) GBM, not otherwise specified (NOS). IDH-1/2-wild type GBMs account for approximately 90% of cases, correspond with the clinically defined primary (or de novo)

GBM, and predominate in patients over 55 years old (Louis et al., 2016). IDH-1/2-mutant

GBMs represent approximately 10% of cases, correspond to secondary GBM (i.e., tumors with history of prior lower grade glioma), and arise predominantly in younger patients

(Louis et al., 2016). Tumors for which evaluation of IDH-1/2 status cannot be performed are designated GBM, NOS. Incidence of GBM is greater in male vs female patients

(57.6% vs 42.4%), with 94.6% of GBMs diagnosed in patients aged ≥ 40 years (Barnholtz-

Sloan et al., 2018). Median survival of patients diagnosed with GBM is 14.6 months, with

37.4% of patients surviving after 1 year and only 4.9% of patients surviving after 5 years

(Barnholtz-Sloan et al., 2018).

1.2. TREATMENT OF GBM

The current standard-of-care for patients with GBM is a multimodal therapeutic regimen consisting of maximal safe surgical resection, radiotherapy (cumulative dose of 60 Gy, 30 fractions over 6 weeks) with concomitant chemotherapy with the DNA alkylating agent temozolomide (TMZ), followed by six cycles of adjuvant maintenance chemotherapy with

TMZ (150-200 mg/m2, 5 days of a 28-day cycle) (Stupp et al., 2005). This regimen was established following a seminal clinical trial in 2005 by Stupp et al., in which patients receiving TMZ concomitant with radiotherapy exhibited a median overall survival of 14.6 months vs 12.1 months in patients receiving radiotherapy alone (Stupp et al., 2005). More recently, a multi-center randomized phase III clinical trial conducted by Stupp et al. assessed the effect of tumor-treating fields (TTFields) plus maintenance TMZ versus TMZ alone on progression-free and overall survival (Stupp et al., 2017). TTFields are an antimitotic treatment that delivers low-intensity, intermediate-frequency (200 kHz) alternating electric fields via transducer arrays applied directly to the scalp, subsequently

2 causing mitotic arrest and apoptosis of rapidly dividing cells (Giladi et al., 2015; Kirson et al., 2007; Stupp et al., 2017). Notably, patients receiving TTFields in conjunction with maintenance TMZ displayed progression-free and overall survival of 6.7 and 20.9 months, respectively, compared to 4.0 and 16.0 months in patients receiving maintenance TMZ alone (Stupp et al., 2017). The Food and Drug Administration approved the use of the

TTFields device for treatment of recurrent GBM in 2011, and for newly diagnosed GBM in 2015 (Fabian et al., 2019). However, despite these recent advances, GBM tumors exhibit nearly universal recurrence and remain incurable.

Several clinical trials aimed at a variety of promising therapeutic targets have sought to improve outcomes for patients with GBM, albeit with limited success. For instance, a phase III clinical trial evaluated the efficacy of a dose-dense TMZ regimen (75-100 mg/m2,

21 days of a 28-day cycle), hypothesizing that prolonged TMZ exposure would deplete intracellular levels of the DNA repair enzyme O6-methylguanine DNA methyltransferase

(MGMT) (Gilbert et al., 2013). Epigenetic silencing via methylation of the MGMT gene has been associated with increased overall survival in patients diagnosed with high-grade glioma, due to the inability to repair DNA damage induced by TMZ (Esteller et al., 2000;

Gilbert et al., 2013). However, dose-dense TMZ did not increase overall survival or progression-free survival in a cohort of GBM patients (Gilbert et al., 2013). A separate randomized phase III clinical trial assessed the addition of bevacizumab, a monoclonal antibody against the angiogenesis-promoting vascular endothelial growth factor (VEGF), to standard-of-care therapy, but found no significant increase in overall survival in the bevacizumab group (Gilbert et al., 2014). More recent reports indicate that addition of the mechanistic target of rapamycin (mTOR) inhibitor everolimus to standard therapy led to

3 increased treatment-related toxicities and decreased overall survival in a randomized phase II study (Chinnaiyan et al., 2018). Moreover, while immunotherapy has continued to gain momentum and has proven an effective treatment strategy for some patients diagnosed with melanoma (Larkin et al., 2015), non-small cell lung (Scott J. Antonia et al., 2016), and renal malignancies (Motzer et al., 2015), high-grade glioma patients treated with immunotherapy have not demonstrated benefits in overall survival in any phase III clinical trials to date (Reardon et al., 2017a).

1.3. GENETIC AND MOLECULAR STRATIFICATION OF GBM

The advent of genomic profiling and the continual refining of nucleotide sequencing capabilities have provided powerful tools for the stratification of tumors in order to identify more personalized, rational, and targeted therapeutic strategies with tumor heterogeneity in mind. The Cancer Genome Atlas (TCGA) has provided a means with which to understand the comprehensive catalog of genomic aberrations driving tumorigenesis.

Indeed, the recent WHO re-classification of GBM tumors to include molecular information

(i.e., IDH-1/2 mutation status) acknowledges the strides taken in our understanding of the genomic landscapes observed across GBM tumors.

Utilizing TCGA data, seminal work by Verhaak et al. laid the foundation for the current understanding of genomic abnormalities driving GBM tumors by stratifying tumors into four different subtypes characterized by distinct genetic aberrations: classical, mesenchymal, proneural, and neural (Verhaak et al., 2010). The classical subtype is defined by chromosome 7 amplification paired with loss of chromosome 10 (observed in

100% of the classical subtype) and high-level amplification of the epidermal growth factor

4 receptor gene (EGFR, observed in 97% of the classical subtype). Notably, regardless of subtype, EGFR amplification is observed in approximately 50% of all GBM tumors and is often accompanied by expression of the tumor-specific, constitutively active EGFR variant

3 (EGFRvIII) mutant, which is characterized by loss of exons 2-7 and truncation of its extracellular ligand-binding domain (Gan, Kaye, & Luwor, 2009). EGFRvIII is expressed in approximately 50% of classical tumors and 25% of GBM tumors regardless of subtype.

The mesenchymal subtype is characterized by a predilection for focal hemizygous deletions of a region on chromosome 17 containing the gene NF1. Proneural subtype tumors featured alterations in PDGFRA and point mutations in IDH1. Neural subtype tumors were unique in their expression of neuronal markers such as NEFL, GABRA1,

SYT1, and SLC12A5.

Epigenetic analyses of GBM tumors has revealed that silencing via promoter methylation of the MGMT gene is a predictor of longer overall survival in patients treated with TMZ and radiotherapy (Brennan et al., 2013; Hegi et al., 2005). Expression of the

DNA-repair enzyme MGMT confers tumor cell resistance by blunting the therapeutic efficacy of DNA-alkylating agents such as TMZ. Methylation of the MGMT promoter is often coupled with the glioma cytosine-phosphate-guanine (CpG) island methylator phenotype (G-CIMP), characterized by hypermethylation at a large number of foci

(Noushmehr et al., 2010). Overall, MGMT promoter methylation, G-CIMP DNA methylation, IDH1 mutation, as well as patient age less than 50 years, are all independent positive prognostic factors (Brennan et al., 2013).

Advances in single-cell gene expression capabilities have recently provided higher resolution for understanding GBM tumor compositions, treatment-induced tumor

5 evolution, and the GBM tumor microenvironment. Single-cell RNA seq of primary GBM tumors has revealed striking inter-sample variability (Patel et al., 2014). Multivariate analysis consisting of hierarchical clustering and principal component analysis revealed four expression signatures enriched for genes related to cell cycle, hypoxia, immune response, and oligodendrocyte function (Patel et al., 2014). Additionally, all five tumors in this study exhibited a subpopulation of cells with a stem-like signature that correlated negatively to the cell cycle signature, indicating that subpopulations of glioma stem cells

(GSCs) divide at lower overall rates (Patel et al., 2014). Intriguingly, all five tumors analyzed consisted of heterogeneous mixtures of individual cells corresponding to different GBM subtypes, as well as some “hybrid” cells exhibiting signatures reminiscent of two subtypes (Patel et al., 2014). Indeed, spatial intratumoral differences in mutational load and clonality of tumor cells have been described, with some mutations being regionally exclusive (R. V. Lukas et al., 2019). Altogether, these studies underscore the vast heterogeneity in intra- and intertumor gene expression landscapes in GBM.

More recent analysis from the Verhaak group using single-cell gene expression analysis of IDH-wt GBMs and derivative neurospheres has revised the original subtype class system to include classical, mesenchymal, and proneural as the three primary subtypes comprising GBM tumors (Q. Wang et al., 2017). The previously observed neural subtype was likely an artifact of sample contamination with neurons related to the tumor margin (Q. Wang et al., 2017). In this analysis, median survival of mesenchymal tumor cases was 11.5 months, compared to 14.7 and 17.0 months for classical and proneural cases, respectively. This analysis provided a snapshot of the GBM immune microenvironment; illustrated the fluid nature of primary and recurrent GBMs, which were

6 demonstrated to be capable of switching molecular subtype upon recurrence; and demonstrated transitions in the microenvironmental immune landscape in response to treatment.

1.4. PATHOLOGICAL CHARACTERISTICS OF GBM TUMORS

As evidenced by studies including those discussed above, tumor heterogeneity and plasticity are major contributing factors to the limited success observed in achieving robust survival benefits in patients with GBM. Histologically, GBMs are distinguished from lower grade gliomas by virtue of a number of defining characteristics, most notably areas of necrosis and microvascular proliferation (Brat et al., 2004; Louis, 2006). Surrounding these necrotic foci are hypercellular zones called pseudopalisades comprised of neoplastic cells migrating away from the hypoxic core (Brat et al., 2004). Cells within pseudopalisades are characterized as less proliferative than adjacent tumor, show nuclear expression of HIF-1α, and exhibit enhanced gelatinase activity (Brat et al., 2004).

The majority of pseudopalisades arise as a direct result of hypoxia arising from a vaso- occlusive event, such as distorted, degenerated, or thrombosed blood vessels, indicating that pseudopalisading cells arise from a sequence of migration away from a hypoxic, necrotic foci suffering from inefficient, though hyperproliferative, neovascularization (Brat et al., 2004; Louis, 2006).

The definitive cell of origin for malignant gliomas remains elusive. While mature glial cells could potentially undergo neoplastic events during reactive proliferation, a paucity of epidemiological evidence linking events evoking reactive proliferation, e.g. trauma, with the development of gliomas undermines this possibility (Louis, 2006). Instead, work in

7 both animal models and primary gliomas suggests that malignant gliomas likely arise from primitive pluripotent cells, including neural stem cells, glial precursor cells, and oligodendrocyte precursor cells (Holland et al., 2000; R. V. Lukas et al., 2019).

Neuroectodermal stem cells have proliferative potential, are migratory, and can undergo diverse paths of differentiation (Louis, 2006). Additionally, malignant gliomas have been demonstrated to harbor cancer stem cell populations responsible for tumor propagation

(Galli et al., 2004; Singh et al., 2004). Work using murine xenograft models has demonstrated that GBM tumors harbor a slow-cycling population of stem-like cells, which give rise to a rapidly dividing progenitor cell population, which in turn gives rise to terminally differentiated daughter cells, altogether supporting a hierarchical model of gliomagenesis and intratumor heterogeneity (Lan et al., 2017). Recent work by Lee et al. indicates that human GBM tumors arise from transformed neural stem cells (NSC) in the subventricular zone (SVZ) (J. H. Lee et al., 2018). Using deep sequencing of triple- matched tissues from 28 patients (tumor tissue, normal SVZ away from tumor, normal cortical tissue), an average of 23.0 somatic mutations in tumor-free SVZ were observed compared to 80.7 in tumor specimens. Some of these somatic mutations were shared between tumor-free SVZ and tumor, an observation that occurred only in IDH-wt GBM

(i.e., primary de novo GBM). Shared alterations included mutations in TERT promoter,

EGFR, PTEN, and TP53. Notably, mouse models of p53/Pten/Egfr mutations targeted to

NSCs in the SVZ generated cortical brain tumors, which developed after migration from the SVZ to distant cortical regions (J. H. Lee et al., 2018).

Regardless of genetic heterogeneity, all malignant gliomas share a cardinal pathological characteristic in cell invasion. GBM tumors are highly invasive and rapidly

8 infiltrate surrounding brain parenchyma. Amplification of EGFR has been associated with fast migratory behavior of cells (Parker et al., 2018). Indeed, the highly invasive nature of

GBM cells is a chief contributor to disease progression and recurrence (Alieva et al.,

2019). Invasion represents an early event in GBM progression, and infiltrating cells which lie beyond the definable margin for tumor resection give rise to inevitable tumor relapse, typically within 1-2 cm of the primary tumor border (Alieva et al., 2019; Louis, 2006).

Similarly to adult stem cells in the mature brain, GBM cells tend to migrate along existing brain structures, such as white matter tracts and blood vessels (Cuddapah, Robel,

Watkins, & Sontheimer, 2014). Additionally, a variety of invasion patterns have been described, including single cell invasion as well as group migration consisting of leading fronts and follower cells (Alieva et al., 2019). Despite being overly adept at infiltration throughout brain, GBM tumors rarely ever metastasize out of the brain (0.4-2%). This paucity of extracranial metastases originating from GBM or other primary brain tumors may be due in part to an inability of tumor cells to breach the blood-brain barrier (BBB) basement membrane and enter vasculature, an unfavorable microenvironment in extraneural tissues, a patient survival time too short to observe extracranial metastasis, or a combination of the aforementioned (Cuddapah et al., 2014).

Migration of tumor cells in ex vivo culture GBM tissue culture slices was observed to be intermittent, characterized by stages of filopodia-mediated monitoring of the surrounding environment, followed by bursts of rapid movement, during which filopodia are retracted and nuclei deformed (Parker et al., 2018). Notably, this mode of migration has also been ascribed to neural progenitor cells (Parker et al., 2018). Other recent work by Alieva et al. utilizing high-resolution, time-lapse intravital imaging provided in-depth

9 characterization of invasive patterns of patient-derived orthotopic GBM tumors in SCID mice (Alieva et al., 2019). Analysis of patient-derived cells stably expressing H2B-

Dendra2 (a fluorescent marker of nuclei) revealed three invasion patterns: 1) “invasive margin,” characterized by protruding multicellular groups originating at the transition between tumor and parenchyma; 2) “well-defined tumor border,” characterized by margins with no protrusions; and 3) “diffuse cell infiltration,” characterized by individual cell migration into surrounding tissue (Alieva et al., 2019). In all patterns (as well as in the tumor core), movement was highly dynamic, and cells were observed moving both toward and away from the tumor. However, the “invasive margin” and diffusively invading cells exhibited the most persistent, directional movement of cohorts of cells or highly motile individual cells, respectively, contributing to tumor expansion. Additionally, perivascular migrating cells moved faster than parenchymal migrating cells. Collectively, this and other studies characterizing tumor invasion point to tumor microenvironmental triggers which influence GBM cell invasion.

The GBM tumor microenvironment is characterized by a distinctive and heterogeneous population of cells, including endothelial cells, pericytes, and immune cells, as well as tissue specific astrocytes, neurons, and microglia (Figure 1.1).

Additionally, gradients in nutrient and oxygen availability, as well as a milieu of biophysical cues, add to the complement of extracellular factors regulating GBM cell behavior, selecting for more invasive and therapeutically resistant cell populations, and supporting

10 tumor progression. The following sections will focus on two aspects of the GBM tumor microenvironment, namely hypoxia and tumor-associated microglia.

Figure 1.1. Characteristics of the GBM tumor microenvironment. The GBM tumor microenvironment is populated with a heterogeneous collection of cells participating in dynamic heterotypic cell-cell and cell-environment interactions. Rapid tumor growth, erratic neovascularization, and vaso-occlusive events result in tumor hypoxia, or physiologically low levels of oxygen, which selects for more malignant cell populations. Additionally, non-neoplastic cells, such as astrocytes, bone marrow-derived macrophages (BMDM), and microglia assist in tumor microenvironment remodeling and immunosuppression, mediated in part by Tregs. Additionally, mechanochemical cues from brain ECM, which contains high amounts of hyalyronic acid, glycosaminoglycans, and proteoglycans, influence cell behavior.

1.5. HYPOXIA

Hypoxia, defined as physiologically low levels of oxygen (O2), presents as a common microenvironmental feature in solid tumors, including GBM (Monteiro, Hill, Pilkington, &

Madureira, 2017). Like other solid tumors, GBM is characterized by a heterogeneous pattern of oxygenation (Richards, Jenkinson, Haylock, & See, 2016). While normal oxygen levels in healthy human brain range from 5-8%, oxygen levels in GBM tumors range from 0.75-2.76% (Richards et al., 2016). Intra-tumor hypoxia occurs as a consequence of rapid cell proliferation and tumor growth coupled with aberrant and inefficient tumor neovascularization, which together result in poor oxygen diffusion throughout the tumor (Bar, Lin, Mahairaki, Matsui, & Eberhart, 2010; Monteiro et al.,

2017). Hypoxia increases GBM cell resistance to radiotherapy and chemotherapeutics and promotes cell invasion, presumably away from the hostile metabolic environment it creates (Monteiro et al., 2017; Richards et al., 2016). Hypoxia-induced migration of GBM cells results in the distinct GBM pathophysiological feature of necrotic foci containing microvascular proliferation surrounded by hyperproliferative cellular pseudopalisades, which form an invasive front (Monteiro et al., 2017). In line with these findings, greater volume of hypoxia and intensity in GBM tumors are both associated with decreased time- to-progression and patient survival (Spence et al., 2008).

Hypoxia induces the activity of the hypoxia-inducible factor (HIF) transcription factors,

HIF-1α and HIF-2α, which are the canonical master regulators of the hypoxia adaptive response (Monteiro et al., 2017). Under normoxic conditions (normal oxygen levels), HIFs are constitutively degraded via binding of the von Hippel-Linau (VHL) protein to prolyl residues within the HIF-α subunit that are hydroxylated by oxygen-dependent prolyl hydroxylases (PHDs) 1-3, subsequently recruiting E3 ubiquitin ligases that target HIF-α for proteasomal degradation (Monteiro et al., 2017; Schito & Rey, 2018). Conversely, hypoxia results in inhibition of PHDs, abrogated proteasomal degradation, and subsequent stabilization of HIF-α subunits. HIFs operate by binding to cis hypoxia responsive elements (HRE) within promoters of their target genes, which include a suite of genes involved in promoting angiogenesis, cell migration and invasion, survival and proliferation, and metabolic reprogramming (Monteiro et al., 2017; Schito & Rey, 2018;

H. Xie & Simon, 2017). Notably, EGFR signaling results in canonical activation of the

PI3K/AKT/mTOR signaling cascade, which itself can result in subsequent upregulation of

HIF-1α (Monteiro et al., 2017). In GBM, amplification of EGFR, constitutive EGFR

12 signaling as a result of the EGFRvIII mutation, and/or phosphatase and tensin homolog

(PTEN) deletion (observed in up to 60% of GBM cases) collectively exacerbate HIF-1 α upregulation. Thus, hypoxia and the molecular signature of GBM synergistically potentiate the progression of tumor growth and dissemination throughout surrounding brain parenchyma.

Indeed, the effect of hypoxia on tumor biology, including GBM, has received tantamount attention. Hierarchical clustering of single cell RNAseq (sc-RNAseq) data revealed a hypoxic gene expression signature within GBM tumors (Patel et al., 2014).

Hypoxia has been demonstrated to result in GBM cell migration, necrosis, and the histopathological hallmark of pseudopalisades (Brat et al., 2004; J. E. Chen, Lumibao,

Blazek, Gaskins, & Harley, 2018; W. Huang et al., 2018). Hypoxia additionally promotes chemoresistance by upregulation of the ATP-binding cassette subfamily B member 1

(ABCB1) transporter, an ATP-dependent drug efflux pump conferring multidrug resistance (X. Zhang, Ding, Wang, Li, & Zhao, 2018). Additionally, hypoxia induces radioresistance of GBM cells (Benej et al., 2018), and recent efforts have targeted mitochondrial respiration and oxygen consumption as a strategy to induce tumor re- oxygenation and sensitization to radiotherapy (Benej et al., 2018; Dewhirst, 2018).

One major consequence of intra-tumor hypoxia is the metabolic reprogramming conducted by cells to maintain cellular bioenergetics in the face of decreasing oxygen tensions. Indeed, HIF-1 activity leads to changes in gene expression that enhance glycolysis and attenuate oxidative phosphorylation and oxygen consumption (Loscalzo,

2016). Accompanying this shift is increased expression of monocarboxylate transporters, including MCT-4, to enhance efflux of lactate and H+ produced as a result of increased

13 glycolytic flux (Schito & Rey, 2018). Notably, constitutive expression of MCT-1 on neighboring cells allows non-hypoxic cell populations to uptake secreted lactate, which is then converted to pyruvate via LDH-B, thus feeding the TCA cycle and oxidative phosphorylation in an elegant tumor microenvironmental metabolic symbiosis.

Additionally, glycolytic and TCA cycle intermediates are diverted into anabolic reactions necessary for cell division and tumor growth. For example, hypoxia induces the entry of glucose-derived carbons to the pentose phosphate pathway to generate reducing power in the form of NADPH needed for antioxidant and biosynthesis reactions (Kathagen et al.,

2013). This shift toward glycolytic flux additionally serves to mitigate the generation of harmful reactive oxygen species (ROS) that are produced by the leakage of electrons from an inefficient mitochondrial electron transport chain (primarily from complexes I and

- III), which in turn react with oxygen to form superoxide anion (O2 ) and, subsequently, hydrogen peroxide (H2O2) (Loscalzo, 2016; Schito & Rey, 2018).

Excess production of ROS in hypoxia that exceeds endogenous antioxidant capacity

(either through antioxidant enzymes or small molecule antioxidants) results in oxidative stress (Loscalzo, 2016). Complementary to this, reductive stress implies the presence of excess reducing equivalents that cannot be quenched by cellular oxidoreductases or oxygen tensions in the surrounding environment (e.g., limited oxygen in hypoxia)

(Loscalzo, 2016). In both cases, resultant redox reactions have the ability to modulate biological phenotype. However, care must be taken to distinguish between “oxidative” and

“reductive” stress in the biological context. For example, sustained hypoxia canonically leads to an increase in ROS production, implying oxidative stress. However, these ROS

- (e.g., superoxide anion, O2 ) are partially reduced forms of oxygen, the consequences of

14 which change depending upon the redox reactions they are engaged in (Loscalzo, 2016).

Redirecting metabolism in favor of increased glycolytic flux during hypoxia produces a mounting pool of reducing equivalents in the form of NADH and NADPH, implying reductive stress. Given the pleiotropic effects that arise from hypoxia-induced redox perturbations, “redox stress” arises as an appropriate term to describe the functional consequences of oxidative and reductive disturbances within cells (Loscalzo, 2016). The presence of structural and functional disulfides which respond to redox stress to modify protein function and influence cellular adaptation has gained increasing attention and is of particular interest to this work and is discussed further in later chapters.

1.6. TUMOR-ASSOCIATED MICROGLIA

Microglia are the primary resident immune cells of the CNS parenchyma and account for

5-12% of cells in the brain (Alliot, Godin, & Pessac, 1999; Hickman et al., 2013;

Nimmerjahn, Kirchhoff, & Helmchen, 2005). As mediators of baseline CNS immunity, microglial plasticity allows them to assume neurotoxic or neuroprotective states that determine their responses to changes in the surrounding microenvironment, invading pathogens and toxins, neurodegenerative diseases and traumatic brain injury, and cellular debris (Hickman et al., 2013; Nimmerjahn et al., 2005). Seminal work conducted by Nimmerjahn et al. (2005) revealed that, in the healthy brain, microglia continuously and dynamically surveil their surroundings via continuous cycles of de novo formation and withdrawal of cellular processes, which interact with cortical elements such as astrocytes, neurons, and blood vessels (Nimmerjahn et al., 2005).

Brain resident microglia develop from embryonic yolk sac progenitor cells (Alliot et al.,

1999). In mouse brains, seeding of cells with myeloid features occurs as early as embryonic day 8.5-9, with the BBB beginning to form at embryonic day 13.5, thus isolating the developing brain from contributions of fetal liver hematopoiesis (Butovsky et al.,

2014). Development of microglia is dependent on TGF-β1, as mice with TGF-β1 deficiency targeted to the CNS display loss of microglia (Butovsky et al., 2014). Stable microglia populations in adult brain are maintained primarily through prolonged cellular longevity and in situ proliferation, as adult microglia are not replenished postnatally via peripheral hematopoiesis (Ajami, Bennett, Krieger, Tetzlaff, & Rossi, 2007).

In adult brain and CNS, microglia mediate baseline immunity via constant physical surveillance of their surroundings (Nimmerjahn et al., 2005). As sentinels of CNS parenchyma, microglia require a complement of proteins for sensing extracellular perturbations, such as microbial invasion, neurodegenerative cell injury or death, chemokines and cytokines, or extracellular matrix changes. Some of these proteins, described collectively as the microglial sensome, are differentially expressed between microglia and bone marrow derived macrophages and consist primarily of transmembrane and cell surface receptors including pattern recognition receptors, chemoattractant and chemokine receptors (e.g. Cxcr4, Cxcr2), Fc receptors (e.g. Fcgr3,

Fcer1g), toll-like receptors (e.g. Tlr2, Tlr7), cytokine receptors, purinergic receptors (e.g.

P2ry12, P2rx7), receptors involved in cell-cell interaction, and receptors for extracellular matrix proteins (Hickman et al., 2013).

The BBB, which is composed of specialized capillary endothelial cells joined by complex tight junctions, establishes a barrier between the blood and CNS parenchyma

(Engelhardt, Vajkoczy, & Weller, 2017). Endothelial cells of the BBB are covered by a basement membrane formed by the fusion of endothelial and glia limitans basement membranes, which are further supported by astrocyte end-feet and in which pericytes are embedded. Given this tight barrier, the brain has conventionally been regarded as an

“immune-privileged” organ, whose interaction with peripheral immunity is tightly regulated

(Galea, Bechmann, & Perry, 2007). However, evidence that activated circulating T cells can cross the BBB in the absence of neuroinflammation and fact that cerebrospinal fluid drains into cervical lymph nodes have challenged this concept (Engelhardt & Ransohoff,

2012; Louveau et al., 2015). Furthermore, disruption of the BBB as a result of rapid glioma outgrowth facilitates entry of tumor-infiltrating lymphocytes from the periphery (Louis,

2006; Wainwright, Sengupta, Han, & Lesniak, 2011). Rather than complete immune cell exclusion, GBM tumors undermine anti-tumor immunity by cultivating a potently immunosuppressive microenviornment, via secretion of immunosuppressive cytokines

(e.g. TGF-B, IL-10), expression of inhibitory molecules (e.g. PD-L1), and recruitment of immunosuppressive cells such as regulatory T cells, myeloid-derived suppressor cells, and tumor-associated microglia and macrophages (Binder, Davis, & Wainwright, 2016;

Dutoit, Migliorini, Dietrich, & Walker, 2016; Wainwright & Lesniak, 2014).

The heterogeneity of the GBM immune landscape across tumors has been highlighted by previous efforts in single cell expression and next generation sequencing techniques, and data indicate that immune responses and cell populations vary among subtypes (Q.

Wang et al., 2017). For instance, mesenchymal tumors were characterized by a greater frequency of IBA1+ tumor-infiltrating microglia and macrophages compared to non- mesenchymal tumors (i.e. classical and proneural) (Q. Wang et al., 2017). An activated

17 natural killer cell gene signature was reduced in the mesenchymal subtype, and the resting memory CD4+ T cell gene signature was reduced in the proneural subtype, while classical subtype tumors displayed greater activated dendritic cell gene signatures (Q.

Wang et al., 2017). These observations suggest that different immunotherapeutic modalities may exhibit differential efficacy against GBM tumors based on subtype.

Tumor-associated microglia (TAM) and bone marrow-derived macrophages (BMDM) can constitute as much as 30%-50% of GBM tumor bulk (Charles, Holland, Gilbertson,

Glass, & Kettenmann, 2011; da Fonseca & Badie, 2013; Hambardzumyan, Gutmann, &

Kettenmann, 2016; Quail & Joyce, 2017; Szulzewsky et al., 2015). Intra-tumor density of

TAM and BMDM is highest in malignant high-grade gliomas compared to lower grade gliomas and is inversely correlated with patient survival (Choi, Stradmann-Bellinghausen,

Yakubov, Savaskan, & Regnier-Vigouroux, 2015; Gieryng, Pszczolkowska,

Walentynowicz, Rajan, & Kaminska, 2017). Originating either from peripheral circulation as BMDMs or representing brain-intrinsic TAM, they are actively recruited to GBM tumors via tumor secretion of chemokines, cytokines, and soluble matrix proteins (da Fonseca &

Badie, 2013; Hambardzumyan et al., 2016). In turn, paracrine signaling networks among

GBM cells, TAM, and BMDMs propagate the survival and proliferation of the tumor, making TAM and BMDMs an increasingly investigated aspect of the GBM tumor microenvironment. TAM commonly produce pro- as well as anti-inflammatory molecules and matrix remodeling proteins (VEGF, MMP-2, MMP-9, MT1-MMP), thus promoting tumor growth and invasion and local immunosuppression (Hambardzumyan et al., 2016).

Accurate identification of TAM versus BMDMs in gliomas has largely been precluded by a lack of definitive markers capable of distinguishing the two very closely related cell

18 populations. Most studies use CD45 expression levels to differentiate between microglia

(CD45low) and peripherally-derived macrophages (CD45high). Other microglia-specific markers have been introduced recently, including Tmem119, P2ry12, Cx3cr1, and Siglec-

H (Bennett et al., 2016; Butovsky et al., 2014). The ability to discriminate between TAMs and BMDMs will enable investigation of relative contributions from each cell type to tumor progression.

The impact TAM and BMDMs have on tumor progression has inspired therapeutic strategies aimed at depleting them from the GBM tumor microenvironment. These efforts have focused primarily on inhibition of the colony-stimulating factor 1 receptor (CSF-1R) signaling axis, the inhibition of which depletes TAM (Coniglio et al., 2012). GL261 glioma cells exhibited significantly reduced invasion in a Matrigel invasion assay when co- cultured with microglia pre-treated with PLX3397 (Coniglio et al., 2012). Preclinical studies in orthotopic xenograft GBM mouse models have demonstrated that ionizing radiation upregulates CSF-1R expression and increases Iba1+ BMDMs in tumors

(Stafford et al., 2016). Administration of PLX3397 depleted Iba1+ BMDMs in developing tumors, sensitized tumors to ionizing radiation, and increased median survival time

(Stafford et al., 2016). However, a phase II multicenter study demonstrated limited efficacy of PLX3397 as a single agent in increasing progression-free survival of patients with recurrent GBM (Butowski et al., 2016). Intriguingly, the only two patients who exhibited improved PFS in this study both displayed tumors similar to a mesenchymal subtype, which have since been demonstrated to harbor a greater frequency of TAM and

BMDM (Q. Wang et al., 2017). More recent preclinical investigations demonstrate robust survival benefits in mice bearing intracranial gliomas treated with trimodal combination

19 therapy consisting of dendritic cell vaccination, anti-PD-1 checkpoint inhibitor, and

PLX3397, underscoring the significance of combination therapies (Antonios et al., 2017).

While therapies such as PLX3397 and BLZ945 (another CSF-1R inhibitor) targeting the TAM and BMDM compartments of the GBM tumor microenvironment show promising preclinical results, long-term regression remains a challenge as a subset of tumors invariably acquires resistance to CSF-1R inhibition in a manner that is dependent upon the tumor microenvironment itself (Quail et al., 2016). Enhanced understanding of the complex milieu of signals tumor cells receive from neighboring malignant, cancer- associated, and co-opted endothelial and tissue-resident cells, as well as how they sense and respond to gradients in extracellular metabolic and biophysical cues, may reveal new approaches to managing tumor burden and improving patient care.

CHAPTER 2: RETROGRADE SIGNALING BETWEEN MITOCHONDRIA AND

NUCLEI

2.1. MITOCHONDRIA

The evolution of mitochondria within modern eukaryotic cells is firmly established as having arisen from the endosymbiotic engulfment of a prokaryotic organism (α- proteobacterium) approximately 2 billion years ago (Archibald, 2015; Muller et al., 2012;

Yang, Oyaizu, Oyaizu, Olsen, & Woese, 1985). It remains to be definitively determined whether this initial endosymbiotic event occurred between a complex Archezoa and prokaryote (“mitochondrion-late” hypothesis) or between two prokaryotes evolving in concert (“mitochondrion-early” hypothesis; hydrogen hypothesis) (Archibald, 2015; Martin

& Muller, 1998). Nevertheless, the horizontal transfer of genes from an α-proteobacterium to host consolidated the symbiotic relationship, and mitochondria have since evolved as organelles critical for proper metabolic functioning and bioenergetics within most eukaryotic cells.

Mitochondria are comprised of two membranes: an outer mitochondrial membrane

(OMM) and inner mitochondrial membrane (IMM), with the IMM housing respiratory complexes I, II, III, IV of the electron transport chain (ETC) and ATP synthase. The surface area of the IMM is increased by virtue of membrane folds called cristae, the structural integrity of which are maintained by multiprotein complexes at cristae junctions called MICOS (mitochondrial contact site and cristae organizing system) (Wollweber, von der Malsburg, & van der Laan, 2017). Seven mammalian MICOS subunits have been identified which comprise the large MICOS multiprotein complex: Mitofilin, CHCHD3,

CHCHD6, APOO, APOOL, QIL1, and MINOS1 (Wollweber et al., 2017), with Mitofilin and

MINOS1 acting as key core components. The space created between the OMM and IMM is termed the inter-membrane space (IMS), while the compartment encapsulated by the

IMM is the mitochondrial matrix, in which a variety of metabolic activities are conducted, including fatty acid oxidation and the tricarboxylic acid (TCA) cycle, which produce reducing equivalents for the electron transport chain.

Within cells, mitochondria are critically responsible for a plethora of functions, including regulation of the intrinsic pathway of apoptosis, calcium signaling, generation of heme and iron-sulfur clusters, carbohydrate and fatty acid metabolism, redox homeostasis, and generation of ROS as signaling molecules (Connolly et al., 2018).

Arguably, the most widely recognized function of mitochondria is the oxygen-dependent generation of energy in the form of ATP, driven by an electrochemical gradient of H+ across the IMM. This gradient of protons is generated by the actions of the ETC, comprised of Complex I (NADH:ubiquinone oxidoreductase/NADH dehydrogenase),

Complex II (succinate-ubiquinone oxidoreductase/succinate dehydrogenase), Complex

III (ubiquinol-cytochrome c oxidoreductase), and the oxygen consuming Complex IV

(cytochrome c oxidase). Electrons are donated initially to Complexes I and II via NADH and FADH2, respectively. The movement of electrons through the electron transport chain is coupled to the pumping of hydrogen ions (H+) against their concentration gradient from the mitochondrial matrix into the IMS. In turn, a chemiosmotic gradient is produced across the IMS, and H+ can only move down its electrochemical gradient and back into the matrix through ATP synthase. This proton motive force drives ATP synthase and is coupled with the production of ATP from ADP.

2.2. A GENERAL OVERVIEW OF MITOCHONDRIA IN CANCER

An emerging hallmark of cancer is cellular reprogramming of bioenergetics in order to provide the biomass and anabolic precursors for rapid cell proliferation and the bioenergetics to participate in cellular events such as migration, invasion, and metastasis

(Hanahan & Weinberg, 2011). The first observation of striking increases in glucose consumption by tumors compared to non-malignant tissue was described by the German physiologist Otto Warburg (Warburg, 1956; Warburg, Wind, & Negelein, 1927). Since, this phenomenon has been observed in a variety of tumor contexts and has been leveraged successfully in the clinic via positron emission tomography (PET)-based imaging of tumor uptake of a radioactive fluorine-labeled glucose analog [18F- fluorodeoxyglucose (18F-FDG)], allowing for tumor diagnosis, staging, and monitoring of treatment responses (P. J. Song, Lu, Li, Li, & Shen, 2016). Cancer metabolism has since greatly evolved into a burgeoning field of research, with six hallmarks of cancer metabolism described to date, including: 1) deregulated uptake of glucose and amino acids; 2) use of opportunistic modes of nutrient acquisition; 3) use of glycolysis and TCA cycle intermediates for biosynthesis and NADPH production; 4) increased nitrogen demand; 5) alterations in metabolite-driven gene regulation; and 6) metabolic interactions with the microenvironment (Pavlova & Thompson, 2016).

The “carbon economy” of a proliferating tumor cell is focused primarily on biosynthesis of biomolecules including fatty acids and cholesterol, nucleotides, pentose and hexose sugar derivatives, non-essential amino acids, reducing power in the form of NADPH, and others (Pavlova & Thompson, 2016). However, assuming that glucose was primarily used to generate ATP in ex vivo tumor slices, Warburg and others concluded, albeit incorrectly,

23 that increased glucose fermentation and tumorigenesis itself was a direct result of mitochondrial respiratory dysfunction (Warburg, 1956). Although this misinterpretation was recognized by some at the time (Weinhouse, 1956), it has nevertheless fueled contemporary investigations into cancer metabolism.

It is now established that mitochondria and respiration are not entirely “deficient” or

“damaged” in cancer, but continue to be critical for maintenance of malignant cell bioenergetics, protection against induction of apoptosis, and generation of ROS as second messengers (Galluzzi et al., 2010). The close interplay between mitochondrial function and hallmarks of cancer have been elegantly reviewed by others (Galluzzi et al.,

2010). Depletion of mitochondrial DNA has been shown to diminish tumorigenic potential of glioblastoma and breast cancer cell lines (Cavalli, Varella-Garcia, & Liang, 1997; A. S.

Tan et al., 2015). Intriguingly, using in vivo mouse models, Tan et al. demonstrated that, breast tumor cells depleted of mitochondrial DNA (mtDNA) and devoid of mitochondrial respiration horizontally acquired mtDNA of host stromal origin in a stepwise recovery of mitochondrial respiration, with primary tumor cells, circulating tumor cells, and metastatic cells showing progressive increases in respiratory capacity (A. S. Tan et al., 2015). In vivo isotope tracing analysis of human brain and lung tumor patients has revealed that these tumors possess an oxidative phenotype, in which glucose is fully oxidized through the TCA cycle, as opposed to primarily converted into lactate (Courtney et al., 2018).

Glioma stem cells have been demonstrated to be less glycolytic than proliferative, differentiated glioma cells (Vlashi et al., 2011). Additionally, mitochondrial spare respiratory capacity has been associated with glioma stem cell resistance to radiotherapy

(Vlashi et al., 2011). Therefore, the Warburg effect represents a regulated state of

24 metabolic reprogramming in which glycolysis is decoupled from TCA cycle flux and oxidative phosphorylation, rather than a consequence of mitochondrial dysfunction.

Mitochondrial function has been highlighted in priming states of invasive tumor cells navigating tumor microenvironments. Existing as tubular, reticular networks, mitochondria constantly undergo dynamic reorganization, i.e. fusion and fission, which controls subcellular compartmentalization of ATP production based on local need. Indeed, formation of lamellipodia by migrating cells is dependent upon reorganization of the actin cytoskeleton and requires an abundance of local energy production at the leading edge, which can be achieved by mitochondrial fission. Metastatic breast cancer cells have been demonstrated to display more fragmented mitochondria coincident with higher levels of the mitochondrial fission protein dynamin-related protein 1 (Drp1) and lower levels of mitochondrial fusion protein 1 (Mfn1) (Zhao et al., 2013). Selection for highly invasive tumor cell populations is accompanied by increased mitochondrial oxygen consumption in SiHA human cervix squamous cell carcinoma cells (Porporato et al., 2014). In this study, mitochondrial production of ROS in the form of H2O2 induced cell migration in a

Src-SMAD-Pyk2-mediated mechanism (Porporato et al., 2014).

Cancer stem cells represent a minority population of tumor cells which are capable of propagating and repopulating tumors (W. S. Song et al., 2016). GBM has been described as following the cancer stem cell hierarchy, with highly plastic brain tumor initiating cells

(BTICs)/glioma stem cells (GSCs) residing at the apex. This population of cells has been characterized as resistant to conventional therapies, elusive of the antitumor immune response, and highly invasive (Fidoamore et al., 2016; W. S. Song et al., 2016).

Therefore, defining qualities that differentiate GSCs from the bulk tumor population has

25 received significant attention. Compared to their non-stem counterparts, GSCs have been described as more respiratory and less glycolytic, a phenotype associated with resistance to radiotherapy (Vlashi et al., 2011). Mitochondria in BTICs from patient-derived xenografts display a more fragmented phenotype, in line with greater expression of activated phosphorylated fission protein Drp1, compared to mitochondria in matched non-

BTIC populations (Q. Xie et al., 2015). shRNA-mediated Drp1 knockdown, as well as pharmacologic inhibition of Drp1-mediated mitochondrial fission with Mdivi-1, significantly decreased BTIC growth (Q. Xie et al., 2015). Notably, increased expression of activated

Drp1, observed in greater levels in tumor vs non-tumor tissue, was associated with poorer

GBM patient survival (Q. Xie et al., 2015).

