!"!#$ CHEMICAL GENETIC INTERROGATION OF NEURAL STEM CELLS:

PHENOTYPE AND FUNCTION OF NEUROTRANSMITTER PATHWAYS IN

NORMAL AND BRAIN TUMOUR INITIATING NEURAL PRECUSOR CELLS

by

Phedias Diamandis

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy.

Department of Molecular Genetics University of Toronto

© Copyright by Phedias Diamandis 2010

Phenotype and Function of Neurotransmitter Pathways in Normal and Brain Tumor Initiating Neural Precursor Cells

Phedias Diamandis

Doctor of Philosophy

Department of Molecular Genetics University of Toronto

2010

&'(!)&*!% The identification of self-renewing and multipotent neural stem cells (NSCs) in the mammalian brain brings promise for the treatment of neurological diseases and has yielded new insight into brain cancer. The complete repertoire of signaling pathways that governs these cells however remains largely uncharacterized. This thesis describes how chemical genetic approaches can be used to probe and better define the operational circuitry of the NSC. I describe the development of a small molecule chemical genetic screen of NSCs that uncovered an unappreciated precursor role of a number of neurotransmitter pathways commonly thought to operate primarily in the mature central nervous system (CNS). Given the similarities between stem cells and cancer, I then translated this knowledge to demonstrate that these neurotransmitter regulatory effects are also conserved within cultures of cancer stem cells. I then provide experimental and epidemiologically support for this hypothesis and suggest that neurotransmitter signals may also regulate the expansion of precursor cells that drive tumor growth in the brain. Specifically, I first evaluate the effects of neurochemicals in mouse models of brain tumors. I then outline a retrospective meta-analysis of brain tumor incidence rates in psychiatric patients presumed to be chronically taking neuromodulators similar to those identified in the initial screen. Lastly, by further exploring the phenotype and function of neurotransmitter pathways in purified populations of human NSCs, I determined that neurotransmitter pathway gene expression exists in a functionally heterogeneous phase-varying state that restricts the responsiveness of these populations to various stimuli. Taken together, this research provides novel insights into the phenotypic and functional landscape of neurotransmitter pathways in both normal and cancer-derived NSCs. In additional to a better fundamental understanding of NSC biology, these results suggest how clinically approved neuromodulators can be used to remodel the mature CNS and find application in the treatment of brain cancer.

! ""%

"!+&,&% % When you set out on your journey to Ithaca, pray that the road is long, full of adventure, full of knowledge. The Lestrygonians and the Cyclops, the angry Poseidon -- do not fear them: You will never find such as these on your path, if your thoughts remain lofty, if a fine emotion touches your spirit and your body. The Lestrygonians and the Cyclops, the fierce Poseidon you will never encounter, if you do not carry them within your soul, if your soul does not set them up before you.

Pray that the road is long. That the summer mornings are many, when, with such pleasure, with such joy you will enter ports seen for the first time; stop at Phoenician markets, and purchase fine merchandise, mother-of-pearl and coral, amber and ebony, and sensual perfumes of all kinds, as many sensual perfumes as you can; visit many Egyptian cities, to learn and learn from scholars.

Always keep Ithaca in your mind. To arrive there is your ultimate goal. But do not hurry the voyage at all. It is better to let it last for many years; and to anchor at the island when you are old, rich with all you have gained on the way, not expecting that Ithaca will offer you riches.

Ithaca has given you the beautiful voyage. Without her you would have never set out on the road. She has nothing more to give you.

And if you find her poor, Ithaca has not deceived you. Wise as you have become, with so much experience, you must already have understood what Ithacas mean.

- Constantine P. Cavafy (1911)

! """%

&*,-./#$01$2$-!(

I thank God for guiding me down the long and laborious journey of graduate studies. For what started off as a simple pursuit of a diploma and future career, soon became a voyage full of adventure, knowledge and perhaps the most significant part of my personal development. I would like to thank my supervisors Dr. Mike Tyers and Dr. Peter Dirks for the years of quality co-supervision. I truly feel you have both shaped the way I think not only about science, but also the way I see the world around me. Thank you for giving me freedom, independence and guidance when I needed it. I am more creative, efficient and a better self-critic because of you two. I hope that in the spirit of mentoring, I too can one day pass these skills down to my own students and that our collaborations and friendship continue to flourish. To my committee members Dr. Peter Roy and Dr. James Ellis, thank you for your valuable time and thoughts. Although our time together was short, your ideas and constructive criticism have left a forever-lasting improvement to this document. To Ian, you have been a model citizen, taking the time to share your endless wealth of patience and knowledge with us graduates students. You’ve been like a father figure in the lab to so many of us and have always led by good example. Thank you for consistently being there for both the good and bad times. You are a “true” team player and role model. To Kevin and Jenny, I also thank you for your constant support in my science and personal life. I consider you both true friends. To Caroline, Erick and Ryan, I have enjoyed embarking on this long PhD journey with you. I hope your graduate training has also brought you more riches that you could ever imagine. And to the remainder of the Dirks Lab, thank you all for your contributions to this work and for your friendship. You have all made coming to work enjoyable the last few years. Lastly, thank you to my mother Anastasia, my father Eleftherios and my sister Maria. I may have encountered many marvels and wonderful people during this long and difficult voyage, but perhaps the most wonderful realization it has brought me is the magnitude of love you have in your hearts. You have all proven that people can unselfishly provide unlimited belief, support and love to help another succeed. You have

! "#%

taken a once troubled adolescent and helped him already surpass the expectations of his given abilities. There are many problems in this world, but if every child had the love you have given me, the earth would be a much better place. I hope that the completion of this journey has given me the insight, skills and experience needed to one day find a more sufficient way to honor the sacrifices you have made for my life. Until then, I dedicate this thesis, the most significant accomplishment of my life so far, to you.

