Jul 24, 2017 - "ZEB1 Loss Drives Radio-resistance in Glioma Stem Cells" Cancer ..... Through human development most cells in the body mature from stem-like precursors ...... We used lentiviral particles to transfect shRNA for ASNS in the.
CEDARS SINAI HEALTH SYSTEM
BIOMEDICAL SCIENCE & TRANSLATIONAL MEDICINE
Asparagine Synthetase (ASNS) Copy-number Gain in Glioma Stem Cells Drives Cellular Plasticity and Resistance to Oxidative Stress via Adaptive Redox Homeostasis
By
Tom M. Thomas
A dissertation presented to the Graduate Program in Biomedical Sciences & Translational Medicine Of the Cedars Sinai Health System In partial fulfillment of the requirements for the degree of Doctor of Philosophy
CEDARS-SINAI MEDICAL CENTER LOS ANGELES, CALIFORNIA JULY 24, 2017 1
© Tom M. Thomas, All Rights Reserved, 2017
2
TABLE OF CONTENTS
Dissertation Committee Publication List Acknowledgements Dedication List of Figures List of Abbreviations Abstract
4 5 6 8 11 13 15
Introduction Glioblastoma Cancer Stem Cells & Tumor Hierarchy Metabolism & Cancer Amino Acid Metabolism in Cancer
17 25 32 41
Chapter I: ASNS COPY NUMBER GAIN IS A POOR PROGNOSTIC FACTOR IN GLIOMA PATIENTS Introduction & Rationale Results & Discussion
45 48
Chapter II: ASNS COPY NUMBER GAIN IMPROVES GROWTH & SURVIVAL OF GSCS Introduction & Rationale Results & Discussion
56 57
Chapter III: ASNS GAIN CONFERS METABOLIC PLASTICITY & REMODELS REDOX HOMEOSTASIS Introduction & Rationale Results & Discussion
77 78
Conclusions & Future Directions Conclusion Future Directions
97 101
Materials & Methods
102
Appendix I
110
References
119
3
Dissertation Thesis Committee
John S. Yu, M.D. Professor of Neurosurgery Director, Surgical Neuro-Oncology, Cedars-Sinai Medical Center Thesis Research Mentor Joshua Breunig, Ph.D. Assistant Professor, Department of Biomedical Sciences Director, Confocal Microscopy Core, Cedars-Sinai Medical Center Chair, Thesis Committee H. Philip Koeffler, M.D. Mark Goodson Chair in Oncology Research Cedars-Sinai Medical Center Roberta Gottlieb, M.D. Dorothy and E. Philip Lyon Chair in Molecular Cardiobiology Director, Molecular Cardiobiology, Cedars-Sinai Medical Center Dhruv Sareen, Ph.D. Assistant Professor, Department of Biomedical Sciences Director, Induced Pluripotent Stem Cell Core, Cedars-Sinai Medical Center Jethro L. Hu, M.D. Attending Physician, Neuro-Oncology Cedars-Sinai Medical Center Clinical Research Mentor Erina Vlashi, Ph.D. Assistant Professor, Department of Radiation Oncology Division of Molecular & Cellular Oncology David Geffen School of Medicine, UCLA External Committee Member
4
Publications During Dissertation Work
Research Articles Thomas TM, Edwards L, Madany M, Gottlieb R, Black K, Yu JS. “Asparagine Synthetase (ASNS) Copy-number Gain in Glioma Stem Cells Drives Cellular Plasticity and Resistance to Oxidative Stress via Adaptive Redox Homeostasis” Molecular Cell (2017, Manuscript in preparation) Edwards LE, Kim S, Madany M, Nuno M, Thomas TM, Li A, Berel D, Lee BS, Lui M, Black K, Fan X, Zhang W, Yu JS. "ZEB1 is a transcription factor that is prognostic and predictive in diffuse gliomas" Clinical Cancer Research [2017, In Revision] Madany M, Edwards LE, Kim S, Thomas TM, Nuno M, Lee BS, Lui M, Black K, Fan X, Zhang W, Yu JS. "ZEB1 Loss Drives Radio-resistance in Glioma Stem Cells" Cancer Research (2017, Manuscript in Preparation) Sun, H., Lin, D. C., Cao, Q., Guo, X., Marijon, H., Zhao, Z., ... & Lee, V. K. M. (2016). CRM1 Inhibition Promotes Cytotoxicity in Ewing Sarcoma Cells by Repressing EWSFLI1–Dependent IGF-1 Signaling. Cancer Research, 76(9), 2687-2697. Review Articles Thomas T.M, Yu JS. “Metabolic Regulation of Glioma Stem-like Cells in the Tumor Micro-environment.” Cancer Letters, 2017. [accepted] Book Chapters Madany, M., Thomas, T. M., Edwards, L., & Yu, J. S. (2015). Immunobiology and Immunotherapeutic Targeting of Glioma Stem Cells. In M. Ehtesham (Ed.), Stem Cell Biology in Neoplasms of the Central Nervous System (pp. 139–166). Springer International Publishing.
5
Acknowledgements
First and foremost, I thank God for all His guidance and blessings that have lead up to this and continue to lead me; every moment, every day.
I came to Cedars-Sinai with a very specific interest in cancer research having seen the disease up front, in my own life. The level of exposure to the clinic and the commitment to really driving at translational research has allowed me to pursue some of the most interesting questions in brain tumor biology today without compromise. Being able to interact with clinicians and researchers in a very open and dynamic setting is very rare and very valuable, but very common here at Cedars. This program was always unique in that way, and my research and training has been better as a result.
The most uncomfortable places in life are filled with unknowns and uncertainties. The Ph.D. process is ripe with these. There are several specific people who were critical to my success. I want to thank Lincoln Edwards, Ph.D., who was the person who taught me about cancer stem cells and inspired much of the early work on the ASNS project. Lincoln’s support and advise was instrumental in the progress of the project, but more importantly in my progress as a scientist. I want to thank Roberta Gottlieb, M.D., who was a huge support when it came to learning and mastering many of the techniques and problems in the lab specifically with regards to cancer metabolism. Dr. Gottlieb was not only a scientific assist on this project but was a very important personal mentor to me, and I will be forever grateful for that. I want to thank Bong Seop Lee, Ph.D., for this
6
continued support and encouragement throughout this process, and this sharing of his nano-particle research applied to my work. I want to thank Mecca Madany, my fellow graduate student in the Yu Laboratory, for being a great friend and help, always in the lab trenches with me. I finally and most especially want to thank my thesis mentor, John Yu M.D., for his guidance and support through this entire process. He was extremely supportive and encouraging both when the project was succeeding and failing at times, and his clinical insights were very useful in developing the most translationally relevant adjustments to the work. Dr. Yu is and will continue to be a scientific ally and personal mentor to me and I am very grateful for his support and friendship.