2.3. RETROGRADE SIGNALING

Mitochondria are semi-autonomous organelles which harbor multiple copies of mtDNA, ranging from 100-1000 copies per cell, depending on cell and tissue type (Guha &

Avadhani, 2013). Approximately 16.5 kb of circular DNA, mtDNA encodes only 13 essential structural subunits out of the 90 protein subunits that comprise the respiratory complexes of the ETC, while the majority of respiratory subunits and mitochondrial proteome are encoded by the nuclear genome (Baughman et al., 2009; Guha & Avadhani,

2013). It is estimated that 1500-2000 nuclear encoded, cytoplasmically translated proteins are imported into mitochondria (Guha & Avadhani, 2013). Therefore, tight coordination between mitochondrial and nuclear gene expression is crucial for the maintenance of appropriate mitochondrial function.

Communication between mitochondria and the nucleus is bidirectional, and can be described by either (1) anterograde or (2) retrograde signaling (Guha & Avadhani, 2013).

Anterograde signaling refers to activation of nuclear transcription and cytoplasmic translation which directly regulates mitochondrial biogenesis. Complementary to this, retrograde signaling originates in mitochondria. Perturbations in mitochondrial function initiate signaling back to the nucleus, resulting in altered expression of nuclear genes

(Eisenberg-Bord & Schuldiner, 2017). This signaling axis thus facilitates a mechanism by which mitochondria communicate with the nuclear genetic compartment to relay metabolic, respiratory, and redox conditions in order to maintain cellular homeostasis

(Guha & Avadhani, 2013).

Three main routes of retrograde signaling have been described by which intracellular proteins mediate communication Figure 2.1. Routes of mitonuclear retrograde signaling. Proteins that participate in mitonuclear communication accomplish their function via three main routes: as participants in the classical retrograde signaling pathway between mitochondria (Protein 1), residing as dually localized “moonlighting” echoforms in both subcellular compartments (Protein 2), or directly translocating from and the nucleus (Figure mitochondria to the nucleus (Protein 3).

2.1). These include: 1) the classical retrograde signaling pathway, in which resident cytosolic proteins sense disruptions in mitochondrial output and translocate to the nucleus from the cytosol; 2) “moonlighting” proteins that are located in both mitochondria and the nucleus, but are differentially shuttled between these compartments based on need; and

3) direct translocation from mitochondria to the nucleus (da Cunha, Torelli, &

Kowaltowski, 2015; Liao & Butow, 1993; Monaghan & Whitmarsh, 2015).

The classical retrograde signaling pathway was first described by Liao and Butow in

Saccharomyces cerevisiae (Liao & Butow, 1993). This pathway is controlled by the

ReTroGrade regulation proteins Rtg1, Rtg2, and Rtg3. Rtg1 and Rtg3 are basic helix- loop-helix/leucine zipper transcription factors that translocate from the cytosol to the nucleus upon activation, which occurs as a result of Rtg2 inhibiting phosphorylation inhibition of Rtg3 by Mks1 (Eisenberg-Bord & Schuldiner, 2017). Activation of the retrograde pathway (e.g. by loss of mtDNA, sudden drops in ATP levels, or spikes in ROS production), leads to genetic reprogramming affecting amino acid metabolism, redox homeostasis, and other metabolic adaptations (Hashim, Mukai, Bamba, & Fukusaki,

2014; Torelli, Ferreira-Junior, Kowaltowski, & da Cunha, 2015). A well-characterized mechanism responsible for maintaining intracellular redox status that could potentially be classified as classical retrograde signaling involves the NRF2-KEAP1 axis (transcription factor NF-E2 p45-related factor 2/Kelch-like ECH-associated protein 1). Under unstressed conditions, KEAP1 sequesters NRF2 to the cytosol and targets it for proteasomal degradation. However, upon redox stress, KEAP1 loses this ability by virtue of cysteine residue oxidation, allowing Nrf2 to localize to the nucleus and regulate the expression of a suite of genes harboring antioxidant response elements (ARE) to bolster defense against redox stress (Cuadrado et al., 2019).

Moonlighting proteins, also known as echoforms, are capable of dual localization in both mitochondria and the nucleus. These are distinct from proteins with multiple isoforms that localize in different compartments, such as the IDH isoforms. Proteins with echoforms across various subcellular compartments include aconitase (mitochondria and the cytosol), Nfs1 (mitochondria and the nucleus), fumarase (mitochondria and the nucleus),

28 and the telomerase component telomerase reverse transcriptase (TERT, mitochondria and the nucleus) (Monaghan & Whitmarsh, 2015; Yogev, Naamati, & Pines, 2011).

Mitochondrial fumarase catalyzes the conversion of fumarate to malate in the TCA cycle, while enhanced localization of fumarase in the nucleus has been observed as a DNA damage response (Yogev et al., 2011). While it has been demonstrated that the enzymatic activity of fumarase is required for its defensive capabilities against DNA damage, how its primary metabolites affect the DNA damage response remains unknown, although it is possible that nuclear fumarate inhibits histone demethylation (Jiang et al.,

2015). TERT has also been demonstrated to localize in both mitochondria and the nucleus to maintain integrity of both nuclear and mitochondrial genomes by protecting against telomere shortening in the nucleus and ROS-induced mtDNA damage in mitochondria (Monaghan & Whitmarsh, 2015).

A third paradigm of mitonuclear communication involves the direct translocation of mitochondrial proteins to the nucleus. A number of proteins have been implicated in retrotranslocation out of mitochondria and import into the nucleus in response to mitochondrial perturbations. Recent work in HEK293 cells has ascribed this ability to

MOTS-c, a 16 amino acid peptide encoded by mtDNA that can regulate cellular metabolism in an AMPK-dependent manner (C. Lee et al., 2015). In HEK293 cells,

MOTS-c translocated from mitochondria to the nucleus in response to metabolic stress imposed by glucose restriction, serum deprivation, and oxidative stress induced by tert- butyl hydrogen peroxide (Kim, Son, Benayoun, & Lee, 2018). In the nucleus, MOTS-c interacts with Nrf2 to regulate expression of a complement of genes involved in a metabolic and oxidative stress response. Similarly to MOTS-c, in differentiating

29 adipocytes and brown adipose tissue, G-Protein Pathway Suppressor 2 (GPS2) translocates from mitochondria to the nucleus in response to mitochondrial depolarization to activate the transcription of nuclear encoded mitochondrial genes coding for the electron transport chain, fatty acid oxidation, and mitochondrial structure (Cardamone et al., 2018). Quite notably, the pyruvate dehydrogenase complex (PDC) has also been suggested to be able to directly translocate from mitochondria to the nucleus during S phase, as well as in response to treatment with epidermal growth factor, complex I inhibition with rotenone, and serum stimulation in A549 cells (Sutendra et al., 2014).

Canonically, mitochondrial PDC catalyzes the irreversible conversion of pyruvate to acetyl-CoA, which can then feed into the TCA cycle. Sutendra et al. concluded that accumulation of nuclear PDC was required for increasing the pool of nuclear acetyl-CoA used for acetylation of key histones important for the progression into S phase of the cell cycle. However, a mechanism describing the export of PDC, a multienzyme complex on the order of thousands of kDa, out of the mitochondrial matrix, across the IMM and OMM, and into the nucleus remains elusive.

The mitochondrial unfolded protein response (UPRmt) is another avenue by which mitochondrial structure, integrity, and function are maintained. Although it is considered distinct from the mitochondrial retrograde stress response heretofore discussed, the two pathways share some overlap with regard to signaling pathways involved. The canonical

UPRmt is driven by the release of unfolded and/or misfolded peptides from mitochondria via the HAF-1 transporter and detection of those peptides in the cytosol, as well as nuclear accumulation and transcriptional activity of activating transcription factor 4 (ATF-4),

C/EBP homologous protein (CHOP), and the dually-localized activating transcription

30 factor associated with stress-1 (ATFS-1), whose import into mitochondria is inhibited upon

UPRmt (Munch, 2018). Ultimately, this results in increased folding capacity either through the induction of chaperones, the reduction of protein uptake into mitochondria, or reduced translation, all in an effort to reduce the load of unfolded proteins. Notably, the UPRIMS is distinct from the canonical UPRmt and is dependent upon estrogen receptor α (ERα) and

ROS-dependent phosphorylation and activation of ERα by AKT (Munch, 2018).

Subsequently, activated ERα induces expression of nuclear respiratory factor 1 (NRF1) and the IMS protease OMI, as well as increased proteasome activity, to maintain mitochondrial IMS proteostasis (Munch, 2018). To what extent classical retrograde signaling pathways involve mitochondrial protein misfolding is an intriguing avenue of future research.

Maintenance of mitochondrial integrity, biogenesis, and respiration has been demonstrated to mediate neoplastic characteristics in a variety of cancer contexts, some described above. The transcription coactivator peroxisome proliferator-activated receptor

γ, coactivator 1 alpha (PGC-1α) is responsive to extracellular stimuli such as nutrient supply and oxygen tension and is regarded as a master regulator of mitochondrial biogenesis and oxidative phosphorylation in mammals (Wu et al., 1999). Although itself possessing no DNA-binding activity, PGC-1α acts in concert with other transcription factors including NRFs and estrogen-related receptors (ERR) to mediate nuclear expression of mitochondrial genes, and is associated with predictive prognosis in a variety of neoplasms including breast, ovarian, prostate, endometrial, and colorectal cancer, among others (Deblois, St-Pierre, & Giguere, 2013). Strong correlations between PGC-

1α expression in invasive breast cancer cells and formation of distant metastases have

31 been observed (LeBleu et al., 2014). Knockdown of PGC-1α abrogated migratory and invasive phenotypes of breast cancer and GBM cells (Bruns et al., 2019; LeBleu et al.,

2014). Thus, mitochondrial functions, as well as their regulation, have deep implications in maintaining and promoting cancer malignancy.

CHAPTER 3: COILED-COIL-HELIX-COILED-COIL-HELIX DOMAIN-CONTAINING

PROTEIN 2

3.1. INTRODUCTION TO CHCH DOMAIN-CONTAINING PROTEINS

Coiled-coil-helix-coiled-coil-helix (CHCH) domain-containing proteins are a family of nucleus-encoded, small mitochondrial proteins with functions in maintenance of mitochondrial structure and function (Modjtahedi, Tokatlidis, Dessen, & Kroemer, 2016;

Z. D. Zhou, Saw, & Tan, 2017). To date, nine members of this protein family have been described (CHCHD1, CHCHD2, CHCHD3, CHCHD4, CHCHD5, CHCHD6, CHCHD7,

CHCHD8, CHCHD10), each of which contains one CHCH domain (with the exception of

CHCHD5, which harbors two) (Z.

D. Zhou et al., 2017). The CHCH domain, in turn, consists of two

CX9C motifs, with each motif defined by two cysteines (Cys) separated by 9 amino acids. The pair of CX9C motifs engages in

Figure 3.1. Import of CX3C and CX9C motif-containing proteins disulfide bond formation, which is is regulated by the Mia40/Erv1 disulfide relay system (CHCHD4/ALR in humans). After translation in the cytosol, necessary for proper oxidative CHCHD2 permeates the OMM via TOM40 in a reduced form. The oxidoreductase CHCHD4, tethered to the IMM via interaction with AIF, catalyzes covalent intermolecular bond disulfide formation, protein folding as well as import ultimately resulting in oxidative disulfide bond formation within incoming substrate. CHCHD4 is then re-oxidized by the ALR, with electrons ultimately transferring to complex IV and H2O. into the mitochondrion.

Each member of the CHCH domain-containing protein family, along with other cysteine motif-containing proteins (such as CX3C proteins), is imported into mitochondria via the redox-regulated Mia40/Erv1 (human equivalents: CHCHD4/ALR) disulfide relay system (Modjtahedi et al., 2016), which features oxidation-driven folding into the IMS

(Figure 3.1). Nucleus-encoded (CX9C)2 proteins are first translated by cytosolic ribosomes, then imported in their reduced form into the IMS through the translocase of the outer membrane (TOM) complex. Central to the relay system is the CHCHD4 oxidoreductase, which is bound to the IMM via its interaction with apoptosis-inducing factor (AIF). CHCHD4 first binds to the MISS motif (mitochondrial IMS-sorting signal) present on incoming protein substrates, followed by covalent intermolecular disulfide bond donation, ultimately resulting in the catalyzed oxidative folding and entrapment of substrate proteins in the IMS (Modjtahedi et al., 2016). The re-oxidation of

Mia40/CHCHD4 is facilitated by the FAD-linked sulfhydryl oxidase Erv1/ALR, with electrons subsequently transferring to FAD, cytochrome c, complex IV of the electron transport chain, and, finally, water.

3.2. CHCHD2

Coiled-coil-helix-coiled-coil-helix domain-containing protein 2 (CHCHD2) is a 16 kDa protein and member of the evolutionarily conserved CHCH domain-containing protein family. The CHCHD2 gene is located on chromosome 7p11.2 and encodes the 151 amino acid CHCHD2 protein, defined by its C-terminal CHCH domain encompassing amino acids 107-147. The CHCH domain itself is defined by two cysteine-X9-cysteine motifs

114 (twin CX9C), each consisting of two cysteines (Cys) separated by 9 amino acids. Cys

124 134 144 and Cys define the first CX9C motif, and Cys and Cys define the second. Two disulfide bonds, formed between Cys114-Cys144 and Cys124-Cys134, stabilize the helix-turn- helix fold of the CHCH domain. These Cys residues are indispensable for proper protein folding and import of CHCHD2 into the mitochondrial inter-membrane space via interaction with CHCHD4 (Aras et al., 2015). In addition, Cys144 is the terminal amino acid in the MISS motif, which interacts with CHCHD4 and is required for CHCHD4-mediated protein import into the IMS via disulfide relay. Furthermore, the N-terminus of CHCHD2 harbors a putative mitochondrial localization sequence defined by amino acid residues 1-

53 (Aras et al., 2015). The sequence of CHCHD2 is highly conserved among human, mouse, and rat, suggesting a critical intracellular function (Funayama et al., 2015).

CHCHD2 was first identified during a computational screen as a protein required for optimal mitochondrial respiration (Baughman et al., 2009). The tight transcriptional regulation of electron transport chain subunits offered a strategy for identifying non- structural regulators based on shared patterns of co-expression in micro-array experiments (Baughman et al., 2009). In a large-scale computational screen, Baughman et al. surveyed 1,427 independent microarray data sets to identify genes that consistently co-expressed with canonical electron transport chain subunits across different biological contexts, and identified CHCHD2 as among these highly co-expressed genes. shRNA- mediated gene silencing of CHCHD2 resulted in reduced oxygen consumption rate of

MCH58 fibroblasts by 40%, decreased levels of the complex IV subunit COX2, and decreased complex IV activity by 40-60% (Baughman et al., 2009). Similar regulation of oxygen consumption by CHCHD2 has since been demonstrated in HEK293 cells (Aras et al., 2017; Aras et al., 2015), non-small cell lung carcinoma (NSCLC) HCC827 cells and

35 xenograft-derived LPC43 lung primary cells (Wei et al., 2015), as well as Drosophila embryonic cells (Meng et al., 2017). This regulation of oxygen consumption has been attributed to physical binding of CHCHD2 with complex IV and positive transcriptional regulation of the tissue specific complex IV subunit 4 isoform 2 (COX4I2) (Aras et al.,

2015). However, other reports indicate a lack of this physical interaction, instead suggesting that CHCHD2 maintains proper electron flow rate through the electron transport chain between complexes III and IV (Meng et al., 2017). Collectively, these studies establish a canonical role for CHCHD2 in mediating efficient electron transport chain function, oxygen consumption, and cellular respiration. However, the precise mechanism by which this is accomplished remains to be fully elucidated.

Functions of CHCHD2 within mitochondria extend beyond regulation of cellular respiration to include maintenance of baseline cellular ROS levels. shRNA-mediated

CHCHD2 knockdown resulted in decreased abundance of ROS scavenging enzymes glutathione peroxidase (GPx), manganese superoxide dismutase (Mn-SOD), and copper/zinc superoxide dismutase (Cu/Zn-SOD), as well as increased total cellular ROS in HEK293 cells (Aras et al., 2015). CHCHD2 knockdown cells also displayed decreased abundance of complex I subunits NDUFB6 and NDUFS3 (Aras et al., 2015). Loss of

CHCHD2 in Drosophila embryonic cells resulted in elevated total ROS production, however without changes in the level of Cu/Zn-SOD (Meng et al., 2017).

More recently, CHCHD2 has been demonstrated to function in the maintenance of inner mitochondrial membrane cristae integrity in HEK293 cells via formation of a tertiary complex with MICS1, a protein associated with maintenance of cristae structure (Oka et al., 2008), and cytochrome c (Meng et al., 2017). Indirect flight muscles of CHCHD2-/-

Drosophila displayed disordered cristae and dilated matrices (Meng et al., 2017). Ectopic overexpression of mutant CHCHD2 protein, as well as siRNA-mediated transient

CHCHD2 knockdown, significantly altered cristae ultrastructure in neural progenitor cells

(NPCs) derived from isogenic human embryonic stem cells (W. Zhou et al., 2018).

Accompanying cristae deformation and loss were decreased levels of MICOS core components Mitofilin and MINOS1, as well as MICOS-associated protein CHCHD10, the latter of which CHCHD2 binds to in a C-terminus-dependent fashion (W. Zhou et al.,

2018). Together, these results provide evidence for a structural role for CHCHD2 in mediating mitochondrial cristae integrity.

CHCHD2 also participates in the negative regulation of the intrinsic pathway of apoptosis that originates at the level of mitochondria. In mouse embryonic fibroblasts, ectopically expressed WT CHCHD2 repressed cytochrome c release and caspase activation, while Parkinson’s disease-associated CHCHD2 mutant isoforms (T61I and

R145Q) failed to do so (Meng et al., 2017). Another study in U2OS osteosarcoma cells demonstrated that CHCHD2 constitutively interacts with the pro-survival B-cell lymphoma

2 (Bcl-2) protein Bcl-xL under unstressed conditions to prevent the oligomerization of pro- apoptotic Bcl-2-associated X protein (Bax) and subsequent mitochondrial outer membrane permeabilization (MOMP), the putative “point of no return” during the intrinsic pathway of apoptosis (Kalkavan & Green, 2018).

Moreove, CHCHD2 mediates extra-mitochondrial cellular functions, such as acting as a hypoxia-sensitive nuclear transcription factor. Early studies focusing on the lung- and trachea-specific isoform of complex IV COX4I2 demonstrated its oxygen-responsive transcriptional regulation, induced maximally at 4% O2 under the control of a HIF-

37 independent cis oxygen responsive element (ORE) (Huttemann, Lee, Liu, & Grossman,

2007). Later work from the same group identified CHCHD2, in concert with recombination signal sequence-binding protein J κ (RBPJ), as transcription factors positively regulating

COX4I2 gene expression in HEK293 and primary pulmonary arterial smooth muscle cells

(Aras et al., 2013).

A subset of literature across diverse biological contexts describes regulatory mechanisms for CHCHD2 at the transcriptional and post-translational levels. Additional work by Aras et al. further demonstrates the ability of CHCHD2 to act as a transcription factor increasing the expression of itself in hypoxia, maximally at 4% O2 (Aras et al.,

2013). Other work has demonstrated cAMP regulatory element binding protein (CREB) as a positive regulator of CHCHD2 gene expression in hepatocellular carcinoma (R. Song et al., 2015). Notably, Zhu et al. demonstrated that CHCHD2 expression in human induced pluripotent stem cells was positively regulated by pluripotency proteins OCT4 and SOX2 (Zhu et al., 2016). One study has provided evidence for post-translational modification of CHCHD2, demonstrating phosphorylation of CHCHD2 at Tyr99 by Abl kinase promoting mitochondrial complex IV activity (Aras et al., 2017).

Additional studies have reported CHCHD2 as a mediator of cell migration. Using a functional genetic screen, Seo et al. identified CHCHD2 as a novel regulator of NIH3T3 cell migration (Seo, Lee, & Suk, 2010). shRNA-mediated knockdown of endogenous

CHCHD2 decreased motility in a Transwell migration assay, while ectopic overexpression increased cell migration in an Akt-dependent manner (Seo et al., 2010). Notably, the physical interaction of CHCHD2 with hyaluronic acid-binding protein 1

(HABP1/C1QBP/p32) suppressed cell migration (Seo et al., 2010). Similar results were

38 observed in 786-O and ACHN renal cell carcinoma (RCC) cells, in which CHCHD2 knockdown led to downregulated cell migration and decreased matrix metalloproteinase

2 (MMP-2) expression (Cheng, Qu, Lu, & Zhang, 2019). Additionally, conditioned media from CHCHD2 knockdown 786-O and ACHN cells failed to stimulate tube formation by

HUVECs, coincident with decreased VEGF secretion (Cheng et al., 2019). A separate study also demonstrated an interaction between CHCHD2 and HABP1, and that knockdown and overexpression of CHCHD2 in lung carcinoma LPC43 cells also reduced and increased cell migration, respectively (Wei et al., 2015). Shimojima et al. demonstrated significantly decreased CHCHD2 expression in induced pluripotent stem cells (iPSCs) from two patients with distinct genetic backgrounds of lissencephaly, one of the most severe forms of cortical migration disorders that occurs from defects in neuronal migration during development (Shimojima et al., 2015). During neuronal differentiation from iPSCs, normal control iPSCs exhibited step-wise increases in CHCHD2 expression over 16 and 48 days, a phenomenon not observed in iPSCs derived from patients with lissencephaly (Shimojima et al., 2015).

In line with results from Shimojima et al., Zhu et al. demonstrated a positive correlation between expression of CHCHD2 in human embryonic stem cells (hESC) and induced pluripotent stem cells (iPSCs) and their ability to efficiently differentiate specifically into neuroectodermal lineages (Zhu et al., 2016). This function was attributed to CHCHD2 binding and sequestering SMAD4 away from the nucleus and in mitochondria. This sequestration in turn suppressed the activity of the TGF-β signaling pathway, which negatively impacted hESC and iPSC neuroectoderm differentiation capability (Sakaki-

Yumoto, Katsuno, & Derynck, 2013). Human iPSCs, which exhibited reduced

39 neuroectodermal differentiation compared to hESCs, also displayed significantly lower levels of CHCHD2 gene expression compared to hESCs (Zhu et al., 2016). Notably, neural stem cells (NSCs) derived from iPSCs exhibited decreased migration capabilities compared to NSCs derived from hESCs, indicating the involvement of CHCHD2 in migration as well as differentiation of pluripotent stem cells (Zhu et al., 2016).

Due to the pleiotropic functions ascribed to CHCHD2 across a variety of biological contexts, the pathological relationship of CHCHD2 with neurodegenerative diseases has gained increasing attention. Indeed, various mutations and polymorphisms within the

CHCHD2 gene have been associated with Parkinson’s disease (PD). A genome-wide linkage analysis on individuals with autosomal dominant PD identified putative pathogenic missense mutations including Thr61Ile and Arg145Gln, as well as one splice-site mutation in a Japanese population (Funayama et al., 2015). Notably, Thr61 and Arg145 are conserved residues among vertebrates (Funayama et al., 2015), suggesting their functional importance. Indeed, Arg145 lies C-terminal adjacent to the CHCH domain. A separate study failed to identify these causal variants in a separate analysis of PD cases of European ancestry, instead noting three separate novel putative pathogenic variants

(Ala32Thr, Pro34Leu, Ile80Val) in 4 cases (Jansen et al., 2015). Foo et al. identified a

Pro2Leu mutation as significantly associated with PD in a Chinese population of sporadic disease (Foo, Liu, & Tan, 2015). A nonsense mutation (Gln126X) was identified in a

German individual with Parkinson’s disease (Koschmidder et al., 2016). However, separate studies in other familial PD cohorts from Arkansas, Sweden, China, and Canada have not identified putative pathogenic variants, or other mutations, in the CHCHD2 gene

(Z. Liu et al., 2015; Puschmann, Dickson, Englund, Wszolek, & Ross, 2015; M. Zhang et

40 al., 2016). The rarity of CHCHD2 variants and their heterogeneity across various populations suggests that mutations in CHCHD2 may contribute to, but do not fully account for, the underlying genetic risk for Parkinson’s disease.

Collectively, these studies highlight the functional relevance of CHCHD2 across a variety of biological contexts. A canonical mediator of mitochondrial function, CHCHD2 has also been ascribed to other functions including regulation of apoptosis, cell migration, stem cell differentiation into neuroectodermal lineages and neural stem cells, and retrograde signaling in hypoxia. Continuing to study and interrogate CHCHD2 in the context of human health and disease is thus an exciting and promising avenue of research.

CHAPTER 4: CRISPR-CAS9-MEDIATED CHCHD2 KNOCKOUT COMPROMISES

MALIGNANT CHARACTERISTICS OF GLIOBLASTOMA CELLS EXPRESSING

EGFRvIII

ABSTRACT

INTRODUCTION

Glioblastoma (GBM) is the most common and malignant primary brain tumor in adults and remains incurable. Although amplification of the epidermal growth factor receptor

(EGFR) is prevalent across GBM tumors, therapies targeting EGFR have failed to robustly increase patient overall survival. Inter-organelle communication between mitochondria and the nucleus may be a potential compensatory signaling route undermining molecular subtype therapies. Previous reports have demonstrated the mitochondrial protein coiled-coil-helix-coiled-coil-helix domain-containing protein 2

(CHCHD2) to mediate nuclear gene expression, cell migration, and apoptosis, but evidence of these capabilities in GBM is lacking. The hypothesis motivating this study was that CHCHD2 served a functional role in mediating GBM malignant phenotype. The objective of the present study was to characterize the functional capacity of CHCHD2 in mitochondrial respiration and glutathione redox homeostasis, cell proliferation and invasion, and therapeutic sensitivity in U87 GBM cells expressing the constitutively active

EGFRvIII mutation (U87vIII).

METHODS

Analysis of CHCHD2 gene amplification patterns and patient survival were conducted on the glioblastoma TCGA Provisional database using cBioPortal. Analysis of CHCHD2 mRNA expression levels was compared among GBM tumors using Gliovis. U87vIII

CHCHD2 knockout (KO) cells were derived using CRISPR-Cas9 genome editing techniques. Oxygen consumption rate (OCR) was measured using a Seahorse XFp

Extracellular Flux Analyzer. Compartmentalized glutathione (GSH) redox state was assessed using genetically encoded fluorescent redox biosensors targeted to the cytosol and mitochondrial matrix. Protein levels of xCT, GPx-1/2, GPx-4, and MMP-2 were analyzed using western blot. The sulforhodamine B assay was used to determine cell proliferation (in normoxia and hypoxia), as well as cell sensitivity to sulfasalazine, erlotinib, temozolomide, and Pac-1. Methacrylated gelatin (GelMA) hydrogels were employed as a biophysical model of the GBM microenvironment to analyze cell invasion in normoxia and hypoxia. CHCHD2 protein levels were analyzed in a panel of GBM patient-derived cells via western blot.

RESULTS

CHCHD2 gene amplification occurred in 9% of GBM tumors, nearly always with EGFR, and was associated with decreased overall survival and progression-free survival.

CHCHD2 mRNA levels were increased in grade IV glioma, IDH-wt GBM tumors, and in tumor versus non-tumor tissue. U87vIII CHCHD2 KO cells displayed decreased OCR and spare respiratory capacity compared to U87vIII CHCHD2 WT cells. The redox state of the

GSH pool was more reduced within the mitochondrial matrix of CHCHD2 KO cells compared to WT cells, an effect independent of GSH synthesis, cystine import via xCT, or GPx-1, GPx-2, or GPx-4 levels. The cytosolic GSH pool was not perturbed by CHCHD2

KO. Furthermore, CHCHD2 KO abrogated the increased proliferation and invasion of

U87vIII cells in hypoxia (1% O2) compared to standard normoxic culture conditions (20%

O2), coupled with decreased expression of MMP-2. U87vIII CHCHD2 KO cells displayed

43 increased sensitivity to sulfasalazine, erlotinib, and temozolomide, but not Pac-1.

CHCHD2 exhibited a heterogeneous protein expression pattern among patient-derived cells investigated, with greater levels observed in more invasive samples.

CONCLUSIONS

CHCHD2 mediates a variety of GBM cell hallmark characteristics, including cell proliferation, cell invasion in response to hypoxia, and cellular resistance to cytotoxic agents.

4.1. INTRODUCTION

Glioblastoma (GBM, WHO grade IV glioma) is the most common, malignant, and aggressive form of primary brain tumor in adults, accounting for approximately 50% of diagnosed gliomas each year (Barnholtz-Sloan et al., 2018). Patients with GBM present with a median survival time of only 15-20 months, with only 5-10% of patients surviving after 5 years (Stupp et al., 2017). Despite the current multimodal standard of care, which consists of maximal surgical resection followed by radiotherapy and chemotherapy with the DNA alkylating agent temozolomide (TMZ), overall prognosis remains poor, underscoring the need for a deeper understanding of tumor biology to inspire new therapeutic targets. Contributing to tumor aggressiveness and virtually universal recurrence are GBM hallmarks, including but not limited to: rapid, diffuse invasion into surrounding brain parenchyma, substantial chemo- and radioresistance, and rapid adaptation to microenvironmental stressors such as hypoxia (Ceccarelli et al., 2016;

Omuro & DeAngelis, 2013).

Genetic aberrations resulting in dysregulated cell signaling perpetuate malignant cell phenotypes. Amplification of the epidermal growth factor receptor (EGFR) is the most common genetic abnormality observed in GBM tumors (Michael Weller et al., 2017).

Increased signaling downstream of EGFR contributes to proliferative and anti-apoptotic signaling, as well as invasive behavior (Gan et al., 2009). Indeed, amplification of EGFR has been associated with fast migratory behavior of GBM cells (Parker et al., 2018).

Tumors that exhibit EGFR amplification also frequently present with the constitutively active EGFR variant 3 (EGFRvIII) mutant (Michael Weller et al., 2017). The EGFRvIII mutant arises from deletion of exons 2-7 of the EGFR gene, leading to deletion of 267 amino acids in the extracellular domain, and resulting in a truncated, yet constitutively active, EGFR protein (Gan et al., 2009). The resultant increased signaling downstream confers enhanced glioma malignancy through multiple mechanisms, and, importantly, the

EGFRvIII mutant is not present on non-malignant tissues (Gan et al., 2009). However, although targeting EGFRvIII is a rational strategy to combat GBM, phase II clinical trials with the EGFR receptor tyrosine kinase inhibitor erlotinib and phase III clinical trials with the EGFRvIII vaccine rindopepimut have failed to robustly increase patient overall survival, highlighting the immense plasticity of GBM cell populations, which dampens treatment efficacies and hinders tumor management (Raizer et al., 2010; Raizer et al.,

2016; Michael Weller et al., 2017).

Mitochondria, in addition to regulating cellular energy conservation, also serve as signaling organelles. Mitochondria participate in retrograde signaling to the nucleus to induce changes in nuclear gene expression in response to intracellular metabolic perturbations in order to maintain metabolic homeostasis (Eisenberg-Bord & Schuldiner,

2017). Several proteins have been demonstrated to participate in inter-organelle signaling between mitochondria and the nucleus, including p53, fumarase, the pyruvate dehydrogenase complex, and coiled-coil-helix-coiled-coil-helix domain-containing protein

2 (CHCHD2) (Aras et al., 2015; Sutendra et al., 2014; Yogev et al., 2011; Zhuang et al.,

2013).

The CHCHD2 gene is located proximal to EGFR on chromosome 7; as such,

CHCHD2 and EGFR are frequently co-amplified in non-small cell lung carcinoma

(NSCLC) (Wei et al., 2015). CHCHD2 belongs to a family of 9 evolutionarily conserved small mitochondrial proteins, all containing at least one CHCH domain (Modjtahedi et al.,

2016; Z. D. Zhou et al., 2017). The CHCH domain, characterized by two CX9C motifs (two cysteines separated by 9 amino acids), is necessary for simultaneous oxidative folding and protein import into the mitochondrial intermembrane space (IMS), where CHCHD2 canonically functions in mediating mitochondrial respiration (Aras et al., 2015; Meng et al., 2017; Modjtahedi et al., 2016). In addition, CHCHD2 has been demonstrated in a variety of biological contexts to regulate other cellular functions, including cell migration and regulation of apoptosis (Y. Liu et al., 2015; Seo et al., 2010). The objective of this study was to characterize the functional capacity of CHCHD2 in GBM cells expressing

EGFRvIII. Our results indicate that CHCHD2 participates in a number of GBM cell functions representing disease hallmarks and may inspire therapeutic strategies which undermine mitochondrial biology to potentially improve GBM tumor management.

4.2. METHODS

CHCHD2 gene amplification and mRNA expression analysis

Analysis of CHCHD2 gene amplification patterns across GBM tumors was performed on

The Cancer Genome Atlas (TCGA) Provisional GBM database using cBioPortal

(www.cbioportal.org) (Cerami et al., 2012; Gao et al., 2013). Analysis was limited to tumor samples with available copy number alteration (CNA) data (n = 577). Analysis of CHCHD2 mRNA expression levels (HG-U133A Array) was compared among GBM tumors using publicly available data via Gliovis (http://gliovis.bioinfo.cnio.es/) (Bowman, Wang, Carro,

Verhaak, & Squatrito, 2017). Tumors were stratified by variables including tumor grade, tumor vs non-tumor tissue, GBM subtype, G-CIMP status, IDH1 mutational status, and

MGMT mutational status.

Cell culture

Human U87 GBM cells transduced to stably express the constitutively active EGFRvIII mutant (U87vIII) were generously provided by Dr. Nathan Price (Institute for Systems

Biology, Seattle, WA). Cells were cultured in Dulbecco’s modified Eagle’s medium

(DMEM) containing 1 mM sodium pyruvate, 15 mM HEPES, non-essential amino acids

(NEAA), 10% fetal bovine serum, and 1% penicillin/streptomycin. For hypoxic experiments, media was pre-equilibrated overnight at designated hypoxic oxygen conditions in order to account for time required for oxygen-saturated culture media to equilibrate with the gas atmosphere (Huttemann et al., 2007). Cells were then incubated in a BiospherixTM hypoxic incubator (BiospherixTM, Parish, NY). All cells were maintained at 37 °C and 5% CO2.

Generation of CHCHD2 knockout cells

U87vIII CHCHD2 KO cells were derived using CRISPR-Cas9 genome engineering following published protocols (Ran et al., 2013). Suitable target sites within exons of the coding sequence for CHCHD2 were identified using the online WU-CRISPR design tool

(Wong, Liu, & Wang, 2015). Potential guide RNA (gRNA) oligonucleotides were obtained from Integrated DNA Technologies (IDT, Coralville, IA). Each gRNA sequence was 20 nucleotides in length and directly upstream of the protospacer adjacent motif (PAM) 5’-

NGG-3’. Three separate gRNA expression constructs were generated by cloning phosphorylated and annealed gRNA oligos into the BbsI (NEB, Ipswich, MA) site of the pSpCas9(BB)-2A-Puro expression vector (Addgene, Watertown, MA) for co-expression of each sgRNA with the Cas9 endonuclease. The integrity of constructs was confirmed by plasmid sequencing (University of Illinois Urbana-Champaign Roy J. Carver

Biotechnology Center).

U87vIII cells were transfected with Cas9-CHCHD2gRNA expression constructs using

Lipofectamine 2000 Reagent according to the manufacturer’s protocol (ThermoFisher,

Waltham, MA). Stably transfected cells were selected with puromycin (10 µg/mL).

Successful CHCHD2 protein knockout was confirmed by western blot as described below.

Assessment of genomic mutational status was conducted via nested PCR of the region containing the induced double-strand break using two primer sets. The PCR product was cloned into the pCR2.1-TOPO vector (TOPO TA Cloning Kit, ThermoFisher, Waltham,

MA) and confirmed by sequencing (UIUC Roy J. Carver Biotechnology Center).

Western blot

Total protein levels of CHCHD2 (1:500, NBP1-94106, Novus Biologicals, Centennial,

CO), the glutamate-cystine antiporter xCT (1:500, ab37185, Abcam, Cambridge, UK),

GPx-1/2 (1:100, sc-133160, Santa Cruz Biotechnology, Dallas, TX), GPx-4 (1:100, NBP2-

75511, Novus Biologicals, Centennial, CO), and matrix metalloproteinase 2 (MMP-2)

(1:500, 10373-2-AP, Proteintech Group, Rosemont, IL) were compared in U87vIII

CHCHD2 WT and KO cells using western blot. β-actin (1:1000, #4967, Cell Signaling

Technology, Danvers, MA) was used as loading control. Total protein concentrations were determined using the Pierce BCA assay (Thermo Fisher, Waltham, MA). Protein lysates were mixed 1:1 with 2X Laemmli Sample Buffer (Bio-Rad, Hercules, CA) (5% β- mercaptoethanol) and heated at 95 °C for 5-10 min. Denatured lysates were loaded into

4-20% Mini-Protean ® TGXTM electrophoresis gels (Bio-Rad), and SDS-PAGE was run at 150 V for 1-1.5 h. Proteins were transferred to a nitrocellulose membrane (Amersham,

Little Chalfont, UK) at 300 mA for 2 h at 4⁰ C. Membranes were then blocked with either

5% BSA or 5% non-fat dry milk for 1 h at RT, then incubated in primary antibody at designated concentrations overnight at 4⁰ C. Membranes were washed in TBS-T for 5 min 3x, then incubated in HRP-linked goat anti-rabbit secondary antibody (1:2500, #7074,

Cell Signaling Technology, Danvers, MA) at RT for 1.5 h. Following TBS-T washes (3x 5 min each), membranes were imaged using the SuperSignalTM West Femto Maximum

Sensitivity Substrate (Thermo Fisher, Waltham, MA) in an ImageQuant LAS 4010 (GE

Life Sciences, Pittsburgh, PA). Band densitometry analysis was performed using ImageJ.

Measurement of oxygen consumption rate

Mitochondrial respiration of U87vIII CHCHD2 WT and KO cells was compared using the

Seahorse XFp Extracellular Flux Analyzer. Cells were seeded at 1 × 104 cells/well five hours prior to conducting the mitochondrial stress test, according to the manufacturer’s protocol (Agilent, Santa Clara, CA). Serial applications of oligomycin (ATP synthase inhibitor, 1 µM), FCCP (protonophore, 0.5 µM), and rotenone and antimycin A (respiratory complex I and III inhibitor, respectively, 0.5 µM) over time enabled calculation of various parameters of mitochondrial respiration in both cell lines.

Measurement of compartmentalized glutathione redox poise

U87vIII CHCHD2 WT and KO cells were transfected with genetically encoded, fluorescent redox biosensors targeted to the cytosol (cyto-Grx1-roGFP2) or mitochondrial matrix

(mito-Grx1-roGFP2), previously described by our lab (Kolossov et al., 2015). Cells expressing cytosolic or mitochondrial Grx1-roGFP2 were seeded at equal densities in standard culture medium without phenol red in µ-Slide eight-well ibiTreat microscopy chambers (Ibidi, Munich, Germany). Time-lapse images were collected with a fluorescence-enabled inverted microscope (Axiovert 200 M, Carl Zeiss, Feldbach,

Switzerland). Dual-excitation imaging of live cells used 395 and 494 nm excitation cubes, and an emission filter at 527 nm was used for both cubes. Exposure times were set to

100-200 ms, and images were taken every 15 s. To assess the effect of GSH synthesis inhibition on GSH:GSSG status, cells were pre-treated with buthionine sulfoximine (BSO,

100 µM) for 24, 48, and 72 h before time-lapse image acquisition. Acquired images were processed with Zeiss Axiovision SE64 Rel6.8 software, via manual selection of three to

50 five individual cells to obtain multiple regions of interest in each time lapse. The means of emission intensities at 527 nm were exported to Excel files and corrected by background subtraction.

Cell proliferation

Proliferation of U87vIII CHCHD2 WT and KO cells was compared over 72 h in normoxic culture conditions (20% O2) and hypoxia (1% O2) using the sulforhodamine B assay according to published protocols (Vichai & Kirtikara, 2006). Briefly, cells were seeded at equal densities in a 96-well plate, followed by fixation with 10% trichloroacetic acid for 1 h at 4⁰ C at 0 h and 72 h timepoints. After washing 4x with water and air-drying at RT,

0.057% sulforhodamine B (SRB) solution (wt/vol in 1% acetic acid) was applied to each well and incubated for 30 min at RT, followed by washing 4x in 1% acetic acid. After drying, bound SRB was solubilized in 10 mM Tris base solution (pH 10.5), and plates were shaken for 30 min at RT. The optical density (OD) of each well was measured at

510 using a BioTek SynergyTM HT microplate reader (BioTek, Winooski, VT). OD values at 72 h were corrected by 0 h OD subtraction to account for possible variations in initial seeding densities.

Hydrogel preparation and measurement of cell invasion

The invasive behavior of U87vIII CHCHD2 WT and KO cells was examined in normoxic culture conditions (20% O2) and hypoxia (1% O2). Invasion was quantified within methacrylamide-gelatin (GelMA) hydrogels via a bead invasion assay described previously (J.-W. E. Chen et al., 2018; J. E. Chen et al., 2018; J. E. Chen, Pedron, &

Harley, 2017). Hydrogels used in this study were made using 5% wt GelMA, ~53% degree of methacrylamide functionalization determined via H1-NMR (data not shown), and photopolymerized under UV light (AccuCure LED 365 nm, 7.1 mW cm-2 for 30 s) in the presence of a lithium acylphophinate photoinitiator. The compressive modulus of 5% wt

GelMA hydrogels was measured using a Instron 5943 mechanical tester with the Young’s modulus obtained from the linear region of the stress-strain curve (0-10% strain) (J.-W.