Thank you. Phedias Diamandis (2010)

! #%

#"(!%.3%&'')$4"&!".-(% (±)-PPHT (±)-2-(N-Phenethyl-N- EGR Early growth response propyl)amino-5-hydroxytetralin 5-HT 5-Hydroxytryptamine (serotonin) Emx2 empty spiracles homeobox 2 5-HTR1A 5-Hydroxytryptamine (serotonin) ENU N-ethyl-N-nitrosourea receptor, 1A AC Adenylate cyclase ER endoplasmic reticulum Ach Acetylcholine ES Embryonic stem ACSF Artificial cerebral spinal fluid FACS fluorescence activated cell sorting AML Acute myeloid leukemia FGF basic fibroblast growth factor (bFGF) AMP adenosine monophosphate FITC fluorescein isothiocyanate B2M beta-2-microglobulin GABA !-aminobutyric acid BDNF Brain-derived neurotrophic factor GABRB1 !-aminobutyric acid receptor, beta 1 BMPs Bone Morphogenetic Proteins GABRB2 !-aminobutyric acid receptor, beta 2 BSA Bovine serum albumin GAD67 Glutamate decarboxylase, 67-kD BTSC brain tumor stem cell GBM glioblastoma multiforme BTX "-bungarotoxin GFAP glial fibrillary acidic protein 2+ CaMK Ca /calmodulin-dependent protein Gi G protein, inhibitory kinases cAMP cyclic adenosine monophosphate GluR Glutamate receptor, metabotropic (mGlu) CD cluster of differentiation GPCR Guanine nucleotide binding protein (G protein), coupled receptor

CHRM3 Cholinergic receptor, muscarinic 3 Gq/11 G protein, q polypeptide (M3) CHRNA7 Cholinergic receptor, neuronal GRIN2B glutamate receptor, ionotropic, N- ("7nAChR) nicotinic, "-polypeptide 7 Methyl-D-Aspartate, subunit 2B CNTF ciliary neurotrophic factor GRIA1 glutamate receptor, ionotropic, AMPA-1

CpG cytosine and guanine separated by a Gs G Protein, stimulatory phosphate CREB cAMP response element binding GSK-3 Glycogen synthase kinase 3 CSC cancer stem cell h Human CNS central nervous system HDACS Histone deacetylases DAG Diacylglycerol HEPES N-2-Hydroxyethelypiperazine-N’-2- ethanesulfonic acid DAPI 4’,6’-diamidino-2-phenylindole Hh Hedgehog hydrochloride Dlk6 Delta-like 1 homolog HRP horseradish peroxidase Dlx5 Distal-less homeobox 5 HSCs Hematopoietic stem cells DMEM Dulbecco’s Modified Essential HTS high throughput screening Medium DMSO Dimethyl sulfoxide IP3 inositol 1,4,5-trisphosphate dNTPs Deoxynucleotide triphosphates JAK Janus kinase 3 DPAT (dipropylamino)tetralin Ki-67 Antigen identified by monoclonal antibody Ki-67 DRD2 Dopamine receptor D2 LIF Leukemia inhibitory factor E Embryonic day Lin Lineage EC# effective concentration needed to LOPAC library of pharmacologically active decrease proliferation by #% compounds ECT electroconvulsive therapy m Mouse EDTA ethylenediaminetetraacetic acid MAO Monoamine oxidase EGF Epidermal growth factor

! #"%

% MAPK Mitogen activated protein kinease RAR retinoic acid receptor MeCP2 Methly- RB1 retinoblastoma protein 1 CpG-binding protein 2 MEK Mitogen-activated protein kinase RBP- recombination signal binding protein kinease Jkappa for immunoglobulin kappa J region MEL mouse erythroid leukemia ROC Receiver operating characteristic mRNA messenger RNA RNA ribonucleic acid MTT 3-(4,5-dimethylthiazol-2-yl)-2,5- RT reverse transcriptase diphenyltetrazolium bromide NA Noradrenaline RT-PCR reverse transcriptase polymerase chain reaction NE Norepinephrine RTT Rett syndrome NGS Normal goat serum SAR structural activity relationship NMDA N-methyl-D-aspartic acid Sca-1 spinocerebellar ataxia-1

NOD- Non-obese diabetic severe combined SCN1A sodium channel, voltage-gated, type SCID immunodeficiency I, alpha subunit NPC neural precursor cell SCN1B sodium channel, voltage-gated, type I, beta subunit NR-1 NMDA receptor 1 SCN2A sodium channel, voltage-gated, type II, alpha subunit NS Neural stem SDS- Sodium dodecylsulfate PAGE polyacrylamide gel electrophoresis NSC neural stem cell Shh sonic hedge hog NSF N-ethylmaleimide-sensitive factor SGZ subgranular zone NT Neurotransmitter SIN3A SIN3 homolog A Nurr1 nuclear receptor-related 1 SIR Standardized Incidence Ratio P postnatal day SOX2 sex determining region Y)-box 2 p16INK4A Cyclin-dependant kinase inhibitor SVZ subventricular zone 2A; CDKN2A P53 Tumor protein 53 TBST Tris-Buffered Saline Tween-20 (TP53) PAPP p-Aminophenethyl-m- TH Tyrosine hydroxylase trifluoromethylphenyl piperazine Pax6 Paired box gene 6 (aniridia, keratitis) Topo Topoisomerase PBGD porphobilinogen deaminase TPH1 tryptophan hydroxylase 1 PBS phosphate buffered saline TRPV Transient receptor potential (VR1) Vanilloid PD Parkinson's disease WHO World Health Organization PFA Paraformaldehyde Wnt Wingless p-F- p-fluoro-hexahydrosila-difenidol HHSiD PHCCC (-)-N-Phenyl-7- (hydroxyimino)cyclopropa[b]chrome n-1a-carboxamide PI propidium iodine PI3K phosphoinositide 3-kinase PIP2 phodphatidylinositol 4,5-bisphate PKA protein kinase A PKC protein kinase C PLC Phospholipase C PLO poly lysine ornithine Ptc1 Patched 1 PTEN Possphatase and tensin homolog

! #""%

%

! #"""%

! "$%

! $%

! $"%

! ! ! !