Additionally, I want to extend a thanks to the graduate program, my fellow students, my fellow lab members, and clinical collaborators.
On personal note, I want to wholeheartedly thank my family for their support, encouragement, and patience through all of it. I am the son of two cancer survivors and this work is extremely personal to me. I’m driven every day by the strength and courage I see in my family, and I continue to work for all those families who continue to face this disease every day, throughout the world.
7
Dedication
I have been supported throughout my life by many wonderful people; friends, family, and teachers. But the real drivers of all my success and all my passion for this work have been my parents, Rose & Raju Thomas. There is not a day that goes by where why I do what I do is lost on me, and that is no small part due to my parents.
This one is for you.
8
“The Spirit of the Lord will rest on him – the Spirit of wisdom and of understanding, the Spirit of counsel and of might, the Spirit of the knowledge and fear of the Lord.” -
Isaiah 11:2
9
“In God we trust; all others bring data.” -
W. Edwards Deming
10
LIST OF FIGURES
FIGURE 1. SIGNIFICANT SOMATIC CHROMOSOMAL ABERRATIONS IN GLIOBLASTOMA (TCGA)
21
FIGURE 2. TREATMENT SCHEMA IN GLIOMA PATIENT CARE
24
FIGURE 3. EVOLVING MODELS OF TUMOR HETEROGENEITY DRIVEN BY CANCER STEM CELLS
27
FIGURE 4. DEFINING FEATURES OF CANCER STEM CELLS
29
FIGURE 5. METABOLIC REPROGRAMMING IN CANCER CELLS BY WARBURG EFFECT
35
FIGURE 6. METABOLIC PLASTICITY EXHIBITED BY CANCER STEM CELLS
39
FIGURE 7. CANCER STEM CELL FOCUSED TREATMENT PARADIGMS
40
FIGURE 8. ASPARAGINE SYNTHETASE (ASNS)
43
FIGURE 9. ASNS CORRELATES TO POOR PATIENT SURVIVAL AND HIGH TUMOR GRADE
45
FIGURE 10. ASNS CORRESPONDS TO MORE AGGRESSIVE BRAIN TUMOR PROFILE
48
FIGURE 11. ASNS CORRESPONDS TO MORE AGGRESSIVE BRAIN TUMOR PROFILE; INVERSELY CORRELATED TO IDH1 MUTANT
50
FIGURE 12. ASNS EXPRESSION IN PATIENT-DERIVED GLIOMA STEM CELL LINES WITH SOME
53
MATCHING CLINICAL DATA
FIGURE 13 ASNS EXPRESSION CONSTRUCTS USING GSC827 AND GSC604
55
FIGURE 14. STEM CHARACTERIZATION IN GSC AND GSC-ASNS CONSTRUCTS
60
FIGURE 15. IN VITRO PROLIFERATION INCREASES WITH GAIN OF ASNS
61
FIGURE 16. COMBINED GROWTH CURVE FOR GSC827 AND GSC604 CONSTRUCTS
62
FIGURE 17. RNA-SEQ DATA OF GSC827 & 604 CONSTRUCTS
63
FIGURE 18. ASNS-HIGH CELLS WERE RESISTANT TO ASN DEPLETION COMPARED TO ASNS-LOW
64
CELLS
FIGURE 19. ASNS HIGH CELLS SENSITIVE TO GLUTAMINE WITHDRAWAL
68
FIGURE 20. ASNS-HIGH CELLS WERE LESS EFFECTED BY GLUCOSE WITHDRAWAL
69
FIGURE 21. KEGG PATHWAY ANALYSIS SHOWS ENRICHMENT FOR CELL CYCLE & DNA
70
REPLICATION PATHWAYS
11
FIGURE 22. ASNS-HIGH CELLS MORE RESISTANT TO NUTRIENT RESTRICTION
72
FIGURE 23. SUBCUTANEOUS TUMOR MODEL SHOWS ASNS GAIN CONFERS SIGNIFICANT GROWTH ADVANTAGE IN VIVO.
75
FIGURE 24. INTRACRANIAL TUMOR IMPLANTATION MODEL IN B6/FOXN1-/- MICE SHOWS SIGNIFICANT SURVIVAL DIFFERENCE BASED ON ASNS STATUS
76
FIGURE 25. CELL MITO STRESS TEST SHOWS HIGHER MITOCHONDRIAL CAPACITY IN ASNS-HIGH
80
CELLS
FIGURE 26. ASNS-HIGH CELLS HAVE HIGHER MITOCHONDRIAL RESERVE CAPACITY
81
FIGURE 27. ASNS-HIGH CELLS DISPLAY MORE ROBUST METABOLIC PLASTICITY
83
FIGURE 28. ASNS-HIGH CELLS SHIFT FROM GLYCOLYSIS TO OXIDATIVE PHOSPHORYLATION MORE
84
QUICKLY
FIGURE 29. DIFFERENTIATION SHIFTS TOWARDS A MORE GLYCOLYTIC PHENOTYPE
85
FIGURE 30. ASNS GAIN ALLOWS FOR METABOLIC PLASTICITY AS WELL AS IMPROVED FUEL
86
FLEXIBILITY
FIGURE 31. ASNS-HIGH CELLS MORE RESISTANT TO OXIDATIVE STRESS
90
FIGURE 32. KEGG PATHWAY ANALYSIS INDICATES FOLATE METABOLISM DOWNSTREAM OF ASNS
94
FIGURE 33. L-ASPARAGINASE TREATMENT INDUCED SYNTHETIC LETHALITY TOWARDS MTX IN ASNS HIGH CELLS
95
LIST OF TABLES TABLE 1. SIGNIFICANT GENOMIC AMPLIFICATIONS IN GBM
112
TABLE 2. SIGNIFICANT GENOMIC DELETIONS IN GBM
113
TABLE 3. CHROMOSOMAL ALTERATIONS IN GBM
114
TABLE 4. BIOLOGICAL PROCESS (GO)
115
TABLE 5. MOLECULAR FUNCTION (GO)
116
TABLE 6. KEGG PATHWAYS
116
12
LIST OF ABBREVIATIONS ASNS- ASPARAGINE SYNTHETASE GBM- GLIOBLASTOMA WHO- WORLD HEALTH ORGANIZATION SV-40- SIMIAN VACUOLATING VIRUS 40 HHV-6- HUMAN HERPES VIRUS 6 CMV- CYTOMEGALOVIRUS TCGA- THE CANCER GENOME ATLAS EGFR- ENDOTHELIAL GROWTH FACTOR RECEPTOR IDH1- ISOCITRATE DEHYDROGENASE 1 PDGFRA- PLATELET DERIVED GROWTH FACTOR RECEPTOR A NF1- NEUROFIBROMATOSIS TYPE 1 PTEN- PHOSPHATASE AND TENSIN HOMOLOG RB1- RETINOBLASTOMA 1 FDA- FOOD & DRUG ADMINISTRATION TMZ- TEMOZOLOMIDE MGMT- O-6-METHYLGUANINE-DNA METHYLTRANSFERASE BBB- BLOOD BRAIN BARRIER CNS- CENTRAL NERVOUS SYSTEM PFS- PROGRESSION FREE SURVIVAL OS- OVERALL SURVIVAL CSC- CANCER STEM CELL GSC- GLIOMA STEM CELL SVZ- SUBVENTRICULAR ZONE VEGF- VASCULAR ENDOTHELIAL GROWTH FACTOR FGF- FIBROBLAST GROWTH FACTOR EGF- ENDOTHELIAL GROWTH FACTOR HIF- HYPOXIA INDUCIBLE FACTOR LDH- LACTATE DEHYDROGENASE ROS- REACTIVE OXYGEN SPECIES NEAA- NON-ESSENTIAL AMINO ACID 13