E. Chen et al., 2018).

To examine invasive behavior, U87vIII CHCHD2 WT and KO cells were seeded onto collagen-coated dextran beads (~200 µm diameter, GE Life Sciences, Pittsburgh, PA) at a density of 2 × 106 cells per 5 × 103 beads in 5 mL DMEM. Cell-bead suspensions were lightly shaken for one min, every 30 min, for 5 h to facilitate cell adhesion to beads. Cell- coated beads were then encapsulated in pre-polymerized GelMA hydrogel solution, and bead-containing hydrogels were cultured in standard DMEM in either normoxic culture conditions (20% O2) or hypoxia (1% O2) for 7 days. Cell invasion distance was measured from the bead surface using ImageJ from images acquired via fluorescent microscopy after fixing and staining cells with DAPI (Invitrogen, Carlsbad, CA) (10 µg/mL in 1X PBS).

Cell invasion is reported as the mean invasion of all cells from the surface of the bead (J.

E. Chen et al., 2018).

Cell cytotoxicity assays

Sensitivity of U87vIII CHCHD2 WT and KO cells to a panel of cytotoxic agents was determined using the sulforhodamine B assay according to published protocols as described above (Vichai & Kirtikara, 2006). The cytotoxicity of sulfasalazine (SSZ, xCT

52 cystine-glutamate antiporter inhibitor), erlotinib (Erl, EGFR receptor tyrosine kinase inhibitor), temozolomide (TMZ, DNA alkylating agent), and Pac-1 (procaspase-3 activator) (Joshi et al., 2017) were assessed using dosages derived from the literature.

CHCHD2 protein levels in patient-derived cells cultured in GelMA hydrogels

Total protein levels of CHCHD2 were analyzed in a panel of GBM patient-derived cells

(PDCs) with disparate EGFR/PTEN status, MGMT methylation state, molecular subtype, invasive characteristics in mouse orthotopic xenografts (0: low, 7: hi), and sensitivity to erlotinib (0: not sensitive, 100: sensitive) (Figure 4.7A). PDCs were cultured in GelMA hydrogels without hyaluronic acid (HA) (7% wt GelMA) or with HA (6% wt GelMA, 1% wt

HA).

Statistical analysis

Differences among means were tested for using Student’s t-test or one-way ANOVA, followed by post-hoc Student-Newman-Keuls analysis where appropriate. Statistical significance was set at p < 0.05. Variance is reported as standard error of the mean. Odds ratios for CHCHD2 co-amplification with EGFR and log-rank tests for overall and progression-free survival were conducted in cBioPortal (www.cbioportal.org). Tukey’s

Honest Sigificant Difference or pairwise t-tests were conducted to compare mRNA expression levels in GlioVis (http://gliovis.bioinfo.cnio.es/).

4.3. RESULTS

CHCHD2 is amplified in a subset of GBM tumors and is associated with decreased patient survival

Figure 4.1. cBioPortal analysis of CHCHD2 gene amplification patterns across GBM tumors. (A) Oncoprint from cBioPortal (www.cbioportal.org) representing tumors with amplification (red), deep deletion (blue), or no alteration (grey) of query genes. Percentages represent percentage of samples analyzed (tumor samples with available copy number alteration data, n = 577) with alteration in given gene. COA4 encodes CHCHD8. (B) Overall survival of patients with (red) and without (blue) CHCHD2 amplification. (C) Progression-free survival of patients with (red) and without (blue) CHCHD2 amplification. * p < 0.05, log-rank test conducted in cBioPortal. Analysis of CHCHD2 amplification patterns across GBM tumors was performed on The

Cancer Genome Atlas (TCGA) Provisional GBM database using cBioPortal

(www.cbioportal.org) (Cerami et al., 2012; Gao et al., 2013). Analysis was performed on tumor samples with available copy number alteration (CNA) data (n = 577). Amplification of CHCHD2 was observed in 9% of GBM tumors (Figure 4.1A). Of tumors with EGFR amplification, the most common genetic aberration across GBM tumors, CHCHD2 was co-amplified in 20% of cases, with a significant tendency to co-occur (log2 odds ratio > 3, p < 0.001) (Figure 4.1A). Of the 9 proteins in the CHCH domain-containing protein family, only CHCHD2 was amplified at an appreciable frequency, with the next most frequently

54 amplified CHCH protein being CHCHD3 (1.2%) (Figure 4.1A). Additionally, patients with

CHCHD2 amplified tumors had decreased overall survival (OS, 12.48 mo vs 14.45 mo, log-rank p = 0.0223) (Figure 4.1B) and progression-free survival (PFS, 4.86 mo vs 7.82 mo, log-rank p = 0.0420) (Figure 4.1C). While this effect could potentially be explained by the propensity for EGFR to co-amplify with CHCHD2, amplification of EGFR was not solely associated with decreased patient survival (not shown). These data indicate that amplification of the CHCHD2 gene occurs in a subset of GBM tumors, is associated with decreased OS and PFS, and is nearly always accompanied by EGFR amplification.

Figure 4.2. GlioVis analysis of CHCHD2 mRNA expression patterns across GBM tumors. CHCHD2 mRNA expression determined using Human Genome U133A Microarray was compared across (A) glioma grade (II, III, and IV), as well as GBM (B) IDH1 mutational status, (C) non-tumor vs tumor tissue, (D) GBM subtype (classical, mesenchymal, and proneural), (E) G-CIMP status, and (F) MGMT methylation status. Data were analyzed using Tukey’s Honest significance difference or pairwise t-test directly in GlioVis. * p < 0.001.

Expression of CHCHD2 across brain tumors was further explored using the GlioVis data portal (http://gliovis.bioinfo.cnio.es/) (Bowman et al., 2017). CHCHD2 mRNA

55 expression was observed to be increased in grade IV gliomas (GBM) relative to grade II and III (Figure 4.2A), as well as in IHD1-wt tumors compared to IDH1-mutant (Figure

4.2B). CHCHD2 expression was also increased in GBM tumor compared to non-tumor tissue (Figure 4.2C). Additionally, CHCHD2 expression was increased in classical subtype tumors (Figure 4.2D), which are defined by chromosome 10 loss paired with chromosome 7 amplification and focused predilection for EGFR amplification. Tumors exhibiting genome-wide promoter hypermethylation, i.e. glioma-CpG island methylator phenotype (G-CIMP), displayed decreased CHCHD2 mRNA expression (Figure 4.2E).

No differences in CHCHD2 expression were observed between MGMT methylated vs non-methylated tumors (Figure 4.2F).

Knockout of CHCHD2 alters mitochondrial respiration in U87vIII cells

U87vIII CHCHD2 KO cells were derived using CRISPR-Cas9 following published protocols, and protein knockout validated using western blot

(Figure 4.3A) (Ran et al., 2013; Wong et al., 2015). To determine if CHCHD2

KO impacted mitochondrial respiration Figure 4.3. Mitochondrial respiration in U87vIII CHCHD2 in GBM cells, a mitochondrial stress WT and KO cells. (A) Western blot of U87vIII and U87vIII CHCHD2 KO cells. (B) Oxygen consumption rate (OCR) traces of U87vIII CHCHD2 WT and KO cells during test was conducted on U87vIII mitochondrial stress test. (C) Basal OCR, (D) maximal respiration, (E) spare respiratory capacity, and (F) OCR coupled to ATP production in U87vIII CHCHD2 WT and KO CHCHD2 WT and CHCHD2 KO cells cells. Data presented as mean ± SE, n = 2 * p < 0.05. using a Seahorse XFp Extracellular Flux Analyzer (Figure 4.3B) (Divakaruni, Paradyse,

Ferrick, Murphy, & Jastroch, 2014). CHCHD2 KO cells displayed decreased basal and maximal oxygen consumption rate (OCR) (Figure 4.3C-D). Spare respiratory capacity was also significantly lower in CHCHD2 KO cells (Figure 4.3E), indicating a decreased ability of U87vIII CHCHD2 KO cells to respond to increased energy demand. Additionally, the amount of oxygen consumed coupled to ATP production by ATP synthase was significantly lower in CHCHD2 KO cells (Figure 4.3F). Collectively, these data indicate that CHCHD2 is required for efficient mitochondrial respiration in U87 GBM cells expressing EGFRvIII.

Knockout of CHCHD2 leads to a more reduced glutathione redox potential in the mitochondrial matrix

It was further hypothesized that perturbed mitochondrial respiration and deficient electron transport chain function would result in a more oxidized intracellular redox environment.

Glutathione (GSH), an enzymatically produced tripeptide of cysteine, glycine, and glutamate, is the main intracellular redox buffer, which, along with thioredoxins and glutaredoxins, maintains thiol redox status (Hanschmann, Godoy, Berndt, Hudemann, &

Lillig, 2013). The glutathione redox couple (reduced and oxidized glutathione, GSH and

GSSG respectively), along with glutathione peroxidase (GPx) and glutathione reductase

(GR), comprises the glutathione system, which maintains thiol redox homeostasis and functions in antioxidant defense (Estrela, Ortega, & Obrador, 2006). The balance of reduced to oxidized glutathione (GSH:GSSG) thus represents intracellular redox status, and can be interrogated within live cells using genetically encoded, fluorescent redox biosensors (Grx1-roGFP2) (Figure 4.4A). Additionally, such probes can be targeted to

57 various subcellular compartments, including cytosol (cyto-Grx1-roGFP2) or mitochondrial matrix (mito-Grx1-roGFP2) to measure compartmentalized GSH redox status in live cells in real time via ratiometric fluorescence intensity measurements (Figure 4.4B) (Kolossov et al., 2015).

The percent of oxidized mito-Grx1-roGFP2 in U87vIII CHCHD2 KO cells was decreased by 15.8% compared to that measured in CHCHD2 WT cells, indicating that the pool of glutathione in the mitochondrial matrix of KO cells is more reduced than WT counterparts (Figure 4.4C). This effect was confined to the mitochondrial matrix, as the

Figure 4.4. Glutathione redox status in U87vIII CHCHD2 WT and KO cells. (A) Schematic of the molecular mechanism of the Grx1-roGFP2 sensor and redox response of the compartmentalized probe to exogenous oxidant • − and reductant. Superoxide ( O2 ) is rapidly converted by superoxide dismutase (SOD) into H2O2, which is then reduced by glutathione peroxidase (GPx) to water. Glutaredoxin (Grx) fused to roGFP2 efficiently and rapidly equilibrates the probe with alterations in the local GSH:GSSG ratio. Additionally, thiol-disulfide equilibration is reversible, as GSH reductase (GR) catalyzes reduction of GSSG to GSH. (B) Representative fluorescence images demonstrate the sensor targeted to mitochondria of U87vIII CHCHD2 WT (left) and KO cells (right). (B’) Corresponding time-lapse responses of the 395/494 nm ratio to treatment with 1 mM diamide (vertical solid line) to the fully oxidized state and 10 mM DTT (vertical dashed line) to the fully reduced state. Each trace designates a separate cell. Arrows represent basal oxidation level of probe. (C) Percent oxidized mito-Grx1-roGFP2 and (D) cyto-Grx1-roGFP2 in U87vIII CHCHD2 WT and KO cells. (E) Percent oxidized mito-Grx1-roGFP2 in U87vIII CHCHD2 WT and KO cells at baseline and after treatment with GSH synthesis inhibitor BSO for 24, 48, and 72 h. (F) Western blot of xCT, GPx-1/2, and GPx-4 in U87vIII CHCHD2 WT and KO cells. Data are presented as mean ± SE, n = 3 *, differing superscript: p < 0.05.

58 cytosolic glutathione pool was not affected by CHCHD2 KO (Figure 4.4D). Treatment with buthionine sulfoximine (BSO), an inhibitor of glutamate-cysteine ligase (GCL, the rate-limiting enzyme in GSH synthesis), led to similar oxidation of the glutathione pool over time in the mitochondrial matrix of both U87vIII CHCHD2 WT and KO cells (Figure

4.4E). Additionally, levels of components of the glutathione system, including the glutamate-cystine antiporter xCT, GPx-1/2, and GPx-4, were all unaltered in CHCHD2

KO cells (Figure 4.4F). These data demonstrate a role for CHCHD2 in mediating glutathione redox balance specifically in the mitochondrial matrix, albeit through a mechanism not involving GSH synthesis, cystine import through xCT, or reduced GSH flux through glutathione peroxidases 1, 2, or 4.

U87vIII cell growth and invasion are negatively impacted by CHCHD2 KO in both normoxia and hypoxia

The observed deficiencies in mitochondrial respiration led to the hypothesis that U87vIII

GBM cell growth would be negatively impacted by CHCHD2 KO. To test this hypothesis,

U87vIII CHCHD2 WT and KO cells were incubated in either standard oxygen culture conditions (normoxia, 20% O2) or pathophysiologically relevant hypoxia (1% O2). Utilizing the sulforhodamine B (SRB) colorimetric assay (Vichai & Kirtikara, 2006) to assess cell growth over 72 h, CHCHD2 WT cell growth was observed to be significantly increased in hypoxia (Figure 4.5A). CHCHD2 KO cell growth was decreased compared to normoxic control (Figure 4.5A). Additionally, the growth-inducing effect of hypoxia on U87vIII cells was abrogated upon CHCHD2 KO (Figure 4.5A). These results demonstrate that

CHCHD2 is involved in not only mediating U87vIII cell growth in normoxia, but also plays a role in the initial increased cell proliferation observed in hypoxic cells.

Rapid, diffuse invasion of GBM cells into tumor margins and into surrounding brain parenchyma represents a major obstacle impeding effective tumor treatment. To determine whether CHCHD2 knockout impacted GBM cell invasion, a bead invasion assay in methacrylated gelatin (GelMA) hydrogels (Young’s modulus 2.9 ± 0.45 kPa) was used as a three-dimensional culture platform to model biophysical aspects of the brain parenchyma and GBM cell invasion (J.-W. E. Chen et al., 2018; J. E. Chen et al., 2018;

J. E. Chen et al., 2017). This technique enables monitoring of cell invasion from a defined starting point in a spatiotemporal manner in normal as well as hypoxic three-dimensional culture conditions (Figure 4.5B). Hypoxia (1% O2) stimulated U87vIII cell invasion over

Figure 4.5. Growth and invasion of U87vIII CHCHD2 WT and KO cells in normoxia and hypoxia. (A) Cell growth of U87vIII CHCHD2 WT and KO cells over 72 h in standard oxygen conditions (21% O2) and hypoxia (1% O2) determined using the SRB assay. (B) Representative images of U87vIII CHCHD2 WT and KO cells stained for DAPI after incubation in 21% or 1% O2 for 7 days during bead invasion assay. Associated ImageJ-derived images for average invasion distance analysis are displayed below corresponding fluorescence images. (C) Quantification of U87vIII CHCHD2 WT and KO invasion as determined by the bead invasion assay. (D) Western blot of pro-MMP- 2 levels in U87vIII CHCHD2 WT and KO cells and fold-change normalized to β-actin loading control. Data are presented as mean ± SE, n = 3. Differing superscript, *: p < 0.05.

60 long-term culture (7 d) (Figure 4.5C). In contrast, U87vIII CHCHD2 KO cells displayed minimal cell invasion in both normoxia and hypoxia, with the hypoxia-induced invasion observed in CHCHD2 WT cells abrogated (Figure 4.5C). Furthermore, CHCHD2 KO cells displayed decreased levels of pro-MMP-2 (matrix metalloproteinase 2) (Figure 4.5D), a key protein involved in the breakdown of extracellular matrix. These data demonstrate a role for CHCHD2 in mediating U87vIII cell invasion, particularly in response to hypoxia.

CHCHD2 knockout increases U87vIII sensitivity to a variety of cytotoxic drugs

To determine the effect of CHCHD2 KO on cellular resistance to various drugs, cells were treated with increasing concentrations of a panel of agents, and cytotoxicity was assessed using the SRB assay (Vichai & Kirtikara, Figure 4.6. Sensitivity of U87vIII CHCHD2 WT and KO cells to various 2006). Included in this cytotoxic agents. Sensitivity of U87vIII and CHCHD2 KO cells to (A) sulfasalazine, (B) erlotinib, (C) temozolomide, and (D) Pac-1 determined using the SRB assay. Data are presented as percent of untreated control, mean ± panel were: SE, n = 3. Differing superscript: p < 0.05. temozolomide (TMZ), a DNA-alkylating agent and the standard-of-care chemotherapy administered to patients with GBM (Stupp et al., 2005); erlotinib (Erl), a receptor tyrosine kinase inhibitor that inhibits EGFR and EGFRvIII tyrosine kinase activity (Raizer et al.,

2010); sulfasalazine (SSZ), an inhibitor of the cell membrane xCT antiporter, which couples the export of the amino acid glutamate with the import of cystine, thus depriving cells of the rate-limiting substrate to synthesize reduced GSH (L. Chen et al., 2015;

Huberfeld & Vecht, 2016); and Pac-1, a novel activator of apoptosis which acts on procaspase-3 (Joshi et al., 2017). Results, shown in Figure 4.6, demonstrate that

CHCHD2 KO cells were more susceptible to treatment with TMZ, Erl, and most significantly, SSZ. However, CHCHD2 KO had no effect on cellular sensitivity to treatment with Pac-1 (Figure 4.6D), consistent with previous work demonstrating that CHCHD2 regulates apoptosis upstream of procaspase-3 activity in the apoptotic cascade (Y. Liu et al., 2015). These results demonstrate a role for CHCHD2 in mediating cell sensitivity to various drugs relevant to GBM treatment, highlighting CHCHD2 as a promising avenue for future investigation.

CHCHD2 protein levels vary across GBM patient-derived cell samples

Figure 4.7. CHCHD2 protein levels in GBM patient-derived cells. (A) Patient-derived xenograft (PDX) subset analyzed in the current study, with descriptors including EGFR/PTEN status, patient status, invasion in mouse orthotopic xenografts, erlotinib sensitivity, MGMT methylation status, and subtype. (B) CHCHD2 total protein levels in patient-derived cell (PDC) samples from each PDX cultured in GelMA hydrogels without hyaluronic acid (HA) (7% wt GelMA). (C) CHCHD2 total protein levels in PDC samples from each PDX cultured in GelMA hydrogels containing matrix-immobilized HA (6% wt GelMA, 1% wt HA). (D) Effect of matrix-immobilized HA on CHCHD2 total protein levels in each PDC. Data are presented as mean ± SE, n = 3 samples. Differing superscript: p < 0.05. * denotes significant effect of HA, p < 0.05 Total protein levels of CHCHD2 were analyzed in a panel of GBM patient-derived cells

(PDCs) cultured in GelMA hydrogels with extracellular HA (6% wt gelatin, 1% wt HA) or without HA (7% wt gelatin). PDCs were characterized by disparate EGFR/PTEN status,

MGMT methylation state, molecular subtype, invasive characteristics in mouse orthotopic xenografts (0: low, 7: hi), and sensitivity to erlotinib (0: not sensitive, 100: sensitive)

(Figure 4.7A). Included in this panel were: GBM10 (EGFR-/PTEN-), GBM44 (EGFR-

/PTEN-), GBM12 (EGFR+/PTENwt), GBM39 (EGFRvIII/PTENwt), and GBM6

(EGFRvIII/PTENwt). When PDCs were cultured in GelMA hydrogels without HA, GBM6 expressed the greatest amount of CHCHD2, followed by GBM44 and GBM12, and

GBM10 and GBM39 (Figure 4.7B). When HA was included in the hydrogel matrix, GBM6

63 and GBM12 expressed the greatest amount of CHCHD2 (Figure 4.7C). GBM39 responded most strikingly to the presence of HA in the extracellular matrix (Figure 4.7D).

These data demonstrate heterogeneity in CHCHD2 expression among tumor samples, as well as in response to the presence of extracellular HA.

4.4 DISCUSSION

The current work investigated the functional capacity of CHCHD2, a dually-localized protein implicated in mitochondria-to-nucleus retrograde signaling (Aras et al., 2015), in the context of GBM. The challenges associated with achieving robust survival benefits in patients with GBM has prompted the need for a deeper understanding of intracellular events mediating cellular plasticity, tumor malignancy, and recurrence. CHCHD2 presented as a promising avenue to pursue, given: 1) its proximity to and frequency of co-amplification with EGFR in NSCLC and GBM (Figure 4.1A) (Wei et al., 2015), both of which have recently been characterized as relatively oxidative versus fermentative tumors

(Courtney et al., 2018); 2) its mRNA expression patterns across glioma grade and tumor vs non-tumor tissue (Figure 4.2A, C); 3) its reported oxygen-sensitive transcription factor activity (Aras et al., 2013); and 4) its pleiotropic roles mediating cellular functions reminiscent of cancer hallmarks, including cell proliferation, migration and invasion, and inhibition of apoptosis (Aras et al., 2015; Y. Liu et al., 2015; Seo et al., 2010). Here, we demonstrate that CHCHD2 is involved in mediating cell growth and invasion in response to hypoxia, as well as therapeutic sensitivity, in GBM cells expressing mutant EGFRvIII.

Knockout of CHCHD2 in U87vIII GBM cells resulted in decreased baseline respiration as well as spare respiratory capacity (Figure 4.3), an effect corroborated by multiple

64 studies which have characterized CHCHD2 as a canonical regulator of mitochondrial respiration (Aras et al., 2015; Baughman et al., 2009; Meng et al., 2017; Wei et al., 2015).

While metabolic reprogramming towards increased glycolytic flux to favor increased cell proliferation is an emerging and increasingly recognized hallmark of cancer (Hanahan &

Weinberg, 2011), functional mitochondria remain essential in maintaining malignant cell bioenergetics in particular tumors, including those of lung and brain (Courtney et al., 2018;

Marin-Valencia et al., 2012; Pavlova & Thompson, 2016). Notably, mitochondrial spare respiratory capacity has been positively associated with glioma stem cell resistance to radiotherapy (Vlashi et al., 2011). Thus, the decreased spare respiratory capacity measured in U87vIII CHCHD2 KO cells (Figure 4.3E) may partially contribute to their increased sensitivity to treatment with TMZ, Erl, and SSZ (Figure 4.6A-C).

Additionally, CHCHD2 has been demonstrated by others to mediate mitochondrial outer membrane permeabilization (MOMP) (Y. Liu et al., 2015), the “point of no return” during the intrinsic pathway of apoptosis. Indeed, U87vIII CHCHD2 KO cells exhibited increased sensitivity to TMZ, Erl, and SSZ. The observation that no increase in cytotoxicity was seen in CHCHD2 KO cells treated with Pac-1, an activator of procaspase-

3 which acts downstream of MOMP in the apoptotic cascade (Joshi et al., 2017), further corroborates the demonstrated role of CHCHD2 early in apoptosis regulation.

Notably, as evidenced by the use of mito-Grx1-roGFP2, the mitochondrial GSH pool was more reduced in response to CHCHD2 KO compared to WT, an effect independent of GSH biosynthesis (Figure 4.4E). It should be noted that roGFP2 itself is not directly oxidized by ROS species, and rather equilibrates with the local GSH redox potential, which is influenced by the GSH:GSSG ratio, and in turn is influenced by activity of various

ROS-scavenging enzymes that use GSH as a cofactor (Figure 4.4A). In HEK293 cells,

CHCHD2 knockdown was accompanied by decreased expression of superoxide dismutase 2 (SOD2/Mn-SOD) and loss of glutathione peroxidase (GPx) expression (Aras et al., 2015). However, we did not observe changes in protein levels of GPx-1/2, GPx-4, nor the glutamate-cystine antiporter xCT (Figure 4.4F).

In the current work, incubation in hypoxia increased U87vIII CHCHD2 WT cell proliferation over 3 days, and increased invasion over 7 days (Figure 4.5A-C). Using the

GelMA hydrogel platforms described here, previous work from our demonstrated that hypoxic U87vIII cells exhibited increased cell proliferation until day 5, at which point cell proliferation stalled, while invasion continued to increase up to day 7 (J. E. Chen et al.,

2018). Notably, CHCHD2 KO abrogated the increased cell growth and invasion in hypoxia exhibited by WT cells (Figure 4.5A-C). The observed decrease in U87vIII CHCHD2 KO cell growth is not likely due to an increase in apoptosis, as evidence has indicated that shRNA-mediated CHCHD2 knockdown does not alter cellular levels of poly (ADP-ribose) polymerase (PARP) (Aras et al., 2015), the cleavage of which by caspases is a characteristic of apoptosis (Chaitanya, Steven, & Babu, 2010). Deficiencies in mitochondrial respiration and ATP production are likely responsible for hampering cell proliferative capacity, and may also be implicated in the repression of cell invasion, as migration and invasion throughout parenchyma is an energy-expensive process, particularly when moving through dense extracellular matrices (Zanotelli et al., 2018). The observed decrease in pro-MMP-2 expression in CHCHD2 KO cells provides further explanation for their decrease in invasive capacity (Figure 4.5D). Other work has

66 highlighted CHCHD2 as an activator of NIH3T3 fibroblast migration via the AKT-

RhoA/ROCK-JNK cascade (Seo et al., 2010).

Analysis of CHCHD2 gene amplification, mRNA expression, and protein levels in a subset of clinical patient-derived samples further revealed intriguing patterns indicating an involvement of CHCHD2 in supporting GBM malignant characteristics. Of note,

CHCHD2 mRNA expression was increased in GBM vs grade II and III gliomas, as well as in GBM tumor compared to non-tumor tissue (Figure 4.2A, C). Additionally, among GBM subtypes, CHCHD2 expression was greatest in the classical subtype (Figure 4.2D), an observation consistent with GBM6 (classical subtype) exhibiting greater levels than all other PDCs analyzed in this study (which were classified as mesenchymal subtype)

(Figure 4.7A, B). Notably, GBM6 also expresses mutant EGFRvIII, is relatively resistant to erlotinib, and is the most invasive PDC sample analyzed (Figure 4.7A, B).

Functional outcomes investigated here relate primarily to mitochondrial functions of

CHCHD2. As a protein implicated in mitonuclear communication in response to hypoxia, the nuclear functions of CHCHD2 in GBM cells and the panel of nuclear genes it regulates are of equal importance and areas of future work. Evidence in the literature indicates that

CHCHD2 acts as a nuclear transcription factor in hypoxia, maximally active at 4% O2, to increase gene expression of cytochrome c oxidase subunit 4 isoform 2 (COX4I2) and itself (Aras et al., 2013), an effect we did not observe in U87vIII cells, as COX4I2 is a tissue-specific isoform highly expressed in lung and trachea (data not shown). It is tempting to speculate that the abrogated proliferative and invasive responses of CHCHD2

KO cells are due to a disturbance of a CHCHD2-mediated mitonuclear communication network. The nuclear function of CHCHD2 was demonstrated to act in concert with other

67 transcription factors, namely recombining binding protein suppressor of hairless (RBPJ)

(Aras et al., 2013), the main downstream effector protein of the Notch signaling pathway, which has been implicated in the maintenance of glioma stem cell maintenance and viability (Q. Xie et al., 2016). Future work should seek to identify the complement of genes regulated by CHCHD2 in both normal oxygen conditions as well as hypoxia to provide a deeper understanding of CHCHD2 function in GBM.

4.5. CONCLUSIONS

This study is the first to describe the functional relevance of CHCHD2 in GBM. These data indicate that CHCHD2 is essential for mitochondrial respiration and maintenance of mitochondrial GSH status in U87vIII cells. Additionally, these studies demonstrate a critical role for CHCHD2 in mediating cell growth and invasion in normoxia and hypoxia and resistance to various cytotoxic agents in U87 GBM cells expressing the mutant

EGFRvIII, underscoring CHCHD2 as a mediator of GBM cell malignant phenotype.

CHAPTER 5: DEFINING THE RELATIONSHIP BETWEEN HYPOXIA, REDOX

STATUS, AND SUBCELLULAR LOCALIZATION OF CHCHD2 IN GLIOBLASTOMA

CELLS

ABSTRACT

INTRODUCTION

Glioblastoma (GBM) is the most common and malignant form of primary brain tumor in adults and remains incurable. Inter-organelle communication between mitochondria and the nucleus represents a potential compensatory signaling pathway that may allow cells to sense and respond to external stress and fluctuations in intracellular conditions, and consequently contribute to therapeutic resistance and cellular plasticity in a hypoxic tumor microenvironment. Coiled-coil-helix-coiled-coil-helix domain-containing protein 2

(CHCHD2) is a protein that colocalizes to the mitochondrial inter-membrane space and the nucleus, the latter wherein it exhibits hypoxia-sensitive transcription factor capabilities

(Aras et al., 2013). The C-terminus of CHCHD2 contains redox-sensitive cysteines in twin

CX9C motifs that are amenable to thiol/disulfide formation, which in turn determines protein folding and import into mitochondria. The objective of this study was to characterize the effects of hypoxia and cellular redox status on CHCHD2 subcellular localization. The hypothesis was that reducing the C-terminal CHCH domain would induce mitochondrial export of CHCHD2.

METHODS

U87 and U87vIII GBM cells (the latter of which are transduced to stably express the constitutively active EGFRvIII mutant) were incubated for 48 h either in standard culture oxygen conditions (20% O2), 7% O2, 4% O2, or 1% O2. Total protein levels of CHCHD2

69 in response to decreasing oxygen tension were analyzed via western blot. Protein turnover of CHCHD2 was determined by incubation of cells in cycloheximide (CHX, 10

µg/mL) followed by total protein lysis at designated time points. Subcellular localization of

CHCHD2 was observed in U87 and U87vIII cells expressing the pDsRed2-Mito fluorescent mitochondrial marker incubated in 20% and 1% O2 for 48 h, buthionine sulfoximine (BSO) or N-acetylcysteine (NAC) for 48 h, or in serum-free media for 30 min.

Acute chemical oxidation and reduction of cells was achieved using hydrogen peroxide

(H2O2) and dithiothreitol (DTT), respectively, at designated concentrations and time courses. The ability of CHCHD2 to induce expression of itself in U87vIII cells was assessed by ectopic expression of 3X-FLAG-tagged CHCHD2, followed by western blot analysis of endogenous CHCHD2 band intensity.

RESULTS

Total CHCHD2 protein levels were not significantly altered by decreasing oxygen tensions in U87 cells. U87vIII cells exhibited constant levels of total CHCHD2 protein at 20%, 7%, and 4% O2; however, incubation in 1% O2 led to a significant decrease in total CHCHD2.

Half-life of CHCHD2 was determined to be approximately 3 h in both U87 and U87vIII cells. U87vIII cells constitutively displayed a greater amount of nuclear CHCHD2 compared to isogenic U87 cells at 20% O2. Metabolic stress induced by incubation in 1%

O2 for 48 h or serum-free media for 30 min both resulted in increased CHCHD2 detected in nuclei of U87 and U87vIII cells. Oxidative stress induced by the glutathione synthesis inhibitor BSO increased nuclear CHCHD2 in U87vIII cells, while the glutathione precursor

NAC did not affect nuclear CHCHD2 levels. Including H2O2 in serum-free media abrogated the amount of nuclear CHCHD2 detected in nuclei of U87 and U87vIII cells.

Reductive unfolding of CHCHD2 with the disulfide reductant DTT consistently resulted in the greatest and most drastic nuclear accumulation of CHCHD2 in both cell lines within

15-30 min. Ectopic expression of exogenous 3X-FLAG-tagged CHCHD2 in U87vIII cells did not result in increased expression of endogenous CHCHD2.

CONCLUSIONS

Collectively, these data demonstrate subcellular redistribution via translocation of

CHCHD2 from mitochondria to the nucleus in response to metabolic stress induced by hypoxia and serum deprivation. Furthermore, the data indicate that reductive unfolding of

CHCHD2 is responsible for mitochondrial export and nuclear translocation. However, the functional relevance of nuclear CHCHD2 remains to be determined, as introduction of exogenous 3x-FLAG-tagged CHCHD2 did not induce expression of endogenous

CHCHD2 in U87vIII cells.

5.1. INTRODUCTION

Mitochondria are critical for a variety of cellular functions, including regulation of the intrinsic pathway of apoptosis, calcium signaling, generation of heme and iron-sulfur clusters, carbohydrate and fatty acid metabolism, redox homeostasis, and generation of reactive oxygen species (ROS) as signaling molecules (Connolly et al., 2018; Guha &

Avadhani, 2013). Arguably, the most widely recognized function of mitochondria is the oxygen-dependent generation of energy in the form of ATP, driven by chemiosmosis of protons down their electrochemical gradient across the inner mitochondrial membrane through ATP synthase, which couples this movement with the phosphorylation of ADP.

This gradient of protons is generated by the actions of the electron transport chain (ETC),

71 comprised of Complex I (NADH:ubiquinone oxidoreductase/NADH dehydrogenase),

Complex II (succinate-ubiquinone oxidoreductase/succinate dehydrogenase), Complex

III (ubiquinol-cytochrome c oxidoreductase), and the oxygen consuming Complex IV

(cytochrome c oxidase).

In addition to regulating cellular energy conservation, mitochondria also serve as crucial signaling organelles. The vast majority of mitochondrial proteins are encoded by nuclear genes, necessitating the ability for mitochondria to communicate their status to the nucleus in response to metabolic perturbations. Indeed, only 13 of the 90 protein subunits that comprise the respiratory complexes that constitute the electron transport chain are encoded by mitochondrial DNA, while the majority of respiratory subunits and mitochondrial proteome are encoded by the nuclear genome (Baughman et al., 2009;

Guha & Avadhani, 2013) Disturbances in ATP and ROS production, damage to mitochondrial DNA, and aberrations in mitochondrial protein folding induce a retrograde signaling pathway by which mitochondria can communicate with the nucleus to induce changes in nuclear gene expression in order to maintain metabolic homeostasis and mitochondrial biogenesis (Eisenberg-Bord & Schuldiner, 2017). Several proteins have been demonstrated to participate in inter-organelle signaling between mitochondria and the nucleus, including p53, fumarase, the pyruvate dehydrogenase complex, and coiled- coil-helix-coiled-coil-helix domain-containing protein 2 (CHCHD2) (Aras et al., 2015;

Sutendra et al., 2014; Yogev et al., 2011; Zhuang et al., 2013).

CHCHD2 is a 16 kDa protein and member of the evolutionarily conserved, mitochondrial CHCH domain-containing protein family. The CHCHD2 gene is located on chromosome 7p11.2 and encodes the 151 amino acid CHCHD2 protein, characterized

72 by its C-terminal CHCH domain. This domain is defined by two cysteine-X9-cysteine motifs (twin CX9C), each consisting of two cysteines (Cys) separated by 9 amino acids.

Two disulfide bonds, formed between Cys114-Cys144 and Cys124-Cys134, stabilize the helix- turn-helix fold of the CHCH domain. These Cys residues are indispensable for proper protein folding and import of CHCHD2 into the mitochondrial inter-membrane space (IMS) via the CHCHD4-mediated disulfide relay system (Aras et al., 2015). Furthermore, the N- terminus of CHCHD2 harbors a putative mitochondrial localization sequence defined by amino acid residues 1-53 (Aras et al., 2015). Moreover, the amino acid sequence of

CHCHD2 is highly conserved among human, chimpanzee, mouse, rat, zebrafish, and other metazoan species, as well as yeast, suggesting a critical function across species

(Modjtahedi et al., 2016).

Previous studies implicate CHCHD2 as a protein involved in mitonuclear communication with the ability to act as a nuclear transcription factor in response to hypoxia, inducing expression of complex IV subunit 4 isoform 2 (COX4I2) and itself maximally at 4% O2 (Aras et al., 2013). Notably, hypoxia is a key component of the glioblastoma (GBM) tumor microenvironment, inducing intracellular signaling, canonically via hypoxia inducible factors (HIFs), which in turn promotes malignant characteristics such as cell invasion and therapeutic resistance (J. E. Chen et al., 2018; W. L. Chen,

Wang, Lin, Wu, & Hsieh, 2015; Monteiro et al., 2017). However, the subcellular localization, distribution, and dynamics of CHCHD2 in GBM cells in response to hypoxia has not been described. Additionally, a mechanism governing its mitochondrial export and subcellular redistribution remains elusive.

Mitochondria-to-nucleus retrograde signaling is a signaling axis that can potentially be undermined to combat progression and malignancy of cancers, including GBM, the plasticity of which has undermined attempts at successful therapies. The objective of this study was to interrogate the intracellular dynamics of CHCHD2 in response to metabolic stressors in U87 and U87vIII GBM cells. The hypothesis was that the cysteines defining the CHCH domain acted as redox-sensitive “switches,” governing CHCHD2 protein folding and subcellular localization in response to metabolic and/or redox stress. Hypoxia and serum deprivation were used to model metabolic stressors imposed by the tumor microenvironment. Treatment with the glutathione synthesis inhibitor buthionine sulfoximine (BSO), as well as with the glutathione precursor N-acetylcysteine (NAC), were used to model oxidative and reductive stress, respectively. Acute chemical oxidation and reduction with H2O2 and the disulfide reductant dithiothreitol (DTT), respectively, were used to investigate the relationship between CHCHD2 protein folding and subcellular localization.

5.2. METHODS

Cell culture

The human U87 parental GBM cell line, as well as U87 GBM cells transduced to stably express the constitutively active EGFRvIII mutant (U87vIII) were generously provided by

Dr. Nathan Price (Institute for Systems Biology, Seattle, WA). U87 and U87vIII cells transfected to express the pDsRed2-Mito fluorescent mitochondrial marker were used for a subset of immunofluorescence experiments. Cells were cultured in Dulbecco’s modified

Eagle’s medium (DMEM) containing 1 mM sodium pyruvate, 15 mM HEPES, non-

74 essential amino acids, 10% fetal bovine serum (FBS), and 1% penicillin/streptomycin.

Media devoid of FBS was used for serum deprivation experiments. Media without phenol red was used for immunofluorescence experiments.

Hypoxic cell culture

For hypoxic experiments, media was pre-equilibrated overnight in a BioSpherixTM hypoxic incubator (BioSpherix, Parish, NY) at designated oxygen concentrations (7%, 4%, and

1%) to account for time required for oxygen-saturated media to equilibrate with the gas atmosphere (Huttemann et al., 2007). Standard culture conditions are designated as normoxia (20% O2). All cells were maintained at 37 °C and 5% CO2.

Chemically-induced oxidation and reduction

To model the relationship between CHCHD2 protein folding and subcellular localization, chemically induced oxidation and reduction of U87 and U87vIII cells was achieved by incubation in hydrogen peroxide (H2O2) or dithiothreitol (DTT) at designated concentrations (1, 5, 10 mM) for 15 or 30 min. Additionally, treatment with the glutamate- cysteine ligase inhibitor L-buthionine sulfoximine (BSO, 100 µM, 48 h), as well as with the glutathione precursor N-acetyl-L-cysteine (NAC, 1 mM, 48 h), were employed as alternative oxidant and reductant, respectively.

Western blot

Total protein levels of CHCHD2 (1:500, NBP1-94106, Novus Biologicals, Centennial, CO) were compared in U87 and U87vIII cells exposed to decreasing oxygen tensions using

75 western blot. β-actin (1:1000, #4967, Cell Signaling Technology, Danvers, MA) was used as loading control. Total protein concentrations were determined using the Pierce BCA assay (Thermo Fisher, Waltham, MA). Protein lysates were mixed 1:1 with 2X Laemmli

Sample Buffer (Bio-Rad, Hercules, CA) (5% β-mercaptoethanol) and heated at 95 °C for

5-10 min. Denatured lysates were loaded into 4-20% Mini-Protean ® TGXTM electrophoresis gels (Bio-Rad, Hercules, CA), and SDS-PAGE was run at 150 V for 1-1.5 h. Proteins were transferred to a nitrocellulose membrane (Amersham, Little Chalfont,

UK) at 300 mA for 2 h at 4⁰ C. Membranes were then blocked with either 5% BSA or 5% non-fat dry milk for 1 h at RT, then incubated in primary antibody at designated concentrations overnight at 4⁰ C. Membranes were washed in TBS-T for 5 min 3x, then incubated in HRP-linked goat anti-rabbit secondary antibody (1:2500, #7074, Cell

Signaling Technology, Danvers, MA) at RT for 1.5 h. Following TBS-T washes (3x 5 min each), membranes were imaged using the SuperSignalTM West Femto Maximum

Sensitivity Substrate (Thermo Fisher, Waltham, MA) in an ImageQuant LAS 4010 (GE

Life Sciences, Pittsburgh, PA). Analysis of bands was conducted using ImageJ.

RT-qPCR

Gene expression of CHCHD2 in hypoxic (1% O2) or physiological (7% O2) oxygen levels was measured in U87 and U87vIII cells using TaqMan gene expression probes

(Hs00853326_g1, Thermo Fisher). Total cellular RNA was isolated from cells using the

RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol.

Complementary DNA (cDNA) was generated by reverse transcriptase polymerase chain reaction using the High Capacity cDNA Reverse Transcription Kit (Applied BiosystemsTM,

Foster City, CA). Transcript levels were measured by real-time PCR using an Applied

Biosystems 7900HT Fast Real-Time PCR System. Analysis was performed using the

ΔΔCt method. Gene expression was normalized to GUSB (Hs02786624_g1).