! $""% "!

CHAPTER 1: INTRODUCTION

PART 1: CANCER 1.1.1 CANCER EPIDEMIOLOGY Presently, malignant neoplasms remain the leading cause of death in Americans under the age of 85. Overall, this accounts for at least one quarter of all deaths reported in developed countries including those in the United States of America (USA) and Canada1. This large societal burden extends globally with an estimated 12 million newly diagnosed cases and 7.6 million deaths (13% overall) from cancer-related illnesses in 2007 alone2. With an economic cost in 2004 of over 72 billion dollars in the USA alone, cancer therapy continues to account for just under 5% of the total healthcare spending; a figure than has not changed in over 40 years3. It is now projected that with the recent increase in the standard of care of cancer and chronic diseases3, the cost of cancer therapy will put an unsustainable strain on our current medical system if novel and more effective therapies are not developed.

1.1.2 CANCER BIOLOGY Cell numbers found in each of the body’s various organs are tightly regulated by a multitude of molecular signaling pathways that balance cellular turnover and expansion with apoptosis. Cancer is thus a cellular state where accumulating transformation events begin to interfere with cells’ ability to properly respond to inhibitory cues, consequently leading to a bias towards cell expansion4. Such defects in the regulation of cell division may arise from a variety of mechanisms (i.e. point mutations, chromosomal translocations, epigenetic alterations) that result in a loss-of-function of tumor suppressor genes (genes suppressing cell growth, migration and survival) and/or a gain-of-function of oncogenes (genes promoting cell growth, migration and survival). At the genetic level, it is thought that cells acquire multiple alterations resulting in the sequential loss of one of more tumor suppressor genes and the activation of one or more oncogenes before it can become truly malignant5. Biologically, the multiplicity of these genetic events should lead to cancer cells that (1) grow in the absence of endogenous growth signals, (2) are

! #! insensitive to growth-inhibitory signals, (3) do not respond to apoptotic cues, (4) have sufficient telomerase activity to prevent senescence, (5) promote angiogenesis, and (6) become highly invasive/metastatic to neighboring/distant tissue6. Although these properties, that have been deemed necessary for malignant growth, have emerged as the key biological targets for the development of novel chemotherapeutic agents in recent years, there have been few documented successes of their selective exploitation in cancer cells. In light of these failures, the development of less toxic and more effective therapies may require a more precise understanding of the molecular and cellular pathways that define cancer biology. Recent work suggests that many of these unexpected obstacles in cancer drug development, such as radio- and chemo-resistance, can be explained by the previously unappreciated functional heterogeneity existing within different tumor types7,8. A better understanding of the underlying functional cellular hierarchy found in these malignant masses may provide alternate and more fruitful targets for drug development.

1.1.3 CANCER STEM CELL BIOLOGY Documentation of histological similarities between stem and cancer cells dates back over 150 years to observations made by Rudolph Virchow and Julius Cohnheim suggesting that cancer may arise from rare tissue remnants left behind from embryonic developement9,10. They suggested that the reactivation of such a embryonic program in these cells could explain the extensive long-term survival, self-renewal and proliferative potential seen in cancer cells11. Over 100 years later, Ernest A. McCulloch and James E. Till strengthened the prospect of a stem cell origin for cancer by demonstrating, through the use of serial dilution assays, that adult organs like the spleen contained rare cells with robust self-renewing and multilineage differentiation potential12,13. When applied to cancer, these assays unexpectedly suggested that malignancies, although hypothesized to be clonal in origin, contained only subpopluations of cells capable of initiating and maintaining tumor growth. For example, less than 1% of cells isolated from murine myeloma could form in vitro colonies on semi-solid agar14. Similarly, in vivo transplantation of mouse lymphoma cells into recipient mice suggested that only a rare population (1-4%) of cells were responsible for the propagation of the

! $! original tumor15. Technical adaptations to this colony forming assay allowed others to investigate this phenomenon in solid tumors, and also suggested that hundreds of cells were required to identify cells with colony forming ability in many solid cancers including human lung, ovarian and neuroblastoma tumors16. In the 1960’s, some controversial and unethical experiments that involved autologous subcutaneous re- injection of disseminated human tumors back into patient also supported the hypothesis that only a rare subpopulation of cells within the tumor drives cancer growth17. Despite this, the idea that most, if not all cells within a primary tumor were functionally equivalent and capable of regenerating the original histological features of the bulk tumor mass persisted. Authorities presumed that these observations, suggesting the inability of every resident cancer cell to re-from the original tumor, could be explained by stochastic models of cancer7,18 (Fig 1.1). Just over a decade ago, this dogma was dispelled by the demonstration that only a very rare sub-fraction of cells found within acute myeloid leukemia (AML) samples, prospectively defined by Lin- CD34+CD38- expression, exclusively retained the tumorigenic potential when injected into NOD-SCID mice19 (Fig 1.1). Through the analysis of a variety of different AML samples, Bonnet and Dick noticed this rare subpopulation of tumorigenic cells found within these tumors, morphologically resembled normal hematopoietic stem cells. When the various subfractions of the tumor were transplanted into immunodeficient mice, only cells with stem cell-like properties could propagate AML in this model. Interestingly, this rare cancer-forming subfraction also faithfully recapitulated the histopathological features and heterogeneity that were characteristic to the primary disease. This suggested AML is a heterogeneous disease that is organized in a functional hierarchy where only rare cancer stem-like cells found within the tumor bulk retain the tumorigenic potential. The importance of the cellular hierarchy seen in AML was soon reinforced in solid tumors where a similar study showed that only breast cancer cells expressing stem cell markers (CD44+CD24-) and possessing stem cell-like properties could propagate the disease and recapitulate the histopatholgical features and cell surface marker profiles of the original tumor20. Since then, the number of tissue specific cancers that have been demonstrated to be maintained by relatively rare primitive cells that express stem cell markers has grown to include the bone marrow19,21-23, mammary gland20, brain24,25,