ALL- ACUTE LYMPHOBLASTIC LEUKEMIA L-ASNASE- L-ASPARAGINASE MSKCC- MEMORIAL SLOAN KETTERING CANCER CENTER LGG- LOWER GRADE GLIOMA HR- HAZARD RATIO IHC- IMMUNOHISTOCHEMISTRY FISH- FLUORESCENCE IN SITU HYBRIDIZATION G-CIMP- GLIOMA CPG ISLAND METHYLATOR PHENOTYPE PD-GSC- PATIENT DERIVED GLIOMA STEM CELL OCR- OXYGEN CONSUMPTION RATE ECAR- EXTRACELLULAR ACIDIFICATION RATE FCCP- CARBONYL CYANIDE P-TRIFLUOROMETHOXY-PHENYLHYDRAZONE 2-DG- 2-DEOXYGLUCOSE BSA- BOVINE SERUM ALBUMIN GSSG- GLUTATHIONE DISULFIDE GSH- GLUTATHIONE KEGG- KYOTO ENCYCLOPEDIA OF GENES AND GENOMES MTX- METHOTREXATE
14
Asparagine Synthetase (ASNS) Copy-number Gain in Glioma Stem Cells Drives Cellular Plasticity and Resistance to Oxidative Stress via Adaptive One-Carbon Metabolism by Tom M. Thomas
Abstract
Glioblastoma multiforme (GBM), also referred to as Grade IV Astrocytoma, is the most aggressive form of high-grade brain cancer and also happens to be the most common type or primary brain malignancy, with median survival ranging between 12-14 months despite multimodal intervention, prompting the need to look elsewhere for novel therapeutic approaches. We investigated the copy number amplification of the amino acid biosynthesis enzyme asparagine synthetase (ASNS) in high grade gliomas and a possible role in regulating metabolic plasticity in high grade gliomas. We identified ASNS through a largescale genomic screen of over 4000 glioma patients searching for important alterations that may have been lost by conventional thresholding from prior analyses. Through use of patient derived glioma stem cells (GSCs), we have demonstrated significant metabolic alterations occur in gliomas when perturbing the expression of ASNS that extends beyond amino acid homeostasis. ASNS-high GSCs maintained a slow-cycling basal metabolic profile while loss of ASNS shifted towards a faster cycling glycolytic phenotype, although ASNS-high cells maintained an overall more robust proliferative capacity. Moreover, ASNS-high GSCs were able to retain a certain level of metabolic plasticity that allowed them to readily switch between either glycolysis or 15
oxidative phosphorylation upon single inhibition of either pathway. These observations were more pronounced under stressed conditions. High copies of ASNS in GSCs appear to confer survival benefits under conditions of high stress which includes hypoxia, nutrient deprivation, and chemo-radiation. A particular resistance to oxidative stress was observed and corresponded to increased activation of 1C/folate cycle mediated glutathione production and redox buffering. Asparaginase treatment, which is normally ineffective against cells with high levels of ASNS, was able to further drive 1C/folate cycle activity and confer synthetic lethality to methotrexate treatment. This study therefore reveals a new role for ASNS in metabolic control and redox homeostasis in glioma stem cells and proposes a new treatment strategy that attempts to exploit an apparent metabolic vulnerability in this hard to target cell population.
Thesis Mentor: John S. Yu, M.D. Title: Professor of Neurosurgery
16
INTRODUCTION
Glioblastoma Glioblastoma (GBM, WHO grade IV gliomas) represents one of the most lethal solid tumors with one of the highest mortality rates, having an overall survival of approximately 12-14 months with the full complement of treatment, which includes tumor debulking surgery followed usually by a combination of temozolomide and radiation14. Negligible progress has been made on that survival figure despite advancements in chemotherapy, radiation therapy, and surgical techniques over the past 30 years. This has required a reassessment of where within our understanding of glioma biology the major gaps in knowledge remain and the sobering reality is that the gaps are not small. The developing view of tumors as a heterogeneous disease comprised of many subpopulations with unique properties has forced us to consider more closely the dynamics within a tumor that play a role in not only driving tumor development forward but also in maintaining tumor survival under severe stress. It is important to start to piece together the specifics of the many different networks within a tumor system including interactions between tumor subpopulations, interactions between the tumor and its stroma surrounding, interactions between the tumor and the immune components, and even interactions between the tumor and the local stem cell niche either within the tumor or the local tumor microenvironment. With the need to begin to address these gaps in understanding gliomagenesis, the tumor microenvironment and the complex metabolic networks within these tumors have come into particular focus in recent research.