Cycloheximide time-course

Half-life of CHCHD2 in U87 and U87vIII cells was determined using the cycloheximide

(CHX) pulse-chase method (Kao et al., 2015). Cells were seeded at equal densities and treated with CHX (10 µg/mL), and total protein lysates collected at 3, 6, 9, and 12 h post- treatment. CHCHD2 levels over time were analyzed using western blot, as described above. Resultant bands were quantified using ImageJ to determine the time at which 50% of total CHCHD2 protein was degraded following translation inhibition with CHX. β-actin was used as loading control.

Immunofluorescence

U87 and U87vIII, as well as cells expressing the pDsRed2-mito fluorescent marker targeted to mitochondria were seeded at equal densities on sterile coverslips. Cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA) in PBS at

RT for 30 min. After washing 3 x 5 min in PBS, cells were permeabilized in 0.5% Triton

X-100 in PBS at RT for 15 min, followed by incubation in ultracold methanol at –20 ⁰C for

15 min, then incubation in PBS at RT for 30 min. Samples were blocked using Image-iT®

FX Signal Enhancer ReadyProbes® Reagent (Thermo Fisher) at RT for 30 min. Samples were then incubated with primary antibody against CHCHD2 (1:100, NBP1-94106, Novus

Biologicals, Centennial, CO) at RT overnight. After washing 3 x 5 min in PBS, samples

77 were incubated with Goat anti-Rabbit IgG (H+L) Secondary Antibody conjugated with

Alexa Fluor 488 (1:200, Thermo Fisher) at 37 °C for 2 h. After washing 3 x 5 min in PBS, coverslips were mounted in ProLong® Gold Antifade Mountant with DAPI (Thermo

Fisher). Samples were imaged using a Zeiss LSM 700 confocal fluorescence-enabled microscope (Carl Zeiss, Oberkochen, Germany).

Amounts of nuclear CHCHD2 detected were analyzed using Axiovision SE.64 and an automated image analysis pipeline therein. Quantities of nuclear CHCHD2 are expressed as fluorescent area detected within nuclei of individual cells normalized to nuclear area, in order to account for varying surface areas of nuclei imaged among cells.

CHCHD2 ectopic expression

Expression vectors for CHCHD2 were generously provided by Dr. Siddhesh Aras (Wayne

State University, Detroit, MI). pCI-Neo CHCHD2 3x-FLAG expression vectors and pCI-

Neo empty vectors were transfected into U87vIII cells using Lipofectamine 3000 Reagent according to manufacturer’s protocols. Successfully transfected cells were selected for with G418 (640 µg/mL). Determination of exogenous (3x-FLAG) and endogenous

CHCHD2 levels was conducted by western blot, relying on the increased molecular weight of exogenous CHCHD2 harboring a 3x-FLAG epitope.

Statistical analysis

5.3. RESULTS

Nuclear CHCHD2 is increased in hypoxic, serum-deprived, and oxidatively stressed GBM cells

U87 and U87vIII cells expressing the fluorescent mitochondrial marker pDsRed2-Mito were incubated in 20% and

1% O2 for 48 h, followed by CHCHD2 Figure 5.1. Subcellular distribution of CHCHD2 in U87 and U87vIII cells in hypoxic and serum-deprived conditions. (A) Representative immunofluorescence images of U87 (left column) and U87vIII cells (right column) immunofluorescence expressing the pDsRed2-mito fluorescent mitochondrial marker. Cells were incubated in 21% O2 (top row) or 1% O2 (middle row) for 48 h, or serum-free analysis. Under media for 30 min (bottom row). (B) Quantification of nuclear CHCHD2 as a ratio of nucleus area occupied in U87 and U87vIII cells expressing pDsRed2-mito incubated in 20% or 1% O2 for 48 h, serum-free media for 30 min, or BSO (100 standard culture µM) or NAC (1 mM) for 48 h. Differing superscript indicates statistically significant effect of treatment on nuclear CHCHD2 levels detected compared to 20% O2 control condition for each cell line. (C) Quantification of nuclear CHCHD2 as a conditions (20% O2),

U87vIII cells consistently exhibited a greater quantity of nuclear CHCHD2 compared to isogenic, EGFRvIII-lacking U87 counterparts (p < 0.0001) (Figure 5.1A, C), indicating a greater propensity for CHCHD2 to reside in nuclei of GBM cells expressing the mutant

EGFRvIII. Upon incubation in 1% O2, the amount of nuclear CHCHD2 detected significantly increased in both U87 and U87vIII cells (Figure 5.1A-B), indicative of a hypoxia-sensitive redistribution of CHCHD2 to nuclei of GBM cells. Notably, incubation of

U87 and U87vIII cells in 7% O2 (representative of average physiological oxygen levels in

79 normal brain) did not result in increased accumulation of nuclear CHCHD2 compared to

20% O2 (Figure 5.1C).

Hypoxia represents a common metabolic stressor imposed by the GBM tumor microenvironment resulting from rapid tumor growth coupled with aberrant and inefficient neovascularization (Louis, 2006). In addition to oxygen constraints, gradients in various nutrients, metabolites, and growth factors commonly occur across solid tumors, along with chronic intracellular redox stress. To mimic the metabolic stress imposed by decreasing nutrient and/or growth factor gradients, U87 and U87vIII pDsRed2-Mito cells were incubated in serum-free media for 30 min. Similarly to chronic hypoxia, acute serum deprivation resulted in a marked increase in nuclear CHCHD2 detected in U87 and

U87vIII cells (Figure 5.1A-B). Additionally, incubation in the pro-oxidant L-buthionine sulfoximine (BSO, 100 µM, 48 h), an inhibitor of glutathione (GSH) synthesis, induced an increase in nuclear CHCHD2 that was only statistically significant in U87vIII cells (Figure

5.1B). Application of the GSH precursor N-acetyl-L-cysteine (NAC, 1 mM, 48 h) did not increase nuclear CHCHD2 (Figure 5.1B). Altogether, these data demonstrate subcellular redistribution of CHCHD2 to nuclei in U87 and U87vIII cells in response to chronic hypoxia, acute serum deprivation, and chronic oxidative stress.

Total CHCHD2 protein levels do not increase in hypoxic GBM cells

Figure 5.2. CHCHD2 expression in hypoxia. (A) RT-qPCR of CHCHD2 in U87 and U87vIII cells in 1% O2 compared to 7% O2. (B) Representative western blot of CHCHD2 in U87 and U87vIII cells incubated in decreasing O2 levels. U87 and U87vIII cells were incubated in standard culture conditions (20% O2), 7% O2, 4% O2, and 1% O2 for 48 h. (C) Quantification of relative band intensity. β-actin was used as loading control. Data are presented as mean ± SE. n = 2. Differing superscript: p < 0.05.

A potential explanation for the increased nuclear CHCHD2 detected in hypoxic U87 and

U87vIII cells is an increase in total CHCHD2 protein levels in response to hypoxia.

Previous work has demonstrated that, in hypoxia, CHCHD2 transactivates gene expression of itself (Aras et al., 2015; Aras et al., 2013). Incubation of U87 and U87vIII cells in 1% O2 for 24 h led to increased gene expression of CHCHD2 only in U87, but not

U87vIII cells, compared to 7% O2 incubation (Figure 5.2A). To determine if increased total CHCHD2 protein levels accounted for increased nuclear CHCHD2 in hypoxic GBM cells, U87 and U87vIII cells were incubated for 48 h in 21% O2, 7% O2 (average physiological oxygen level in brain), 4% O2 (level demonstrated to induce maximal

CHCHD2 expression) (Aras et al., 2015), and 1% O2. A slight, though non-statistically significant increase in U87 CHCHD2 protein levels was observed at 4% O2 (Figure 5.2B-

C). Overall, U87 CHCHD2 protein levels did not change with decreasing oxygen tensions

(Figure 5.2B-C). On the other hand, CHCHD2 protein levels in U87vIII cells decreased when cells were incubated in 1% O2 (Figure 5.2B-C). Together, these data indicate that

81 the increase in nuclear CHCHD2 observed under hypoxia is not due to an increase in total cellular CHCHD2 expression.

CHCHD2 exhibits relatively rapid turnover in GBM cells

A second possible explanation for increased CHCHD2 in nuclei of hypoxic U87 and U87vIII cells is that, in hypoxia, newly translated

CHCHD2 is preferentially shuttled to the nucleus versus to mitochondria.

To begin to address this hypothesis, Figure 5.3. CHCHD2 turnover analyzed using cycloheximide chase. (A) Representative western blot of the turnover rate of CHCHD2 in U87 CHCHD2 levels in U87 and U87vIII cells treated with CHX (10 µg/mL) for 0, 3, 6, 9, and 12 h. Top band is CHCHD2. (B) and U87vIII cells was first Quantification of CHCHD2 band intensity following CHX treatment in U87 cells. Data are presented as percentage of the band intensity at 0 h. (C) Quantification of CHCHD2 band determined in a time-course intensity following CHX treatment in U87vIII cells. Data are presented as percentage of the band intensity at 0 h. experiment with the translation inhibitor cycloheximide (CHX), followed by western blotting of total protein lysates collected over time after CHX administration. CHX inhibits total intracellular protein translation machinery, thus allowing for the tracking of degradation rate for proteins of interest and determination of protein half-life, representative of protein turnover in cells at rest. Total protein lysates were collected at

3, 6, 9, and 12 h after administration of CHX (10 µg/mL) to U87 and U87vIII cells. The half-life of CHCHD2 was approximately 3 h (U87: 2.79 h; U87vIII: 3.41 h) (Figure 5.3).

Extrapolating these data, an entire pool of CHCHD2 would be recycled in approximately

15 h. Levels of β-actin remained relatively stable throughout the 12 h time-course. These

82 results suggest a regulatory (versus structural) role for CHCHD2, as well as establish a time frame in which to interpret subcellular redistribution of CHCHD2 under various conditions.

Disulfide reduction induces rapid nuclear accumulation of CHCHD2

Figure 5.4. Subcellular distribution of CHCHD2 in U87 and U87vIII cells acutely reduced with DTT. (A) Representative images of U87 (left column) and U87vIII cells (right column). Cells were treated with 10 mM DTT for 15 min (middle row) or 30 min (bottom row). (B) Quantification of nuclear CHCHD2 as a ratio of nucleus area occupied in U87 and U87vIII cells incubated in 10 mM DTT for 15 and 30 min. Data are presented as mean ± SE, n ≥ 10 cells. Differing superscript: p < 0.05. Import of CHCHD2 into the mitochondrial IMS requires disulfide relay-mediated import via interactions with CHCHD4, which result in oxidative formation of CHCH disulfide bonds, protein folding, and trapping in the IMS (Modjtahedi et al., 2016). Additionally, mounting pools of reducing intermediates produced by increased glycolytic flux as a result of hypoxia and/or malignant transformation can result in reductive stress, i.e. the presence of excess reducing equivalents that cannot be quenched by cellular oxidoreductases or oxygen tensions in the surrounding environment (e.g., limited oxygen

83 in hypoxia) (Loscalzo, 2016). It was hypothesized that reductive unfolding of CHCHD2 was the primary driver for its export out of mitochondria and nuclear translocation. To test this hypothesis, U87 and U87vIII cells were treated with the disulfide-reducing agent dithiothreitol (DTT) under a variety of conditions. Incubation of cells in 10 mM DTT resulted in a striking increase of nuclear CHCHD2 within 15 min (Figure 5.4). By 30 min,

CHCHD2 could only be detected in the nuclei of both cell lines (Figure 5.4). This observation was not due to malformations in cellular shape, as evidenced by differential interference contrast (DIC) images in which cell shape was still intact (Figure 5.5A-C).

Treatment with lower concentrations of DTT (1 mM) was sufficient to induce nuclear accumulation of CHCHD2 in both cell lines within 15 min (Figure 5.5A).

Figure 5.5. Subcellular distribution of CHCHD2 in U87 and U87vIII cells acutely reduced with DTT is not a result of cell shape deformation. Representative images of U87 (top row) and U87vIII cells (bottom row) treated with (A) 1 mM DTT for 15 min, (B) 10 mM DTT for 15 min, and (C) 10 mM DTT for 30 min. DIC images are included for qualitative assessment of cell morphology.

Additional treatments in serum-free media were conducted which mirrored oxidized and reduced redox western blot conditions reported in the literature (Go & Jones, 2009).

Treatment of U87 and U87vIII cells expressing the pDsRed2-Mito fluorescent mitochondrial marker with 5 mM DTT in serum free media for 30 min again resulted in rapid nuclear accumulation of CHCHD2 (Figure 5.6). Notably, addition of the oxidant

H2O2 (5 mM) abrogated the increase in nuclear CHCHD2 observed with serum

Figure 5.6. Subcellular distribution of CHCHD2 in U87 and U87vIII cells treated with H2O2 or DTT in serum- free media. (A) Representative images of U87 (left column) and U87vIII cells (right column). Cells were treated with 5 mM H2O2 for 30 min (middle row) or 5 mm DTT for 30 min (bottom row). (B) Quantification of nuclear CHCHD2 as a ratio of nucleus area occupied in U87 and U87vIII cells incubated in 5 mM H2O2 or DTT for 30 min. Data are presented as mean ± SE, n ≥ 10 cells. Differing superscript: p < 0.05. deprivation (Figure 5.6). Collectively, these data are consistent with reductive unfolding of CHCHD2 being a primary driver of its rapid accumulation in nuclei.

Ectopic expression of CHCHD2 does not result in increased levels of endogenous

CHCHD2

Previous work demonstrated that CHCHD2 acts as a transcription factor to transactivate gene expression of itself via physical interaction with recombination signal binding protein for immunoglobulin κ J (RBPJ), the main downstream effector of the Notch signaling pathway, which has been implicated in maintenance of brain tumor initiating cells (Aras et al., 2015; Aras et al., 2013; Q. Xie et al., 2016). To determine if CHCHD2 transactivates expression of itself in GBM cells expressing EGFRvIII, U87vIII cells were transiently transfected with a CHCHD2 3x-FLAG expression vector (pCI-Neo CHCHD2 3x-FLAG), and protein levels of CHCHD2 were analyzed using western blot. Notably, presence of the 3x-FLAG epitope allows for the distinction between endogenous and exogenously expressed CHCHD2 by virtue of increased molecular weight of exogenous CHCHD2

(approximately 3 kDa) (Figure 5.7). No increase in endogenous CHCHD2 levels were detected in U87vIII cells expressing the pCI-Neo CHCHD2 3x-FLAG vector (Figure 5.7), indicating that, although CHCHD2 is consistently detected at higher levels in nuclei of

U87vIII cells, it does not transactivate expression of itself.

Figure 5.7. Ectopic overexpression of CHCHD2 in U87vIII cells. (A) Representative western blot of U87vIII cells transfected to express CHCHD2 3x-FLAG and pCI-Neo empty expression vector. (B) Quantification of relative endogenous CHCHD2 band intensity (16 kDa) in WT, CHCHD2 3x-FLAG, and pCI-Neo empty vector expressing cells. Data are presented as mean ± SE. n = 2.

5.4. DISCUSSION

The dual localization of CHCHD2 in mitochondria and the nucleus implicates CHCHD2 as a protein involved in mitonuclear communication and signaling. The ability of mitochondria to communicate their status to the nucleus under stressful conditions represents a key signaling axis in tumor cell plasticity and adaptive capabilities. The current work investigated the dynamics of CHCHD2 in U87 and U87vIII cells as a function of metabolic stressors including chronic hypoxia and acute serum deprivation, as well as chronic redox stress imposed by chemical oxidation or reduction induced by BSO or NAC, respectively. It was found that, under normoxic oxygen tensions, U87vIII cells displayed a greater propensity to harbor nuclear CHCHD2. Chronic hypoxia, acute serum deprivation, chronic oxidative stress with BSO, and reductive unfolding with DTT were all able to induce nuclear accumulation of CHCHD2 in both cell lines. Given the striking increases in nuclear CHCHD2 under acute treatment conditions (30 min) relative to the half-life and turnover rate for a pool of cellular CHCHD2 (3 h and 15 h, respectively), as well as the lack of an increase in total CHCHD2 protein levels in hypoxia, the observed subcellular redistribution most likely arises due to direct translocation of a reduced form of CHCHD2 from mitochondria to the nucleus.

The CHCHD2 protein harbors a number of motifs directing its localization to the mitochondria, including: 1) a putative N-terminal mitochondrial targeting sequence; 2) a

MISS motif required for interaction with the CHCHD4 import machinery into the mitochondrial IMS; and 3) a CHCH domain required for oxidative folding, import, and trapping in the IMS. However, a putative nuclear localization sequence is not present within the 151 amino acid sequence of CHCHD2. Shuttling of proteins between the

87 nucleus and cytoplasm is mediated by nuclear pore complexes (NPC), one of the largest macromolecular complexes in eurkaryotic cells at ~120 million Daltons (Timney et al.,

2016). Canonically, nuclear localization or export sequences present on proteins destined for nuclear import or export are recognized by importins, which chaperone proteins through the NPC via facilitated diffusion in a GTP-dependent manner (Wente & Rout,

2010). However, passive diffusion of proteins through the NPC has also been described, limited by a molecular mass size threshold of 30-60 kDa (Timney et al., 2016). Notably, at only 16 kDa, CHCHD2 may well passively permeate into the nucleus under certain favorable conditions, such as those described in this study.

Although real-time monitoring of CHCHD2 shuttling in live cells using GFP-fused

CHCHD2 may have directly demonstrated direct translocation in response to various stressors, use of a CHCHD2-GFP fusion expression vector did not yield appropriate controls. Specifically, U87vIII cells transfected with a pEGFP CHCHD2 expression vector displayed cytosolic localization of CHCHD2 (not shown). It is likely that the addition of the globular, 27 kDa GFP protein to the C-terminus of CHCHD2 interfered with its ability to be appropriately imported into mitochondria, precluding the ability to observe its subcellular dynamics.

Nevertheless, acute treatments to model the relationship between CHCHD2 protein folding and subcellular localization in U87 and U87vIII cells provide indirect evidence supportive of direct translocation. This conclusion is supported by various lines of evidence. First, neither U87 nor U87vIII cells increased their expression of total CHCHD2 protein under decreasing oxygen tensions (Figure 5.2), indicating that increased total

CHCHD2 expression was not responsible for increased nuclear CHCHD2 observed in

88 hypoxia. Notably, in U87vIII cells, incubation in 1% O2 decreased total CHCHD2 levels in conjunction with increased CHCHD2 detected in nuclei (Figure 5.2). Second, acute treatment for 30 min with serum-free media, as well as 30 and 15 min application of various concentrations of the thiol-reducing agent DTT to model protein reductive unfolding (1, 5, and 10 mM), resulted in significant increases in nuclear CHCHD2 (Figure

5.4, 5.5). The possibility that this is explained by preferential shuttling of newly translated

CHCHD2 is unlikely, given i) the short time frame of subcellular redistribution (15-30 min) relative to the 3 hour half-life of CHCHD2, as well as ii) the likely disruption of protein translation machinery resulting from DTT treatment. Furthermore, environmental stresses such as nutrient starvation (simulated by serum-deprivation) generally reduce rates of protein translation (B. Liu & Qian, 2014).

Because import of CHCHD2 into the mitochondrial IMS depends upon oxidation of its twin CX9C motifs (resulting in formation of disulfide bonds that stabilize the helix-turn- helix fold of the CHCH domain and trap CHCHD2 within the IMS), it was hypothesized that reduction of these disulfide bonds, modeled via treatment with the disulfide-reducing agent DTT, would induce reductive unfolding and retro-translocation out of the mitochondrial IMS. The observation that reduced CHCHD2 localized rapidly to the nucleus is consistent with mechanisms of redox regulation of nuclear transcription factors.

Redox-dependent modulation of DNA binding has been described for a number of transcription factors, including NF-kB, p53, and Nrf-2, all of which contain at least one critical cysteine residue that must be reduced in nuclei for DNA binding, a function inhibited by oxidation (Go & Jones, 2010). Functional domain analysis of CHCHD2 demonstrated that the amino acids immediately upstream of Cys114 (aa 101-113) are

89 required for CHCHD2 interaction with RBPJ and transactivation of a COX4I2 luciferase reporter (Aras et al., 2015). Notably, site-directed mutagenesis of all four cysteine residues, which would preclude proper oxidative folding of CHCHD2 and inhibit its import into the IMS, did not inhibit its ability to transactivate the COX4I2 luciferase reporter (Aras et al., 2015), indicating that CHCHD2 could perform its gene transactivation function in a reduced, unfolded state.

Notably, other than the reductive protein unfolding modeled using DTT, stimulants of

CHCHD2 nuclear accumulation, such as serum deprivation, treatment with BSO, and hypoxia, are conventionally regarded as oxidative stressors. Additionally, U87 GBM cells expressing EGFRvIII have been reported to exhibit greater levels of reactive oxygen species (Nitta et al., 2010; Sangar et al., 2014), and CHCHD2 was consistently detected at greater levels in resting U87vIII versus U87 cells (Figure 5.1). It should be noted that redox poise across organelles is independently maintained and compartmentalized, and distinctly altered by environmental insults. For instance, hypoxia induces oxidation of the cytosolic and IMS glutathione pools, yet reduces the mitochondrial matrix glutathione pool

(Byrne, Leslie, Patel, Gaskins, & Kenis, 2017; Waypa et al., 2010). It is possible that IMS glutathione oxidation in hypoxia could lead to glutathionylation of CHCHD2 thiol(s).

Alternatively, redox stress could also potentially perturb interactions between CHCHD2 and its mitochondrial binding partners CHCHD10 and p32 (Imai, Meng, Shiba-Fukushima,

& Hattori, 2019; Purandare, Somayajulu, Huttemann, Grossman, & Aras, 2018), facilitating disulfide modification of its CHCH domain and translocation. However, the binding mode of IMS-resident CHCHD2 with matrix-resident p32 and the orientation of

CHCHD2 within the IMS (soluble or membrane-embedded) remains unclear. Future

90 studies are warranted to dissect these, or other, possibilities, to obtain a clearer picture of CHCHD2’s apparent redox-sensitive translocation activity.

It was further demonstrated that CHCHD2 did not transactivate expression of itself in

U87vIII cells. Despite displaying increased nuclear localization in hypoxia, gene expression of CHCHD2 was unaltered and protein levels were decreased in hypoxic

U87vIII cells (Figure 5.2). Additionally, ectopic expression of CHCHD2 with the pCI-Neo

CHCHD2 3x-FLAG expression vector in U87vIII cells did not result in increased levels of endogenous CHCHD2 (Figure 5.7). These data contrast with previous studies demonstrating transcription factor activity of CHCHD2 inducing expression of itself maximally at 4% O2 via luciferase reporter assays in HEK293 and rat pulmonary cells, as well as increased protein levels of endogenous CHCHD2 in HEK293 cells ectopically overexpressing CHCHD2 with a 3x-FLAG epitope (Aras et al., 2013). Aras et al. previously demonstrated that CHCHD2 enhances gene expression via interactions with

RBPJ, the main downstream effector of the Notch signaling pathway. Notably, recent work by Xie et al. has implicated RBPJ as critical for maintenance of brain tumor initiating cell viability and stemness (Q. Xie et al., 2016). Given the studies implicating CHCHD2 in neuroectodermal differentiation capacities of human embryonic stem cells (Zhu et al.,

2016), its association with the migratory capabilities of neural stem cells in patients with lissencephaly (Shimojima et al., 2015), as well as recent work highlighting neural stem cells as a GBM cell of origin (J. H. Lee et al., 2018), it is tempting to speculate that the

CHCHD2-RBPJ interaction may play a role in maintaining glioma stem cells and tumor progression. Future studies should investigate the extent to which CHCHD2 interacts with

RBPJ in GBM cells, as well as determine the suite of genes CHCHD2 may regulate, which

91 could reveal novel gene targets to further explain the contribution of nuclear CHCHD2 to

GBM malignancy.

5.5. CONCLUSIONS

These data indicate that CHCHD2 accumulates in the nucleus of U87 and U87vIII GBM cells in response to metabolic and redox stress, and that reductive unfolding drives rapid nuclear localization of CHCHD2. This is the first work to provide evidence that this subcellular redistribution from mitochondria to the nucleus is likely driven by translocation of reduced mitochondrial CHCHD2, versus other potential mechanisms.

CHAPTER 6: CHARACTERIZING MICROGLIA-GLIOBLASTOMA CROSSTALK

ABSTRACT

INTRODUCTION

Glioblastoma (GBM) is the most common and deadly form of primary malignant brain tumor in adults. Microglia, the primary resident immune cells in the brain, represent a substantial portion of GBM tumors, and paracrine signaling networks between microglia and GBM play a critical role in cultivating a microenvironment that facilitates angiogenesis, cell invasion, immunosuppression, and tumor expansion. Additionally, glutamate, the main excitatory neurotransmitter in the central nervous system, is released in excess by GBM tumors and promotes cell proliferation and invasion. Export of

- glutamate and import of cystine through system xc serves to both remodel the extracellular environment and preserve intracellular glutathione redox status. The objective of the current study was to characterize the potential impact of the tumor- specific antigen, EGFRvIII, on immunologic and angiogenic crosstalk between GBM cells

- and microglia, as well as the impact of microglia on system xc -mediated glutathione redox status in GBM cells. We hypothesized that GBM-microglia crosstalk would facilitate secretion of soluble factors that promote angiogenesis and extracellular matrix remodeling, as well as alter GBM cell glutathione redox balance.

METHODS

The U87 parental GBM cell line, U87 cells transduced to stably express EGFRvIII

(U87vIII), patient-derived EGFRvIII-expressing GBM6 short-term explants, HMC3 microglia cell line, and primary neonatal murine microglia were employed to model tumor- microglia crosstalk. The influence of tumor-microglia crosstalk on secretion of soluble

93 factors related to angiogenesis and matrix remodeling was assessed using the R&D

Systems Proteome Profiler Angiogenesis Cytokine Array. Cell proliferation in response to incubation in conditioned-media was analyzed via Iba1 and Ki-67 staining. 2D cell migration in response to tumor-microglia crosstalk was characterized using GBM:HMC3 co-culture conditions in 2-well adhesive cell culture inserts. Mitochondrial GSH:GSSG status in U87 and U87vIII cells was determined using fluorescent redox biosensors

- targeted to the mitochondrial matrix (mito-Grx1-roGFP2). Expression of the system xc glutamate-cystine antiporter was analyzed using western blot of xCT, as well as RT-qPCR of the SLC7A11 gene.

RESULTS

The expression of EGFRvIII on U87 cells resulted in an altered profile of secreted angiogenesis-related factors under basal culture conditions. Similarly, primary neonatal murine microglia, but not HMC3 cells, exhibited differential angiogenesis-related secretomes when cultured in U87 or U87vIII conditioned media. Primary murine microglia additionally displayed increased cell proliferation when cultured in U87 and U87vIII conditioned media, while HMC3 cells did not. Cell migration of U87 and U87vIII cells was not increased in the presence of HMC3 co-culture, although differences in cellular spatial organization and adhesion were observed. GBM6 conditioned medium led to a striking increase in HMC3 secretion of angiogenesis-related factors, inducing the greatest secretion of CCL2, IGFBP-2, TIMP-4, angiogenin, and IGFBP-3. HMC3 conditioned

- medium decreased expression of the system xc antiporter and increased glutathione oxidation in mitochondria of U87 cells, while its xCT expression and mitochondrial glutathione redox status in U87vIII cells remained stable.

CONCLUSIONS

GBM cells stimulate the secretion of cytokines, chemokines, and soluble factors by microglia to cultivate an immunosuppressive microenvironment promoting angiogenesis

- and cell invasion. Additionally, system xc expression and, consequently, intracellular glutathione redox balance remain stable in U87vIII cells exposed to HMC3 conditioned media.

6.1. INTRODUCTION

Glioblastoma (WHO grade IV glioma) is the most common and malignant primary brain tumor in adults and represents approximately 50% of all gliomas diagnosed each year

(Barnholtz-Sloan et al., 2018). Median survival for patients with GBM is only 14.6 months, with less than 5% of patients surviving after 5 years (Stupp et al., 2005). The current standard of care, which is multimodal and consists of surgical resection, radiotherapy, and chemotherapy with temozolomide, is only palliative and does little to prevent tumor dissemination into surrounding brain parenchyma and, eventually, inevitable tumor relapse. Numerous features of the surrounding tumor microenvironment, such as tumor- associated microglia, play a dynamic role in modulating cell behavior and thus selecting for more malignant cell populations.

Microglia are the primary resident immune cells of the brain and spinal cord and account for 5-12% of cells in the brain (Alliot et al., 1999; Hickman et al., 2013;

Nimmerjahn et al., 2005). As sentinels of baseline immunity in the brain, microglia exhibit profound plasticity which allows them to assume neurotoxic or neuroprotective states that determine their responses to changes in the surrounding microenvironment, invading

95 pathogens and toxins, neurodegenerative diseases and traumatic brain injury, and cellular debris (Hickman et al., 2013; Nimmerjahn et al., 2005). Seminal work conducted by Nimmerjahn et al. (2005) revealed that, in the healthy brain, microglia continuously and dynamically surveil their surroundings via continuous cycles of de novo formation and withdrawal of cellular processes, which interact with cortical elements such as astrocytes, neurons, and blood vessels (Nimmerjahn et al., 2005). Further facilitating microglial monitoring of brain parenchyma is a repertoire of cell surface proteins and receptors including chemoattractant and chemokine receptors, cytokine receptors, receptors involved in cell-cell interaction, and receptors for extracellular matrix proteins, collectively termed the microglial sensome (Hickman et al., 2013).

Previous studies have demonstrated various avenues of synergistic crosstalk between GBM tumors and tumor associated microglia and macrophages (TAMs).

Notably, TAMs constitute up to 30% of the GBM tumor bulk, promoting tumor progression, angiogenesis, and invasion (Coniglio et al., 2012). Intratumoral microglia density is higher than in surrounding normal brain and increases with tumor grade (Bettinger, Thanos, &

Paulus, 2002). TAMs are not merely a nonspecific reaction to tissue injury, but rather are actively recruited to GBMs via tumor secretion of chemotactic factors such as granulocyte-macrophage stimulating factor (GM-CSF), monocyte chemoattractant protein-1 (MCP-1 aka CCL2), glial cell-derived neurotrophic factor (GDNF), and hepatocyte growth factor (HGF) (Bettinger et al., 2002; Hambardzumyan et al., 2016). In turn, recruited microglia release several anti-inflammatory (e.g. TGF-β, IL-10), pro- inflammatory (e.g. TNF-α, IL-1β), and extracellular matrix remodeling, invasion promoting factors (e.g. EGF, MMP-9, VEGF).

Glutamate, the main excitatory neurotransmitter in the central nervous system, is released in excess by GBM tumors and promotes cell proliferation and invasion (de Groot

- & Sontheimer, 2011). System xc is a sodium-independent membrane transporter which couples the influx of extracellular cystine to the efflux of glutamate (Robert et al., 2015).

- Previous studies have reported upregulation of system xc expression conferring chemoresistance and stem-like properties in glioma cells (Polewski, Reveron-Thornton,

Cherryholmes, Marinov, & Aboody, 2017; Polewski et al., 2016). Upregulation of xCT confers a tumor growth advantage by 1) increased import of cystine, which is rapidly reduced within the cell to cysteine, the rate-limiting substrate for synthesis of glutathione

(GSH), the main intracellular redox buffer; and 2) increased efflux of glutamate, which can result in excitotoxic damage to neurons, thus clearing room for tumor growth and resulting in tumor-associated seizures (Robert et al., 2015).

The objective of the current study was to exsamine inter-cellular crosstalk between microglia and GBM cells, and to profile its effects on cell proliferation, invasion, and glutathione redox status. Conditioned-media and co-culture experiments were conducted to interrogate the impact of GBM-microglia crosstalk on cell proliferation, cell migration, and microglia secretion of a repertoire of angiogenesis-related factors. A variety of models were employed, including the human HMC3 microglia cell line, primary neonatal murine microglia, U87 and U87vIII GBM cells, and primary EGFRvIII-expressing GBM6 patient- derived cells.

6.2. METHODS

Cell culture

The human U87 parental GBM cell line, as well as U87 GBM cells transduced to stably express the constitutively active EGFRvIII mutant (U87vIII) were generously provided by

Dr. Nathan Price (Institute for Systems Biology, Seattle, WA). Cells were cultured in

Dulbecco’s modified Eagle’s medium (DMEM) containing 1 mM sodium pyruvate, 15 mM

HEPES, non-essential amino acids, 10% fetal bovine serum (FBS), and 1% penicillin/streptomycin.

GBM6 (EGFRvIII/PTEN-wt, highly invasive) primary patient-derived cells obtained from Dr. Jann Sarkaria (Mayo Clinic) were cultured as spheroids in methacrylated gelatin

(GelMA) hydrogels. Spheroids were generated by seeding cells in equal densities (5,000 cells/well) in a low-adhesion, round-bottom 96-well plate, followed by shaking at 60 rpm for 48 h. Resultant spheroids were cultured in GelMA hydrogels photopolymerized under

UV light (AccuCure LED 365 nm, 7.1 mW cm-2 for 30 s) in the presence of a lithium acylphophinate photoinitiator.

The human microglia cell line HMC3 was obtained from American Type Culture

Collection (ATCC, CRL-3304) and cultured in DMEM containing 1 mM sodium pyruvate,

15 mM HEPES, non-essential amino acids, 10% FBS, and 1% penicillin/streptomycin.

Primary murine microglia were isolated from neonatal mice and generously provided by Dr. Andrew Steelman and cultured in DMEM/F12. Briefly, brains from neonatal

C57BL/6 pups (1-3 d old) were harvested, meninges removed, and tissue incubated in

Accutase for 30 min at 37 °C. Tissue was then homogenized using a 70 µm filter and 3 mL syringe, and the filter washed with 5 mL DMEM containing 10% FBS. Cell suspension

98 was then centrifuged at 300 x g for 5 min, and cells resuspended in fresh media, counted, and plated in a T-75 flask at a concentration of 1 x 106 cells/mL. Medium was changed every other day until cells reached confluency, at which point the flask was shaken at 170 rpm for 1 h at 37 °C. Microglia were isolated as floating cells in media. All cells were maintained at 37 °C and 5% CO2.

Cytokine array

U87 and U87vIII cells were seeded at equal densities (150,000 cells/well) in two 6-well plates and allowed to condition media for 72 h. The resulting conditioned media (CM) from each cell line was centrifuged for 5 min at 267 x g, and supernatant directly applied to HMC3 cells seeded at equal densities (100,000 cells/well) in a 12-well plate. HMC3 cells were incubated in either U87 or U87vIII CM for 48 h. Media from untreated HMC3 cells, HMC3 cells incubated in U87 or U87vIII CM, as well as U87 and U87vIII CM was profiled for the presence of secreted angiogenesis-related factors using the R&D Systems

Proteome Profiler Angiogenesis Cytokine Array according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN). Similarly, U87 and U87vIII cells were cultured for 72 h before application of conditioned media from each cell line to primary neonatal murine microglia plated in a 24-well culture plate. Additionally, the influence of

GBM6-conditioned media (GBM6 CM) on HMC3 secretion of angiogenesis-related factors was examined. Conditioned media were collected directly from the culture flask in which GBM6 cells were plated and shipped from Mayo Clinic.

Cell proliferation

Cell proliferation of HMC3 cells in response to U87 and U87vIII CM was assessed using

Ki-67 staining and analysis. U87 and U87vIII cells were seeded at equal densities

(150,000 cells/well) in a 6-well plate and allowed to condition media for 72 h. Afterwards, conditioned media was centrifuged at 267 x g for 5 min, and applied to HMC3 cells seeded at equal densities in an µ-Slide eight-well ibiTreat microscopy chambers (Ibidi, Munich,

Germany). Cells were incubated in either control DMEM or GBM-conditioned media for

48 h before fixation and staining for Ki-67 (1:400, D3B5, Cell Signaling Technology,

Danvers, MA). Stained samples were imaged using the tile scan function at 20x on a

Zeiss LSM 700 confocal fluorescence microscope (Carl Zeiss, Oberkochen, Germany), and Ki-67+ cells manually counted. To assess U87 and U87vIII cell proliferation in response to HMC3 CM, the cell culture workflow above was conducted in reverse.

Proliferation of primary neonatal murine microglia in response to U87 and U87vIII CM was observed. Microglia treated with U87 and U87vIII CM for 48 h were fixed and stained with anti-Iba1 antibody and Hoechst. Iba-1+ cells were counted in three 20x fields per well, and averages of each well and condition were calculated.

Cell migration

Cell migration of U87 and U87vIII cells in response to co-culture with HMC3 cells was assessed using wound-healing assays. U87 and U87vIII cells expressing the fluorescent mitochondrial marker pDsRed2-Mito were seeded at equal densities in each well of iBidi

2-well silicone adhesive culture inserts (iBidi) placed in a 12-well plate. Use of these inserts allows for a reproducible, defined 500 µm cell-free gap to assess cell migration.

U87 and U87vIII cells were seeded either independently or co-cultured 2:1 with HMC3

100 cells (2 GBM:1 HMC3), and time-lapse images of three 4x fields per well were acquired.

Percent of initial wound closure was tracked over time in each condition.

RT-qPCR

Expression of SLC7A11 in U87 and U87vIII cells incubated for 48 h in 72 h HMC3 CM was measured using TaqMan gene expression probes (Hs00921938_m1, Thermo

Fisher, Waltham, MA). Total cellular RNA was isolated from cells using the RNeasy Mini

Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Complementary

DNA (cDNA) was generated by reverse transcriptase polymerase chain reaction using the High Capacity cDNA Reverse Transcription Kit (Applied BiosystemsTM, Foster City,

CA). Transcript levels were measured by real-time PCR using an Applied Biosystems

7900HT Fast Real-Time PCR System. Analysis was performed using the ΔΔCt method.

Gene expression was normalized to GAPDH (Hs02786624_g1).

Western blot

Total protein levels of the glutamate-cystine antiporter xCT (1:500, ab37185, Abcam,

Cambridge, UK) were compared in U87 and U87vIII cells incubated for 48 h in 72 h HMC3

CM using western blot. β-actin (1:1000, #4967, Cell Signaling Technology, Danvers, MA) was used as loading control. Total protein concentrations were determined using the

Pierce BCA assay (Thermo Fisher). Protein lysates were mixed 1:1 with 2X Laemmli

Sample Buffer (Bio-Rad, Hercules, CA) (5% β-mercaptoethanol) and heated at 95 °C for

5-10 min. Denatured lysates were loaded into 4-20% Mini-Protean ® TGXTM electrophoresis gels (Bio-Rad, Hercules, CA), and SDS-PAGE was run at 150 V for 1-1.5

101 h. Proteins were transferred to a nitrocellulose membrane (Amersham, Little Chalfont,

UK) at 300 mA for 2 h at 4⁰ C. Membranes were then blocked with 5% BSA for 1 h at RT, then incubated in primary antibody at designated concentrations overnight at 4⁰ C.

Membranes were washed in TBS-T for 5 min 3x, then incubated in HRP-linked goat anti- rabbit secondary antibody (1:2500, #7074, Cell Signaling Technology) at RT for 1.5 h.

Following TBS-T washes (3x 5 min each), membranes were imaged using the

SuperSignalTM West Femto Maximum Sensitivity Substrate (Thermo Fisher) in an

ImageQuant LAS 4010 (GE Life Sciences, Pittsburgh, PA). Band densitometry analysis was performed using ImageJ.

Measurement of mitochondrial glutathione redox poise

U87 and U87vIII cells were transfected with genetically encoded, fluorescent redox biosensors targeted to the mitochondrial matrix (mito-Grx1-roGFP2), previously described by our lab (Kolossov et al. 2015). Cells expressing mitochondrial Grx1-roGFP2 were seeded at equal densities in standard culture medium without phenol red in µ-Slide eight-well ibiTreat microscopy chambers (Ibidi). Time-lapse images were collected with a fluorescence-enabled inverted microscope (Axiovert 200 M, Carl Zeiss, Feldbach,

Switzerland). Dual-excitation imaging of live cells used 395 and 494 nm excitation cubes, and an emission filter at 527 nm was used for both cubes. Exposure times were set to

100-200 ms, and images were taken every 15 s. To assess the effect of 72 h HMC3 conditioned media on mitochondrial GSH:GSSG status, U87 and U87vIII cells were pre- treated with conditioned media 48 h before time-lapse image acquisition. Acquired images were processed with Zeiss Axiovision SE64 Rel6.8 software, via manual selection

102 of three to five individual cells to obtain multiple regions of interest in each time lapse.

The means of emission intensities at 527 nm were exported to Excel files and corrected by background subtraction.

Statistical analysis

6.3. RESULTS

EGFRvIII expression influences the secretion of pro-angiogenic and ECM remodeling factors by U87 cells and primary murine microglia

The milieu of angiogenesis-related factors secreted by

U87 and isogenic

U87vIII cells under standard basal culture conditions was profiled via cytokine Figure 6.1. Effect of EGFRvIII on U87 and microglia secretion of array. Proteins angiogenesis-related factors. Secreted chemokines, cytokines, and soluble factors related to angiogenesis were profiled in culture media using a R&D Systems Cytokine Array. (A) Angiogenesis-related factors secreted > 1.5 fold by secreted greater than U87 vs U87vIII cells. (B) Heatmap of angiogenesis-related factors secreted > 1.5 fold by U87vIII cells vs U87 cells. (C) Angiogenesis-related factors secreted > 2 1.5 fold by U87 versus fold by primary neonatal murine microglia incubated in U87 and U87vIII CM for 48 h.