! %! prostate26,27, colon28,29, pancreas30, head and neck30, mesenchyme31, skin32 and ovaries33 and liver34. This fundamental switch in our understanding of how cancer is organized suggests that only a rare subpopulation of cells is responsible for generating the heterogeneous array of cells types seen in histological sections of human tumors. Moreover, these finding suggest that anti-cancer therapies that may initially reduce tumor load by targeting the majority of non-cancer stem cells found within the tumor bulk, are flawed and may be unable to completely eliminate malignancies on their own. Alternatively, the cancer stem cell hypothesis suggests that the development of strategies that target and deplete these rare and relatively quiescent stem cell-like cells may provide novel and effective approaches to eradicating many forms of aggressive malignancies.

1.1.4 NORMAL TISSUE SPECIFIC STEM CELLS In addition to providing new hopes for regenerative medicine, the existence of resident stem cell populations in adult organs and tissues has created excitement in those investigating cancer biology. As was the case in leukemia, there is hope that a better understanding of the pathways that regulate normal stem cell biology will shed new insight into the mechanisms and biology that initiate and propagate cancer at additional sites of the body7. Stem cells initiate fetal development, generate the different tissues found in the adult body and persist throughout the lifetime of the organism. Following the fusion of an oocyte and sperm cell during reproduction, the diploid cells begins to divide multiple times until it generates a blastocyst containing a group of cells termed the inner cells mass. Each cell found within this colony of cells has the capacity to generate all the different germ layer of the organism and are thus termed to be “pluripotent”. During the development of the fetus, the differentiation potential of the cells of the inner cells mass segregates to give rise to different subpopulations of adult tissue-specific stem cells capable of forming a more restricted repertoire of tissue specific cell types (multi-potent). Although the tissue-specific stem cells were postulated to exist over 100 years ago, their existence was first definitively demonstrated in the spleen by the pioneering work of Ernest A. McCulloch and James E. Tillin in 1963 through the development of a colony- forming assay12,13. Similar in vitro colony forming assays have thereafter helped in the

! &! identification of other somatic stem cells residing in other tissues35-38. These colony- forming assays assess the presence of stem cells as defined by the their two cardinal properties; the ability of a single cultured cell to expand and produce differentiated cells from all the cellular lineages found in that specific organ (multipotent) and their ability to maintain their undifferentiated state over multiple cellular divisions (self-renewal)36.

PART II: BRAIN CANCER, STEM CELLS AND BRAIN CANCER STEM CELLS 1.2.1 BRAIN TUMORS Primary brain tumors encompass a diverse array of neoplasms in the CNS of both children and adults, continue to have a poor prognosis and lack any known modifiable risk factors39. As a whole, cancers of the CNS have a five year survival rate of 35%, ranking them among the most aggressive malignancies and are surpassed only by cancers of the esophagus, pancreases, lung, liver and myeloma1. They account for 2.4% of all cancer deaths, represent 20% of all childhood malignancies and remain the second leading cause of cancer deaths in the pediatric population40. In adults, high-grade anaplastic astrocytomas and glioblastoma multiforme (GBMs) represent at least one third of all primary brain tumours diagnosed. Even with intensive radio- and chemotherapy following surgical resection, the median survival of these patients remains at 9-12 months, with only 8-12% of patients surviving past 2 years41-45. In spite of this, the recent introduction of the DNA alkylating agent temozolamide, remains the only significant chemotherapeutic advancement for the management of GBMs in the last 30 years46. Even with all the promise and attention it has received, it humbly prolongs the median survival time from 12.1 to 14.6 months47. With such a grim prognosis and so few, if any, documented examples of complete remission48, brain tumor treatment strategies must shift away from traditional paradigms and look for alternative solutions. The failure to develop more effective therapies suggests that our understanding of the cellular and molecular mechanism governing the pathogenesis and maintenance of this disease requires reinvestigation. For example, tumor heterogeneity in brain tumors has been a long-observed phenomenon, yet until recently it has been assumed that the tumorigenic ability of different subpopulations was equivalent. Understanding how this heterogeneity

! '! arises and which cells within the tumor drive and maintain disease may help lend way to new and more effective therapies. Recent data suggests that these failures and unanswered questions may be explained by the cancer stem cell hypothesis49. A firmer understanding the molecular and cellular pathways that regulate the fates of cancer derived precursor and stem cells of the brain is therefore of immediate interest.

1.2.2 NEURAL STEM CELLS One of the longest standing dogmas in neuroscience suggested by even neuroanatomical pioneers like Santiago Ramony y Cajal was that the brain was a static organ lacking regenerative capacity. In 1983, investigation into anatomical changes occurring in the CNS of female songbird canaries challenged this long-standing assumption. These seminal experiments demonstrated that a remodeling process that includes both the proliferation of cells (assessed via [3H]thymidine incorporation) in the ventricular zones of canaries, as well as the migration and differentiation of these cells into newly integrated neurons (neurogenesis) occurs even throughout adulthood50. Although these experiments began to change the scientific community’s stance on adult neurogenesis, it was not until 1992, when the definitive existence of neural stem cells in both the embryonic37 and adult51 mammalian brains was first demonstrated. By isolating different regions of the mouse brains and culturing them in serum free media supplemented with epidermal growth factor (EGF) and fibroblast growth factor 2 (FGF2), Reynolds and Weiss showed that the mammalian CNS does indeed contain proliferative cells capable of expanding to form floating clonigenic clusters of cells they termed neurospheres. Consistent with the definition of bonifide stem cells, these neurospheres retained the capacity to self-renew following repeated dissociations and re-plating as single cells. The ability of a number of daughter cells derived from a single sphere colony to reform neurospheres over long-term passage also suggested that these cells symmetrically divide and expand to generate additional stem cells. Furthermore, given the number of cells found within a neurosphere, it is assumed that only the most primitive stem cells could retain such a robust proliferative capacity. These clusters thus provide a convenient index of the neural stem cells frequency within a given culture. Secondly,