17
Disease Pathology & Risk Factors – Gliomas represent a largely sporadic tumor type that is rarely dictated by any specific genetic predisposition. Despite the lack of any specific environmental or behavioral factors linked to brain tumor incidence, glioblastoma has been linked with certain viral elements including SV4015, HHV-616,17, and CMV18. A key feature of gliomas is the significant level of vascular involvement in tumor development along with a high degree of cellular necrosis. Centrally focused necrotic cores surrounded by rapidly developing hyperplastic lesions are called psuedopalisading necrosis and is a key factor in the diagnostic criteria for GBM19. These tumors with highly necrotic cores in the later stages of development tend to display a high level of heterogeneity on multiple levels, and it is common to find cellular and molecular gradients throughout a tumor mass. Oxygen and nutrient levels fluctuate across a glioma mass and so then does the expression of many signaling factors such as hypoxia inducible factors (HIFs) and many angiogenic promoters such as VEGF, often immediately adjacent to the necrotic cores. These gradients within gliomas allow for cells to survive across a spectrum and thus gliomas are very good at promoting the development of highly adaptable cellular sub-compartments within a single tumor. Cells adapted to the hypoxic, necrotic, and perinecrotic regions of the tumor mass help signal angiogenic development into these poorly vascularized spaces and help promote rapid tumor development.20,21 These cellular adaptations in hostile tumor regions usually overlap with areas of the enrichment for stem-like cells within tumors and have led some to suggest a significant role for cancer stem cells in tumor plasticity and heterogeneity.22 This cycle of newly developing necrotic cores driven by a very proliferative tumor edge followed by highly adapted cells
18
promoting neovascular supplies to these regions is a hallmark of these highly aggressive tumors and lends to the poor patient prognosis in the clinic. Molecular & Genomic Subtyping in GBM – A lot of work in recent years through the Cancer Genome Atlas (TCGA) and its many collaborators have started to suggest a genomic landscape in GBM comprised of distinct molecular subtypes.1,23-26 What has subsequently developed is the current understanding of four major molecular groups in gliomas linked to patient outcome: classical, proneural, mesenchymal, and neural. The classical subtype is usually associated with significant amplification of epidermal growth factor receptor (EGFR), which is less common in the other three subtypes. Additionally, classical GBMs do not harbor the most common glioma associated mutation in the TP53 gene. Proneural tumors conversely do not have significant EGFR amplification but show enrichment for TP53 mutations. Proneural tumors also have the most frequent incidence of IDH1 mutation, a common metabolic dysregulation in gliomas that is associated with tumor progression. These tumors also have significant amplification in the PDGFRA gene that is not seen in the other subtypes. Patients who have more proneural tumors tend to be younger than the other groups and survive better overall; however, patients did not respond any better with very aggressive treatment compared to less aggressive treatment.27 IDH1 mutant patients harboring the R132H mutation tend to have actually better overall survival, although this mutation is still considered oncogenic and a precursor for other more lethal genetic transformations. Furthermore, IDH1 mutants tend to enrich in lower grade gliomas, so the survival benefit and age discrepancies seen in the proneural subtype are likely skewed by the preponderance of lower grade gliomas in this group. The high grade glioma subgroup within proneural tumors which harbor the
19
PDGFRA amplification represent a relatively aggressive and lethal form of disease. The mesenchymal subgroup contains mutations in the NF1, PTEN, and TP53 genes. These patients actually benefited more from aggressive treatment when compared with the other groups.27 Finally, the neural subgroup represented the least distinct of the four groups, while sharing generally many of the same mutational aberrations as the other groups but with no particular enrichments. The neural subgroup tended to have the oldest patients and did respond to aggressive treatment but not as well as the mesenchymal subgroup.1 There is much debate on what to make of all of this information with regards to molecular subtyping in gliomas. Certainly when correlated to clinical outcomes, doctors can begin to make more informed decisions regarding treatment plans. From a scientific perspective, the debate is largely focused on how these subtypes might inform gliomagenesis and cell of origin questions. One idea for instance suggests that there are really only two major subtypes comprised of the proneural and mesenchymal groups, and everything else observed runs along a spectrum of development and differentiation from proneural towards mesenchymal.28,29 With respect to the general somatic landscape of glioblastoma (Figure 1.), the most commonly observed alteration involves the copynumber gain of chromosome 7 and the corresponding loss of chromosome 10, with the most commonly mutated genes being TP53, PTEN, EGFR, PDGFRA, NF1, and RB1 (Table 1-3, Appendix).1
20
Figure 1. Significant somatic chromosomal aberrations in glioblastoma (TCGA), Taylor et al. 2008, PLoS One. GBM is defined by some common genetic changes, including focal amplifications and whole chromosomal amplifications and/or deletions. The most common of these alterations is the gain of Chr. 7 accompanied by the loss of Chr. 10. There is also a focal amplification of EGFR on Chr. 7 but this amplification alone has not been shown to be sufficient for tumorigenesis. ASNS is located at 7q21.3, very near CD6 and MCM7. The gain of ASNS does not seem to be a driver of Chr. 7 amplification, and seems to follow the general observation of metabolic oncogenes often being passenger alterations during tumorigenic development.
21
Clinical Course & Treatment – Standard of care for glioblastoma involves surgical tumor resection for maximal debulking followed by an aggressive course of chemotherapy and radiation (Figure 2). The only FDA approved chemotherapeutic with any demonstrated efficacy today is temozolomide (TMZ), a DNA alkylating agent. TMZ derived DNA damage is often overcome in GBM patients who have an active and intact O6methylguanine-DNA methyltransferase (MGMT), involved in one of the cell’s endogenous DNA repair mechanism. The MGMT gene is silenced in a subset of glioma patients and as such remains a good predictor of clinical response to TMZ.30,31 The problem however remains that a full course of aggressive treatment including TMZ and heavy ionizing radiation treatment, overall survival remains approx. 14.6 months, compared to 10 months without chemotherapy, and 3 months with surgery alone.32 Historically, glioblastoma has been considered a fairly chemoresistant disease, largely due to the difficulty in overcoming delivery problems with respect to the blood-brain barrier (BBB). However, in higher grade tumors there is sufficient disruption of the BBB to allow for CNS penetration of medication, but even still, clinically there is approximately only a 15-40% objective response to any chemotherapy, including nitrosoureas, procarbazine, and platinum complexes.32 Anti-angiogenic therapies have been utilized in recent years as adjuvant treatment with the introduction of bevacizumab (Avastin) in the clinic with varying degrees of success. In some patients with highly vascularized tumors, Avastin along with a regimen of Irinotecan achieved nearly 10-fold improvement in radiological response and doubling of progression free survival (PFS) and overall survival (OS). However, such reports are not uniformly seen and often times,
22
anti-angiogenic treatment in glioma serves more for quality of life by helping reduce cerebral swelling around the tumor, rather than actually induce tumor regression33. Salvage and palliative care is the only option after patients fail two consecutive rounds of chemotherapy, and in gliomas radio-resistance is usually confirmation of no further clinical options. All of this further emphasizes the need for more dynamic, multi-modal treatment schemes that address the dynamic and heterogeneous nature of this disease.
23
CLINICAL/RADIOLOGICAL PRESENTATION
Surgery/Debulking • • •
SURGICAL DEBULKING
SALVAGE
Steroids to reduce edema Margins sometimes guided by awake surgery Often required for recurrence, usually around original site
Chemotherapy • • • •
Temozolomide 200mg/m2 CCNU 110 mg/m2 VP-16 50 mg/m2 CPT-11 125 mg/m2
**Objective response 15-40% Radiotherapy
RADIOTHERAPY & ADJUVANT TREATMENT
• •
30-60 Gy over 2-6 wks. Standard 59.4Gy over 6 wks.