103

U87vIII cells included coagulation factor III (F3), endoglin (ENG), GM-CSF, angiopoietin-

2 (ANGPT2), amphiregulin (AREG), IL-1β, thrombospondin-2 (THBS2), IL-8, GDNF, activin A, and CXCL16 (Figure 6.1A). Conversely, proteins secreted greater than 1.5 fold by U87vIII cells included insulin growth factor binding proteins 2 and 3 (IGFBP-3, IGFBP-

2), endocrine gland derived vascular endothelial growth factor (EG-VEGF), platelet- derived endothelial cell growth factor (PD-ECGF), vasohibin, angiogenin, MMP-8, and

TGFβ-1 (Figure 6.1B).

The effect of EGFRvIII expression

extended beyond secretion patterns of

U87 cells. Primary murine microglia

exhibited differential secretion patterns of

angiogenesis-related factors in response

to incubation in U87 or U87vIII conditioned

media (CM). Compared to U87 CM,

U87vIII CM induced greater microglia

secretion of various factors including

MMP-3, CXCL1, CCL2, IGFBP-3, and IL-

Figure 6.2. Effect of U87 and U87vIII conditioned 10 (Figure 6.1C). However, this effect was media on angiogenesis-related factors released by HMC3. Secreted chemokines, cytokines, and soluble factors related to angiogenesis were profiled in culture not observed by HMC3 cells incubated in media using a R&D Systems Cytokine Array. (A) Heatmap of secreted factors detected in HMC3 culture media, U87 culture media, and HMC3 culture media U87 or U87vIII CM (Figure 6.2A-B). No after incubation in U87 conditioned media. (B) Heatmap of secreted factors detected in HMC3 culture media, factors were secreted greater than two-fold U87vIII culture media, and HMC3 culture media after incubation in U87vIII conditioned media. by HMC3 cells in response to incubation in

U87 or U87vIII CM. Only secretion of angiogenin in response to U87 CM and FGF-7 in

104 response to U87vIII CM exceeded a fold change greater than 1.5 (1.85 and 1.88 fold, respectively). Collectively, these data indicate differential protein secretion by U87 GBM cells, as well as primary murine microglia, as a result of GBM EGFRvIII expression.

U87 and U87vIII conditioned media increase proliferation of primary neonatal murine microglia

Proliferation of microglia is key in their response to damage in the brain and central nervous system pathologies. The number of Iba1+ primary murine microglia in vitro was significantly increased when

Figure 6.3. Effect of conditioned media on cell proliferation. cells were incubated in U87 CM (A) Representative 20x field images of primary neonatal murine microglia incubated in U87 or U87vIII conditioned media for 48 h compared to basal control and stained for Iba1 and Hoechst. (B) Average Iba1+ cell counts in wells treated with CTL media, U87 CM, or U87vIII CM for 48 h. (C) Average fold change in Ki-67+ HMC3 cells incubated in CTL conditions, and more so when media, U87 CM, or U87vIII CM for 48 h. (C) Average fold change in Ki-67+ U87 and U87vIII cells incubated in CTL media or HMC3 CM. Data are presented as mean ± SE, n = 3-4. n.s.: not incubated in U87vIII CM (Figure significant. * p < 0.05

6.3A-B). Additionally, microglia treated with U87vIII CM exhibited apparent morphological changes over the 48 h incubation period. These data indicate the presence of soluble factors constitutively released by U87 and U87vIII cells under basal culture conditions that stimulate microglia proliferation.

Similarly, the impact of U87 and U87vIII conditioned media on HMC3 proliferation was assessed via staining for Ki-67, a nuclear protein and marker of actively proliferating cells whose expression markedly increases during progression through S phase of the cell

105 cycle. An apparent pattern of increased HMC3 proliferation in response to U87 and

U87vIII CM occurred similar to murine microglia; however, this was not statistically significant over multiple experiments (Figure 6.3C). Similarly, U87 and U87vIII cells displayed a non-statistically significant pattern of increased proliferation in response to

HMC3 CM (Figure 6.3D).

Cell migration of U87 and U87vIII cells is not enhanced by co-culture with HMC3 cells

The influence of U87 and

U87vIII co-culture with

HMC3 cells on cell migration was investigated using wound healing assays. As tumor-associated microglia and macrophages can Figure 6.4. Effect of HMC3 co-culture on cell migration. (A) Representative images of U87 and U87vIII cells seeded independently or constitute 30-50% of GBM co-cultured 2:1 with HMC3 cells at 0 h and 12 h timepoints of cell migration wound healing assay. The propensity for U87vIII cells to participate in spontaneous spheroid formation and anchorage-independent growth tumors (Hambardzumyan et precluded the ability to observe efficient wound closure. However, co- culture with HMC3 cells, while not limiting spheroid formation, improved al., 2016), U87 and U87vIII cell adhesion throughout the assay. (B) Representative images of U87 cells expressing the pDsRed2-Mito marker highlight a patterned reorganization during co-culture with HMC3 cells. This was not observed in cells were co-cultured 2:1 U87vIII cells. (C) Quantification of wound area percent decrease normalized to 0 h. Note y-axis lower bound. (D) Quantification of wound area change for U87vIII cells. Note y-axis lower bound. Data are presented with HMC3 microglia. as mean ± SE, n = 3. * p < 0.05 between independent and co-culture conditions at given time point. Scale bar = 0.5 mm. Representative images of

U87 and U87vIII cells seeded independently or co-cultured with HMC3 cells are displayed in Figure 6.4A. U87vIII cells display a propensity to participate in spontaneous spheroid formation and adhesion-independent growth, which resulted in gradual increases in

106 wound area over time and precluded the ability to observe efficient wound closure (Figure

6.4A, D). Notably, however, when cultured with HMC3 cells, U87vIII maintained adhesion throughout the assay, although wound closure was not altered (Figure 6.4D). U87 cells, which do not readily form spheroids in traditional cell culture, maintained adhesion throughout the assay, though wound closure over 12 h was marginal (5-6%) and not significantly improved by co-culture with HMC3 (Figure 6.4A, C). However, co-culture with HMC3 appeared to induce a spatial re-organization of U87 cells, an observation not seen with U87vIII cells (Figure 6.4B).

107

HMC3 cells exhibit a striking secretory response when stimulated with primary patient- derived GBM cell conditioned media

Figure 6.5. Effect of GBM6-conditioned media on HMC3 secretion of angiogenesis-related factors. (A) Descriptive table of patient-derived xenograft (PDX) GBM6, with descriptors including EGFR/PTEN status, patient status, invasion in mouse orthotopic xenografts, erlotinib sensitivity, MGMT methylation status, and subtype. (B) Heatmap of secreted factors detected in GBM6 culture media, HMC3 culture media, and HMC3 culture media after incubation in GBM6 conditioned media. (C) Angiogenesis-related factors secreted > 2 fold (> 1 log2 fold change, blue), or downregulated (< -1 log2 fold change, grey) by HMC3 cells treated with GBM6 CM vs resting HMC3 cells. (D) STRING protein-protein interaction map of angiogenesis-related factors secreted > 2 fold by HMC3 cells in response to GBM6 CM. (E) Gene ontology biological processes associated with upregulated secreted factors.

The GBM6 patient-derived xenograft is characterized by expression of the EGFRvIII mutant, a highly invasive phenotype, significant resistance to the receptor tyrosine kinase inhibitor erlotinib, expression of the MGMT DNA repair enzyme, and classical GBM subtype (Figure 6.5A). Compared to U87vIII CM, HMC3 cells responded strikingly to

GBM6 CM (Figure 6.5B), producing and secreting significantly increased amounts of factors including CCL2, IGFBPs 1, 2, and 3, TIMP-4, angiogenin, TGFβ-1, and others

(Figure 6.5C). Protein interaction networks among upregulated secreted proteins clustered around VEGF (Figure 6.5D), with gene ontology analysis revealed the most

108 enriched biological functions of upregulated secreted factors being related to regulation of cell proliferation, migration, and endothelial cell proliferation (Figure 6.5E).

HMC3 conditioned media decreases expression of xCT in U87, but not U87vIII cells

Tumor cells display an increased demand for amino acids to meet the biosynthetic demands of rapidly proliferating cells and often upregulate expression of cell surface amino acid

- transporters, such as the system xc glutamate-cystine antiporter, to meet this need (Bhutia, Babu, Figure 6.6. Expression of the glutamate-cystine antiporter in U87 and U87vIII cells exposed to HMC3- Ramachandran, & Ganapathy, 2015; conditioned media. RT-qPCR of SLC7A11 in (A) U87 and (B) U87vIII cells incubated in control or HMC3-conditioned media for 48 h. (C) Representative western blot of xCT in Pavlova & Thompson, 2016). Both U87 U87 and U87vIII cells incubated in control or 72 h HMC3- conditioned media for 48 h. Quantification of xCT relative band intensity in (D) U87 and (E) U87vIII cells. β-actin was and U87vIII cells displayed baseline used as loading control. Data are presented as mean ± SE, n = 2. n.s.: not significant. * p < 0.05. expression of the SLC7A11 gene, which

- encodes the system xc catalytic subunit xCT. When incubated in HMC3 CM, U87 cells displayed decreased expression of SLC7A11 with a concomitant decrease in xCT protein levels (Figure 6.6A, C, D). In contrast, expression of SLC7A11 and xCT in U87vIII cells was not significantly altered (Figure 6.6B, C, E). These data indicate that HMC3 cells

- induce downregulation of system xc in U87 cells, and that xCT expression displays greater stability in U87 cells expressing EGFRvIII.

109

HMC3 conditioned medium oxidizes the mitochondrial glutathione pool in U87, but not

U87vIII cells

- The system xc transporter mediates the import of extracellular cystine, the rapid reduction of which into cysteine provides the rate-limiting precursor for the synthesis of reduced glutathione (GSH);

Figure 6.7. Mitochondrial matrix glutathione redox status in U87 - thus, system xc expression is and U87vIII cells incubated in HMC3 conditioned media. (A) Schematic of the molecular mechanism of the Grx1-roGFP2 sensor and redox response of the compartmentalized probe to exogenous an important determinant of • − oxidant and reductant. Superoxide ( O2 ) is rapidly converted by superoxide dismutase (SOD) into H2O2, which is then reduced by intracellular redox balance. glutathione peroxidase (GPx) to water. Glutaredoxin (Grx) fused to roGFP2 efficiently and rapidly equilibrates the probe with alterations in the local GSH:GSSG ratio. Additionally, thiol-disulfide equilibration is Using genetically encoded reversible, as GSH reductase (GR) catalyzes reduction of GSSG to GSH. (B) Representative fluorescence images demonstrate the sensor targeted to mitochondria of U87 cells incubated in CTL (left) or fluorescent redox biosensors HMC3 conditioned media (right). (B’) Corresponding time-lapse responses of the 395/494 nm ratio to treatment with 1 mM diamide targeted to the mitochondrial (vertical solid line) to the fully oxidized state and 10 mM DTT (vertical dashed line) to the fully reduced state. Each trace designates a separate cell. Arrows represent basal oxidation level of probe. Percent matrix (Figure 6.7A-B), HMC3 oxidized mito-Grx1-roGFP2 in (C) U87 and (D) U87vIII cells incubated in CTL or HMC3 conditioned media. Data are presented as mean ± conditioned media increased SE, n = 5. n.s.: not significant. * p < 0.05. mitochondrial glutathione oxidation in U87 cells (Figure 6.7B’-C). In contrast, glutathione oxidation in U87vIII mitochondrial matrix was unaltered by incubation in HMC3 CM

(Figure 6.7D).

110

6.4. DISCUSSION

Tumor-associated microglia and macrophages represent a key component of the GBM tumor microenvironment, which promotes glioma cell proliferation and invasion, angiogenesis, extracellular matrix degradation, and immunosuppression

(Hambardzumyan et al., 2016). The current work examined crosstalk between EGFRvIII- expressing GBM cells and microglia by profiling secreted chemokines, cytokines, and

- soluble factors related to angiogenesis, as well as examining system xc expression and compartmentalized glutathione redox balance of U87 GBM cells incubated in HMC3 conditioned media.

Notably, the expression of EGFRvIII influenced secretion patterns of U87 GBM cells

(Figure 6.1A-B). U87vIII cells secreted greater levels of insulin-like growth factor-binding proteins 2 and 3. Canonically, IGFBPs bind to IGF in the extracellular compartment and can both enhance as well as hamper IGF binding to its receptor (Chesik, De Keyser, &

Wilczak, 2004). IGFBP-2 was demonstrated recently to stimulate vasculogenic mimicry of glioma cells, i.e. the construction of fluid-conducting channels formed by aggressive tumor cells rather than endothelial cells, both in vitro and in vivo (Y. Liu et al., 2018).

IGFBP-2 also enhanced expression of MMP-2 to promote cell invasion, promote self- renewal of neural stem cells during CNS development in mice, and regulate expression of Notch pathway and neural stemness genes including Nestin, OLIG2, and SOX2 (Shen,

Song, Liu, Zhang, & Wei Song, 2018; H. Wang et al., 2003). Similarly, IGFBP-3 expression correlated with glioma grade and tumor histology, with siRNA-mediated

IGFBP-3 knockdown suppressing tumor growth and longer survival of tumor-bearing mice

(C. H. Chen et al., 2019). Notably, the secreted factors profiled in the current work are

111 limited in scope to those available for analysis by the angiogenesis array. U87, U87vIII, and GBM6 cells likely secrete a milieu of other proteins stimulating cell invasion and extracellular matrix remodeling. Indeed, Sangar et al. previously demonstrated the striking effect of EGFRvIII expression in U87 GBM cell secretion of pro-invasion proteins, observing that EGFRvIII-expressing cells secreted a profile of proteins enriched in the promotion of epithelial-mesenchymal transition (e.g. CD44), angiogenesis (e.g. TIMP1), proteolysis (e.g. cathepsins, MMPs), and oxidative stress (Sangar et al., 2014).

The presence of EGFRvIII on U87 cells additionally influenced differential protein secretion of ECM-remodeling proteins by murine microglia (Figure 6.1C), leading to greater microglia secretion of (among others) MMP-3, MMP-9, IGFBP-2, and CCL2.

Although HMC3 cells did not respond significantly in their secretion of soluble factors in response to U87 or U87vIII CM (Figure 6.2), they exhibited a striking response to incubation in GBM6 CM (Figure 6.4). Secretion of all IGFBPs analyzed was significantly upregulated, as were chemotactic factors including HGF, ECM-remodeling proteins including MMP-9 and VEGF, and proteins implicated in cell growth and invasion such as

EGF and TGF-β1. Of note, the most highly induced protein secreted was CCL2. Levels of the C-C chemokine CCL2, also known as monocyte chemoattractant protein-1 (MCP-

1), are elevated in GBM and drive macrophage infiltration as well as recruitment of regulatory T cells to promote immunosuppression (Chang et al., 2016). Recent work demonstrated that EGFR and EGFRvIII cooperatively induce tumor macrophage infiltration via upregulation of GBM cell CCL2 expression (An et al., 2018). Receptor tyrosine kinase inhibition with lapatinib decreased CCL2 expression in GBM6 and GBM39

EGFRvIII-expressing patient-derived xenografts. Notably, CCL2 expression and

112 macrophage infiltration were both significantly enhanced by co-expression of both EGFR and EGFRvIII in U87 cells, compared to when EGFR or EGFRvIII were expressed independently, indicating a synergistic signaling effect between wild-type and mutant receptors.

The presence of EGFRvIII additionally led to differential responses in intracellular glutathione redox balance in U87 cells incubated in HMC3 CM, as the mitochondrial glutathione pool became more oxidized in U87, but not U87vIII cells (Figure 6.6). It is conceivable that this oxidation of glutathione in U87 cells resulted from decreased

- expression of system xc (Figure 6.5), which in turn would hamper cellular uptake of cystine and thus lead to decreased rates of glutathione synthesis. Notably, U87vIII cells did not display an increase in oxidized glutathione, coinciding with stable expression of

- system xc in HMC3 CM (Figure 6.5-6.6). Previous work has elegantly demonstrated that

EGFR and EGFRvIII stabilize xCT cell surface expression on GBM cells via a kinase- independent, physical interaction between the EGFR intracellular domain and transmembrane domains of xCT (Tsuchihashi et al., 2016). Our results are also corroborated by evidence that ectopic overexpression of EGFR in U87 cells increases surface xCT expression and total glutathione content (Tsuchihashi et al., 2016).

Numerous studies have detailed the stimulatory effect of microglia on GBM cell proliferation and invasion (Bettinger et al., 2002; Coniglio et al., 2012; Coniglio & Segall,

2013; Hambardzumyan et al., 2016; Poon, Sarkar, Yong, & Kelly, 2017; Ye et al., 2012); however, experiments described here making use of the HMC3 human microglia cell line were unable to detect significant differences in these outcomes. Compared to primary neonatal murine microglia, immortalized HMC3 cells constitutively proliferate, which may

113 have confounded the ability to detect changes in proliferative index in response to U87 or

U87vIII CM. However, in a broader sense, microglia harbor distinctive ontogeny, morphology, gene expression patterns, and functional characteristics that can vary widely depending on cellular context and upon the model of microglia used (Butovsky et al.,

2014; Gosselin et al., 2017). Previous studies demonstrate disparities between expression of microglia signature genes among primary microglia and immortalized microglia cell lines, including HMC3, on which there are conflicting reports regarding its expression of microglia signature genes such as C1Q, P2RY12, and TMEM119 (Butovsky et al., 2014; Dello Russo et al., 2018; Rawat & Spector, 2017). HMC3 cells used in this study stained strongly for IBA1 but weakly for TMEM119 (not shown), yet also exhibited a strong secretory response when stimulated with primary GBM patient-derived cells

(Figure 6.5). Future work may benefit from incorporating more representative, primary models of disease, such as the primary murine microglia and GBM6 samples this study benefitted from.

Interfering with microglia activation, attempting to inhibit the tumor-promoting effects of TAMs, and depleting the TAM compartments within GBM tumor microenvironments have received attention as potential strategies to decrease tumor burden. While therapies such as PLX3397 and BLZ945 (CSF-1R inhibitors) targeting the TAM compartments of the GBM tumor microenvironment show promising preclinical results, long-term regression remains a challenge as a subset of tumors invariably acquires resistance to

CSF-1R inhibition in a manner that is dependent upon the tumor microenvironment itself

(Quail et al., 2016). A phase II multicenter study demonstrated limited efficacy of PLX3397 as a single agent in increasing progression-free survival of patients with recurrent GBM

114

(Butowski et al., 2016). A phase I clinical trial using the antibiotic minocycline to impair microglia activation as an adjuvant therapy alongside chemoradiation in patients with newly diagnosed GBM is currently underway (NCT02272270). Enhancing our understanding of the complex milieu of signals tumor cells receive from neighboring malignant, cancer-associated, and co-opted endothelial and tissue-resident cells, as well as how they sense and respond to gradients in extracellular metabolic and biophysical cues, may reveal new approaches to managing tumor burden and improving patient care.

6.5 CONCLUSIONS

The current work demonstrated a modulatory effect of EGFRvIII on i) U87 GBM cell secretion of pro-angiogenic soluble factors, ii) microglia secretion of ECM-remodeling chemokines and cytokines, and iii) subcellular glutathione redox balance in U87 cells exposed to HMC3 secreted factors. Incorporation of additional patient-derived GBM cell specimens in addition to GBM6, developing culture platforms including endothelial cells to more truly measure angiogenesis as a result of GBM-microglia paracrine signaling, as well as exploring the mechanism by which microglia-secreted factors modulate system

- xc expression, will add to the current understanding of microglia-GBM crosstalk.

115

CHAPTER 7: OVERALL CONCLUSIONS AND FUTURE DIRECTIONS

The work described herein provides evidence supporting the relevance and functional capacity of the mitochondrial protein CHCHD2 in mediating GBM cell malignant characteristics as they navigate transitions in the tumor microenvironment. Previous studies have ascribed a variety of functions to CHCHD2, including regulation of mitochondrial respiration and redox status (Aras et al., 2015; Baughman et al., 2009), cell migration (Seo et al., 2010), regulation of pluripotent stem cell differentiation into neuroectoderm (Zhu et al., 2016), negative regulation of apoptosis (Y. Liu et al., 2015), and hypoxia-sensitive transcription factor activity (Aras et al., 2013). Inspired by these findings, the current work is the first to describe the role of CHCHD2 in GBM cells.

Leveraging CRISPR-Cas9 techniques, the work described demonstrates a role for

CHCHD2 in mediating mitochondrial respiration and glutathione redox status, cell sensitivity to select cytotoxic agents, and cell proliferation and invasion in both standard culture conditions and in response to hypoxia, highlighting CHCHD2 as a functional node in GBM cells expressing the frequently observed mutant EGFRvIII. It should be noted that

GBM tumors of various subtypes indeed consist of a heterogenous cell population, which harbor a variety of cell surface receptor tyrosine kinases with various intra-tumor expression patterns, and are not limited by EGFR or EGFRvIII expression. Profiling of total CHCHD2 protein levels in a panel of patient-derived GBM specimens, which themselves are comprised of both tumor and parenchymal cells, revealed that CHCHD2 expression was generally greatest in GBM6, which expresses EGFRvIII, is defined as a classical subtype tumor, and is the most invasive specimen analyzed. A contextualized analysis of CHCHD2 function in an in vivo setting will be critical to understanding how

116 tumor phenotype is mediated by CHCHD2 in the context of other GBM hallmarks, such as angiogenesis and suppression of anti-tumor immunity, particularly in light of the paucity of effective immunotherapies against gliomas to date.

Previous studies had demonstrated the ability of CHCHD2 to act as a hypoxia- sensitive transcription factor (Aras et al., 2013). Plasticity in the face of hypoxia provides a key growth advantage to GBM tumors. Indeed, CHCHD2 KO U87vIII cells displayed a diminished ability to proliferate and invade in hypoxic conditions. Motivated by the literature and results from Chapter 4, Chapter 5 provides an analysis of CHCHD2 subcellular distribution, specifically between mitochondria and the nucleus, as a function of a variety of imposed stressors. Chronic hypoxia, chronic glutathione redox stress, and acute serum starvation all resulted in nuclear accumulation of CHCHD2. Furthermore, it is likely that reductive unfolding and translocation of mitochondrially localized CHCHD2 is the mechanism by which immediate nuclear accumulation of CHCHD2 occurs in the face of imposed stress. Such a mechanism wound underscore the importance of intra- molecular disulfide bonds which perform not just structural, but also regulatory roles in protein function. Subcellular fractionation followed by CHCHD2 redox western blots of

CHCHD2 in mitochondrial and nuclear fractions in response to imposed stressors would provide an additional level of mechanistic understanding. However, It is still possible that, over longer periods of metabolic stress (e.g. chronic hypoxia), newly translated CHCHD2 is preferentially shuttled to the nucleus versus mitochondria. Additionally, discrete details of CHCHD2 protein interactions in various subcellular compartments in GBM cells during its translocation remain to be described. Previous literature has highlighted the ability of

CHCHD2 to physically interact with CHCHD10 and p32 in mitochondria, and bind to RBPJ

117 to transactivate expression of nuclear target genes (Aras et al., 2013; Imai et al., 2019;

Purandare et al., 2018). Notably, work by Xie et al. has implicated RBPJ as critical for maintenance of brain tumor initiating cell viability and stemness (Q. Xie et al., 2016).

Given the studies implicating CHCHD2 in neuroectodermal differentiation capacities of human embryonic stem cells (Zhu et al., 2016), its association with the migratory capabilities of neural stem cells in patients with lissencephaly (Shimojima et al., 2015), as well as recent work highlighting neural stem cells as a GBM cell of origin (J. H. Lee et al.,

2018), it is tempting to speculate that the CHCHD2-RBPJ interaction may play a central role in maintaining glioma stem cells and tumor progression. Future studies should investigate whether CHCHD2 interacts with RBPJ in GBM cells, as well as determine the suite of genes CHCHD2 may regulate, which could reveal novel gene targets to further explain the contribution of nuclear CHCHD2 to GBM malignancy.

Tumor-associated microglia represent another key component of the GBM tumor microenvironment. Chapter 6 explored GBM-microglia crosstalk by employing a variety of cell models, including U87 and U87vIII GBM cell lines, GBM6 primary patient-derived specimens, the HMC3 microglia cell line, and primary neonatal murine microglia. Overall, it was found that the expression of EGFRvIII on GBM cells influenced the secretion of chemokines, cytokines, and soluble factors involved in angiogenesis released by microglia activated by GBM-conditioned media. Additionally, EGFRvIII expression stabilized the glutamate-cystine antiporter xCT on U87 cells incubated in HMC3- conditioned media, likely contributing to enhanced defense against oxidative stress suffered by U87 cells lacking EGFRvIII. Future work would largely benefit from

118 incorporating more representative, primary models of disease, as well as translating work to an in vivo setting.

Collectively, the studies conducted herein are all motivated by the understanding that

GBM tumors (and truly, all tumors) are shaped by dynamic crosstalk with their surrounding microenvironment. Indeed, tumor progression can be considered an accelerated microcosm of Darwinian evolution, in which cues from the tumor microenvironment inform the selection of increasingly malignant cell populations that compete, survive, and reproduce, altogether contributing to disease progression.

Advancing our understanding of the unique aspects of the GBM tumor microenvironment, as well as how tumors sense and respond to these microenvironmental cues, will be key in combatting a disease marked by highly plastic and adaptive cells. In turn, the tumor microenvironment itself is influenced by patient physiology. For instance, a number of studies have demonstrated the impact of chronic psychological distress on immunosuppression and anti-tumor immunity (Appendix A). Therefore, an appreciation for, and continued understanding of the tight, dynamic relationships between tumor and its surrounding microenvironment will be crucial in identifying aspects of tumor biology that may be leveraged to improve patient well-being and prognosis.

119

References

Ajami, B., J. L. Bennett, C. Krieger, W. Tetzlaff, and F. M. Rossi. 2007. Local self-renewal can sustain CNS microglia maintenance and function throughout adult life, Nat Neurosci, 10:1538-43. Alieva, M., V. Leidgens, M. J. Riemenschneider, C. A. Klein, P. Hau, and J. van Rheenen. 2019. Intravital imaging of glioma border morphology reveals distinctive cellular dynamics and contribution to tumor cell invasion, Sci Rep, 9:2054. Alliot, F., I. Godin, and B. Pessac. 1999. Microglia derive from progenitors, originating from the yolk sac, and which proliferate in the brain, Brain Res Dev Brain Res, 117: 145-52. Amidei, C. 2018. Symptom-based interventions to promote quality survivorship, Neuro Oncol, 20: vii27-vii39. An, Z., C. B. Knobbe-Thomsen, X. Wan, Q. W. Fan, G. Reifenberger, and W. A. Weiss. 2018. EGFR cooperates with EGFRvIII to recruit macrophages in glioblastoma, Cancer Res, 78(24):6785-6794. Antonia, S. J., J. A. Lopez-Martin, J. Bendell, P. A. Ott, M. Taylor, J. P. Eder, D. Jager, M. C. Pietanza, D. T. Le, F. de Braud, M. A. Morse, P. A. Ascierto, L. Horn, A. Amin, R. N. Pillai, J. Evans, I. Chau, P. Bono, A. Atmaca, P. Sharma, C. T. Harbison, C. S. Lin, O. Christensen, and E. Calvo. 2016. Nivolumab alone and nivolumab plus ipilimumab in recurrent small-cell lung cancer (CheckMate 032): a multicentre, open-label, phase 1/2 trial, Lancet Oncol, 17:883-95. Antonios, J. P., H. Soto, R. G. Everson, D. Moughon, J. R. Orpilla, N. P. Shin, S. Sedighim, J. Treger, S. Odesa, A. Tucker, W. H. Yong, G. Li, T. F. Cloughesy, L. M. Liau, and R. M. Prins. 2017. Immunosuppressive tumor-infiltrating myeloid cells mediate adaptive immune resistance via a PD-1/PD-L1 mechanism in glioblastoma, Neuro Oncol, 19:796-807. Aras, S., H. Arrabi, N. Purandare, M. Huttemann, J. Kamholz, S. Zuchner, and L. I. Grossman. 2017. Abl2 kinase phosphorylates bi-organellar regulator MNRR1 in mitochondria, stimulating respiration, Biochim Biophys Acta, 1864:440-48. Aras, S., M. Bai, I. Lee, R. Springett, M. Huttemann, and L. I. Grossman. 2015. MNRR1 (formerly CHCHD2) is a bi-organellar regulator of mitochondrial metabolism, Mitochondrion, 20:43-51. Aras, S., O. Pak, N. Sommer, R. Finley, Jr., M. Huttemann, N. Weissmann, and L. I. Grossman. 2013. Oxygen-dependent expression of cytochrome c oxidase subunit 4-2 gene expression is mediated by transcription factors RBPJ, CXXC5 and CHCHD2, Nucleic Acids Res, 41:2255-66. Archibald, J. M. 2015. Endosymbiosis and eukaryotic cell evolution, Curr Biol, 25:R911- 21. Assad Kahn, S., S. L. Costa, S. Gholamin, R. T. Nitta, L. G. Dubois, M. Feve, M. Zeniou, P. L. Coelho, E. El-Habr, J. Cadusseau, P. Varlet, S. S. Mitra, B. Devaux, M. C. Kilhoffer, S. H. Cheshier, V. Moura-Neto, J. Haiech, M. P. Junier, and H. Chneiweiss. 2016. The anti-hypertensive drug prazosin inhibits glioblastoma growth via the PKC-delta-dependent inhibition of the AKT pathway, EMBO Mol Med, 8:511-26.

120

Bar, E. E., A. Lin, V. Mahairaki, W. Matsui, and C. G. Eberhart. 2010. Hypoxia increases the expression of stem-cell markers and promotes clonogenicity in glioblastoma neurospheres, Am J Pathol, 177:1491-502. Barnholtz-Sloan, J. S., Q. T. Ostrom, and D. Cote. 2018. Epidemiology of brain tumors, Neurol Clin, 36:395-419. Baughman, J. M., R. Nilsson, V. M. Gohil, D. H. Arlow, Z. Gauhar, and V. K. Mootha. 2009. A computational screen for regulators of oxidative phosphorylation implicates SLIRP in mitochondrial RNA homeostasis, PLoS Genet, 5:e1000590. Bellmunt, J., R. de Wit, D. Vaughn, Y. Fradet, J. Lee, L. Fong, N. Vogelzang, M. Climent, D. Petrylak, T. Choueiri, A. Necchi, W. Gerritsen, H. Gurney, D. Quinn, S. Culine, C. Sternberg, Y. Mai, C. Poehlein, R. Perini, and D. Bajorin. 2017. Pembrolizumab as second-line therapy for advanced urothelial carcinoma, N Engl J Med, 376:1015-26. Benej, M., X. Hong, S. Vibhute, S. Scott, J. Wu, E. Graves, Q. T. Le, A. C. Koong, A. J. Giaccia, B. Yu, C. S. Chen, I. Papandreou, and N. C. Denko. 2018. Papaverine and its derivatives radiosensitize solid tumors by inhibiting mitochondrial metabolism, Proc Natl Acad Sci U S A, 115:10756-61. Bennett, M. L., F. C. Bennett, S. A. Liddelow, B. Ajami, J. L. Zamanian, N. B. Fernhoff, S. B. Mulinyawe, C. J. Bohlen, A. Adil, A. Tucker, I. L. Weissman, E. F. Chang, G. Li, G. A. Grant, M. G. Hayden Gephart, and B. A. Barres. 2016. New tools for studying microglia in the mouse and human CNS, Proc Natl Acad Sci U S A, 113:E1738- 46. Bettinger, I., S. Thanos, and W. Paulus. 2002. Microglia promote glioma migration, Acta Neuropathol, 103:351-5. Bhutia, Y. D., E. Babu, S. Ramachandran, and V. Ganapathy. 2015. Amino acid transporters in cancer and their relevance to "glutamine addiction": novel targets for the design of a new class of anticancer drugs, Cancer Res, 75:1782-8. Binder, D. C., A. A. Davis, and D. A. Wainwright. 2016. Immunotherapy for cancer in the central nervous system: current and future directions, Oncoimmunology, 5: e1082027. Bloch, Orin, Qian Shi, S. Keith Anderson, Michael Knopp, Jeffrey Raizer, Jennifer Clarke, Allen Waziri, Howard Colman, Jeffrey Bruce, Jeffrey J. Olson, John Schwerkoske, Andrew Parsa, Joon Uhm, Priya Kumthekar, Evanthia Galanis, and Ian Parney. 2017. ATIM-14. ALLIANCE A071101: A phase II randomized trial comparint the efficacy of heat shock protein peptide complex-96 (HSPPC-96) vaccine given with bevacizumab versus bevacizumab alone in the treatment of surgically resectable recurrent glioblastoma, Neuro-Oncol, 19:vi29-vi29. Bodnar, C. N., J. M. Morganti, and A. D. Bachstetter. 2018. Depression following a traumatic brain injury: uncovering cytokine dysregulation as a pathogenic mechanism, Neural Regen Res, 13:1693-704. Bowman, R. L., Q. Wang, A. Carro, R. G. Verhaak, and M. Squatrito. 2017. GlioVis data portal for visualization and analysis of brain tumor expression datasets, Neuro- Oncol, 19:139-41. Brat, D. J., A. A. Castellano-Sanchez, S. B. Hunter, M. Pecot, C. Cohen, E. H. Hammond, S. N. Devi, B. Kaur, and E. G. Van Meir. 2004. Pseudopalisades in glioblastoma

121

are hypoxic, express extracellular matrix proteases, and are formed by an actively migrating cell population, Cancer Res, 64:920-7. Brennan, C. W., R. G. Verhaak, A. McKenna, B. Campos, H. Noushmehr, S. R. Salama, S. Zheng, D. Chakravarty, J. Z. Sanborn, S. H. Berman, R. Beroukhim, B. Bernard, C. J. Wu, G. Genovese, I. Shmulevich, J. Barnholtz-Sloan, L. Zou, R. Vegesna, S. A. Shukla, G. Ciriello, W. K. Yung, W. Zhang, C. Sougnez, T. Mikkelsen, K. Aldape, D. D. Bigner, E. G. Van Meir, M. Prados, A. Sloan, K. L. Black, J. Eschbacher, G. Finocchiaro, W. Friedman, D. W. Andrews, A. Guha, M. Iacocca, B. P. O'Neill, G. Foltz, J. Myers, D. J. Weisenberger, R. Penny, R. Kucherlapati, C. M. Perou, D. N. Hayes, R. Gibbs, M. Marra, G. B. Mills, E. Lander, P. Spellman, R. Wilson, C. Sander, J. Weinstein, M. Meyerson, S. Gabriel, P. W. Laird, D. Haussler, G. Getz, L. Chin, and Tcga Research Network. 2013. The somatic genomic landscape of glioblastoma, Cell, 155:462-77. Bruns, I., B. Sauer, M. C. Burger, J. Eriksson, U. Hofmann, Y. Braun, P. N. Harter, A. L. Luger, M. W. Ronellenfitsch, J. P. Steinbach, and J. Rieger. 2019. Disruption of peroxisome proliferator-activated receptor gamma coactivator (PGC)-1alpha reverts key features of the neoplastic phenotype of glioma cells, J Biol Chem, 294:3037-50. Bucsek, M. J., G. Qiao, C. R. MacDonald, T. Giridharan, L. Evans, B. Niedzwecki, H. Liu, K. M. Kokolus, J. W. Eng, M. N. Messmer, K. Attwood, S. I. Abrams, B. L. Hylander, and E. A. Repasky. 2017. Beta-adrenergic signaling in mice housed at standard temperatures suppresses an effector phenotype in CD8(+) T cells and undermines checkpoint inhibitor therapy, Cancer Res, 77:5639-51. Butovsky, O., M. P. Jedrychowski, C. S. Moore, R. Cialic, A. J. Lanser, G. Gabriely, T. Koeglsperger, B. Dake, P. M. Wu, C. E. Doykan, Z. Fanek, L. Liu, Z. Chen, J. D. Rothstein, R. M. Ransohoff, S. P. Gygi, J. P. Antel, and H. L. Weiner. 2014. Identification of a unique TGF-beta-dependent molecular and functional signature in microglia, Nat Neurosci, 17:131-43. Butowski, N., H. Colman, J. F. De Groot, A. M. Omuro, L. Nayak, P. Y. Wen, T. F. Cloughesy, A. Marimuthu, S. Haidar, A. Perry, J. Huse, J. Phillips, B. L. West, K. B. Nolop, H. H. Hsu, K. L. Ligon, A. M. Molinaro, and M. Prados. 2016. Orally administered colony stimulating factor 1 receptor inhibitor PLX3397 in recurrent glioblastoma: an Ivy Foundation Early Phase Clinical Trials Consortium phase II study, Neuro Oncol, 18:557-64. Byrne, M. B., M. T. Leslie, H. S. Patel, H. R. Gaskins, and P. J. A. Kenis. 2017. Design considerations for open-well microfluidic platforms for hypoxic cell studies, Biomicrofluidics, 11:054116. Cardamone, M. D., B. Tanasa, C. T. Cederquist, J. Huang, K. Mahdaviani, W. Li, M. G. Rosenfeld, M. Liesa, and V. Perissi. 2018. Mitochondrial retrograde signaling in mammals is mediated by the transcriptional cofactor GPS2 via direct mitochondria- to-nucleus translocation, Mol Cell, 69:757-72.e7. Cardwell, C. R., A. Pottegard, E. Vaes, H. Garmo, L. J. Murray, C. Brown, P. A. Vissers, M. O'Rorke, K. Visvanathan, D. Cronin-Fenton, H. De Schutter, M. Lambe, D. G. Powe, M. P. van Herk-Sukel, A. Gavin, S. Friis, L. Sharp, and K. Bennett. 2016. Propranolol and survival from breast cancer: a pooled analysis of European breast cancer cohorts, Breast Cancer Res, 18:119.

122

Catt, S., A. Chalmers, and L. Fallowfield. 2008. Psychosocial and supportive-care needs in high-grade glioma, Lancet Oncol, 9:884-91. Caudill, Jonathan S., Paul D. Brown, Jane H. Cerhan, and Teresa A. Rummans. 2011. Selective serotonin reuptake inhibitors, glioblastoma multiforme, and impact on toxicities and overall survival: the Mayo Clinic experience, Am J Clin Oncol, 34(4):385-7. Cavalli, L. R., M. Varella-Garcia, and B. C. Liang. 1997. Diminished tumorigenic phenotype after depletion of mitochondrial DNA, Cell Growth Differ, 8:1189-98. Ceccarelli, M., F. P. Barthel, T. M. Malta, T. S. Sabedot, S. R. Salama, B. A. Murray, O. Morozova, Y. Newton, A. Radenbaugh, S. M. Pagnotta, S. Anjum, J. Wang, G. Manyam, P. Zoppoli, S. Ling, A. A. Rao, M. Grifford, A. D. Cherniack, H. Zhang, L. Poisson, C. G. Carlotti, Jr., D. P. Tirapelli, A. Rao, T. Mikkelsen, C. C. Lau, W. K. Yung, R. Rabadan, J. Huse, D. J. Brat, N. L. Lehman, J. S. Barnholtz-Sloan, S. Zheng, K. Hess, G. Rao, M. Meyerson, R. Beroukhim, L. Cooper, R. Akbani, M. Wrensch, D. Haussler, K. D. Aldape, P. W. Laird, D. H. Gutmann, Tcga Research Network, H. Noushmehr, A. Iavarone, and R. G. Verhaak. 2016. Molecular profiling reveals biologically discrete subsets and pathways of progression in diffuse glioma, Cell, 164:550-63. Cerami, E., J. Gao, U. Dogrusoz, B. E. Gross, S. O. Sumer, B. A. Aksoy, A. Jacobsen, C. J. Byrne, M. L. Heuer, E. Larsson, Y. Antipin, B. Reva, A. P. Goldberg, C. Sander, and N. Schultz. 2012. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data, Cancer Discov, 2:401-4. Chaitanya, G. V., A. J. Steven, and P. P. Babu. 2010. PARP-1 cleavage fragments: signatures of cell-death proteases in neurodegeneration, Cell Commun Signal, 8:31. Chang, A. L., J. Miska, D. A. Wainwright, M. Dey, C. V. Rivetta, D. Yu, D. Kanojia, K. C. Pituch, J. Qiao, P. Pytel, Y. Han, M. Wu, L. Zhang, C. M. Horbinski, A. U. Ahmed, and M. S. Lesniak. 2016. CCL2 Produced by the glioma microenvironment is essential for the recruitment of regulatory T cells and myeloid-derived suppressor cells, Cancer Res, 76:5671-82. Charles, N. A., E. C. Holland, R. Gilbertson, R. Glass, and H. Kettenmann. 2011. The brain tumor microenvironment, Glia, 59:1169-80. Chen, C. H., P. Y. Chen, Y. Y. Lin, L. Y. Feng, S. H. Chen, C. Y. Chen, Y. C. Huang, C. Y. Huang, S. M. Jung, L. Y. Chen, and K. C. Wei. 2019. Suppression of tumor growth via IGFBP3 depletion as a potential treatment in glioma, J Neurosurg:1-12. Chen, J. E., J. Lumibao, A. Blazek, H. R. Gaskins, and B. Harley. 2018. Hypoxia activates enhanced invasive potential and endogenous hyaluronic acid production by glioblastoma cells, Biomater Sci, 6:854-62. Chen, J. E., S. Pedron, and B. A. C. Harley. 2017. The combined influence of hydrogel stiffness and matrix-bound hyaluronic acid content on glioblastoma invasion, Macromol Biosci, 17(8). Chen, Jee-Wei E., Sara Pedron, Peter Shyu, Yuhang Hu, Jann N. Sarkaria, and Brendan A. C. Harley. 2018. Influence of hyaluronic acid transitions in tumor microenvironment on glioblastoma malignancy and invasive behavior, Front Mater, 5.