! (! consistent with the classical definition of stem cells, these colonies are comprised of a variety of non-stem cell progeny. Furthermore, although a large proportion of these cells express precursor marker nestin, an intermediate filament expressed in the developing and immature brain; upon dissociation and re-plating of single neurospheres, only 50-100 colonies are derived from the thousands of plated cells. This suggests that although a single NSC can expand in vitro to form many more neurosphere forming cells, only a fraction of their progeny retain self-renewing capacity52. The remaining cells in these cultures were shown to represent more committed progenitors. Although, these mature CNS cellular phenotypes do not survive in serum-free conditions, they are supported when grown adherently on the extracellular matrix laminin and the stem cell growth factors are sequential withdrawn from the media; a process mimicking later stages of CNS development. Differentiation of these single cell derived colonies show co-expression both glial and neuronal cell markers demonstrating the multipotency of these cells. The ability of rare CNS cells to proliferate through symmetric divisions for extended periods of time (self-renewal capacity) and to retain the multipotentiality form the various lineages seen in the mature CNS (astrocytes, neurons and oligodentrocytes) demonstrated that there were in fact bonefide neural stem cells present in the adult mammalian brain53,54. The existence of stem cells in the CNS and culture systems that have allowed us to interrogate their function may one day provide us with the ability to control their mitotic rate and cell fate choices. Such breakthroughs may provide us with novel strategies to repair CNS damage resulting from a variety of trauma or disease processes. Anatomically, NSCs of the postnatal brain are thought to reside in the subventricular zone (SVZ) throughout the entire neuroaxis and persist during adulthood and into old age53,55-57. Although, the dentate gyrus of the hippocampus remains one of the most neurogenic regions of the adult brain, it is still debated if the identity of these proliferative cells represent a mixed population of lineage restricted progenitors or are true self-renewing and multipotent NSCs58,59. Rare cells with an extensive self-renewing, proliferative and differentiation capacity (termed NSCs60-63) are now recognized to exist in not only the embryonic and adult brain of mice51,64 but also in higher order mammals including primates65 and humans66-73. Although the gross anatomical location of

! )! proliferating immature cells has been well delineated, the exact identity of these precursor cells is not well defined. For example, the cytoplasmic expression of intermediate filament nestin has long been recognized not only to mark NSCs both in vitro and in vivo, but is also a marker also expressed by more committed neural progenitors74-76. Similarly, although the expression of the Sox2 transcription factor is thought to be a more stringent marker of NSCs, it is also found to overlap into other fractions of early progenitors77-80. Furthermore, it is thought that different subpopulations of neural precursors exist that are regionally specified to respond to cues unique to their neighboring cells and precise anatomical location. Therefore, labeling particular subpopulations of immature cells as the definitive neural stem cells found at the top of the hierarchy has become a difficult task81. Although the pro-proliferative in vitro conditions used to study neural precursors could be one of the key reasons for the conflicting data regarding the identity of neurosphere forming cells, in vivo experimentation has provided more definitive answers regarding the hierarchical organization of stem cells and their precursors. Using electron microscopy, Alvarez- Buylla and colleagues identified 3 morphologically different proliferating species in the adult brain of mice (Type A, B, and C cells)82. In a series of elegant ablation experiments, it was demonstrated that the slowly dividing cells of this complex (Type B cell) give rise to rapidly proliferative progenitors cells (Type C cells) that then go on to differentiate into glial and neuronal progeny (Type A cells). The Type A immature neuroblast arising from these cells then migrate from this progenitor region and through the rostal migratory stream where a fraction of the surviving cells go on to integrate into the olfactory bulb82,83. Their data suggests that in vivo, NSCs can be best characterized as the slowly dividing Type B cells, as they are the only cells capable of repopulating the other two mitotic SVZ populations69,82,84. NSCs are thus described as being relatively quiescent bipolar radial glial cells residing of the subventricular zone with processes extending towards the pial surface of the developing neural tube. These cells are thought to be the most primitive stem cells source indentified in the adult brain83. Analysis of these different subpopulations suggests that radial glial cells can be selectively marked by their expression of the radial glial markers BLBP, GLAST, and RC2 while the cells surface antigens A2B5 and PSA-NCAM mark their more committed neuronal and glial

! *! progeny85-89. It is interesting to note that unlike their embryonic counterparts, both rodent and human radial glials cells of the adult brain can also be identified by their expression the mature astrocytic marker GFAP69,90. Although further work is needed to better resolve this hierarchical organization of precursors, prospective enrichment of neural stem cells has been achieved by florescent activated cell sorting (FACS) of CD133 (AC133, prominin 1) positive cells91. This antigen, originally shown to be expressed by hematopoietic stem cells, was also found to be present on the apical membrane of a subpopulation of nestin expressing cells lining the SVZ of both developing mouse and human brains92-97. Using antibodies against this antigen, Uchida and colleagues were able to prospectively enrich the fraction of neurosphere forming cells isolated from human fetal CNS tissues. These human CD133+ fetal CNS cells also demonstrated a remarkable potency to engraft, migrate, proliferate and differentiate in vivo98. Taken together, although there is plenty of evidence to support the existence of normal dividing cells in the CNS, there are still many questions that need to be answered regarding the precise identity, cellular characteristics and physiological purpose of these cells in higher species. Additional advances in our understanding of NSCs cannot simply rely on the potential similarities with other stem cell pools and will likely require a multitude of non-biased approaches that specifically probe the biology of neural precursor populations.