(Lyman 2009)
RECURRENCE
CONTINUED CARE/FOLLOW-UP
Figure 2. Treatment Schema in Glioma Patient Care. In most clinical cases glioma is first diagnosed by radiological presentation followed by pathological grading. Standard of care involves tumor debulking surgery when applicable followed by a regimen of chemo-radiation plus any adjuvant treatments. Follow-up is usually every 2-4 weeks following radiotherapy every 2 months. In the case of recurrence, whether local or diffuse, will repeat the treatment cycle starting with further debulking if resectable followed by systemic chemotherapy and focused radiation. Palliative care is the focus after two consecutive failures with systemic chemotherapy. (Adapted from Chicago Institute of Neurosurgery and Neuroresearch. Nov. 2008) [20]
24
Cancer Stem Cells & Tumor Hierarchy Through human development most cells in the body mature from stem-like precursors towards more differentiated cellular fates. These differentiation events are functionally important and tend to result in committed cellular steps towards terminal cell states. However, it is an important aspect of tissue homeostasis to maintain certain subpopulations of stem-like precursors that can give rise to functionally mature progeny in the event of cellular turnover or wound healing. In cancer, it has been proposed that elements of this homeostatic mechanism have been hijacked for cancer propagation.
Cancer Stem Cell Hypothesis –The cancer stem cell hypothesis is a concept that has been around for nearly 150 years but has gained more attention in recent years due to growing understand in the area of basic stem cell biology. The basic concept revolves around the idea that in any given tumor there may be a small subpopulation of specialized cells that retain a more stem-like or progenitor phenotype. This stem-like characteristic in the context of cancer would play a significant role in allowing tumors to adapt to evolving environmental pressures and possibly repopulating a tumor after treatment, much in the same way stem cells function in basic wound repair across various organ systems in the body.4,34-36 Therefore, fundamentally determining how CSCs are maintained and how to specially target these cells could potentially address serious questions related to tumor recurrence and resistance to therapy in the clinic. This concept is not without controversy as there is no consensus as to the details of how this mechanism of cancer stemness is managed in the tumor, or if true “stem” cells actually exist in tumors at all.37-39
25
The original cancer stem cell (CSC) hypothesis proposed a model of tumor propagation via stem cell precursors using the hierarchical model of stem cell division. The traditional hierarchical model of cancer stem cells states that distinct stem-like populations exist from the beginning of the tumor inception and are in fact responsible for the propagation of various more differentiated cell populations that will go on to make up the heterogeneous tumor pool.34 In this model, treatment resistance is at least in part explained but the maintenance of the parental cancer stem cells which can then repopulate the tumor bulk once the treatment insult is removed.40 An alternative idea being developed with regards to cancer stem cell propagation posits the idea of clonal evolution, where the accumulation of a series of mutations, in time, will drive cells away from their assigned cell fates and slowly dedifferentiate into a more progenitor state. In theory, a tumor will eventually develop one or more distinct stem-like clonal populations that have recaptured self-renewal capacity that can then be implemented towards tumor survival and growth. In light of current understandings of tumor heterogeneity and tumor resistance/recurrence, it is more likely that both of these models may in fact describe different elements of a central process and therefore both explain the cancer stem cell model to a point, as some have proposed a hybrid of the two theories to explain the complex dynamics involved (Figure 3).35 The important fact remains that regardless of the origin of these cells, the elimination of the cancer stem cell population in any tumor model represents one of the most important hurdles to cancer research and treatment today. If tumor heterogeneity is one of the more problematic complexities in the clinic today when dealing with cancer, then addressing the underlying mechanisms of this diversity is paramount to making progress towards more intelligent treatments in
26
Figure 3. Evolving models of tumor heterogeneity driven by cancer stem cells. (A) traditional cancer stem cell hypothesis suggested hierarchal model where only certain stem-like cells (red) retained the capacity to repopulate a tumor and drive continued tumor progression, (B) the clonal model of stem cell propagation suggests many cells within the tumor retains this stem-like repopulation capacity, (C) the newest models suggest a hybrid model that allows for dedifferentiation and transdifferentiation within a tumor from multiple stem precursors in response to intra- and extratumoral pressures, driving tumor recurrence and heterogeneity (colors indicate independent clones)
cancer. As our conceptions of tumorigenesis develops across different tumor types throughout the body, the idea that the same cell fate machinery that is in place during embryogenesis and wound healing later through life may in fact also define the principles that govern tumor cell evolution is becoming more appreciated. Understanding tumor hierarchy therefore should not be thought of that much more differently than the organizational structures involved in organ development and tissue homeostasis, both of which rely on and derive from stem-like progenitor populations. It should be noted the hierarchical organization in normal tissue is more tightly regulated than in the tumor, so not all normal cellular control structures translate over into the tumor space equally well.
27
Glioma Stem Cells –Cancer stem cells in gliomas, commonly referred to as glioma stem cells (GSCs) have been demonstrated in vitro to have self-renewal capacity, differentiate into multiple cell lineages, form neurospheres, and express specific neural stem cell markers such as Nestin, Sox2, Prom1/CD133, and Nanog.41 These characteristics are basic to the concept of what defines a cancer stem cell (Figure 4). Several more markers have been suggested over the years and there is unlikely to be a specific expression profile that encompasses every stem-like glioma subpopulation. GSCs have been shown to be more resistant to both chemotherapy and radiation above differentiated tumor cells and several studies have specifically shown GSC ability to repopulate a tumor and drive secondary tumor recurrence post-treatment.8,42,43 To further confound things, as with tumors in general, there has been shown to be great heterogeneity even within the GSC pools, which is consistent with the models of CSC maintenance and propagation.21,44 Various different expression subtypes have been described in glioma patients as discussed earlier (proneural, mesenchymal, classical, and neural) and several of these subtypes have also been attributed to GSCs as well (proneural and mesenchymal). Distinct GSCs clones even from the same tumor can display variability in gene expression profile and metabolic dependencies.3,4,45,46 There is evidence to suggest that variability in GSC clones is at least in part driven by the tumor microenvironment itself and that all of this transcriptional and metabolic heterogeneity allows cancer stem cells to be very adaptable in order to maintain high rates of self-renewal and differentiation.47-49
28
These observations in gliomas have likewise been shown in breast, pancreatic, lung, colon, ovarian, and prostate cancer as well.50-54
Figure 4. Defining features of cancer stem cells. For a tumor cell to be considered a cancer stem cell it must satisfy all three of the required features: the capacity to perpetually self-renew, the capacity to sustain proliferation, and the capacity to form tumors in vivo. Additionally, CSCs will also be able to differentiate into multiple cell lineages, maintain rarity within a tumor, and express an array of stem markers. These are common but not defining characteristics of CSCs.