123

Chen, L., X. Li, L. Liu, B. Yu, Y. Xue, and Y. Liu. 2015. Erastin sensitizes glioblastoma cells to temozolomide by restraining xCT and cystathionine-gamma-lyase function, Oncol Rep, 33:1465-74. Chen, R., P. L. Zinzani, M. A. Fanale, P. Armand, N. A. Johnson, P. Brice, J. Radford, V. Ribrag, D. Molin, T. P. Vassilakopoulos, A. Tomita, B. von Tresckow, M. A. Shipp, Y. Zhang, A. D. Ricart, A. Balakumaran, and C. H. Moskowitz. 2017. Phase II study of the efficacy and safety of pembrolizumab for relapsed/refractory classic Hodgkin lymphoma, J Clin Oncol, 35:2125-32. Chen, Vincent Chin‐Hung, Yi‐Hsien Hsieh, Li‐Jeng Chen, Tsai‐Ching Hsu, and Bor‐Show Tzang. 2018. Escitalopram oxalate induces apoptosis in U‐87MG cells and autophagy in GBM8401 cells, J Cell Mol Med, 22:1167-78. Chen, W. L., C. C. Wang, Y. J. Lin, C. P. Wu, and C. H. Hsieh. 2015. Cycling hypoxia induces chemoresistance through the activation of reactive oxygen species- mediated B-cell lymphoma extra-long pathway in glioblastoma multiforme, J Transl Med, 13:389. Cheng, Q., D. Qu, Z. Lu, and L. Zhang. 2019. Knockdown of CHCHD2 inhibits migration and angiogenesis of human renal cell carcinoma: a potential molecular marker for treatment of RCC, Oncol Lett, 17:765-72. Chesik, D., J. De Keyser, and N. Wilczak. 2004. Involvement of insulin-like growth factor binding protein-2 in activated microglia as assessed in post mortem human brain, Neurosci Lett, 362:14-6. Chinnaiyan, P., M. Won, P. Y. Wen, A. M. Rojiani, M. Werner-Wasik, H. A. Shih, L. S. Ashby, H. H. Michael Yu, V. W. Stieber, S. C. Malone, J. B. Fiveash, N. A. Mohile, M. S. Ahluwalia, M. M. Wendland, P. J. Stella, A. Y. Kee, and M. P. Mehta. 2018. A randomized phase II study of everolimus in combination with chemoradiation in newly diagnosed glioblastoma: results of NRG Oncology RTOG 0913, Neuro Oncol, 20:666-73. Chiorean, R., I. Berindan-Neagoe, C. Braicu, I. S. Florian, D. Leucuta, D. Crisan, and V. Cernea. 2014. Quantitative expression of serum biomarkers involved in angiogenesis and inflammation, in patients with glioblastoma multiforme: correlations with clinical data, Cancer Biomark, 14:185-94. Choi, J., B. Stradmann-Bellinghausen, E. Yakubov, N. E. Savaskan, and A. Regnier- Vigouroux. 2015. Glioblastoma cells induce differential glutamatergic gene expressions in human tumor-associated microglia/macrophages and monocyte- derived macrophages, Cancer Biol Ther, 16:1205-13. Coniglio, S. J., E. Eugenin, K. Dobrenis, E. R. Stanley, B. L. West, M. H. Symons, and J. E. Segall. 2012. Microglial stimulation of glioblastoma invasion involves epidermal growth factor receptor (EGFR) and colony stimulating factor 1 receptor (CSF-1R) signaling, Mol Med, 18:519-27. Coniglio, S. J., and J. E. Segall. 2013. Review: molecular mechanism of microglia stimulated glioblastoma invasion, Matrix Biol, 32:372-80. Connolly, N. M. C., P. Theurey, V. Adam-Vizi, N. G. Bazan, P. Bernardi, J. P. Bolanos, C. Culmsee, V. L. Dawson, M. Deshmukh, M. R. Duchen, H. Dussmann, G. Fiskum, M. F. Galindo, G. E. Hardingham, J. M. Hardwick, M. B. Jekabsons, E. A. Jonas, J. Jordan, S. A. Lipton, G. Manfredi, M. P. Mattson, B. McLaughlin, A. Methner, A. N. Murphy, M. P. Murphy, D. G. Nicholls, B. M. Polster, T. Pozzan, R.

124

Rizzuto, J. Satrustegui, R. S. Slack, R. A. Swanson, R. H. Swerdlow, Y. Will, Z. Ying, A. Joselin, A. Gioran, C. Moreira Pinho, O. Watters, M. Salvucci, I. Llorente- Folch, D. S. Park, D. Bano, M. Ankarcrona, P. Pizzo, and J. H. M. Prehn. 2018. Guidelines on experimental methods to assess mitochondrial dysfunction in cellular models of neurodegenerative diseases, Cell Death Differ, 25:542-72. Courtney, K. D., D. Bezwada, T. Mashimo, K. Pichumani, V. Vemireddy, A. M. Funk, J. Wimberly, S. S. McNeil, P. Kapur, Y. Lotan, V. Margulis, J. A. Cadeddu, I. Pedrosa, R. J. DeBerardinis, C. R. Malloy, R. M. Bachoo, and E. A. Maher. 2018. Isotope tracing of human clear cell renal cell carcinomas demonstrates suppressed glucose oxidation in vivo, Cell Metab, 28:793-800.e2. Cuadrado, A., A. I. Rojo, G. Wells, J. D. Hayes, S. P. Cousin, W. L. Rumsey, O. C. Attucks, S. Franklin, A. L. Levonen, T. W. Kensler, and A. T. Dinkova-Kostova. 2019. Therapeutic targeting of the NRF2 and KEAP1 partnership in chronic diseases, Nat Rev Drug Discov, 18(4):295-317. Cuddapah, V. A., S. Robel, S. Watkins, and H. Sontheimer. 2014. A neurocentric perspective on glioma invasion, Nat Rev Neurosci, 15:455-65. da Cunha, F. M., N. Q. Torelli, and A. J. Kowaltowski. 2015. Mitochondrial Retrograde Signaling: Triggers, Pathways, and Outcomes, Oxid Med Cell Longev, 2015:482582. da Fonseca, A. C., and B. Badie. 2013. Microglia and macrophages in malignant gliomas: recent discoveries and implications for promising therapies, Clin Dev Immunol, 2013:264124. Darefsky, Amy S., Joseph T. King, and Robert Dubrow. 2012. Adult glioblastoma multiforme survival in the temozolomide era: a population-based analysis of surveillance, epidemiology, and end results registries, Cancer, 118:2163-72. de Groot, J., and H. Sontheimer. 2011. Glutamate and the biology of gliomas, Glia, 59:1181-9. Deblois, G., J. St-Pierre, and V. Giguere. 2013. The PGC-1/ERR signaling axis in cancer, Oncogene, 32:3483-90. Dello Russo, C., N. Cappoli, I. Coletta, D. Mezzogori, F. Paciello, G. Pozzoli, P. Navarra, and A. Battaglia. 2018. The human microglial HMC3 cell line: where do we stand? A systematic literature review, J Neuroinflammation, 15:259. Dewhirst, M. W. 2018. A potential solution for eliminating hypoxia as a cause for radioresistance, Proc Natl Acad Sci U S A, 115(42):10548-10550. Dhabhar, F. S., and B. S. McEwen. 1997. Acute stress enhances while chronic stress suppresses cell-mediated immunity in vivo: a potential role for leukocyte trafficking, Brain Behav Immun, 11:286-306. Diez Valle, R., V. Becerra Castro, M. Marigil Sanchez, J. Gallego Perez-Larraya, J. M. Nunez-Cordoba, and S. Tejada Solis. 2018. Results of a policy of fast tapering of steroids after resection surgery in glioblastoma, World Neurosurg, 109:e845-e52. Divakaruni, A. S., A. Paradyse, D. A. Ferrick, A. N. Murphy, and M. Jastroch. 2014. Analysis and interpretation of microplate-based oxygen consumption and pH data, Methods Enzymol, 547:309-54. Dubrow, Robert, Amy S. Darefsky, Daniel I. Jacobs, Lesley S. Park, Michal G. Rose, Maxwell S. H. Laurans, and Joseph T. King. 2013. Time trends in glioblastoma multiforme survival: the role of temozolomide, Neuro-Oncology, 15:1750-61.

125

Dutoit, V., D. Migliorini, P. Y. Dietrich, and P. R. Walker. 2016. Immunotherapy of malignant tumors in the brain: how different from other sites?, Front Oncol, 6:256. Edelstein, K., L. Coate, C. Massey, N. C. Jewitt, W. P. Mason, and G. M. Devins. 2016. Illness intrusiveness and subjective well-being in patients with glioblastoma, J Neurooncol, 126:127-35. Eisenberg-Bord, M., and M. Schuldiner. 2017. Ground control to major TOM: mitochondria-nucleus communication, FEBS J, 284:196-210. Eng, J. W., K. M. Kokolus, C. B. Reed, B. L. Hylander, W. W. Ma, and E. A. Repasky. 2014. A nervous tumor microenvironment: the impact of adrenergic stress on cancer cells, immunosuppression, and immunotherapeutic response, Cancer Immunol Immunother, 63:1115-28. Engelhardt, B., and R. M. Ransohoff. 2012. Capture, crawl, cross: the T cell code to breach the blood-brain barriers, Trends Immunol, 33:579-89. Engelhardt, B., P. Vajkoczy, and R. O. Weller. 2017. The movers and shapers in immune privilege of the CNS, Nat Immunol, 18:123-31. Esteller, M., J. Garcia-Foncillas, E. Andion, S. N. Goodman, O. F. Hidalgo, V. Vanaclocha, S. B. Baylin, and J. G. Herman. 2000. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents, N Engl J Med, 343:1350-4. Estrela, J. M., A. Ortega, and E. Obrador. 2006. Glutathione in cancer biology and therapy, Crit Rev Clin Lab Sci, 43:143-81. Fabian, D., M. D. P. Guillermo Prieto Eibl, I. Alnahhas, N. Sebastian, P. Giglio, V. Puduvalli, J. Gonzalez, and J. D. Palmer. 2019. Treatment of glioblastoma (GBM) with the addition of tumor-treating fields (TTF): a review, Cancers (Basel), 11. Feinstein, Anthony, Sandra Magalhaes, Jean-Francois Richard, Blair Audet, and Craig Moore. 2014. The link between multiple sclerosis and depression, Nat Rev Neurol, 10(9): 507-17. Fenn, A. M., J. C. Gensel, Y. Huang, P. G. Popovich, J. Lifshitz, and J. P. Godbout. 2014. Immune activation promotes depression 1 month after diffuse brain injury: a role for primed microglia, Biol Psychiatry, 76:575-84. Fidoamore, A., L. Cristiano, A. Antonosante, M. d'Angelo, E. Di Giacomo, C. Astarita, A. Giordano, R. Ippoliti, E. Benedetti, and A. Cimini. 2016. Glioblastoma stem cells microenvironment: the paracrine roles of the niche in drug and radioresistance, Stem Cells Int, 2016:6809105. Foo, J. N., J. Liu, and E. K. Tan. 2015. CHCHD2 and Parkinson's disease, Lancet Neurol, 14:681-2. Frenel, J. S., C. Le Tourneau, B. O'Neil, P. A. Ott, S. A. Piha-Paul, C. Gomez-Roca, E. M. J. van Brummelen, H. S. Rugo, S. Thomas, S. Saraf, R. Rangwala, and A. Varga. 2017. Safety and efficacy of pembrolizumab in advanced, programmed death ligand 1-positive cervical cancer: results from the phase Ib KEYNOTE-028 trial, J Clin Oncol, 35:4035-41. Funayama, Manabu, Kenji Ohe, Taku Amo, Norihiko Furuya, Junji Yamaguchi, Shinji Saiki, Yuanzhe Li, Kotaro Ogaki, Maya Ando, Hiroyo Yoshino, Hiroyuki Tomiyama, Kenya Nishioka, Kazuko Hasegawa, Hidemoto Saiki, Wataru Satake, Kaoru Mogushi, Ryogen Sasaki, Yasumasa Kokubo, Shigeki Kuzuhara, Tatsushi Toda, Yoshikuni Mizuno, Yasuo Uchiyama, Kinji Ohno, and Nobutaka Hattori. 2015.

126

CHCHD2 mutations in autosomal dominant late-onset Parkinson's disease: a genome-wide linkage and sequencing study, Lancet Neurol, 14(3):274-82. Galli, R., E. Binda, U. Orfanelli, B. Cipelletti, A. Gritti, S. De Vitis, R. Fiocco, C. Foroni, F. Dimeco, and A. Vescovi. 2004. Isolation and characterization of tumorigenic, stem- like neural precursors from human glioblastoma, Cancer Res, 64:7011-21. Galluzzi, L., E. Morselli, O. Kepp, I. Vitale, A. Rigoni, E. Vacchelli, M. Michaud, H. Zischka, M. Castedo, and G. Kroemer. 2010. Mitochondrial gateways to cancer, Mol Aspects Med, 31:1-20. Gan, H. K., A. H. Kaye, and R. B. Luwor. 2009. The EGFRvIII variant in glioblastoma multiforme, J Clin Neurosci, 16:748-54. Gao, J., B. A. Aksoy, U. Dogrusoz, G. Dresdner, B. Gross, S. O. Sumer, Y. Sun, A. Jacobsen, R. Sinha, E. Larsson, E. Cerami, C. Sander, and N. Schultz. 2013. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal, Sci Signal, 6:pl1. Gathinji, M., M. J. McGirt, F. J. Attenello, K. L. Chaichana, K. Than, A. Olivi, J. D. Weingart, H. Brem, and A. Quinones-Hinojosa. 2009. Association of preoperative depression and survival after resection of malignant brain astrocytoma, Surg Neurol, 71:299-303. Gieryng, A., D. Pszczolkowska, K. A. Walentynowicz, W. D. Rajan, and B. Kaminska. 2017. Immune microenvironment of gliomas, Lab Invest, 97:498-518. Giladi, M., R. S. Schneiderman, T. Voloshin, Y. Porat, M. Munster, R. Blat, S. Sherbo, Z. Bomzon, N. Urman, A. Itzhaki, S. Cahal, A. Shteingauz, A. Chaudhry, E. D. Kirson, U. Weinberg, and Y. Palti. 2015. Mitotic spindle disruption by alternating electric fields leads to improper chromosome segregation and mitotic catastrophe in cancer cells, Sci Rep, 5:18046. Gilbert, M. R., J. J. Dignam, T. S. Armstrong, J. S. Wefel, D. T. Blumenthal, M. A. Vogelbaum, H. Colman, A. Chakravarti, S. Pugh, M. Won, R. Jeraj, P. D. Brown, K. A. Jaeckle, D. Schiff, V. W. Stieber, D. G. Brachman, M. Werner-Wasik, I. W. Tremont-Lukats, E. P. Sulman, K. D. Aldape, W. J. Curran, Jr., and M. P. Mehta. 2014. A randomized trial of bevacizumab for newly diagnosed glioblastoma, N Engl J Med, 370:699-708. Gilbert, M. R., M. Wang, K. D. Aldape, R. Stupp, M. E. Hegi, K. A. Jaeckle, T. S. Armstrong, J. S. Wefel, M. Won, D. T. Blumenthal, A. Mahajan, C. J. Schultz, S. Erridge, B. Baumert, K. I. Hopkins, T. Tzuk-Shina, P. D. Brown, A. Chakravarti, W. J. Curran, Jr., and M. P. Mehta. 2013. Dose-dense temozolomide for newly diagnosed glioblastoma: a randomized phase III clinical trial, J Clin Oncol, 31:4085-91. Giles, A. J., M. N. D. Hutchinson, H. M. Sonnemann, J. Jung, P. E. Fecci, N. M. Ratnam, W. Zhang, H. Song, R. Bailey, D. Davis, C. M. Reid, D. M. Park, and M. R. Gilbert. 2018. Dexamethasone-induced immunosuppression: mechanisms and implications for immunotherapy, J Immunother Cancer, 6:51. Go, Y. M., and D. P. Jones. 2009. Thioredoxin redox western analysis, Curr Protoc Toxicol, Chapter 17: Unit17.12. Gosselin, D., D. Skola, N. G. Coufal, I. R. Holtman, J. C. M. Schlachetzki, E. Sajti, B. N. Jaeger, C. O'Connor, C. Fitzpatrick, M. P. Pasillas, M. Pena, A. Adair, D. D. Gonda, M. L. Levy, R. M. Ransohoff, F. H. Gage, and C. K. Glass. 2017. An environment-

127

dependent transcriptional network specifies human microglia identity, Science, 356(6344):pii:eaaI3222.. Gourgouvelis, J., P. Yielder, S. T. Clarke, H. Behbahani, and B. A. Murphy. 2018. Exercise leads to better clinical outcomes in those receiving medication plus cognitive behavioral therapy for major depressive disorder, Front Psychiatry, 9:37. Guha, M., and N. G. Avadhani. 2013. Mitochondrial retrograde signaling at the crossroads of tumor bioenergetics, genetics and epigenetics, Mitochondrion, 13:577-91. Hambardzumyan, D., D. H. Gutmann, and H. Kettenmann. 2016. The role of microglia and macrophages in glioma maintenance and progression, Nat Neurosci, 19:20- 7. Hanahan, D., and R. A. Weinberg. 2011. Hallmarks of cancer: the next generation, Cell, 144:646-74. Hanschmann, E. M., J. R. Godoy, C. Berndt, C. Hudemann, and C. H. Lillig. 2013. Thioredoxins, glutaredoxins, and peroxiredoxins--molecular mechanisms and health significance: from cofactors to antioxidants to redox signaling, Antioxid Redox Signal, 19:1539-605. Hartung, T. J., E. Brahler, H. Faller, M. Harter, A. Hinz, C. Johansen, M. Keller, U. Koch, H. Schulz, J. Weis, and A. Mehnert. 2017. The risk of being depressed is significantly higher in cancer patients than in the general population: Prevalence and severity of depressive symptoms across major cancer types, Eur J Cancer, 72:46-53. Hashim, Z., Y. Mukai, T. Bamba, and E. Fukusaki. 2014. Metabolic profiling of retrograde pathway transcription factors rtg1 and rtg3 knockout yeast, Metabolites, 4:580-98. He, Jing-Jing, Wen-Hua Zhang, Shi-Ling Liu, Yi-Fang Chen, Chen-Xi Liao, Qian-Qing Shen, and Ping Hu. 2017. Activation of β-adrenergic receptor promotes cellular proliferation in human glioblastoma, Oncology Letters, 14:3846-52. Hegi, M. E., A. C. Diserens, T. Gorlia, M. F. Hamou, N. de Tribolet, M. Weller, J. M. Kros, J. A. Hainfellner, W. Mason, L. Mariani, J. E. Bromberg, P. Hau, R. O. Mirimanoff, J. G. Cairncross, R. C. Janzer, and R. Stupp. 2005. MGMT gene silencing and benefit from temozolomide in glioblastoma, N Engl J Med, 352:997-1003. Hickman, S. E., N. D. Kingery, T. K. Ohsumi, M. L. Borowsky, L. C. Wang, T. K. Means, and J. El Khoury. 2013. The microglial sensome revealed by direct RNA sequencing, Nat Neurosci, 16:1896-905. Hickmann, A. K., M. Hechtner, M. Nadji-Ohl, M. Janko, A. K. Reuter, K. Kohlmann, M. Haug, S. Gruninger, M. Deininger, O. Ganslandt, J. Konig, C. R. Wirtz, J. Coburger, and M. Renovanz. 2017. Evaluating patients for psychosocial distress and supportive care needs based on health-related quality of life in primary brain tumors: a prospective multicenter analysis of patients with gliomas in an outpatient setting, J Neurooncol, 131:135-51. Hoffmann, K., M. Kamp, H. J. Steiger, M. Sabel, and M. Rapp. 2017. Correlation of psychooncological distress- screening and quality of life assessment in neurosurgical patients, Oncotarget, 8:111396-404. Holland, E. C., J. Celestino, C. Dai, L. Schaefer, R. E. Sawaya, and G. N. Fuller. 2000. Combined activation of Ras and Akt in neural progenitors induces glioblastoma formation in mice, Nat Genet, 25:55-7.

128

Huang, J., C. Zeng, J. Xiao, D. Zhao, H. Tang, H. Wu, and J. Chen. 2017. Association between depression and brain tumor: a systematic review and meta-analysis, Oncotarget, 8:94932-43. Huang, W., X. Ding, H. Ye, J. Wang, J. Shao, and T. Huang. 2018. Hypoxia enhances the migration and invasion of human glioblastoma U87 cells through PI3K/Akt/mTOR/HIF-1alpha pathway, Neuroreport, 29(18):1578-85. Huberfeld, G., and C. J. Vecht. 2016. Seizures and gliomas--towards a single therapeutic approach, Nat Rev Neurol, 12:204-16. Huttemann, M., I. Lee, J. Liu, and L. I. Grossman. 2007. Transcription of mammalian cytochrome c oxidase subunit IV-2 is controlled by a novel conserved oxygen responsive element, FEBS J, 274:5737-48. Imai, Y., H. Meng, K. Shiba-Fukushima, and N. Hattori. 2019. Twin CHCH proteins, CHCHD2, and CHCHD10: key molecules of parkinson's disease, amyotrophic lateral sclerosis, and frontotemporal dementia, Int J Mol Sci, 20. Jansen, I. E., J. M. Bras, S. Lesage, C. Schulte, J. R. Gibbs, M. A. Nalls, A. Brice, N. W. Wood, H. Morris, J. A. Hardy, A. B. Singleton, T. Gasser, P. Heutink, and M. Sharma. 2015. CHCHD2 and Parkinson's disease, Lancet Neurol, 14:678-9. Jiang, Y., X. Qian, J. Shen, Y. Wang, X. Li, R. Liu, Y. Xia, Q. Chen, G. Peng, S. Y. Lin, and Z. Lu. 2015. Local generation of fumarate promotes DNA repair through inhibition of histone H3 demethylation, Nat Cell Biol, 17:1158-68. Joshi, A. D., R. C. Botham, L. J. Schlein, H. S. Roth, A. Mangraviti, A. Borodovsky, B. Tyler, S. Joslyn, J. S. Looper, M. Podell, T. M. Fan, P. J. Hergenrother, and G. J. Riggins. 2017. Synergistic and targeted therapy with a procaspase-3 activator and temozolomide extends survival in glioma rodent models and is feasible for the treatment of canine malignant glioma patients, Oncotarget, 8:80124-38. Kalkavan, H., and D. R. Green. 2018. MOMP, cell suicide as a BCL-2 family business, Cell Death Differ, 25:46-55. Kathagen, A., A. Schulte, G. Balcke, H. S. Phillips, T. Martens, J. Matschke, H. S. Gunther, R. Soriano, Z. Modrusan, T. Sandmann, C. Kuhl, A. Tissier, M. Holz, L. A. Krawinkel, M. Glatzel, M. Westphal, and K. Lamszus. 2013. Hypoxia and oxygenation induce a metabolic switch between pentose phosphate pathway and glycolysis in glioma stem-like cells, Acta Neuropathol, 126:763-80. Kim, K. H., J. M. Son, B. A. Benayoun, and C. Lee. 2018. The mitochondrial-encoded peptide MOTS-c translocates to the nucleus to regulate nuclear gene expression in response to metabolic stress, Cell Metab, 28:516-24.e7. Kirson, E. D., V. Dbaly, F. Tovarys, J. Vymazal, J. F. Soustiel, A. Itzhaki, D. Mordechovich, S. Steinberg-Shapira, Z. Gurvich, R. Schneiderman, Y. Wasserman, M. Salzberg, B. Ryffel, D. Goldsher, E. Dekel, and Y. Palti. 2007. Alternating electric fields arrest cell proliferation in animal tumor models and human brain tumors, Proc Natl Acad Sci U S A, 104:10152-7. Kmiecik, J., A. Poli, N. H. Brons, A. Waha, G. E. Eide, P. O. Enger, J. Zimmer, and M. Chekenya. 2013. Elevated CD3+ and CD8+ tumor-infiltrating immune cells correlate with prolonged survival in glioblastoma patients despite integrated immunosuppressive mechanisms in the tumor microenvironment and at the systemic level, J Neuroimmunol, 264:71-83.

129

Kokolus, K. M., Y. Zhang, J. M. Sivik, C. Schmeck, J. Zhu, E. A. Repasky, J. J. Drabick, and T. D. Schell. 2018. Beta blocker use correlates with better overall survival in metastatic melanoma patients and improves the efficacy of immunotherapies in mice, Oncoimmunology, 7:e1405205. Kolossov, V. L., J. N. Beaudoin, N. Ponnuraj, S. J. DiLiberto, W. P. Hanafin, P. J. Kenis, and H. R. Gaskins. 2015. Thiol-based antioxidants elicit mitochondrial oxidation via respiratory complex III, Am J Physiol Cell Physiol, 309:C81-91. Kore, R. A., and E. C. Abraham. 2014. Inflammatory cytokines, interleukin-1 beta and tumor necrosis factor-alpha, upregulated in glioblastoma multiforme, raise the levels of CRYAB in exosomes secreted by U373 glioma cells, Biochem Biophys Res Commun, 453:326-31. Koschmidder, E., A. Weissbach, N. Bruggemann, M. Kasten, C. Klein, and K. Lohmann. 2016. A nonsense mutation in CHCHD2 in a patient with Parkinson disease, Neurology, 86:577-9. Ladwig, S., Z. Zhou, Y. Xu, X. Wang, C. K. Chow, K. Werheid, and M. L. Hackett. 2018. Comparison of treatment rates of depression after stroke versus myocardial infarction: a systematic review and meta-analysis of observational data', Psychosom Med, 80:754-63. Lan, X., D. J. Jorg, F. M. G. Cavalli, L. M. Richards, L. V. Nguyen, R. J. Vanner, P. Guilhamon, L. Lee, M. M. Kushida, D. Pellacani, N. I. Park, F. J. Coutinho, H. Whetstone, H. J. Selvadurai, C. Che, B. Luu, A. Carles, M. Moksa, N. Rastegar, R. Head, S. Dolma, P. Prinos, M. D. Cusimano, S. Das, M. Bernstein, C. H. Arrowsmith, A. J. Mungall, R. A. Moore, Y. Ma, M. Gallo, M. Lupien, T. J. Pugh, M. D. Taylor, M. Hirst, C. J. Eaves, B. D. Simons, and P. B. Dirks. 2017. Fate mapping of human glioblastoma reveals an invariant stem cell hierarchy, Nature, 549:227- 32. Langbecker, D., S. Ekberg, and P. Yates. 2017. Don't need help, don't want help, can't get help: How patients with brain tumors account for not using rehabilitation, psychosocial and community services, Patient Educ Couns, 100:1744-50. Langbecker, D., and P. Yates. 2016. Primary brain tumor patients' supportive care needs and multidisciplinary rehabilitation, community and psychosocial support services: awareness, referral and utilization, J Neurooncol, 127:91-102. Larkin, J., V. Chiarion-Sileni, R. Gonzalez, J. J. Grob, C. L. Cowey, C. D. Lao, D. Schadendorf, R. Dummer, M. Smylie, P. Rutkowski, P. F. Ferrucci, A. Hill, J. Wagstaff, M. S. Carlino, J. B. Haanen, M. Maio, I. Marquez-Rodas, G. A. McArthur, P. A. Ascierto, G. V. Long, M. K. Callahan, M. A. Postow, K. Grossmann, M. Sznol, B. Dreno, L. Bastholt, A. Yang, L. M. Rollin, C. Horak, F. S. Hodi, and J. D. Wolchok. 2015. Combined nivolumab and ipilimumab or monotherapy in untreated melanoma, N Engl J Med, 373:23-34. LeBleu, V. S., J. T. O'Connell, K. N. Gonzalez Herrera, H. Wikman, K. Pantel, M. C. Haigis, F. M. de Carvalho, A. Damascena, L. T. Domingos Chinen, R. M. Rocha, J. M. Asara, and R. Kalluri. 2014. PGC-1alpha mediates mitochondrial biogenesis and oxidative phosphorylation in cancer cells to promote metastasis, Nat Cell Biol, 16:992-1003, 1-15. Lee, C., J. Zeng, B. G. Drew, T. Sallam, A. Martin-Montalvo, J. Wan, S. J. Kim, H. Mehta, A. L. Hevener, R. de Cabo, and P. Cohen. 2015. The mitochondrial-derived

130

peptide MOTS-c promotes metabolic homeostasis and reduces obesity and insulin resistance, Cell Metab, 21:443-54. Lee, J. H., J. E. Lee, J. Y. Kahng, S. H. Kim, J. S. Park, S. J. Yoon, J. Y. Um, W. K. Kim, J. K. Lee, J. Park, E. H. Kim, J. H. Lee, J. H. Lee, W. S. Chung, Y. S. Ju, S. H. Park, J. H. Chang, S. G. Kang, and J. H. Lee. 2018. Human glioblastoma arises from subventricular zone cells with low-level driver mutations, Nature, 560:243-47. Levada, O. A., and A. S. Troyan. 2018. Poststroke depression biomarkers: a narrative review', Front Neurol, 9. Levin, G. T., K. M. Greenwood, F. Singh, D. Tsoi, and R. U. Newton. 2016. Exercise improves physical function and mental health of brain cancer survivors: two exploratory case studies, Integr Cancer Ther, 15:190-6. Liao, X., and R. A. Butow. 1993. RTG1 and RTG2: two yeast genes required for a novel path of communication from mitochondria to the nucleus, Cell, 72:61-71. Liu, B., and S. B. Qian. 2014. Translational reprogramming in cellular stress response, Wiley Interdiscip Rev RNA, 5:301-15. Liu, F., J. Huang, L. Zhang, F. Fan, J. Chen, K. Xia, and Z. Liu. 2018. Screening for distress in patients with primary brain tumor using distress thermometer: a systematic review and meta-analysis, BMC Cancer, 18:124. Liu, Kao-Hui, Shun-Tai Yang, Yen-Kuang Lin, Jia-Wei Lin, Yi-Hsuan Lee, Jia-Yi Wang, Chaur-Jong Hu, En-Yuan Lin, Shu-Mei Chen, Chee-Kin Then, and Shing-Chuan Shen. 2015. Fluoxetine, an antidepressant, suppresses glioblastoma by evoking AMPAR-mediated calcium-dependent apoptosis, Oncotarget, 6:5088-101. Liu, R., M. Page, K. Solheim, S. Fox, and S. M. Chang. 2009. Quality of life in adults with brain tumors: current knowledge and future directions, Neuro Oncol, 11:330-9. Liu, Y., H. V. Clegg, P. L. Leslie, J. Di, L. A. Tollini, Y. He, T. H. Kim, A. Jin, L. M. Graves, J. Zheng, and Y. Zhang. 2015. CHCHD2 inhibits apoptosis by interacting with Bcl- x L to regulate Bax activation, Cell Death Differ, 22:1035-46. Liu, Y., F. Li, Y. T. Yang, X. D. Xu, J. S. Chen, T. L. Chen, H. J. Chen, Y. B. Zhu, J. Y. Lin, Y. Li, X. M. Xie, X. L. Sun, and Y. Q. Ke. 2018. IGFBP2 promotes vasculogenic mimicry formation via regulating CD144 and MMP2 expression in glioma, Oncogene, 38(11):1815-1831. Liu, Z., J. Guo, K. Li, L. Qin, J. Kang, L. Shu, Y. Zhang, Y. Wei, N. Yang, Y. Luo, Q. Sun, Q. Xu, X. Yan, and B. Tang. 2015. Mutation analysis of CHCHD2 gene in Chinese familial Parkinson's disease, Neurobiol Aging, 36:3117.e7-17.e8. Lohr, J., T. Ratliff, A. Huppertz, Y. Ge, C. Dictus, R. Ahmadi, S. Grau, N. Hiraoka, V. Eckstein, R. C. Ecker, T. Korff, A. von Deimling, A. Unterberg, P. Beckhove, and C. Herold-Mende. 2011. Effector T-cell infiltration positively impacts survival of glioblastoma patients and is impaired by tumor-derived TGF-beta, Clin Cancer Res, 17:4296-308. Loscalzo, J. 2016. Adaptions to hypoxia and redox stress: essential concepts confounded by misleading terminology, Circ Res, 119:511-3. Louis, D. N. 2006. Molecular pathology of malignant gliomas, Annu Rev Pathol, 1:97-117. Louis, D. N., A. Perry, G. Reifenberger, A. von Deimling, D. Figarella-Branger, W. K. Cavenee, H. Ohgaki, O. D. Wiestler, P. Kleihues, and D. W. Ellison. 2016. The 2016 World Health Organization classification of tumors of the central nervous system: a summary, Acta Neuropathol, 131:803-20.

131

Louveau, A., I. Smirnov, T. J. Keyes, J. D. Eccles, S. J. Rouhani, J. D. Peske, N. C. Derecki, D. Castle, J. W. Mandell, K. S. Lee, T. H. Harris, and J. Kipnis. 2015. Structural and functional features of central nervous system lymphatic vessels, Nature, 523:337-41. Lukas, R. V., D. A. Wainwright, E. Ladomersky, S. Sachdev, A. M. Sonabend, and R. Stupp. 2019. Newly diagnosed glioblastoma: a review on clinical management, Oncology (Williston Park), 33. Lukas, Rimas V., Jordi Rodon, Kevin Becker, Eric T. Wong, Kent Shih, Mehdi Touat, Marcella Fassò, Stuart Osborne, Luciana Molinero, Carol O’Hear, William Grossman, and Joachim Baehring. 2018. Clinical activity and safety of atezolizumab in patients with recurrent glioblastoma, J Neurooncol, 140(2):317- 28. Marin-Valencia, I., C. Yang, T. Mashimo, S. Cho, H. Baek, X. L. Yang, K. N. Rajagopalan, M. Maddie, V. Vemireddy, Z. Zhao, L. Cai, L. Good, B. P. Tu, K. J. Hatanpaa, B. E. Mickey, J. M. Mates, J. M. Pascual, E. A. Maher, C. R. Malloy, R. J. Deberardinis, and R. M. Bachoo. 2012. Analysis of tumor metabolism reveals mitochondrial glucose oxidation in genetically diverse human glioblastomas in the mouse brain in vivo, Cell Metab, 15:827-37. Martin, W., and M. Muller. 1998. The hydrogen hypothesis for the first eukaryote, Nature, 392:37-41. Meng, H., C. Yamashita, K. Shiba-Fukushima, T. Inoshita, M. Funayama, S. Sato, T. Hatta, T. Natsume, M. Umitsu, J. Takagi, Y. Imai, and N. Hattori. 2017. Loss of Parkinson's disease-associated protein CHCHD2 affects mitochondrial crista structure and destabilizes cytochrome, Nat Commun, 8:15500. Miller, Andrew H., and Charles L. Raison. 2016. The role of inflammation in depression: from evolutionary imperative to modern treatment target, Nat Rev Immunol, 16:22- 34. Mitchell, A. J., M. Chan, H. Bhatti, M. Halton, L. Grassi, C. Johansen, and N. Meader. 2011. Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies, Lancet Oncol, 12:160-74. Modjtahedi, N., K. Tokatlidis, P. Dessen, and G. Kroemer. 2016. Mitochondrial proteins containing coiled-coil-helix-coiled-coil-helix (CHCH) domains in health and disease, Trends Biochem Sci, 41:245-60. Monaghan, R. M., and A. J. Whitmarsh. 2015. Mitochondrial proteins moonlighting in the nucleus, Trends Biochem Sci, 40:728-35. Monteiro, A. R., R. Hill, G. J. Pilkington, and P. A. Madureira. 2017. The role of hypoxia in glioblastoma invasion, Cells, 6. Montoya, A., C. N. Amaya, A. Belmont, N. Diab, R. Trevino, G. Villanueva, S. Rains, L. A. Sanchez, N. Badri, S. Otoukesh, A. Khammanivong, D. Liss, S. T. Baca, R. J. Aguilera, E. B. Dickerson, A. Torabi, A. K. Dwivedi, A. Abbas, K. Chambers, B. A. Bryan, and Z. Nahleh. 2017. Use of non-selective beta-blockers is associated with decreased tumor proliferative indices in early stage breast cancer, Oncotarget, 8:6446-60. Morris, G., E. M. V. Reiche, A. Murru, A. F. Carvalho, M. Maes, M. Berk, and B. K. Puri. 2018. Multiple immune-inflammatory and oxidative and nitrosative stress pathways

132

explain the frequent presence of depression in multiple sclerosis, Mol Neurobiol, 55:6282-306. Motzer, R. J., B. Escudier, D. F. McDermott, S. George, H. J. Hammers, S. Srinivas, S. S. Tykodi, J. A. Sosman, G. Procopio, E. R. Plimack, D. Castellano, T. K. Choueiri, H. Gurney, F. Donskov, P. Bono, J. Wagstaff, T. C. Gauler, T. Ueda, Y. Tomita, F. A. Schutz, C. Kollmannsberger, J. Larkin, A. Ravaud, J. S. Simon, L. A. Xu, I. M. Waxman, P. Sharma, and Investigators CheckMate. 2015. Nivolumab versus everolimus in advanced renal-cell carcinoma, N Engl J Med, 373:1803-13. Munch, C. 2018. The different axes of the mammalian mitochondrial unfolded protein response, BMC Biol, 16:81. Muro, K., H. C. Chung, V. Shankaran, R. Geva, D. Catenacci, S. Gupta, J. P. Eder, T. Golan, D. T. Le, B. Burtness, A. J. McRee, C. C. Lin, K. Pathiraja, J. Lunceford, K. Emancipator, J. Juco, M. Koshiji, and Y. J. Bang. 2016. Pembrolizumab for patients with PD-L1-positive advanced gastric cancer (KEYNOTE-012): a multicentre, open-label, phase 1b trial, Lancet Oncol, 17:717-26. Nimmerjahn, A., F. Kirchhoff, and F. Helmchen. 2005. Resting microglial cells are highly dynamic surveillants of brain parenchyma in vivo, Science, 308:1314-8. Nissen, M. D., E. K. Sloan, and S. R. Mattarollo. 2018. beta-Adrenergic Signaling Impairs Antitumor CD8(+) T-cell Responses to B-cell Lymphoma Immunotherapy, Cancer Immunol Res, 6:98-109. Nitta, M., D. Kozono, R. Kennedy, J. Stommel, K. Ng, P. O. Zinn, D. Kushwaha, S. Kesari, M. M. Inda, J. Wykosky, F. Furnari, K. A. Hoadley, L. Chin, R. A. DePinho, W. K. Cavenee, A. D'Andrea, and C. C. Chen. 2010. Targeting EGFR induced oxidative stress by PARP1 inhibition in glioblastoma therapy, PLoS One, 5:e10767. Noushmehr, H., D. J. Weisenberger, K. Diefes, H. S. Phillips, K. Pujara, B. P. Berman, F. Pan, C. E. Pelloski, E. P. Sulman, K. P. Bhat, R. G. Verhaak, K. A. Hoadley, D. N. Hayes, C. M. Perou, H. K. Schmidt, L. Ding, R. K. Wilson, D. Van Den Berg, H. Shen, H. Bengtsson, P. Neuvial, L. M. Cope, J. Buckley, J. G. Herman, S. B. Baylin, P. W. Laird, and K. Aldape. 2010. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma, Cancer Cell, 17:510-22. Oka, T., T. Sayano, S. Tamai, S. Yokota, H. Kato, G. Fujii, and K. Mihara. 2008. Identification of a novel protein MICS1 that is involved in maintenance of mitochondrial morphology and apoptotic release of cytochrome c, Mol Biol Cell, 19:2597-608. Omuro, A., and L. M. DeAngelis. 2013. Glioblastoma and other malignant gliomas: a clinical review, JAMA, 310:1842-50. Osborn, A. J., J. L. Mathias, and A. K. Fairweather-Schmidt. 2014. Depression following adult, non-penetrating traumatic brain injury: a meta-analysis examining methodological variables and sample characteristics, Neurosci Biobehav Rev, 47:1-15. Ostrom, Quinn T., Haley Gittleman, Peter Liao, Toni Vecchione-Koval, Yingli Wolinsky, Carol Kruchko, and Jill S. Barnholtz-Sloan. 2017. CBTRUS statistical report: primary brain and other central nervous system tumors diagnosed in the United States in 2010–2014, Neuro-Oncology, 19:v1-v88. Ownsworth, T., S. Chambers, E. Damborg, L. Casey, D. G. Walker, and D. H. Shum. 2015. Evaluation of the making sense of brain tumor program: a randomized

133

controlled trial of a home-based psychosocial intervention, Psychooncology, 24:540-7. Pace, A., L. Dirven, J. A. F. Koekkoek, H. Golla, J. Fleming, R. Ruda, C. Marosi, E. Le Rhun, R. Grant, K. Oliver, I. Oberg, H. J. Bulbeck, A. G. Rooney, R. Henriksson, H. R. W. Pasman, S. Oberndorfer, M. Weller, and M. J. B. Taphoorn. 2017. European Association for Neuro-Oncology (EANO) guidelines for palliative care in adults with glioma, Lancet Oncol, 18:e330-e40. Parker, J. J., P. Canoll, L. Niswander, B. K. Kleinschmidt-DeMasters, K. Foshay, and A. Waziri. 2018. Intratumoral heterogeneity of endogenous tumor cell invasive behavior in human glioblastoma, Sci Rep, 8:18002. Partecke, L. I., S. Speerforck, A. Kading, F. Seubert, S. Kuhn, E. Lorenz, S. Schwandke, M. Sendler, W. Kessler, D. N. Trung, S. Oswald, F. U. Weiss, J. Mayerle, C. Henkel, P. Menges, K. Beyer, M. M. Lerch, C. D. Heidecke, and W. von Bernstorff. 2016. Chronic stress increases experimental pancreatic cancer growth, reduces survival and can be antagonised by beta-adrenergic receptor blockade, Pancreatology, 16:423-33. Patel, A. P., I. Tirosh, J. J. Trombetta, A. K. Shalek, S. M. Gillespie, H. Wakimoto, D. P. Cahill, B. V. Nahed, W. T. Curry, R. L. Martuza, D. N. Louis, O. Rozenblatt-Rosen, M. L. Suva, A. Regev, and B. E. Bernstein. 2014. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma', Science, 344:1396-401. Pavlova, N. N., and C. B. Thompson. 2016. The Emerging Hallmarks of Cancer Metabolism, Cell Metab, 23:27-47. Philip, J., Collins, A., Brand, C. A., Moore, G., Lethborg, C., Sundararajan, V., Murphy, V.A., Gold, M. 2014. “I’m just waiting…”: an exploration of the experience of living and dying with primary malignant glioma, Support Care Cancer, 22(2):389-97. Pitter, K. L., I. Tamagno, K. Alikhanyan, A. Hosni-Ahmed, S. S. Pattwell, S. Donnola, C. Dai, T. Ozawa, M. Chang, T. A. Chan, K. Beal, A. J. Bishop, C. A. Barker, T. S. Jones, B. Hentschel, T. Gorlia, U. Schlegel, R. Stupp, M. Weller, E. C. Holland, and D. Hambardzumyan. 2016. Corticosteroids compromise survival in glioblastoma, Brain, 139:1458-71. Polewski, M. D., R. F. Reveron-Thornton, G. A. Cherryholmes, G. K. Marinov, and K. S. Aboody. 2017. 'SLC7A11 overexpression in glioblastoma is associated with increased cancer stem cell-like properties, Stem Cells Dev, 26:1236-46. Polewski, M. D., R. F. Reveron-Thornton, G. A. Cherryholmes, G. K. Marinov, K. Cassady, and K. S. Aboody. 2016. Increased expression of system xc- in glioblastoma confers an altered metabolic state and temozolomide resistance, Mol Cancer Res, 14:1229-42. Poon, C. C., S. Sarkar, V. W. Yong, and J. J. P. Kelly. 2017. Glioblastoma-associated microglia and macrophages: targets for therapies to improve prognosis, Brain, 140:1548-60. Porporato, P. E., V. L. Payen, J. Perez-Escuredo, C. J. De Saedeleer, P. Danhier, T. Copetti, S. Dhup, M. Tardy, T. Vazeille, C. Bouzin, O. Feron, C. Michiels, B. Gallez, and P. Sonveaux. 2014. A mitochondrial switch promotes tumor metastasis, Cell Rep, 8:754-66. Powell, N. D., A. J. Tarr, and J. F. Sheridan. 2013. Psychosocial stress and inflammation in cancer, Brain Behav Immun, 30 Suppl: S41-7.