1.2.3 BRAIN TUMOR STEM CELLS Progress into the development of methodologies used to identify, culture and assay normal stem cells of the CNS soon opened the door to properly investigate if cancer stem-like cells, similar to those found in leukemia and breast carcinomas, maintain and drive brain tumor growth. Using the premise that leukemic cancer stem cells share many similarities to their normal hematopoietic stem cell counterparts, the assay conditions originally developed for the identification and characterization of normal NSCs were used to address if self-renewing and multipotent cells could be isolated from primary brain tumors. These assay revealed that indeed, like their normal NSCs, subpopulations of cells isolated from cancers of the CNS also form passagable neurosphere colonies when cultured at low densities in serum-free media supplemented with FGF and EGF99-104. Importantly, clonally derived neurosphere-like aggregates

! "+! exhibit the cardinal properties of self-renewal and the ability to differentiate down multiple CNS lineages24,99,102-104. These characteristics were prospectively found to exist exclusively in the subpopulation of cells expressing the CD133+ stem cell marker, ruling out stochastic explanations for rare cancer initiating cells in CNS tumors24,103,105 (Fig 1.1). In addition to maintaining the cardinal properties of stem-like cells, these cancer cells have been shown to molecularly mimic their normal counterparts. Cancer stem cells isolated from the brain express transcription factors (Sox2, Bmi1, Pax6, Emx2), intracellular filaments (nestin), signaling molecules (Notch, Jagged1) and cell membrane molecules (CD133) that mark and/or associated with a neural stem cell state24,99,100,102,103. Although there may be some differences in gene expression patterns that lead to their tumorigenic ability, it is clear that the profiles of the cells that drive cancer in the brain resemble those of normal NSCs.

1.2.4 CD133+ SUBPOPULATIONS OF BRAIN TUMOR CANCER STEM CELLS The culture of primary brain tumors in serum-free conditions that foster the growth of precursor cells marked a major leap in the understanding of brain tumor biology. It not only suggested that cancer cells with stem-like properties exist within the bulk tumor, but it also began to provide a great deal of new insights into brain tumor biology. For example, it has now helped explain why the development of novel and effective brain tumor chemotherapeutics has been challenging with little, if any, clinical success. Before the appreciation of cancer stem cells in CNS tumors, the majority of our understanding of brain tumor biology stemmed from work done on serum-derived lines. Drug development and high throughput screening efforts thus relied on these cell populations for the identification of novel brain tumor chemotherapies. The inability of precursor cells to grow in the differentiation promoting conditions of serum cultures suggests that these previous strategies and research done in brain tumors was performed under conditions that do not permit the analysis of the very cells that maintain and drive tumorigenesis in vivo. In support of this, it was recently demonstrated that culturing tumors in serum free conditions containing EGF and FGF more faithfully recapitulate both the phenotype and genotype of tumors, when compared to tumors lines derived in serum conditions106. In this study, culturing fractions of the same tumor under serum and serum free conditions

! ""! showed that the mRNA expression patterns of the two culture systems were significantly different from one another. More importantly, when compared to the gene expression patterns of uncultured tumor cells, only the serum-free cultures grown in EGF and FGF conditions maintained the expression signatures of uncultured tumors106. This finding suggests that molecular pathways and their corresponding drug targets identified using serum cultures may not be relevant to tumor cells exhibiting a precursor phenotype. Research using serum-derived cultures may thus have little clinical utility in identifying compounds that specifically target molecular signaling pathways important for in vivo tumorigenesis. Perhaps an even stronger argument for the importance of focusing on the rare stem cell-like cells rather than the entire tumor bulk comes from prospective isolation of cancer initiating cells. Using the premise that cancer stem cell have properties that closely resemble those of normal neural stem cells, Dirks and colleagues demonstrated that only cancer cells expressing the stem cell antigen CD133 had the capacity to self-renew and undergo multipotent differentiation in vitro103. These primitive cells expressed stem cells markers like nestin and were completely devoid of mature markers associated with neurons, astrocytes and oligodendrocytes. When induced to differentiate, CD133 and nestin expression decreased and was accompanied by multipotent differentiation; a property similar to that seen in their normal NSC counterparts. When uncultured tumor cells were injected in vivo, only cells expressing the CD133 antigen could form tumors (at densities as low as 100 cells), whereas the injection of many more CD133- cells (105 cells) did not show tumorigenic potential in NOD-SCID mice24. Importantly, the tumors formed in the CD133+ fraction accurately recapitulated the histopathological features of the patients’ original tumors and showed signs of invasions into neighboring mouse brain tissue; a hallmark of aggressive brain tumors. Moreover, like the original tumor, the resulting xenografts regenerated the heterogeneous expression of CD133 and generated differentiated cells of different lineages. These experiments demonstrated that only specific subpopulations, that comprise only a relatively small number of cells within brain tumors, are involved in driving and maintaining the characteristics of the tumor seen as a whole. Since these tumorigenic populations can be prospectively identified based on their expression of

! "#!