29
The Glioma Stem Cell Niche – GSCs, like most stem cells, demonstrate a very dynamic relationship with their niche, constantly maintaining bi-directional cross-talk with the tumor microenvironment. This interaction is essential in the understanding of gliomagenesis, propagation, and treatment resistance. First it is important to understand the distinct sub-localizations within the tumor space. GSCs have been enriched in the perivascular/proliferative niche of most gliomas, where there is both an abundance of stem maintenance signaling from the surrounding endothelium and exchange of nutrients.7,55 This is often specifically located around the subventricular zone (SVZ) and the hippocampus, which has also been described as a stem cell niche for normal neural stem cells43,56. There are many different soluble factors that play important roles in the vascular niche with respect to GSC maintenance. As mentioned before, glioblastoma is one of the most vascularized of the solid tumors and microvascular hyperplasia has been described as a key feature in glioma initiation and progression43,57,58. The development of this hyperplastic microvasculature is in large part dictated by the interaction of the GSCs and the tumor microenvironment. GSCs have been shown to promote angiogenesis and express factors such as vascular endothelial growth factor (VEGF), attracting endothelial cells to the tumor bulk and driving neovascular growth57,59. These endothelial cells in turn have been shown to express high levels of Sonic Hedghog (SHh), which plays an important role in the recruitment and activation of cancer stem cells and helps maintain GSC self-renewal and growth60,61. GSC enriched compartments of gliomas have been seen to co-localize with endothelial cells that express high levels of SHh and in turn drive SHh-GLI1 signaling in these GSCs.
30
Another important factor found in the vascular niche is the FGF-2, a growth factor commonly used in vitro to help maintain GSCs in culture. FGF-2 helps GSC maintenance in the niche and has also been demonstrated to bolster the blood-brain-barrier function of associated endothelial cells62-64. FGF-2 withdrawal tends to drive GSCs towards differentiation and FGF-2 signaling has also been associated with more robust Nestin expression, further contributing to stemness. FGF-2 has also been shown to combine with EGF, another factor used in vitro for GSC propagation, and that FGF-2 and EGF may be involved in an autocrine feedback loop to help retain GSC self-renewal.65,66 The general architecture of the glioma tumor space is comprised of normoxic cells that generally drive the tumor at its leading edges along the periphery and most of the hypoxic cells located in the poorly oxygenated necrotic core of the tumor. Therefore, most tumors, small or large, display this gradient of oxygenation throughout the tumor and this results in the development of distinct cellular compartments. The plasticity of GSCs allows them to reside in most compartments of the tumor, but as mentioned earlier, the perivascular niche and vascular niche is often enriched for GSCs due to the interaction with the endothelium. Many studies have shown that GSCs may contribute directly to the vasculature and in fact be able to drive endothelial cells through a trans-differentiation process22,49,67. Endothelial cells surrounding GSCs have a been shown to have expression profiles very similar to the cancer cells themselves and have suggested they might in fact be of neoplastic origin. This microenvironment therefore maintains the GSCs in order to preserve their potential to proliferate and differentiate, and can protect them from any treatment insults that they may encounter68. Cellular responses to hypoxia are often modulated through the family of hypoxia-inducible factors (HIFs). In stem cell biology
31
and GSC biology, HIF1a and HIF2a have been implicated in oxygen dependent signaling and stem cell maintenance, and HIF1/2a expression has been demonstrated to be very high in the hypoxic compartments of gliomas42,69. HIF1a expression is high in GSCs even under mild hypoxia and has been shown to prevent GSC apoptosis through NFкB stabilization and expression of anti-apoptotic NFкB target genes70. All these studies demonstrate that the hypoxic microenvironment plays an important function in maintaining GSCs and promoting recruitment of vascular components to the tumor bulk to sustain proliferation and growth.
Metabolism & Cancer Cancer metabolism has emerged as one of the most interesting old ideas being revisited from a new perspective. In the early 20th century Otto Warburg declared metabolism the prime cause in a disease of many secondary causes, and this statement seems more prescient in view of modern expositions into the true nature of tumor evolution. As the complexity of tumor heterogeneity becomes more clear from a genetic perspective, it is important to consider the inevitably heterogeneous metabolic components of the tumor and the tumor microenvironment.
Tumor metabolism fundamentally discusses two major points of cell behavior: [1] the specific sourcing of macromolecules and metabolites, and [2] the different cellular mechanisms used to deal with different nutrients for either anabolic construction or catabolic breakdown. Many tumors have been shown to augment its microenvironment in order to more optimally acquire nutrients, which is of particular importance to solid
32
tumors as the tumor core becomes more isolated from the native vascular infrastructure. Microvascular hyperplasia is one of the important hallmarks in glioma development and in fact most gliomas maintain extensive, proliferative vascular endothelium43,71,72. Although this vasculature is required for most of the tumor bulk, solid tumors will have variability in access to oxygen and nutrients in different tumor compartments, and adaptations to this variability is also important for tumor growth. As solid glioma bulk grows in mass, the core tumor space will begin to form necrotic and hypoxic regions and a significant amount of necrotic buildup ensues.43 However, there will also be cellular compartments within the tumor bulk that adapt to the oxygen and glucose gradients and may thrive in this space. Within a single tumor one will find many different cellular compartments with variations in oxygenation and fuel source availability and there will be cells enriched in these compartments that have made the suitable adjustments to accommodate these conditions. This again harkens back to the complex heterogeneity of solid tumors which only gets worse with tumor progression. Furthermore, the body of research over the past decade regarding glioma stem-cell (GSC) populations have indicated certain highly resistant and tumorigenic sub-populations are maintained in specific microenvironmental niches, particularly enriched in these perinecrotic/hypoxic/perivascular compartments.43,73
The Warburg effect describes the predominantly observed behavior of most malignantly transformed cells with respect to their metabolism. While normal cells will largely undergo oxidative phosphorylation in the presence of glucose and oxygen, in many cancer cells the large proportion of glucose is diverted away from mitochondrial
33
oxidation and into glycolysis and the production of lactate by lactate dehydrogenase (LDH) even in the presence of oxygen (Figure 5).12 This at first seems paradoxical since the mechanism of anaerobic glycolysis, although important in moments of low or no oxygen, is much less efficient for ATP production than mitochondrial oxidation of glucose. In part this adaptation may provide a sufficient balance between providing the necessary resources for biomass production and growth while allowing plasticity to adapt
34
Figure 5. Metabolic reprogramming in cancer cells by Warburg Effect. Cancer cells demonstrate distinct metabolic differences compared to normal proliferating cells, where transformed cells utilize glycolysis to process glucose even in the presence of oxygen in what is referred to as aerobic glycolysis. Normal cells would the more efficient oxidative phosphorylation when oxygen is available, but cancer cells overcome this apparent energy discrepancy by simply increasing glucose uptake feeding into aerobic glycolysis. CSCs tend to retain the ability to switch readily between oxidative phosphorylation and/or glycolysis depending on environmental selective pressures.
to various stresses, physiological and clinically driven. There have been many described exceptions to the Warburg phenomenon and the importance of this metabolic alteration is yet to be fully elucidated, especially with regards to cancer stem cell metabolism. 35
Cancer Stem Cell Maintenance & Metabolism – The plasticity and adaptability of most cancer stem cells to different metabolic conditions and requirements is considered an important hallmark in cancer development. Like the tumor itself, the metabolic landscape of the tumor is very heterogeneous and cells will metabolize differently depending on the environment they are in. CSCs have the ability to generally maintain a very quiescent profile but can rapidly switch into a more proliferative state if there is a need to repopulate the tumor, for instance in response to radiation derived tumor regression.10,74 When most cells would die under radiation, the CSCs can remain hidden from the stress and enter the cell cycle afterwards in order to replenish the tumor. This is partly adaptive on the part of the tumor but is in large part allowed by the microenvironmental space.