134

Purandare, N., M. Somayajulu, M. Huttemann, L. I. Grossman, and S. Aras. 2018. The cellular stress proteins CHCHD10 and MNRR1 (CHCHD2): partners in mitochondrial and nuclear function and dysfunction, J Biol Chem, 293:6517-29. Puschmann, A., D. W. Dickson, E. Englund, Z. K. Wszolek, and O. A. Ross. 2015. CHCHD2 and Parkinson's disease, Lancet Neurol, 14:679. Qiao, G., M. Chen, M. J. Bucsek, E. A. Repasky, and B. L. Hylander. 2018. Adrenergic signaling: a targetable checkpoint limiting development of the antitumor immune response, Front Immunol, 9:164. Quail, D. F., R. L. Bowman, L. Akkari, M. L. Quick, A. J. Schuhmacher, J. T. Huse, E. C. Holland, J. C. Sutton, and J. A. Joyce. 2016. The tumor microenvironment underlies acquired resistance to CSF-1R inhibition in gliomas, Science, 352:aad3018. Quail, D. F., and J. A. Joyce. 2017. The microenvironmental landscape of brain tumors, Cancer Cell, 31:326-41. Raizer, J. J., L. E. Abrey, A. B. Lassman, S. M. Chang, K. R. Lamborn, J. G. Kuhn, W. K. Yung, M. R. Gilbert, K. A. Aldape, P. Y. Wen, H. A. Fine, M. Mehta, L. M. Deangelis, F. Lieberman, T. F. Cloughesy, H. I. Robins, J. Dancey, M. D. Prados, and Consortium North American Brain Tumor. 2010. A phase II trial of erlotinib in patients with recurrent malignant gliomas and nonprogressive glioblastoma multiforme postradiation therapy, Neuro Oncol, 12:95-103. Raizer, J. J., P. Giglio, J. Hu, M. Groves, R. Merrell, C. Conrad, S. Phuphanich, V. K. Puduvalli, M. Loghin, N. Paleologos, Y. Yuan, D. Liu, A. Rademaker, W. K. Yung, B. Vaillant, J. Rudnick, M. Chamberlain, N. Vick, S. Grimm, I. W. Tremont-Lukats, J. De Groot, K. Aldape, M. R. Gilbert, and Collaborative Brain Tumor Trials. 2016. A phase II study of bevacizumab and erlotinib after radiation and temozolomide in MGMT unmethylated GBM patients, J Neurooncol, 126:185-92. Ran, F. A., P. D. Hsu, J. Wright, V. Agarwala, D. A. Scott, and F. Zhang. 2013. Genome engineering using the CRISPR-Cas9 system, Nat Protoc, 8:2281-308. Randazzo, D., and K. B. Peters. 2016. Psychosocial distress and its effects on the health- related quality of life of primary brain tumor patients, CNS Oncol, 5:241-9. Rawat, P., and S. A. Spector. 2017. Development and characterization of a human microglia cell model of HIV-1 infection, J Neurovirol, 23:33-46. Reardon, D. A., A. Omuro, A. A. Brandes, J. Rieger, A. Wick, J. Sepulveda, S. Phuphanich, P. de Souza, M. S. Ahluwalia, M. Lim, G. Vlahovic, and J. Sampson. 2017a. OS10.3 Randomized Phase 3 Study Evaluating the Efficacy and Safety of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: CheckMate 143, Neuro Oncol 19(Suppl 3):iii21. Renovanz, M., M. Hechtner, M. Janko, K. Kohlmann, J. Coburger, M. Nadji-Ohl, J. Konig, F. Ringel, S. Singer, and A. K. Hickmann. 2017. Factors associated with supportive care needs in glioma patients in the neuro-oncological outpatient setting, J Neurooncol, 133:653-62. Renovanz, M., A. K. Hickmann, J. Coburger, K. Kohlmann, M. Janko, A. K. Reuter, N. Keric, M. Nadji-Ohl, J. Konig, S. Singer, A. Giese, and M. Hechtner. 2018. Assessing psychological and supportive care needs in glioma patients - feasibility study on the use of the Supportive Care Needs Survey Short Form (SCNS-SF34-

135

G) and the Supportive Care Needs Survey Screening Tool (SCNS-ST9) in clinical practice, Eur J Cancer Care (Engl), 27. Renovanz, M., H. Tsakmaklis, S. Soebianto, I. Neppel, M. Nadji-Ohl, M. Beutel, A. Werner, F. Ringel, and A. K. Hickmann. 2018. Screening for distress in patients with intracranial tumors during the first 6 months after diagnosis using self- reporting instruments and an expert rating scale (the basic documentation for psycho-oncology short form - PO-Bado SF)', Oncotarget, 9:31133-45. Richards, R., M. D. Jenkinson, B. J. Haylock, and V. See. 2016. Cell cycle progression in glioblastoma cells is unaffected by pathophysiological levels of hypoxia, PeerJ, 4:e1755. Robert, S. M., S. C. Buckingham, S. L. Campbell, S. Robel, K. T. Holt, T. Ogunrinu- Babarinde, P. P. Warren, D. M. White, M. A. Reid, J. M. Eschbacher, M. E. Berens, A. C. Lahti, L. B. Nabors, and H. Sontheimer. 2015. SLC7A11 expression is associated with seizures and predicts poor survival in patients with malignant glioma, Sci Transl Med, 7:289ra86. Rooney, Alasdair G., Alan Carson, and Robin Grant. 2011. Depression in cerebral glioma patients: a systematic review of observational studies, JNCI: Journal of the National Cancer Institute, 103:61-76. Ruden, Emily, David A. Reardon, April D. Coan, James E. Herndon, Whitney E. Hornsby, Miranda West, Diane R. Fels, Annick Desjardins, James J. Vredenburgh, Emily Waner, Allan H. Friedman, Henry S. Friedman, Katherine B. Peters, and Lee W. Jones. 2011. Exercise behavior, functional capacity, and survival in adults with malignant recurrent glioma, J Clin Oncol, 29:2918-23. Rundle‐Thiele, Dayle, Richard Head, Leah Cosgrove, and Jennifer H. Martin. 2016. Repurposing some older drugs that cross the blood–brain barrier and have potential anticancer activity to provide new treatment options for glioblastoma, Br J Clin Pharmacol, 81:199-209. Sakaki-Yumoto, M., Y. Katsuno, and R. Derynck. 2013. TGF-beta family signaling in stem cells, Biochim Biophys Acta, 1830:2280-96. Sampson, J. H., A. B. Heimberger, G. E. Archer, K. D. Aldape, A. H. Friedman, H. S. Friedman, M. R. Gilbert, J. E. Herndon, 2nd, R. E. McLendon, D. A. Mitchell, D. A. Reardon, R. Sawaya, R. J. Schmittling, W. Shi, J. J. Vredenburgh, and D. D. Bigner. 2010. Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant III peptide vaccination in patients with newly diagnosed glioblastoma, J Clin Oncol, 28:4722-9. Sangar, V., C. C. Funk, U. Kusebauch, D. S. Campbell, R. L. Moritz, and N. D. Price. 2014. Quantitative proteomic analysis reveals effects of epidermal growth factor receptor (EGFR) on invasion-promoting proteins secreted by glioblastoma cells, Mol Cell Proteomics, 13:2618-31. Sardi, I., L. Giunti, C. Bresci, A. M. Buccoliero, D. Degl'innocenti, S. Cardellicchio, G. Baroni, F. Castiglione, M. D. Ros, P. Fiorini, S. Giglio, L. Genitori, M. Arico, and L. Filippi. 2013. Expression of beta-adrenergic receptors in pediatric malignant brain tumors, Oncol Lett, 5:221-25. Schito, L., and S. Rey. 2018. Cell-autonomous metabolic reprogramming in hypoxia, Trends Cell Biol, 28:128-42.

136

Schmidt, Dominic, Daniel Peterlik, Stefan O. Reber, Anja Lechner, and Daniela N. Männel. 2016. Induction of suppressor cells and increased tumor growth following chronic psychosocial stress in male mice, PLoS ONE, 11:e0159059. Schuster, J., R. K. Lai, L. D. Recht, D. A. Reardon, N. A. Paleologos, M. D. Groves, M. M. Mrugala, R. Jensen, J. M. Baehring, A. Sloan, G. E. Archer, D. D. Bigner, S. Cruickshank, J. A. Green, T. Keler, T. A. Davis, A. B. Heimberger, and J. H. Sampson. 2015. A phase II, multicenter trial of rindopepimut (CDX-110) in newly diagnosed glioblastoma: the ACT III study, Neuro Oncol, 17:854-61. Seo, M., W. H. Lee, and K. Suk. 2010. Identification of novel cell migration-promoting genes by a functional genetic screen, FASEB J, 24:464-78. Shen, F., C. Song, Y. Liu, J. Zhang, and S. Wei Song. 2018. IGFBP2 promotes neural stem cell maintenance and proliferation differentially associated with glioblastoma subtypes, Brain Res, 1704:174-186. Shi, C., N. Lamba, L. J. Zheng, D. Cote, Q. R. Regestein, C. M. Liu, Q. Tran, S. Routh, T. R. Smith, R. A. Mekary, and M. L. D. Broekman. 2018. Depression and survival of glioma patients: A systematic review and meta-analysis, Clin Neurol Neurosurg, 172:8-19. Shields, L. B., B. J. Shelton, A. J. Shearer, L. Chen, D. A. Sun, S. Parsons, T. D. Bourne, R. LaRocca, and A. C. Spalding. 2015. Dexamethasone administration during definitive radiation and temozolomide renders a poor prognosis in a retrospective analysis of newly diagnosed glioblastoma patients, Radiat Oncol, 10:222. Shimojima, K., A. Okumura, M. Hayashi, T. Kondo, H. Inoue, and T. Yamamoto. 2015. CHCHD2 is down-regulated in neuronal cells differentiated from iPS cells derived from patients with lissencephaly, Genomics, 106:196-203. Silverman, M. N., and E. M. Sternberg. 2012. Glucocorticoid regulation of inflammation and its behavioral and metabolic correlates: from HPA axis to glucocorticoid receptor dysfunction, Ann N Y Acad Sci, 1261:55-63. Singer, S., J. Roick, H. Danker, R. D. Kortmann, K. Papsdorf, S. Taubenheim, M. Renovanz, K. Jahne, and J. Meixensberger. 2018. Psychiatric co-morbidity, distress, and use of psycho-social services in adult glioma patients-a prospective study, Acta Neurochir (Wien), 160:1187-94. Singh, S. K., C. Hawkins, I. D. Clarke, J. A. Squire, J. Bayani, T. Hide, R. M. Henkelman, M. D. Cusimano, and P. B. Dirks. 2004. Identification of human brain tumour initiating cells, Nature, 432:396-401. Slavich, George M., and Michael R. Irwin. 2014. From stress to inflammation and major depressive disorder: a social signal transduction theory of depression, Psychol Bull, 140:774-815. Song, P. J., Q. Y. Lu, M. Y. Li, X. Li, and F. Shen. 2016. Comparison of effects of 18F- FDG PET-CT and MRI in identifying and grading gliomas, J Biol Regul Homeost Agents, 30:833-38. Song, R., B. Yang, X. Gao, J. Zhang, L. Sun, P. Wang, Y. Meng, Q. Wang, S. Liu, and J. Cheng. 2015. Cyclic adenosine monophosphate response element-binding protein transcriptionally regulates CHCHD2 associated with the molecular pathogenesis of hepatocellular carcinoma, Mol Med Rep, 11:4053-62. Song, W. S., Y. P. Yang, C. S. Huang, K. H. Lu, W. H. Liu, W. W. Wu, Y. Y. Lee, W. L. Lo, S. D. Lee, Y. W. Chen, P. I. Huang, and M. T. Chen. 2016. Sox2, a stemness

137

gene, regulates tumor-initiating and drug-resistant properties in CD133-positive glioblastoma stem cells, J Chin Med Assoc, 79:538-45. Spence, A. M., M. Muzi, K. R. Swanson, F. O'Sullivan, J. K. Rockhill, J. G. Rajendran, T. C. Adamsen, J. M. Link, P. E. Swanson, K. J. Yagle, R. C. Rostomily, D. L. Silbergeld, and K. A. Krohn. 2008. Regional hypoxia in glioblastoma multiforme quantified with [18F]fluoromisonidazole positron emission tomography before radiotherapy: correlation with time to progression and survival, Clin Cancer Res, 14:2623-30. Stafford, J. H., T. Hirai, L. Deng, S. B. Chernikova, K. Urata, B. L. West, and J. M. Brown. 2016. Colony stimulating factor 1 receptor inhibition delays recurrence of glioblastoma after radiation by altering myeloid cell recruitment and polarization, Neuro Oncol, 18:797-806. Sterckx, W., A. Coolbrandt, P. Clement, S. Borgenon, M. Decruyenaere, S. De Vleeschouwer, A. Mees, and B. Dierckx de Casterle. 2015. Living with a high- grade glioma: A qualitative study of patients' experiences and care needs, Eur J Oncol Nurs, 19:383-90. Stupp, R., W. P. Mason, M. J. van den Bent, M. Weller, B. Fisher, M. J. Taphoorn, K. Belanger, A. A. Brandes, C. Marosi, U. Bogdahn, J. Curschmann, R. C. Janzer, S. K. Ludwin, T. Gorlia, A. Allgeier, D. Lacombe, J. G. Cairncross, E. Eisenhauer, and R. O. Mirimanoff. 2005. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, N Engl J Med, 352:987-96. Stupp, R., S. Taillibert, A. Kanner, W. Read, D. Steinberg, B. Lhermitte, S. Toms, A. Idbaih, M. S. Ahluwalia, K. Fink, F. Di Meco, F. Lieberman, J. J. Zhu, G. Stragliotto, D. Tran, S. Brem, A. Hottinger, E. D. Kirson, G. Lavy-Shahaf, U. Weinberg, C. Y. Kim, S. H. Paek, G. Nicholas, J. Bruna, H. Hirte, M. Weller, Y. Palti, M. E. Hegi, and Z. Ram. 2017. Effect of tumor-treating fields plus maintenance temozolomide vs maintenance temozolomide alone on survival in patients with glioblastoma: a randomized clinical trial, JAMA, 318:2306-16. Sutendra, G., A. Kinnaird, P. Dromparis, R. Paulin, T. H. Stenson, A. Haromy, K. Hashimoto, N. Zhang, E. Flaim, and E. D. Michelakis. 2014. A nuclear pyruvate dehydrogenase complex is important for the generation of acetyl-CoA and histone acetylation, Cell, 158:84-97. Szulzewsky, F., A. Pelz, X. Feng, M. Synowitz, D. Markovic, T. Langmann, I. R. Holtman, X. Wang, B. J. Eggen, H. W. Boddeke, D. Hambardzumyan, S. A. Wolf, and H. Kettenmann. 2015. Glioma-associated microglia/macrophages display an expression profile different from M1 and M2 polarization and highly express Gpnmb and Spp1, PLoS One, 10:e0116644. Tan, A. S., J. W. Baty, L. F. Dong, A. Bezawork-Geleta, B. Endaya, J. Goodwin, M. Bajzikova, J. Kovarova, M. Peterka, B. Yan, E. A. Pesdar, M. Sobol, A. Filimonenko, S. Stuart, M. Vondrusova, K. Kluckova, K. Sachaphibulkij, J. Rohlena, P. Hozak, J. Truksa, D. Eccles, L. M. Haupt, L. R. Griffiths, J. Neuzil, and M. V. Berridge. 2015. Mitochondrial genome acquisition restores respiratory function and tumorigenic potential of cancer cells without mitochondrial DNA, Cell Metab, 21:81-94.

138

Tan, Sze Kiat, Anna Jermakowicz, Adnan K. Mookhtiar, Charles B. Nemeroff, Stephan C. Schürer, and Nagi G. Ayad. 2018. Drug repositioning in glioblastoma: a pathway perspective, Front Pharmacol, 9:218. Then, Chee-Kin, Kao-Hui Liu, Ming-Hsuan Liao, Kuo-Hsuan Chung, Jia-Yi Wang, and Shing-Chuan Shen. 2017. Antidepressants, sertraline and paroxetine, increase calcium influx and induce mitochondrial damage-mediated apoptosis of astrocytes, Oncotarget, 8:115490-502. Timney, B. L., B. Raveh, R. Mironska, J. M. Trivedi, S. J. Kim, D. Russel, S. R. Wente, A. Sali, and M. P. Rout. 2016. Simple rules for passive diffusion through the nuclear pore complex, J Cell Biol, 215:57-76. Tondorf, T., A. Grossert, S. I. Rothschild, M. T. Koller, C. Rochlitz, A. Kiss, R. Schaefert, G. Meinlschmidt, S. Hunziker, and D. Zwahlen. 2018. Focusing on cancer patients' intentions to use psychooncological support: A longitudinal, mixed-methods study, Psychooncology, 27:1656-63. Torelli, N. Q., J. R. Ferreira-Junior, A. J. Kowaltowski, and F. M. da Cunha. 2015. RTG1- and RTG2-dependent retrograde signaling controls mitochondrial activity and stress resistance in Saccharomyces cerevisiae, Free Radic Biol Med, 81:30-7. Tsuchihashi, K., S. Okazaki, M. Ohmura, M. Ishikawa, O. Sampetrean, N. Onishi, H. Wakimoto, M. Yoshikawa, R. Seishima, Y. Iwasaki, T. Morikawa, S. Abe, A. Takao, M. Shimizu, T. Masuko, M. Nagane, F. B. Furnari, T. Akiyama, M. Suematsu, E. Baba, K. Akashi, H. Saya, and O. Nagano. 2016. The EGF receptor promotes the malignant potential of glioma by regulating amino acid transport system xc(-)', Cancer Res, 76:2954-63. United States Cancer Statistics: 1999–2011 Incidence, WONDER Online Database.In. 2015. United States Department of Health and Human Services, Centers for Disease Control and Prevention and National Cancer Institute. Centers for Disease Control and Prevention National Center for Health Statistics. Vehling, S., D. W. Kissane, C. Lo, H. Glaesmer, T. J. Hartung, G. Rodin, and A. Mehnert. 2017. The association of demoralization with mental disorders and suicidal ideation in patients with cancer, Cancer, 123:3394-401. Verhaak, R. G., K. A. Hoadley, E. Purdom, V. Wang, Y. Qi, M. D. Wilkerson, C. R. Miller, L. Ding, T. Golub, J. P. Mesirov, G. Alexe, M. Lawrence, M. O'Kelly, P. Tamayo, B. A. Weir, S. Gabriel, W. Winckler, S. Gupta, L. Jakkula, H. S. Feiler, J. G. Hodgson, C. D. James, J. N. Sarkaria, C. Brennan, A. Kahn, P. T. Spellman, R. K. Wilson, T. P. Speed, J. W. Gray, M. Meyerson, G. Getz, C. M. Perou, D. N. Hayes, and Network Cancer Genome Atlas Research. 2010. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1, Cancer Cell, 17:98-110. Vichai, V., and K. Kirtikara. 2006. Sulforhodamine B colorimetric assay for cytotoxicity screening, Nat Protoc, 1:1112-6. Vlashi, E., C. Lagadec, L. Vergnes, T. Matsutani, K. Masui, M. Poulou, R. Popescu, L. Della Donna, P. Evers, C. Dekmezian, K. Reue, H. Christofk, P. S. Mischel, and F. Pajonk. 2011. Metabolic state of glioma stem cells and nontumorigenic cells, Proc Natl Acad Sci U S A, 108:16062-7.

139

Wainwright, D. A., and M. S. Lesniak. 2014. Menage a trois: sustained therapeutic antitumor immunity requires multiple partners in malignant glioma, Oncoimmunology, 3:e28927. Wainwright, D. A., S. Sengupta, Y. Han, and M. S. Lesniak. 2011. Thymus-derived rather than tumor-induced regulatory T cells predominate in brain tumors, Neuro Oncol, 13:1308-23. Wang, F., H. Liu, F. Wang, R. Xu, P. Wang, F. Tang, X. Zhang, Z. Zhu, H. Lv, and T. Han. 2018. Propranolol suppresses the proliferation and induces the apoptosis of liver cancer cells, Mol Med Rep, 17:5213-21. Wang, H., H. Wang, W. Shen, H. Huang, L. Hu, L. Ramdas, Y. H. Zhou, W. S. Liao, G. N. Fuller, and W. Zhang. 2003. Insulin-like growth factor binding protein 2 enhances glioblastoma invasion by activating invasion-enhancing genes, Cancer Res, 63:4315-21. Wang, Q., B. Hu, X. Hu, H. Kim, M. Squatrito, L. Scarpace, A. C. deCarvalho, S. Lyu, P. Li, Y. Li, F. Barthel, H. J. Cho, Y. H. Lin, N. Satani, E. Martinez-Ledesma, S. Zheng, E. Chang, C. G. Sauve, A. Olar, Z. D. Lan, G. Finocchiaro, J. J. Phillips, M. S. Berger, K. R. Gabrusiewicz, G. Wang, E. Eskilsson, J. Hu, T. Mikkelsen, R. A. DePinho, F. Muller, A. B. Heimberger, E. P. Sulman, D. H. Nam, and R. G. W. Verhaak. 2017. Tumor evolution of glioma-intrinsic gene expression subtypes associates with immunological changes in the microenvironment, Cancer Cell, 32:42-56.e6. Warburg, O. 1956. On the origin of cancer cells, Science, 123:309-14. Warburg, O., F. Wind, and E. Negelein. 1927. The metabolism of tumors in the body, J Gen Physiol, 8:519-30. Waypa, G. B., J. D. Marks, R. Guzy, P. T. Mungai, J. Schriewer, D. Dokic, and P. T. Schumacker. 2010. Hypoxia triggers subcellular compartmental redox signaling in vascular smooth muscle cells, Circ Res, 106:526-35. Wei, Y., R. N. Vellanki, E. Coyaud, V. Ignatchenko, L. Li, J. R. Krieger, P. Taylor, J. Tong, N. A. Pham, G. Liu, B. Raught, B. G. Wouters, T. Kislinger, M. S. Tsao, and M. F. Moran. 2015. CHCHD2 is coamplified with EGFR in NSCLC and regulates mitochondrial function and cell migration, Mol Cancer Res, 13:1119-29. Weinhouse, S. 1956. On respiratory impairment in cancer cells, Science, 124:267-9. Weller, M., N. Butowski, D. D. Tran, L. D. Recht, M. Lim, H. Hirte, L. Ashby, L. Mechtler, S. A. Goldlust, F. Iwamoto, J. Drappatz, D. M. O'Rourke, M. Wong, M. G. Hamilton, G. Finocchiaro, J. Perry, W. Wick, J. Green, Y. He, C. D. Turner, M. J. Yellin, T. Keler, T. A. Davis, R. Stupp, J. H. Sampson, and Act Iv trial investigators. 2017. Rindopepimut with temozolomide for patients with newly diagnosed, EGFRvIII- expressing glioblastoma (ACT IV): a randomised, double-blind, international phase 3 trial, Lancet Oncol, 18:1373-85. Weller, Michael, Nicholas Butowski, David D. Tran, Lawrence D. Recht, Michael Lim, Hal Hirte, Lynn Ashby, Laszlo Mechtler, Samuel A. Goldlust, Fabio Iwamoto, Jan Drappatz, Donald M. O'Rourke, Mark Wong, Mark G. Hamilton, Gaetano Finocchiaro, James Perry, Wolfgang Wick, Jennifer Green, Yi He, Christopher D. Turner, Michael J. Yellin, Tibor Keler, Thomas A. Davis, Roger Stupp, John H. Sampson, Nicholas Butowski, Jian Campian, Lawrence Recht, Michael Lim, Lynn Ashby, Jan Drappatz, Hal Hirte, Fabio Iwamoto, Laszlo Mechtler, Samuel Goldlust,

140

Kevin Becker, Gene Barnett, Garth Nicholas, Annick Desjardins, Tara Benkers, Naveed Wagle, Morris Groves, Santosh Kesari, Zsolt Horvath, Ryan Merrell, Richard Curry, James O'Rourke, David Schuster, Mark Wong, Maciej Mrugala, Randy Jensen, John Trusheim, Glenn Lesser, Karl Belanger, Andrew Sloan, Benjamin Purow, Karen Fink, Jeffrey Raizer, Michael Schulder, Suresh Nair, Scott Peak, James Perry, Alba Brandes, Michael Weller, Nimish Mohile, Joseph Landolfi, Jon Olson, Gaetano Finocchiaro, Ross Jennens, Paul DeSouza, Bridget Robinson, Marka Crittenden, Kent Shih, Alexandra Flowers, Shirley Ong, Jennifer Connelly, Costas Hadjipanayis, Pierre Giglio, Frank Mott, David Mathieu, Nathalie Lessard, Sanchez Juan Sepulveda, József Lövey, Helen Wheeler, Po-Ling Inglis, Claire Hardie, Daniela Bota, Maciej Lesniak, Jana Portnow, Bruce Frankel, Larry Junck, Reid Thompson, Lawrence Berk, John McGhie, David Macdonald, Frank Saran, Riccardo Soffietti, Deborah Blumenthal, Sá Barreto Costa Marcos André de, Anna Nowak, Nimit Singhal, Andreas Hottinger, Andrea Schmid, Gordan Srkalovic, David Baskin, Camilo Fadul, Louis Nabors, Renato LaRocca, John Villano, Nina Paleologos, Petr Kavan, Marshall Pitz, Brian Thiessen, Ahmed Idbaih, Jean Sébastien Frenel, Julien Domont, Oliver Grauer, Peter Hau, Christine Marosi, Jan Sroubek, Elizabeth Hovey, P. S. Sridhar, Lawrence Cher, Erin Dunbar, Thomas Coyle, Jane Raymond, Kevin Barton, Michael Guarino, Sumul Raval, Baldassarre Stea, Jorge Dietrich, Kirsten Hopkins, Sara Erridge, Joachim-Peter Steinbach, Losada Estela Pineda, Quintero Carmen Balana, Barco Berron Sonia del, Miklós Wenczl, Katalin Molnár, Katalin Hideghéty, Alexander Lossos, Linde Myra van, Ana Levy, Rosemary Harrup, William Patterson, Zarnie Lwin, Sith Sathornsumetee, E. Jian Lee, Jih-Tsun Ho, Steven Emmons, J. Paul Duic, Spencer Shao, Hani Ashamalla, Michael Weaver, Jose Lutzky, Nicholas Avgeropoulos, Wahid Hanna, Mukund Nadipuram, Gary Cecchi, Robert O'Donnell, Susan Pannullo, Jennifer Carney, Mark Hamilton, Mary MacNeil, Ronald Beaney, Michel Fabbro, Oliver Schnell, Rainer Fietkau, Guenther Stockhammer, Bela Malinova, Karel Odrazka, Martin Sames, Gil Miguel Gil, Evangelia Razis, Konstantin Lavrenkov, Guillermo Castro, Francisco Ramirez, Clarissa Baldotto, Fabiana Viola, Suzana Malheiros, Jason Lickliter, Stanislaw Gauden, Arunee Dechaphunkul, Iyavut Thaipisuttikul, Ziad Thotathil, Hsin- I. Ma, Wen-Yu Cheng, Chin-Hong Chang, Fernando Salas, Pierre-Yves Dietrich, Christoph Mamot, Lakshmi Nayak, and Shona Nag. 2017. Rindopepimut with temozolomide for patients with newly diagnosed, EGFRvIII-expressing glioblastoma (ACT IV): a randomised, double-blind, international phase 3 trial, Lancet Oncol, 18:1373-85. Wente, S. R., and M. P. Rout. 2010. The nuclear pore complex and nuclear transport, Cold Spring Harb Perspect Biol, 2:a000562. Wollweber, F., K. von der Malsburg, and M. van der Laan. 2017. Mitochondrial contact site and cristae organizing system: a central player in membrane shaping and crosstalk, Biochim Biophys Acta Mol Cell Res, 1864:1481-89. Wong, N., W. Liu, and X. Wang. 2015. WU-CRISPR: characteristics of functional guide RNAs for the CRISPR/Cas9 system, Genome Biol, 16:218. Woods, J. A., K. R. Wilund, S. A. Martin, and B. M. Kistler. 2012. Exercise, inflammation and aging', Aging Dis, 3:130-40.

141

Wu, Z., P. Puigserver, U. Andersson, C. Zhang, G. Adelmant, V. Mootha, A. Troy, S. Cinti, B. Lowell, R. C. Scarpulla, and B. M. Spiegelman. 1999. Mechanisms controlling mitochondrial biogenesis and respiration through the thermogenic coactivator PGC-1, Cell, 98:115-24. Xie, H., and M. C. Simon. 2017. Oxygen availability and metabolic reprogramming in cancer, J Biol Chem, 292:16825-32. Xie, Jun‐Chao, Shuai Yang, Xue‐Yuan Liu, and Yan‐Xin Zhao. 2018. Effect of marital status on survival in glioblastoma multiforme by demographics, education, economic factors, and insurance status, Cancer Med, 7:3722-42. Xie, Q., Q. Wu, C. M. Horbinski, W. A. Flavahan, K. Yang, W. Zhou, S. M. Dombrowski, Z. Huang, X. Fang, Y. Shi, A. N. Ferguson, D. F. Kashatus, S. Bao, and J. N. Rich. 2015. Mitochondrial control by DRP1 in brain tumor initiating cells, Nat Neurosci, 18:501-10. Xie, Q., Q. Wu, L. Kim, T. E. Miller, B. B. Liau, S. C. Mack, K. Yang, D. C. Factor, X. Fang, Z. Huang, W. Zhou, K. Alazem, X. Wang, B. E. Bernstein, S. Bao, and J. N. Rich. 2016. RBPJ maintains brain tumor-initiating cells through CDK9-mediated transcriptional elongation, J Clin Invest, 126:2757-72. Ye, X. Z., S. L. Xu, Y. H. Xin, S. C. Yu, Y. F. Ping, L. Chen, H. L. Xiao, B. Wang, L. Yi, Q. L. Wang, X. F. Jiang, L. Yang, P. Zhang, C. Qian, Y. H. Cui, X. Zhang, and X. W. Bian. 2012. Tumor-associated microglia/macrophages enhance the invasion of glioma stem-like cells via TGF-beta1 signaling pathway, J Immunol, 189:444-53. Yeung, Y. T., K. L. McDonald, T. Grewal, and L. Munoz. 2013. Interleukins in glioblastoma pathophysiology: implications for therapy, Br J Pharmacol, 168:591-606. Yogev, O., A. Naamati, and O. Pines. 2011. Fumarase: a paradigm of dual targeting and dual localized functions, FEBS J, 278:4230-42. Zanotelli, M. R., Z. E. Goldblatt, J. P. Miller, F. Bordeleau, J. Li, J. A. VanderBurgh, M. C. Lampi, M. R. King, and C. A. Reinhart-King. 2018. Regulation of ATP utilization during metastatic cell migration by collagen architecture, Mol Biol Cell, 29:1-9. Zhai, L., E. Ladomersky, K. L. Lauing, M. Wu, M. Genet, G. Gritsina, G. Balaz, P.K. Brastianos, D. C. Binder, J. A. Sosman, F. J. Giles, C. D. James, C. Horbinski, R. Stupp, and D. A. Wainwright. 2017. Infiltrating T cells increase IDO1 expression in glioblastoma and contribute to decreased patient survival, Clin Cancer Res, 23(21):6650-6660. Zhang, M., Z. Xi, S. Fang, M. Ghani, C. Sato, D. Moreno, Y. Liang, A. E. Lang, and E. Rogaeva. 2016. Mutation analysis of CHCHD2 in Canadian patients with familial Parkinson's disease, Neurobiol Aging, 38:217.e7-17.e8. Zhang, X., K. Ding, J. Wang, X. Li, and P. Zhao. 2018. Chemoresistance caused by the microenvironment of glioblastoma and the corresponding solutions, Biomed Pharmacother, 109:39-46. Zhao, J., J. Zhang, M. Yu, Y. Xie, Y. Huang, D. W. Wolff, P. W. Abel, and Y. Tu. 2013. Mitochondrial dynamics regulates migration and invasion of breast cancer cells, Oncogene, 32:4814-24. Zhou, W., D. Ma, A. X. Sun, H. D. Tran, D. L. Ma, B. K. Singh, J. Zhou, J. Zhang, D. Wang, Y. Zhao, P. M. Yen, E. Goh, and E. K. Tan. 2018. PD-linked CHCHD2 mutations impair CHCHD10 and MICOS complex leading to mitochondria dysfunction, Hum Mol Genet, 28(7):1100-1116.

142

Zhou, Z. D., W. T. Saw, and E. K. Tan. 2017. Mitochondrial CHCHD-containing proteins: physiologic functions and link with neurodegenerative diseases, Mol Neurobiol, 54:5534-46. Zhu, L., A. Gomez-Duran, G. Saretzki, S. Jin, K. Tilgner, D. Melguizo-Sanchis, G. Anyfantis, J. Al-Aama, L. Vallier, P. Chinnery, M. Lako, and L. Armstrong. 2016. The mitochondrial protein CHCHD2 primes the differentiation potential of human induced pluripotent stem cells to neuroectodermal lineages, J Cell Biol, 215:187- 202. Zhuang, J., P. Y. Wang, X. Huang, X. Chen, J. G. Kang, and P. M. Hwang. 2013. Mitochondrial disulfide relay mediates translocation of p53 and partitions its subcellular activity, Proc Natl Acad Sci U S A, 110:17356-61. Zinzani, P. L., V. Ribrag, C. H. Moskowitz, J. M. Michot, J. Kuruvilla, A. Balakumaran, Y. Zhang, S. Chlosta, M. A. Shipp, and P. Armand. 2017. Safety and tolerability of pembrolizumab in patients with relapsed/refractory primary mediastinal large B- cell lymphoma, Blood, 130:267-70.

143

APPENDIX A

THE INTERPLAY AMONG PSYCHOLOGICAL DISTRESS, THE IMMUNE SYSTEM,

AND BRAIN TUMOR PATIENT OUTCOMES1

Sebastian Otto-Meyer1, Jan Lumibao, Eugene Kim, Erik Ladomersky1, Lijie Zhai1, Kristen

L. Lauing1, Denise M. Scholtens2, Frank Penedo6,7,8, Christina Amidei1, Rimas V.

Lukas1,3,4,5, Derek A. Wainwright1

1Department of Neurological Surgery, 2Department of Preventative Medicine-

Biostatistics, 3Department of Molecular Genetics, 4Department of Medicine-Division of

Hematology and Oncology, 5Robert H. Lurie Comprehensive Cancer Center of

Northwestern University, 6Department of Biochemistry and Molecular Genetics,

7Department of Neurology, 8Department of Microbiology-Immunology, Northwestern

University Feinberg School of Medicine, Chicago, IL 60611

ABSTRACT

A malignant brain tumor diagnosis is often accompanied with intense feelings and can be associated with psychosocial conditions including depression, anxiety, and/or increased distress levels. Previous work has highlighted the impact of uncontrolled psychological distress among brain tumor patients. Given the negative impact of maladaptive psychosocial and biobehavioral factors on normal immune system functions, the question remains as to how psychological conditions potentially affect the brain tumor patient antitumor immune response. Since immunotherapy has yet to show efficacy at increasing malignant glioma patient survival in all randomized, phase III clinical trials to-

1 Otto-Meyer S, Lumibao J, Kim E, Ladomersky E, Zhai L, Lauing KL, Scholtens DM, Penedo F, Amidei C, Lukas RV, Wainwright DA. The interplay among psychological distress, the immune system, and brain tumor patient outcomes. Current Opinion in Behavioral Sciences. 2019;28:44-50. doi: 10.1016/j.cobeha.2019.01.009. 144 date, this review provides new insights into the potential negative effects of chronic distress on brain tumor patient immune functions and outcomes.

Challenges of immunotherapy for malignant glioma

Brain tumors are a relatively rare but potentially life-threatening diagnosis, accounting for approximately 1-2% of cancers in the United States ("United States Cancer

Statistics: 1999–2011 Incidence, WONDER Online Database.," 2015). Nearly 80% of primary malignant brain tumors are a form of glioma, with glioblastoma (GBM, WHO grade

IV) representing the most common, deadly subtype, and accounting for over half of all malignant glioma cases (Ostrom et al., 2017). Despite the current standard of care, GBM remains incurable with a median overall survival (OS) ranging from 9 to 21 months, and a five-year survival of only 5% to 14% (Darefsky, King, & Dubrow, 2012; Dubrow et al.,

2013; Stupp et al., 2017). Standard of care for GBM patients consists of surgical resection, radiation, chemotherapy, and tumor treating fields, which prolongs survival and may transiently reduce symptom burden. However, these treatments also impair the quality of life since patients are presented by a diagnosis that induces significant psychological distress (R. Liu, Page, Solheim, Fox, & Chang, 2009).

Recent research efforts aimed at enhancing treatment efficacy for GBM, have focused on immunotherapy, which is an effective treatment strategy for a subset of cancer patients diagnosed with melanoma (Larkin et al., 2015), non-small cell lung (S. J. Antonia et al., 2016), and renal cell (Motzer et al., 2015) malignancies, as well as many others that originate outside of the central nervous system (CNS) (Bellmunt et al., 2017; R. Chen et al., 2017; Frenel et al., 2017; Muro et al., 2016; Zinzani et al., 2017). In contrast, high-

145 grade glioma patients treated with immunotherapy have not demonstrated an OS benefit in all phase III clinical trials to-date (Reardon et al., 2017b; M. Weller et al., 2017). Several immunotherapeutic strategies have been investigated, including: (i) vaccines (ie. rindopepimut, HSPPC96, or dendritic cells); (ii) checkpoint inhibitors [ie. nivolumab (anti-

PD-1 mAb)]; and (iii) chimeric antigen receptor (CAR) T cells. However, despite the excitement from early phase trials, immunotherapy has yet to translate into an effective clinical solution for patients, as demonstrated in late phase, randomized, multi-site clinical evaluation (Bloch et al., 2017; Sampson et al., 2010; Schuster et al., 2015; M. Weller et al., 2017).