CD133, it lends support, that similar to AML, the functional heterogeneity seen in brain tumors is derived from hierarchical rather than stochastic mechanisms (Fig 1.1). Taken together, these studies highlight the biological significance of cancer cells with stem cell- like properties. Focusing research and drug development towards these cells will likely yield information that is more relevant to the human disease and may lead to the development of therapies that specifically target pathways responsible for maintaining stem cell properties like self-renewal that are required for tumor growth. Interestingly, it was also noted that percentage of CD133+ cells found among various samples of primary uncultured brain tumors correlated with the clinical tumor grade. In low-grade tumors, CD133 was found to mark less than 1% of the residing cells within the tumor, while the same antigen labeled fractions as high as 30% in highly aggressive glioblastomas. The aggressiveness of the tumor and the fraction of CD133+ cells also closely correlated with the number of sphere forming units counted in vitro24,103,105. These experimental findings suggest that designing drugs that deplete the self-renewing ability of tumor cells, read out by CD133+ expressing or the ability to proliferate as neurosphere in vitro, may have clinical utility in the treatment of malignant gliomas.

! "$! "#$%&'!()(! !"#$#%&"'()*(#+,!!

*)! +)! Stochastic model Hierarchical model of cancer of cancer

CD133

Self-renewing cancer stem cell

Figure 1.1 | Models explaining the observed heterogeneous tumorigenic potential of cancer cells Limiting dilution and xenotransplantation experiments suggest that tumor growth is a property not retained among all cells found in the tumor bulk. This heterogeneous potential can be explained by either stochastic (a) or hierarchical (b) mechanisms of cancer growth. (a) The traditional “stochastic model” of cancer tumorigenesis implies that the phenotypic and functional heterogeneity seen within a tumor results from random stochastic events during processes such as cell division. In this model, all cells within a tumor have an equal ability to regenerate the original tumor, and the heterogeneous tumorigenic potential is derived from cell-to-cell variability that affects all cell subpopulations. In this model, tumorgenicity is thus a property not confined to a single subpopluation. (b) The hierarchical (cancer stem cell) model suggests that the heterogeneous tumorigenic potential of cancer cells is derived from sub-populations with discrete functional identities. The subset of cells with self-renewing and multi-lineage differentiation potential, which can be prospectively identified by stem cell markers (i.e. CD133 expression), retain the proliferative and differentiation capacity needed to generate heterogeneous secondary tumors that resemble the histopathological features of the original tumor.

! "%!

1.2.5 DEVELOPMENTAL SIGNALING PATHWAYS REGULATING NSCS The development of defined assays to assess the self-renewing and differentiation capacity of neural precursor both in vitro and in vivo has allowed for investigation into the molecular mechanisms and signaling pathways governing NSC biology37,107-109. Due to their importance on proper CNS development, there has been a great deal of focus on the role of the Notch110-112, Sonic Hedgehog (Shh)113-119 and Wingless (Wnt)120-122 pathways in NSCs. By taking advantage of the ability to convert ES cells into NSC in vitro, Van der Kooy and collaborators demonstrated that although Notch signaling is important for development, deletion of the key Notch signaling molecule RBP-Jkappa in ES cells did not alter their ability to generate neurosphere cultures111. Consistent with this finding in neurospheres, they showed that although homozygous deletion of either RBP- Jkappa or Notch1 alone did not lead to developmental CNS defects, these mice experienced depletions in their NSCs pools later on in development and adulthood. Similar work suggests that deletion of another Notch signaling molecule presenilin 1 (PS1) also negatively affects the NSC pool in developing embryonic brains. Also in line with these observations, the inhibition of Notch signaling correlated with a more robust differentiation response of NSCs. Taken together these in vitro and in vivo readouts of stem cell function suggest that although Notch signaling may not be required for the generation of NSCs, it plays a key role in their maintenance and differentiation potential once these cells are formed111. Shh is another key developmental signaling pathway that is thought to play an important role in the development of the entire brain. During early brain development, Shh is vital for proper ventral patterning, while in later stages, it plays a key role in expanding the precursor pool in the neocortex, tectum and cerebellum of the dorsal brain114,116,117,119. By using the activation of its downstream Gli genes as a surrogate index of Shh pathway activation, Palma and colleagues demonstrated that Gli activation was confined to the proliferative zone of the developing brain. Consistent with this, overexpression of Gli genes in tadpoles led to a robust hyperproliferation of the CNS. The selective pro-proliferative effects this pathway has on precursor regions of the adult CNS, suggests a possible role of Shh on neural stem cell self-renewal114. Activation of the Shh pathway has also been shown to affect proliferation of precursors in the adult

! "&! ventral forebrain and hippocampus. Close examination of cells in the SVZ activated by Shh signaling revealed that in fact, activation of Gli1 signaling and cell proliferation occurs in both the GFAP+ radial glial stem cells and GFAP- neural precursor116. Similar experiments have also helped dissect out an important role of Wnt in NSC biology. For example, selective hyperactivation of Wnt signaling in neural precursors, through the generation of a stabilized !-catenin molecule, leads to an enlarged cerebral cortical surface. These animals display an enlarged ventricular system and an expansion in the SVZ neural precursor populations lining them. Moreover, these precursors showed an increase in mitotic activity compared to control mice and thus intimately link NSC proliferation and self-renewal to Wnt signaling120. The self-renewing effects of Wnt in NSCs are also conserved in vitro. In addition to promoting NSC expansion, specific overexpression of the Wnt agonist Wnt3 leads to increased neurogenesis in adult hippocampal progenitors both in vivo and in in vitro assays. Consistent with this observation, blockade of Wnt signaling in neural progenitors reduces neurogenesis both in vivo and in vitro123. These experiments provide strong support for multiple role of Wnt in both stem cell self-renewal and fate commitment. Finally, these experiments highlight how the in vitro assay used to study NSCs faithfully recapitulate the role of three major developmental in in vivo neural development. In addition to these well studied pathways, NSC proliferation and self-renewal have also been shown to be regulated by other genes including classical oncogenes and tumors suppressors124-129. For example, through its inhibitory actions on the tumors suppressor genes INK4a and ARK, the polycomb gene Bmi1 regulates both the in vivo and in vitro proliferation and self-renewal capacity of NSCs124,126,128,129. Similarly, the tumor suppressor PTEN has also been shown to play a key role in the in vitro and in vivo regulation of proliferation of NSCs and their progeny125,127. As we continue to assess the roles of genes involved in CNS development or genes of fundamental importance to cell cycle regulation, our overall understanding of NSC biology may be biased to only these previously described pathways. Although there is no refuting their importance in the field, such a candidate gene approach based on previously identified development pathways may overlook the presence of other important pathways that regulate NSC biology in subtler and more contextually specific

! "'! circumstances. Less biased genome-wide approaches for the identification of additional pathways regulating NSC function are needed to properly understand the complete repertoire of signals that dictate decisions regarding self-renewal and differentiation in these cells.