The literature on cancer stem cell metabolism is ripe with discrepancies, as there have been nearly as many papers describing glycolytic CSC dependencies as there have been those with oxidative dependencies. In all these reports however, there is at least an acknowledgment that the CSC population maintains a distinct metabolic phenotype compared to the tumor bulk, but the exact profile is not known. There have been multiple studies that have described CSCs being more glycolytic than the differentiated progeny across different tumor types.75,76 In these studies, glucose uptake, glycolytic enzymes, lactate and ATP production are much higher in CSCs compared to when they were differentiated. These observations corresponded to diminished metabolic contribution from mitochondrial oxidation. In many cases CSCs were seen to have lower levels of mitochondrial DNA and differentiation was shown to subsequently increase mitochondrial DNA copy number. Mitochondrial copy number was also inversely
36
correlated with the expression of many genes associated with pluripotency such as Oct4, TERT, and Myc. This however all stands in contrast to a growing body of research that indicates CSCs may in fact rely more on mitochondrial function and oxidative phosphorylation. Studies have shown CSCs actually being less glycolytic than their differentiated counterparts and consume less glucose, produce less lactate, and maintain higher ATP levels. In these studies, CSC mitochondria tend to have larger mass and increased membrane potential, which has been related to increased mitochondrial ROS and enhanced oxygen consumption rates compared to differentiated progeny. Mitochondrial mass was particularly important in these studies and correlated highly with metastatic potential and resistance to DNA damage.77 There is evidence showing mitochondrial biogenesis to be higher in circulating/migratory tumor cells and that inhibition of the transcription co-activator peroxisome proliferator-activated receptor gamma co-activator 1 alpha (PGC1a) can reduce stemness in breast and pancreatic CSCs.77 Mitochondria in these CSCs as mentioned before have higher levels of ROS but the total intracellular ROS levels remain lower, suggesting a strong antioxidant mechanism being utilized in these CSCs.78 This antioxidant mediated ROS buffering not only seems to help maintain stemness in CSCs but promotes treatment resistance, and further underscores the importance of elucidating metabolism in CSCs to help address problems with non-responders in the clinic.
These discrepancies however may not be a problem in our understanding but rather indicative of what CSC metabolism really looks like. Cells undergo metabolic fluctuations during differentiation under hypoxia, going between glycolysis and oxidative
37
phosphorylation to overcome mitochondrial deficiencies.74 It is quite possible that CSCs are able to similarly modulate redox status in order to regulate their own stem maintenance. CSCs may simply be demonstrating a level of metabolic plasticity that normal cells and more differentiated tumor cells cannot accomplish in response to the microenvironment (Figure 6). The true metabolic condition of the CSC population would require analysis of cells directly from patients and/or maintained at a very low passage. In these cases, GSCs from patient-derived gliomas tended towards mitochondrial oxidative phosphorylation compared to the differentiated progeny.10,74 Furthermore, patient derived GSCs maintained at low passage also maintain very high metabolic plasticity since blocking mitochondrial metabolism simply forces these cells to switch to a more glycolytic profile. This plasticity could allow CSCs to deal with fluctuating conditions and survive in unfavorable conditions whether it be stress from treatment or the harshness of metastatic sites.10,79 This also presents a new problem wherein single inhibition of one metabolic pathway may not be effective in vivo despite presumably being effective in vitro, since the true CSC population may be able to modify its metabolic preferences depending on what is available to them. This requires that any therapeutic strategy should take into consideration not only models to eliminate specific tumor populations but also models that prevent the recurrence of progenitor populations that could drive disease recurrence and eventually resistance to further treatment (Figure 7).
38
Figure 6. Metabolic plasticity exhibited by cancer stem cells. Cancer stem cell metabolism may demonstrate another element of cellular plasticity that provides a basis for therapeutic resistance and tumor recurrence after treatment interventions. More terminally differentiated tumor cells may not be as metabolically flexible as stem cells that can switch to different metabolic pathways and different fuel sources more readily.
39
Figure 7. Cancer Stem Cell Focused Treatment Paradigm. Standard of care treatment in many cancers involves surgery, chemo-radiation, and occasionally targeted therapeutics with varying efficacy. Treatments focusing on non-CSCs will invariably result in relapse derived from CSC driven tumor repopulation. CSC targeting alone is insufficient since non-CSCs and transitionary tumor cells may dedifferentiate or transdifferentiate into CSC-like cells and also repopulate the tumor. Effective treatment involves eradication of the CSC compartment of the tumor and preventing de novo CSC formation in order to prevent tumor recurrence2,8-10
40
Amino Acid Metabolism With all of the alterations that cancer cells undergo with respect to their metabolic processes utilizing glucose, similarly complex dynamics are involved regulating and reprogramming non-glucose metabolites. Amino acids represent one of the more highly consumed nutrient in cancer cells and the biosynthetic pathways necessary for rapid proliferation and biomass production rely significantly on a steady and constant supply of these substrates. The most common amino acids required by cancer cells include glutamine, serine, glycine, histidine, proline, and glycine among others, in no particular order. Cancer cells are constantly making adjustments in order to maintain access to these substrates. The mechanistic target of rapamycin (mTOR), one of the most central regulators of cell metabolism, is highly active in many transformed cells and is able to sense with good precision slight alterations to amino acid and protein levels in the cell. As such, many cancer cells are able to coordinate a very complicated network of amino acid sensing and sourcing to support the biochemical pathways involved in biomass production and proliferation. Glutamine, a key amino acid and the most abundant metabolite in the body, is a very important metabolite involved in cancer growth and progression.80,81 Many tumors rely on glutamine and glutamine-dependent processes as much or even more than glucose, even in the context of metabolic demand. In fact, many cancers including glioblastoma have been shown to be so dependent on glutamine that any glutamine withdrawal can induce rapid apoptosis.82 In many of these same studies, one of the only ways to remedy this withdrawal mediated-apoptosis is supplementation with asparagine, another closely related non-essential amino acid (NEAA).82,83 Asparagine is a very interesting amino acid to be involved in cancer since it is a non-
41
essential amino acid that is very tightly controlled by stress response mechanisms in the cell, but rarely contributes directly to processes linked to central carbon metabolism. In recent years the regulation of amino acid metabolism has become more critical in understanding cancer metabolic remodeling beyond the classical Warburg-glucose circuitry. Asparagine is the most recent of these substrates to become relevant in the context of cellular adaptations in cancer.