Additional challenges likely contributing to the effective application of immunotherapy for treating GBM include the: (i) highly infiltrating nature of the disease, which prevents surgical resection from completely removing all associated tumor burden;

(ii) potently immunosuppressive microenvironment, characterized by a relatively low level of tumor-infiltrating effector T cells, combined with an accumulation of immunosuppressive regulatory T cells (Tregs; CD4+CD25+FoxP3+) and myeloid derived suppressor cells; (iii) older age of the patient; and (iv) anatomical location that impedes a routine use of tumor biopsies for monitoring the response to therapy. A critical factor that may be overlooked during immunotherapeutic treatment considerations is the impact of high psychosocial and biobehavioral levels of distress. The subject of distress, specifically its prevalence and influence on the immune system and outcomes in patients with a malignant brain tumor, forms the basis of our review. Research in the past two years will be emphasized, with attention to the associations between increased distress levels, the

146 negative impact of systemic immune/neuroendocrine functions, and preclinical and clinical outcomes in the setting of primary brain cancer.

Psychological distress in patients with brain tumors

Depression and anxiety are common in patients with a terminal illness (Mitchell et al., 2011).Meta-analyses have found that, cancer patients in the palliative care setting have a rate of mood disorders, including depression, anxiety, and adjustment disorder, of

29% (Mitchell et al., 2011). Notably, this rate of psychosocial conditions is likely underappreciated and/or underreported, as cancer patients can experience chronic distress and demoralization without the presence of a psychiatric diagnosis (Philip et al.;

Sterckx et al., 2015; Vehling et al., 2017).

Psychological distress, depression, and anxiety may be particularly enhanced in patients with primary brain tumors. A recent analysis found that 24% of a 4,000 mixed- cancer patient cohort, possessed clinical signs of depression, significantly higher than the general population (Hartung et al., 2017). Strikingly, the rate among patients with primary brain cancer was markedly higher, with 36% showing signs of depression. In contrast, patients diagnosed with malignant melanoma and prostate cancer, had reduced rates of depression, at 16% and 10%, respectively. Additional meta-analyses and prospective studies have discovered an increased rate of psychological conditions and distress among brain tumor patients as compared to the general population (J. Huang et al., 2017;

F. Liu et al., 2018) and as compared to patients diagnosed with non-CNS tumors

(Edelstein et al., 2016).

147

Furthermore, psychological distress in brain tumor patients is not always appropriately acknowledged or well-managed. Although attention to the mental health care of malignant brain tumor patients has increased (Catt, Chalmers, & Fallowfield,

2008; Randazzo & Peters, 2016; Rooney, Carson, & Grant, 2011), a recent review by the

European Association of Neuro-Oncology emphasized the lack of evidence for appropriately treating depression and anxiety in glioma patients and, as such, was unable to provide strong recommendations on the subject (Pace et al., 2017). The direct neurological involvement of primary brain tumors further complicates management guidance from psychosocial standard of care for the treatment of non-CNS cancers.

Recent studies have found that psychological needs are highly unmet among brain tumor patients (Langbecker & Yates, 2016; Renovanz et al., 2017). A longitudinal investigation of mostly high grade glioma patients found that, despite increased distress in approximately half of the patients during various time points, only 14% of the distressed patients received psychological care as an in-patient, and less than half of the distressed patients received psychological care in an outpatient setting (Singer et al., 2018).

Collectively, the above studies indicate psychological distress and depression as common concerns for patients with primary brain cancer.

The etiology of distress among primary brain cancer patients may be related to the inherently high load of stressors accompanying their diagnosis. However, neurological insult, regardless of tumor burden, has been shown to be correlated with increased depression and distress. Patients who present with a stroke have an increased rate of depression as compared to those with a myocardial infarction (Ladwig et al., 2018), and the prevalence of depression is also higher for individuals diagnosed with the CNS

148 autoimmune disease, multiple sclerosis (MS), as well as for traumatic brain injury (TBI) patients (Feinstein, Magalhaes, Richard, Audet, & Moore, 2014; Osborn, Mathias, &

Fairweather-Schmidt, 2014). Although potentially resulting from the reduced quality of life suffered by patients with neurological insult(s), increased depression may be the direct result of enhanced inflammation. There is a well-established link between the development of depression and stressors that trigger inflammatory phenotypes and increased immune activity through the sympathetic nervous system (SNS) (Miller &

Raison, 2016; Slavich & Irwin, 2014). Indeed, inflammation resulting from stroke, MS, and

TBI is associated with depression, further supporting a role for generalized inflammation in in the pathogenesis of depression (Bodnar, Morganti, & Bachstetter, 2018; Fenn et al.,

2014; Levada & Troyan, 2018; Morris et al., 2018). The connection between inflammation and depression is not well-characterized for patients with primary brain cancer. However, levels of inflammatory cytokines are significantly increased in GBM patients and may play a role in outcomes dependent on levels of depression, anxiety, and distress (Chiorean et al., 2014; Kore & Abraham, 2014; Yeung, McDonald, Grewal, & Munoz, 2013) (Fig. A.1).

Additional work is necessary to comprehensively profile the incidence of psychosocial conditions in brain tumor patients and its effects on OS outcomes and responses to therapy.

Interactions between distress, immunosuppression, and cancer

Untreated distress may affect brain tumor patient outcomes through the interactions between psychosocial distress and the immune system. The relationship between excess adrenergic signaling and immune dysregulation, inflammation, and tumor growth is well- established (Partecke et al., 2016; Powell, Tarr, & Sheridan, 2013;

149

Qiao, Chen, Bucsek, Repasky, & Hylander, 2018). Increased distress promotes the release of stress hormones, primarily in the form of catecholamines including epinephrine and norepinephrine via the sympathetic nervous system (SNS), as well as glucocorticoids via the hypothalamic-pituitary-adrenal axis. Chronic release of these molecules, in turn, increases immunosuppression that facilitates uncontrolled tumor growth by inhibiting the anti-tumor immune response as demonstrated in preclinical models of lymphoma and fibrosarcoma (Nissen, Sloan, & Mattarollo, 2018; Schmidt, Peterlik, Reber, Lechner, &

Männel, 2016). Interestingly, beta blockers, which inhibit catecholamine-induced β- adrenergic receptor signaling, inhibit proliferation and tumor growth in models of liver cancer, early stage breast cancer, and melanoma (Kokolus et al., 2018; Montoya et al.,

2017; F. Wang et al., 2018); although this effect is not universal (Cardwell et al., 2016).

Additionally, recent studies found that the inhibition of β-adrenergic signaling with the pan- beta blocker, propranolol, enhances PD-1 checkpoint inhibitor efficacy in preclinical, syngeneic, mouse melanoma models (Bucsek et al., 2017; Kokolus et al., 2018). Notably,

β-adrenergic blockade increased the frequency of CD8+ tumor infiltrating lymphocytes, as well as the ratio of CD8+ effector T cells to Tregs in B16-OVA tumors (Bucsek et al.,

2017). Altogether, these studies suggest a potential role for targeting β-adrenergic signaling to decrease tumor cell proliferation and enhance synergy with immunotherapy.

Although once thought to be an immuno-privileged site, it’s now well-established that leukocytes infiltrate the brain upon appropriate activation signals (Kmiecik et al.,

2013; Lohr et al., 2011; Zhai et al., 2017). However, potent immunosuppressive factors are normally present in the brain, including TGF-β, IL-10, and VEGF, which dampen cytotoxic T lymphocyte migration while enhancing Treg accumulation (Dutoit et al., 2016).

150

Parenchymal cells of the CNS also express FasL, which can facilitate brain-infiltrating

Fas+ T cell apoptosis. Collectively, these characteristics contribute to the profoundly immunosuppressive environment under normal circumstances that likely further contribute to the poor immune response against malignant glioma. Increased β- adrenergic signaling in response to psychosocial distress would only exacerbate the profoundly immunosuppressive microenvironment and facilitate immune escape of GBM cells.

To counteract immunosuppressive β-adrenergic signaling, propranolol has been considered for adjuvant therapy in patients diagnosed with malignant glioma. However, this consideration is primarily based on the results observed in other cancers, or as direct evidence in GBM cells (Rundle‐Thiele, Head, Cosgrove, & Martin, 2016). Accordingly, isoproterenol, a β-adrenergic receptor agonist, enhances GBM U251 cell proliferation and metalloproteinase expression – an effect inhibited by propranolol treatment (He et al.,

2017). In contrast, β-adrenergic receptor blockade with prazosin, inhibits glioblastoma tumor growth in an orthotopic, immunocompetent, syngeneic mouse model, highlighting the potential for an immunosuppressive contribution by β-adrenergic signaling mechanisms (Assad Kahn et al., 2016). Beyond adult brain tumors, β-adrenergic receptors are expressed by malignant pediatric brain tumor cells (Sardi et al., 2013), raising the possible generalizability of targeting the β-adrenergic signaling axis in both the pediatric and adult brain tumor settings.

Cortisol, a glucocorticoid released during periods of distress, participates in the neuroendocrine signaling axis and is associated with suppression of immune cell effector functions, may also play a role (Silverman & Sternberg, 2012). The chronic release of

151 glucocorticoids, as mediated by mechanisms associated with chronically high levels of distress, may promote tumor progression by suppressing the anti-GBM immune response

(Dhabhar & McEwen, 1997; Eng et al., 2014). While the impact of endogenous glucocorticoid signaling has not been characterized among brain tumor patients and their associated outcomes, a large body of evidence suggests that the use of exogenously- administered dexamethasone (trade name, Decadron), is potently immunosuppressive

(Giles et al., 2018). Dexamethasone is a corticosteroid commonly used to relieve symptomatic effects and intracranial edema during therapy for high-grade glioma.

Treatment with dexamethasone is associated with a worse prognosis of patients with

GBM (Pitter et al., 2016; Shields et al., 2015). In contrast, the fast tapering of dexamethasone is associated with significantly better outcomes (Diez Valle et al., 2018).

Furthermore, phase I trial evaluation of GBM patients treated with the immune checkpoint inhibitor, atezolizumab, had a decreased level of circulating lymphocytes and a trend toward decreased OS, when concurrently treated with corticosteroids (Rimas V. Lukas et al., 2018). Strikingly, long-term survivors in the trial were not treated with corticosteroids.

While it’s inappropriate to assume that the effects of synthetic steroids are directly comparable to endogenous glucocorticoid signaling, it is possible that the consistent release of cortisol, due to high distress levels, suppress leukocyte functions and result in a worse patient prognosis, similar to those effects associated with dexamethasone treatment.

Impact of distress on outcomes in brain cancer patients

Psychological distress has been linked to a reduced quality of life among individuals diagnosed with glioma and in other brain tumor patients (Hartung et al., 2017;

152

Hoffmann, Kamp, Steiger, Sabel, & Rapp, 2017). A recent meta-analysis found a decrease in glioma patient OS when also diagnosed with depression (Shi et al., 2018).

Similarly, a retrospective analysis of 1,003 patients undergoing surgical management of high-grade glioma found that, a clinical diagnosis of depression, prior to surgical intervention, was associated with a decreased landmark rate of survival (Gathinji et al.,

2009). Although there was significant heterogeneity within the studies, these collective results suggest a potential for depression to negatively impact the physical health of glioma patients. In contrast, demographic factors associated with the potential strengthening of a glioma patient’s support system, including marriage, are predictors for improved GBM patient OS (J. C. Xie, Yang, Liu, & Zhao, 2018). Exercise also improves glioma outcomes (Levin, Greenwood, Singh, Tsoi, & Newton, 2016; Ruden et al., 2011) potentially by decreasing the effects of inflammation and/or inflammatory mediator- induced depression (Gourgouvelis, Yielder, Clarke, Behbahani, & Murphy, 2018; Woods,

Wilund, Martin, & Kistler, 2012). Therefore, interventions that decrease the level of psychological distress may improve future prognoses among patient with malignant glioma (Fig. A.2).

Standardized surveys have been proposed as mechanisms to better assess psychological distress levels in patients with brain cancer (Hickmann et al., 2017;

Renovanz, Hickmann, et al., 2018; Renovanz, Tsakmaklis, et al., 2018). Reducing the stigma, to better increase the participation in psychosocial assistance programs, is also necessary to properly care for patients with elevated distress levels (Langbecker, Ekberg,

& Yates, 2017; Tondorf et al., 2018). After identification and connection with psychosocial services, proper treatment can be initiated. Currently, home-based psychosocial

153 interventions and other therapies have been shown to address distress in brain cancer patients (Amidei, 2018; Ownsworth et al., 2015).

Few studies have previously evaluated the role of medications in addressing psychosocial distress in malignant brain tumor patients. Although beta blockers may help improve survival in distressed glioma patients, based on evidence from the treatment of patients diagnosed with non-CNS cancers, the use of selective serotonin reuptake inhibitors (SSRIs) or tricyclic antidepressants (TCAs) may be another beneficial option.

Exposure of tumor cells to SSRIs causes anti-proliferative effects in vitro, and in mouse models of GBM. Theoretically, SSRIs would also decrease distress levels in glioma patients (V. C. H. Chen, Hsieh, Chen, Hsu, & Tzang, 2018; K.-H. Liu et al., 2015; Then et al., 2017). A study of 160 individuals diagnosed with glioma showed an OS benefit for 35 patients treated with an SSRI, while several reports have advocated for their use as potential adjunct therapy during standard of care (Caudill, Brown, Cerhan, & Rummans,

2011; S. K. Tan et al., 2018). Ultimately, more investigation is needed to objectively evaluate the efficacy of psychosocial modifiers in patients with malignant brain cancer.

Conclusions

Malignant brain tumors are a profoundly devastating disease associated with significant increases in depression, anxiety, and distress among patients. We propose a direct biological link between psychosocial stressors and poor outcomes in malignant glioma patients. High adrenergic signaling and endogenous steroid activity may impair immune system functions, resulting in the uncontrolled proliferation of tumor cells, as observed in non-CNS cancer patients. Notably, high levels of distress are not always appropriately identified or managed among brain tumor patients. Therefore, additional

154 work is required for understanding the role of distress on prognosis, immune suppression, tumor growth, and clinical care. Ultimately, this future work may substantially improve the quality, and potentially, quantity of life.

Figure A.1: Proposed interactions between inflammation, depression, and immune cell dysfunction in subjects with brain cancer. Brain tumors and psychosocial stress independently contribute to brain inflammation, which provide favorable conditions for the development of depression. High distress levels lead to a cycle of increased sympathetic signaling and dysfunctional HPA signaling. Together, signaling due to the sympathetic and HPA axes, decrease immune effector control of tumor cell proliferation, resulting in worse outcomes for patients. Breaking these cycles and addressing increased distress / depression in brain cancer patients may improve survival outcomes.

155

Figure A.2: Proposed interactions between inflammation, depression, and immune cell dysfunction in subjects with brain cancer. Brain tumors and psychosocial stress independently contribute to brain inflammation, which provide favorable conditions for the development of depression. High distress levels lead to a cycle of increased sympathetic signaling and dysfunctional HPA signaling. Together, signaling due to the sympathetic and HPA axes, decrease immune effector control of tumor cell proliferation, resulting in worse outcomes for patients. Breaking these cycles and addressing increased distress/depression in brain cancer patients may improve survival outcomes.

156

REFERENCES Amidei, C. 2018. Symptom-based interventions to promote quality survivorship, Neuro Oncol, 20: vii27-vii39. Antonia, S. J., Lopez-Martin, J. A., Bendell, J., Ott, P. A., Taylor, M., Eder, J. P., Calvo, E. (2016). Nivolumab alone and nivolumab plus ipilimumab in recurrent small-cell lung cancer (CheckMate 032): a multicentre, open-label, phase 1/2 trial. Lancet Oncol, 17(7), 883-895. Assad Kahn, S., Costa, S. L., Gholamin, S., Nitta, R. T., Dubois, L. G., Feve, M., Chneiweiss, H. (2016). The anti-hypertensive drug prazosin inhibits glioblastoma growth via the PKCdelta-dependent inhibition of the AKT pathway. EMBO Mol Med, 8(5), 511-526. Bellmunt, J., de Wit, R., Vaughn, D., Fradet, Y., Lee, J., Fong, L., Bajorin, D. (2017). Pembrolizumab as second-line therapy for advanced urothelial carcinoma. N Engl J Med, 376(11), 1015-1026. Bloch, Orin, Qian Shi, S. Keith Anderson, Michael Knopp, Jeffrey Raizer, Jennifer Clarke, Allen Waziri, Howard Colman, Jeffrey Bruce, Jeffrey J. Olson, John Schwerkoske, Andrew Parsa, Joon Uhm, Priya Kumthekar, Evanthia Galanis, and Ian Parney. 2017. ATIM-14. ALLIANCE A071101: A phase II randomized trial comparint the efficacy of heat shock protein peptide complex-96 (HSPPC-96) vaccine given with bevacizumab versus bevacizumab alone in the treatment of surgically resectable recurrent glioblastoma, Neuro-Oncol, 19:vi29-vi29. Bodnar, C. N., Morganti, J. M., & Bachstetter, A. D. (2018). Depression following a traumatic brain injury: uncovering cytokine dysregulation as a pathogenic mechanism. Neural Regen Res, 13(10), 1693-1704. Bucsek, M. J., Qiao, G., MacDonald, C. R., Giridharan, T., Evans, L., Niedzwecki, B., Repasky, E. A. (2017). beta-adrenergic signaling in mice housed at standard temperatures suppresses an effector phenotype in CD8(+) T cells and undermines checkpoint inhibitor therapy. Cancer Res, 77(20), 5639-5651. Cardwell, C. R., Pottegard, A., Vaes, E., Garmo, H., Murray, L. J., Brown, C., Bennett, K. (2016). Propranolol and survival from breast cancer: a pooled analysis of European breast cancer cohorts. Breast Cancer Res, 18(1), 119. Catt, S., Chalmers, A., & Fallowfield, L. (2008). Psychosocial and supportive-care needs in high-grade glioma. Lancet Oncol, 9(9), 884-891. Caudill, J. S., Brown, P. D., Cerhan, J. H., & Rummans, T. A. (2011). Selective Serotonin Reuptake Inhibitors, Glioblastoma Multiforme, and Impact on Toxicities and Overall Survival: The Mayo Clinic Experience. American Journal of Clinical Oncology, 34(4). Chen, R., Zinzani, P. L., Fanale, M. A., Armand, P., Johnson, N. A., Brice, P., Moskowitz, C. H. (2017). Phase II study of the efficacy and safety of pembrolizumab for relapsed/refractory classic hodgkin lymphoma. J Clin Oncol, 35(19), 2125-2132. Chen, V. C. H., Hsieh, Y. H., Chen, L. J., Hsu, T. C., & Tzang, B. S. (2018). Escitalopram oxalate induces apoptosis in U‐87MG cells and autophagy in GBM8401 cells. Journal of Cellular and Molecular Medicine, 22(2), 1167-1178. Chiorean, R., Berindan-Neagoe, I., Braicu, C., Florian, I. S., Leucuta, D., Crisan, D., & Cernea, V. (2014). Quantitative expression of serum biomarkers involved in

157

angiogenesis and inflammation, in patients with glioblastoma multiforme: correlations with clinical data. Cancer Biomark, 14(2-3), 185-194. Darefsky, A. S., King, J. T., & Dubrow, R. (2012). Adult glioblastoma multiforme survival in the temozolomide era: a population-based analysis of Surveillance, Epidemiology, and End Results registries. Cancer, 118(8), 2163-2172. Dhabhar, F. S., & McEwen, B. S. (1997). Acute stress enhances while chronic stress suppresses cell-mediated immunity in vivo: a potential role for leukocyte trafficking. Brain Behav Immun, 11(4), 286-306. Diez Valle, R., Becerra Castro, V., Marigil Sanchez, M., Gallego Perez-Larraya, J., Nunez-Cordoba, J. M., & Tejada Solis, S. (2018). Results of a Policy of Fast Tapering of Steroids After Resection Surgery in Glioblastoma. World Neurosurg, 109, e845-e852. Dubrow, R., Darefsky, A. S., Jacobs, D. I., Park, L. S., Rose, M. G., Laurans, M. S. H., & King, J. T. (2013). Time trends in glioblastoma multiforme survival: the role of temozolomide. Neuro-Oncology, 15(12), 1750-1761. doi:10.1093/neuonc/not122 Dutoit, V., Migliorini, D., Dietrich, P. Y., & Walker, P. R. (2016). Immunotherapy of Malignant Tumors in the Brain: How Different from Other Sites? Front Oncol, 6, 256. Edelstein, K., Coate, L., Massey, C., Jewitt, N. C., Mason, W. P., & Devins, G. M. (2016). Illness intrusiveness and subjective well-being in patients with glioblastoma. J Neurooncol, 126(1), 127-135. Eng, J. W., Kokolus, K. M., Reed, C. B., Hylander, B. L., Ma, W. W., & Repasky, E. A. (2014). A nervous tumor microenvironment: the impact of adrenergic stress on cancer cells, immunosuppression, and immunotherapeutic response. Cancer Immunol Immunother, 63(11), 1115-1128. Feinstein, A., Magalhaes, S., Richard, J.-F., Audet, B., & Moore, C. (2014). The link between multiple sclerosis and depression. Nature Reviews Neurology, 10, 507. Fenn, A. M., Gensel, J. C., Huang, Y., Popovich, P. G., Lifshitz, J., & Godbout, J. P. (2014). Immune activation promotes depression 1 month after diffuse brain injury: a role for primed microglia. Biol Psychiatry, 76(7), 575-584. Frenel, J. S., Le Tourneau, C., O'Neil, B., Ott, P. A., Piha-Paul, S. A., Gomez-Roca, C., Varga, A. (2017). Safety and Efficacy of Pembrolizumab in Advanced, Programmed Death Ligand 1-Positive Cervical Cancer: Results From the Phase Ib KEYNOTE-028 Trial. J Clin Oncol, 35(36), 4035-4041. Gathinji, M., McGirt, M. J., Attenello, F. J., Chaichana, K. L., Than, K., Olivi, A., . . . Quinones-Hinojosa, A. (2009). Association of preoperative depression and survival after resection of malignant brain astrocytoma. Surg Neurol, 71(3), 299- 303, discussion 303. Giles, A. J., Hutchinson, M. N. D., Sonnemann, H. M., Jung, J., Fecci, P. E., Ratnam, N. M., Gilbert, M. R. (2018). Dexamethasone-induced immunosuppression: mechanisms and implications for immunotherapy. J Immunother Cancer, 6(1), 51. Gourgouvelis, J., Yielder, P., Clarke, S. T., Behbahani, H., & Murphy, B. A. (2018). Exercise Leads to Better Clinical Outcomes in Those Receiving Medication Plus Cognitive Behavioral Therapy for Major Depressive Disorder. Front Psychiatry, 9, 37.

158

Hartung, T. J., Brahler, E., Faller, H., Harter, M., Hinz, A., Johansen, C., . . . Mehnert, A. (2017). The risk of being depressed is significantly higher in cancer patients than in the general population: Prevalence and severity of depressive symptoms across major cancer types. Eur J Cancer, 72, 46-53. He, J.-J., Zhang, W.-H., Liu, S.-L., Chen, Y.-F., Liao, C.-X., Shen, Q.-Q., & Hu, P. (2017). Activation of β-adrenergic receptor promotes cellular proliferation in human glioblastoma. Oncology Letters, 14(3), 3846-3852. Hickmann, A. K., Hechtner, M., Nadji-Ohl, M., Janko, M., Reuter, A. K., Kohlmann, K., Renovanz, M. (2017). Evaluating patients for psychosocial distress and supportive care needs based on health-related quality of life in primary brain tumors: a prospective multicenter analysis of patients with gliomas in an outpatient setting. J Neurooncol, 131(1), 135-151. Hoffmann, K., Kamp, M., Steiger, H. J., Sabel, M., & Rapp, M. (2017). Correlation of psychooncological distress- screening and quality of life assessment in neurosurgical patients. Oncotarget, 8(67), 111396-111404. Huang, J., Zeng, C., Xiao, J., Zhao, D., Tang, H., Wu, H., & Chen, J. (2017). Association between depression and brain tumor: a systematic review and meta-analysis. Oncotarget, 8(55), 94932-94943. Kmiecik, J., Poli, A., Brons, N. H., Waha, A., Eide, G. E., Enger, P. O., . . . Chekenya, M. (2013). Elevated CD3+ and CD8+ tumor-infiltrating immune cells correlate with prolonged survival in glioblastoma patients despite integrated immunosuppressive mechanisms in the tumor microenvironment and at the systemic level. J Neuroimmunol, 264(1-2), 71-83. Kokolus, K. M., Zhang, Y., Sivik, J. M., Schmeck, C., Zhu, J., Repasky, E. A., . . . Schell, T. D. (2018). Beta blocker use correlates with better overall survival in metastatic melanoma patients and improves the efficacy of immunotherapies in mice. Oncoimmunology, 7(3), e1405205. Kore, R. A., & Abraham, E. C. (2014). Inflammatory cytokines, interleukin-1 beta and tumor necrosis factor-alpha, upregulated in glioblastoma multiforme, raise the levels of CRYAB in exosomes secreted by U373 glioma cells. Biochem Biophys Res Commun, 453(3), 326-331. Ladwig, S., Zhou, Z., Xu, Y., Wang, X., Chow, C. K., Werheid, K., & Hackett, M. L. (2018). Comparison of Treatment Rates of Depression After Stroke Versus Myocardial Infarction: A Systematic Review and Meta-Analysis of Observational Data. Psychosom Med, 80(8), 754-763. Langbecker, D., Ekberg, S., & Yates, P. (2017). Don't need help, don't want help, can't get help: How patients with brain tumors account for not using rehabilitation, psychosocial and community services. Patient Educ Couns, 100(9), 1744-1750. Langbecker, D., & Yates, P. (2016). Primary brain tumor patients' supportive care needs and multidisciplinary rehabilitation, community and psychosocial support services: awareness, referral and utilization. J Neurooncol, 127(1), 91-102. Larkin, J., Chiarion-Sileni, V., Gonzalez, R., Grob, J. J., Cowey, C. L., Lao, C. D., . . . Wolchok, J. D. (2015). Combined Nivolumab and Ipilimumab or Monotherapy in Untreated Melanoma. N Engl J Med, 373(1), 23-34. Levada, O. A., & Troyan, A. S. (2018). Poststroke Depression Biomarkers: A Narrative Review. Front Neurol, 9.

159

Levin, G. T., Greenwood, K. M., Singh, F., Tsoi, D., & Newton, R. U. (2016). Exercise Improves Physical Function and Mental Health of Brain Cancer Survivors: Two Exploratory Case Studies. Integr Cancer Ther, 15(2), 190-196. Liu, F., Huang, J., Zhang, L., Fan, F., Chen, J., Xia, K., & Liu, Z. (2018). Screening for distress in patients with primary brain tumor using distress thermometer: a systematic review and meta-analysis. BMC Cancer, 18(1), 124. Liu, K.-H., Yang, S.-T., Lin, Y.-K., Lin, J.-W., Lee, Y.-H., Wang, J.-Y., . . . Shen, S.-C. (2015). Fluoxetine, an antidepressant, suppresses glioblastoma by evoking AMPAR-mediated calcium-dependent apoptosis. Oncotarget, 6(7), 5088-5101. Liu, R., Page, M., Solheim, K., Fox, S., & Chang, S. M. (2009). Quality of life in adults with brain tumors: current knowledge and future directions. Neuro Oncol, 11(3), 330-339. Lohr, J., Ratliff, T., Huppertz, A., Ge, Y., Dictus, C., Ahmadi, R., . . . Herold-Mende, C. (2011). Effector T-cell infiltration positively impacts survival of glioblastoma patients and is impaired by tumor-derived TGF-beta. Clin Cancer Res, 17(13), 4296-4308. Lukas, R. V., Rodon, J., Becker, K., Wong, E. T., Shih, K., Touat, M., . . . Baehring, J. (2018). Clinical activity and safety of atezolizumab in patients with recurrent glioblastoma. J Neurooncol. Miller, A. H., & Raison, C. L. (2016). The role of inflammation in depression: from evolutionary imperative to modern treatment target. Nat Rev Immunol, 16(1), 22- 34. Mitchell, A. J., Chan, M., Bhatti, H., Halton, M., Grassi, L., Johansen, C., & Meader, N. (2011). Prevalence of depression, anxiety, and adjustment disorder in oncological, haematological, and palliative-care settings: a meta-analysis of 94 interview-based studies. Lancet Oncol, 12(2), 160-174. Montoya, A., Amaya, C. N., Belmont, A., Diab, N., Trevino, R., Villanueva, G., . . . Nahleh, Z. (2017). Use of non-selective beta-blockers is associated with decreased tumor proliferative indices in early stage breast cancer. Oncotarget, 8(4), 6446-6460. Morris, G., Reiche, E. M. V., Murru, A., Carvalho, A. F., Maes, M., Berk, M., & Puri, B. K. (2018). Multiple Immune-Inflammatory and Oxidative and Nitrosative Stress Pathways Explain the Frequent Presence of Depression in Multiple Sclerosis. Mol Neurobiol, 55(8), 6282-6306. Motzer, R. J., Escudier, B., McDermott, D. F., George, S., Hammers, H. J., Srinivas, S., . . . CheckMate, I. (2015). Nivolumab versus Everolimus in Advanced Renal-Cell Carcinoma. N Engl J Med, 373(19), 1803-1813. Muro, K., Chung, H. C., Shankaran, V., Geva, R., Catenacci, D., Gupta, S., . . . Bang, Y. J. (2016). Pembrolizumab for patients with PD-L1-positive advanced gastric cancer (KEYNOTE-012): a multicentre, open-label, phase 1b trial. Lancet Oncol, 17(6), 717-726. Nissen, M. D., Sloan, E. K., & Mattarollo, S. R. (2018). beta-Adrenergic Signaling Impairs Antitumor CD8(+) T-cell Responses to B-cell Lymphoma Immunotherapy. Cancer Immunol Res, 6(1), 98-109. Osborn, A. J., Mathias, J. L., & Fairweather-Schmidt, A. K. (2014). Depression following adult, non-penetrating traumatic brain injury: a meta-analysis examining

160

methodological variables and sample characteristics. Neurosci Biobehav Rev, 47, 1-15. Ostrom, Q. T., Gittleman, H., Liao, P., Vecchione-Koval, T., Wolinsky, Y., Kruchko, C., & Barnholtz-Sloan, J. S. (2017). CBTRUS Statistical Report: Primary brain and other central nervous system tumors diagnosed in the United States in 2010–2014. Neuro-Oncology, 19(suppl_5), v1-v88. Ownsworth, T., Chambers, S., Damborg, E., Casey, L., Walker, D. G., & Shum, D. H. (2015). Evaluation of the making sense of brain tumor program: a randomized controlled trial of a home-based psychosocial intervention. Psychooncology, 24(5), 540-547. Pace, A., Dirven, L., Koekkoek, J. A. F., Golla, H., Fleming, J., Ruda, R., . . . Taphoorn, M. J. B. (2017). European Association for Neuro-Oncology (EANO) guidelines for palliative care in adults with glioma. Lancet Oncol, 18(6), e330-e340. Partecke, L. I., Speerforck, S., Kading, A., Seubert, F., Kuhn, S., Lorenz, E., . . . von Bernstorff, W. (2016). Chronic stress increases experimental pancreatic cancer growth, reduces survival and can be antagonised by beta-adrenergic receptor blockade. Pancreatology, 16(3), 423-433. Philip, J., Collins, A., Brand, C. A., Moore, G., Lethborg, C., Sundararajan, V., Murphy, V.A., Gold, M. 2014. “I’m just waiting…”: an exploration of the experience of living and dying with primary malignant glioma, Support Care Cancer, 22(2):389-97. Pitter, K. L., Tamagno, I., Alikhanyan, K., Hosni-Ahmed, A., Pattwell, S. S., Donnola, S., Hambardzumyan, D. (2016). Corticosteroids compromise survival in glioblastoma. Brain, 139(Pt 5), 1458-1471. Powell, N. D., Tarr, A. J., & Sheridan, J. F. (2013). Psychosocial stress and inflammation in cancer. Brain Behav Immun, 30 Suppl, S41-47. Qiao, G., Chen, M., Bucsek, M. J., Repasky, E. A., & Hylander, B. L. (2018). Adrenergic Signaling: A Targetable Checkpoint Limiting Development of the Antitumor Immune Response. Front Immunol, 9, 164. Randazzo, D., & Peters, K. B. (2016). Psychosocial distress and its effects on the health- related quality of life of primary brain tumor patients. CNS Oncol, 5(4), 241-249. Reardon, D. A., Omuro, A., Brandes, A. A., Rieger, J., Wick, A., Sepulveda, J., Sampson, J. (2017). OS10.3 Randomized Phase 3 Study Evaluating the Efficacy and Safety of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: CheckMate 143. Neuro-Oncology, 19(suppl_3), iii21-iii21. Renovanz, M., Hechtner, M., Janko, M., Kohlmann, K., Coburger, J., Nadji-Ohl, M., . . . Hickmann, A. K. (2017). Factors associated with supportive care needs in glioma patients in the neuro-oncological outpatient setting. J Neurooncol, 133(3), 653- 662. Renovanz, M., Hickmann, A. K., Coburger, J., Kohlmann, K., Janko, M., Reuter, A. K., Hechtner, M. (2018). Assessing psychological and supportive care needs in glioma patients - feasibility study on the use of the Supportive Care Needs Survey Short Form (SCNS-SF34-G) and the Supportive Care Needs Survey Screening Tool (SCNS-ST9) in clinical practice. Eur J Cancer Care (Engl), 27(1). Renovanz, M., Tsakmaklis, H., Soebianto, S., Neppel, I., Nadji-Ohl, M., Beutel, M., . . . Hickmann, A. K. (2018). Screening for distress in patients with intracranial tumors during the first 6 months after diagnosis using self-reporting instruments and an

161

expert rating scale (the basic documentation for psycho-oncology short form - PO- Bado SF). Oncotarget, 9(57), 31133-31145. Rooney, A. G., Carson, A., & Grant, R. (2011). Depression in Cerebral Glioma Patients: A Systematic Review of Observational Studies. JNCI: Journal of the National Cancer Institute, 103(1), 61-76. Ruden, E., Reardon, D. A., Coan, A. D., Herndon, J. E., Hornsby, W. E., West, M., . . . Jones, L. W. (2011). Exercise Behavior, Functional Capacity, and Survival in Adults With Malignant Recurrent Glioma. Journal of Clinical Oncology, 29(21), 2918-2923. Rundle‐Thiele, D., Head, R., Cosgrove, L., & Martin, J. H. (2016). Repurposing some older drugs that cross the blood–brain barrier and have potential anticancer activity to provide new treatment options for glioblastoma. British Journal of Clinical Pharmacology, 81(2), 199-209. Sampson, J. H., Heimberger, A. B., Archer, G. E., Aldape, K. D., Friedman, A. H., Friedman, H. S., . . . Bigner, D. D. (2010). Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant III peptide vaccination in patients with newly diagnosed glioblastoma. J Clin Oncol, 28(31), 4722-4729. Sardi, I., Giunti, L., Bresci, C., Buccoliero, A. M., Degl'innocenti, D., Cardellicchio, S., . . . Filippi, L. (2013). Expression of beta-adrenergic receptors in pediatric malignant brain tumors. Oncol Lett, 5(1), 221-225. Schmidt, D., Peterlik, D., Reber, S. O., Lechner, A., & Männel, D. N. (2016). Induction of Suppressor Cells and Increased Tumor Growth following Chronic Psychosocial Stress in Male Mice. PLoS ONE, 11(7), e0159059. Schuster, J., Lai, R. K., Recht, L. D., Reardon, D. A., Paleologos, N. A., Groves, M. D., . . . Sampson, J. H. (2015). A phase II, multicenter trial of rindopepimut (CDX-110) in newly diagnosed glioblastoma: the ACT III study. Neuro Oncol, 17(6), 854-861. Shi, C., Lamba, N., Zheng, L. J., Cote, D., Regestein, Q. R., Liu, C. M., . . . Broekman, M. L. D. (2018). Depression and survival of glioma patients: A systematic review and meta-analysis. Clinical Neurology and Neurosurgery, 172, 8-19 Shields, L. B., Shelton, B. J., Shearer, A. J., Chen, L., Sun, D. A., Parsons, S., . . . Spalding, A. C. (2015). Dexamethasone administration during definitive radiation and temozolomide renders a poor prognosis in a retrospective analysis of newly diagnosed glioblastoma patients. Radiat Oncol, 10, 222. Silverman, M. N., & Sternberg, E. M. (2012). Glucocorticoid regulation of inflammation and its behavioral and metabolic correlates: from HPA axis to glucocorticoid receptor dysfunction. Ann N Y Acad Sci, 1261, 55-63. Singer, S., Roick, J., Danker, H., Kortmann, R. D., Papsdorf, K., Taubenheim, S., . . . Meixensberger, J. (2018). Psychiatric co-morbidity, distress, and use of psychosocial services in adult glioma patients-a prospective study. Acta Neurochir (Wien), 160(6), 1187-1194. Slavich, G. M., & Irwin, M. R. (2014). From Stress to Inflammation and Major Depressive Disorder: A Social Signal Transduction Theory of Depression. Psychological bulletin, 140(3), 774-815. Sterckx, W., Coolbrandt, A., Clement, P., Borgenon, S., Decruyenaere, M., De Vleeschouwer, S., . . . Dierckx de Casterle, B. (2015). Living with a high-grade

162

glioma: A qualitative study of patients' experiences and care needs. Eur J Oncol Nurs, 19(4), 383-390. Stupp, R., Taillibert, S., Kanner, A., Read, W., Steinberg, D., Lhermitte, B., . . . Ram, Z. (2017). Effect of Tumor-Treating Fields Plus Maintenance Temozolomide vs Maintenance Temozolomide Alone on Survival in Patients With Glioblastoma: A Randomized Clinical Trial. JAMA, 318(23), 2306-2316. Tan, S. K., Jermakowicz, A., Mookhtiar, A. K., Nemeroff, C. B., Schürer, S. C., & Ayad, N. G. (2018). Drug Repositioning in Glioblastoma: A Pathway Perspective. Frontiers in Pharmacology, 9, 218. Then, C.-K., Liu, K.-H., Liao, M.-H., Chung, K.-H., Wang, J.-Y., & Shen, S.-C. (2017). Antidepressants, sertraline and paroxetine, increase calcium influx and induce mitochondrial damage-mediated apoptosis of astrocytes. Oncotarget, 8(70), 115490-115502. Tondorf, T., Grossert, A., Rothschild, S. I., Koller, M. T., Rochlitz, C., Kiss, A., . . . Zwahlen, D. (2018). Focusing on cancer patients' intentions to use psychooncological support: A longitudinal, mixed-methods study. Psychooncology, 27(6), 1656-1663. United States Cancer Statistics: 1999–2011 Incidence, WONDER Online Database. (2015). Retrieved September 9, 2018, from Centers for Disease Control and Prevention National Center for Health Statistics. Vehling, S., Kissane, D. W., Lo, C., Glaesmer, H., Hartung, T. J., Rodin, G., & Mehnert, A. (2017). The association of demoralization with mental disorders and suicidal ideation in patients with cancer. Cancer, 123(17), 3394-3401. Wang, F., Liu, H., Wang, F., Xu, R., Wang, P., Tang, F., Han, T. (2018). Propranolol suppresses the proliferation and induces the apoptosis of liver cancer cells. Mol Med Rep, 17(4), 5213-5221. Weller, M., Butowski, N., Tran, D. D., Recht, L. D., Lim, M., Hirte, H. (2017). Rindopepimut with temozolomide for patients with newly diagnosed, EGFRvIII-expressing glioblastoma (ACT IV): a randomised, double-blind, international phase 3 trial. Lancet Oncol, 18(10), 1373-1385. Woods, J. A., Wilund, K. R., Martin, S. A., & Kistler, B. M. (2012). Exercise, Inflammation and Aging. Aging Dis, 3(1), 130-140. Xie, J. C., Yang, S., Liu, X. Y., & Zhao, Y. X. (2018). Effect of marital status on survival in glioblastoma multiforme by demographics, education, economic factors, and insurance status. Cancer Medicine, 7(8), 3722-3742. Yeung, Y. T., McDonald, K. L., Grewal, T., & Munoz, L. (2013). Interleukins in glioblastoma pathophysiology: implications for therapy. Br J Pharmacol, 168(3), 591-606. Zhai, L., E. Ladomersky, K. L. Lauing, M. Wu, M. Genet, G. Gritsina, G. Balaz, P.K. Brastianos, D. C. Binder, J. A. Sosman, F. J. Giles, C. D. James, C. Horbinski, R. Stupp, and D. A. Wainwright. 2017. Infiltrating T cells increase IDO1 expression in glioblastoma and contribute to decreased patient survival, Clin Cancer Res, 23(21):6650-6660. Zinzani, P. L., Ribrag, V., Moskowitz, C. H., Michot, J. M., Kuruvilla, J., Balakumaran, A., . . . Armand, P. (2017). Safety and tolerability of pembrolizumab in patients with relapsed/refractory primary mediastinal large B-cell lymphoma. Blood, 130(3), 267-270.

163