Asparagine Synthetase and Cancer– ASNS is the ATP dependent enzyme responsible for catalyzing the reaction converting aspartate and glutamine into asparagine. This enzyme is found throughout the body, in all organ systems, and is very tightly regulated since it is an exergonic reaction that also consumes glutamine, therefore metabolically costly.13 ASNS is regulated transcriptionally by the C/EBP-ATF response element (CARE) within the promoter and can be activated through the amino acid response (AAR) under conditions of protein limitation or imbalance, in addition to the ER stress via the unfolded protein response (UPR) pathway through PERK-eIF2-ATF4. In either case, ATF4 activation leads to binding of the CARE and transcriptional upregulation of ASNS.82 The normal biological need for asparagine does not necessitate constant and elevated levels of ASNS activation, but this is exactly what is observed in many cancers, which suggests this mechanism may in fact play a crucial role in cellular adaptations to nutritional or other stress in cancer cells.
42
Figure 8. Asparagine Synthetase (ASNS). (a) 3-Dimensional modeling of ASNS with the active site residue indicated. The V210F polymorphism is the most common mutation in this gene but has not been shown to functionally relevant [modeling done by Dr. Ramachandran from the CSMC Molecular Therapeutics Core] (b) the biochemical pathway for ASNS involves an ATP-dependent conversion of aspartate and glutamine, and this can be pharmacologically reversed using L-Asparaginase. 13
ASNS was identified most notably in the context of acute lymphoblastic leukemia (ALL) where many ALL cells had very low levels to ASNS and were specifically sensitive to LAsparaginase (L-ASNase), a pharmacological agent that can reverse the ASNS reaction by converting asparagine back to aspartate (Figure 8). ALL cells with low ASNS levels could not compensate endogenously for the rapid exogenous depletion of asparagine when administering L-ASNase systemically. This was however easily overcome if cells overexpressed ASNS, but nevertheless a sizable portion of patients were able to respond to this therapy, albeit with significant associated toxicities.13
43
ASNS and asparagine in the context of solid tumor biology is somewhat different since even solid tumors with varying levels of ASNS expression may not be as sensitive to external asparagine flux, and L-ASNase in its current form does not have any penetrative intracellular functions. That being said many studies in recent years have shown ASNS inhibition in solid tumors resulted in diminished cell growth, sometimes cell cycle arrest, and even increased sensitivity to specific stresses.13,82-84 The exact role ASNS and asparagine plays in the maintenance of growth and survival in solid tumors however has yet to be fully elucidated, although it is apparent that its role goes well beyond simple amino acid biosynthesis.
44
Figure A. Interacting Protein Network for ASNS gene. The network of interacting proteins surrounding ASNS includes stress-mediated ATF4 signaling elements, downstream enzymes involved in energy metabolism & 1C cycling, de novo pyrimidine synthesis, and aminoacyl-tRNA synthetases. Expectedly, there are many enzymes linked to amino acid synthesis & interconversion (GADL1, ASPA, GAD1/2, GOT1/2...) but its noteworthy that the other implicated downstream interactions include elements of the 1C/folate cycle and nucleotide biosynthesis. (STRING v10.0, 2017)
45
CHAPTER 1: ASNS COPY-NUMBER GAIN IS A POOR PROGNOSTIC FACTOR IN GLIOMA PATIENTS
Introduction & Rationale When we began this project we sought to identify alterations in high grade gliomas that may have fallen below the standard thresholding for significance in many online data mining resources, as such thresholds are relatively arbitrary starting points for assessment. Prior work had begun compiling patient data from multiple publically available online datasets for glioblastoma from TCGA which included information for both high and low grade gliomas. One of the genes identified in this process was that which encoded for asparagine synthetase (ASNS), and enzyme responsible for the biosynthesis of asparagine from aspartate and glutamine. With a focus on glioma stem cell biology and an interest in metabolic alterations in cancer, I set out to determine whether there was any real clinical significance to perturbations to ASNS in glioma patients. We have been working with patient-derived glioma stem cell models in the lab for many years both in the context of studying treatment resistance and targeted immunotherapeutic vaccine development. This xenograft passage of patient specimens was a more clinically relevant model of studying this disease compared to conventional cell line models, but this was still an in vitro model system. Before we spent any time studying any functional characteristics in vitro we reached into the wealth of patient data available publically. Glioblastoma patient data was especially available as it has been prioritized by many groups including the Cancer Genome Atlas (TCGA) due to overall
46
poor patient landscape and limited therapeutic resources for this cancer subtype, in addition to the easy availability of patient samples.
Results & Discussion ASNS gain correlates with shorter patient survival – We first examined a large database of patient data from the TCGA Cell 2016, Merged Cohort of LGG and GBM downloaded through cBioPortal (MSKCC) which included 1102 total samples, of which 564 high grade gliomas and 454 lower grade gliomas were separated out. We then generated Kaplan-Meier curves with the overall survival data and ASNS copy number data, which was separated into ASNS gain (at least 3+ copies of ASNS) or ASNS diploid (only 2 copy of ASNS). Any deletions were excluded from this analysis. Our prior analysis had indicated that ASNS was altered in a significant portion of GBM patients, and looking only at expression readouts in many of these data sets would not show any significant correlation to patient outcome. However, when looking at copy-number alterations of ASNS we found a significant survival difference in both the GBM [16.5 vs. 13.3 mos., HR=1.480, p=0.0006] and the LGG [95.50 vs. 50.80 mos., HR=2.248, p=0.0009] (Figure 9a,b). In the GBM set, the approximate 3-month difference may not seem significant but is comparable to the survival benefit of patients undergoing an aggressive regimen of temozolomide after surgery and radiation. More importantly, the survival benefit in this context is not curative in any way but provides a wider therapeutic window for adjuvant treatment. In lower grade gliomas the survival benefit was much more significant (44.7 months) as indicated by the pronounced hazard ration (2.248). For the purposes of our study we will generally focus on ASNS in high grade gliomas only, but it would be
47
Figure 9. ASNS correlates with poor patient survival and high tumor grade. Survival variance is observed in both high grade (a) and lower grade (b) patient cohorts, with hazard ratio of 1.48 and 2.248, respectively. KM plots generated using copy number data from 2016 Merged TCGA Cohort, cBioPortal (MSKCC). (c) ASNS expression in patient tissue can be observed in a grade dependent manner via immunohistochemistry. Representative images from patient tissue microarray shown, with quantification from at least 3 tissue cores per grade (Gr 3 p