1.2.6 CELL OF ORIGIN With so many cellular, phenotypic and molecular characteristics shared between normal and cancer stem cells of the CNS, the prospect that the transformation of resident stem cells is the initiating event common to these cancers remains open. Although lessons from the recently developed induced pluripotent stem cells (iPS) have demonstrated that the acquisition of embryonic stem cell gene expression profiles in mature fibroblast cells following alterations in only a few genes is possible, there is a body of evidence that strongly favors a stem cell origin for cancer130-132. For example, the majority of clinically diagnosed gliomas are found to concentrate in the vicinity of the SVZ rather than be randomly distributed throughout the entire CNS133. When mice are subjected to mutation- inducing carcinogens or oncogenic viruses, tumors of the CNS develop with a strong bias for the proliferating germinal regions compared to the overwhelming larger non- proliferative regions of the brain parenchyma133. Similarly, tumor incidence is much higher when carcinogens are injected into the SVZ rather the peripheral cortical regions134. Interestingly, tumors found in the white matter of dogs exposed to avian sarcoma virus earlier on in life have been shown to be derived from migrating tumor cells initially derived and transformed in the subventricular zone of these animals135. Further support for a stem cell origin of brain tumors arises from works reporting a higher frequency of tumors when tumor suppressors or oncogenes are deleted or activated respectively in nestin-expressing progenitor cells rather than in more mature and differentiated astrocytic cells136,137. In another study, although tumors were seen in both nestin-positive precursors and more mature astrocytes following overexpression of platelet-derived growth factor-B (PDGFB), there was a similar preferential induction of tumorigenesis in the more undifferentiated population of nestin expressing progenitors138. It is important to caution that the physiological relevance of the chemical and genetic dosages used in these animal models of CNS tumors may limit the ability to make

! "(! presumptions and hypotheses regarding the precise cell of origin of human brain cancers. Although these questions are hard to address in humans given how late the disease typically presents clinically, there is work in ependymomas that although correlative, also lends support to a NSC origin for humans tumors. Recently, it has been shown that the rare CD133+ cells of human ependymomas also exclusively retain the tumorigenic ability of the bulk tumor25. Interestingly, the gene expression profiles of these human cancer cells closely resemble the profiles of radial glial cells isolated from relatively similar rostral-caudal co-ordinates in the neuro-axis of mice. This finding can be interpreted as evidence that at least for human ependymomas, human CNS tumors may have origins in resident populations of radial glial cells. An observation supporting a more mature cellular origin of CNS tumors is that efficient formation of glioma-like lesions derived from astrocytes transduced to over- express the NSC growth factor EGF Receptor (EGFR)139. These results may imply that it is merely the inherent proliferative capacity of NSC, rather than their precursor properties that makes them more prone to the accumulation of genetic damage and transformation over time. The low incidence of brain tumors in more differentiated regions may thus not be attributed to a lack of these cells to self-renew, but rather to a lack of proliferation, and subsequent opportunity to accumulate genetic damage. Therefore, if giving enough time, the cell of origin of human CNS tumors may in fact be any cell that accumulates enough genetics changes that permits reprogramming into a highly proliferative and immature state. Although this event may occur at a higher frequency in the proliferative NSC compartments of various mouse models, it may have limited applicability to the natural history of the disease in humans. Although this is a valid argument, it is weakened substantially by the unremarkable change in the frequency of tumors in other proliferative regions of the brain such as the hippocampus. Although proliferative, these brain regions are thought to lack self-renewing potential and represent committed precursors rather than population of bonifide stem cells59,140. Therefore, at least in mouse models of brain tumors, glioma-like lesions seem to preferentially originate in region of the brain containing self-renewing stem cells compared to other dormant or more committed proliferating regions of the brain.

! ")!

The different possibilities for the cell of origin in cancer do not need exist as mutually exclusive events. Although, precisely defining the initiating events of cancer may help us one day prevent its occurrence in high-risk populations, this knowledge may have little therapeutic benefit to those already diagnosed. What is important to note, is that whatever the cell of origin may be, many highly aggressive tumors not only express stem cell markers but also behave like these immature cells at both the cellular and molecular level. This suggests that even without a firm understanding of the initiating events in CNS tumors, identifying and translating knowledge of vital signaling pathways operational in NSCs to the study of cancer stem cell still holds great promise for the development of novel cancer stem cell specific therapies for brain cancers.

1.2.7 DEVELOPMENTAL PATHWAYS REGULATE CANCER STEM CELLS Accumulating evidence supports the notion that brain tumors arise, and are maintained; by only a subset of cancer cells with stem cell-like properties found within the heterogeneous tumor mass24,99,100. The inability of traditional therapeutics to eliminate these rare cells may account for past failures and clinical relapses151. The development of agents that act on cancer stem cells (CSCs) and the specific pathways that maintain their stem-like properties may thus provide attractive avenues to effectively treat brain tumors. The concept that cancer is maintained by stem-like cells is reinforced by the self- renewing and multipotent properties of a sub-fraction of cells within the bulk tumor. Furthermore, this is also supported by their expression of multiple stem cell markers including nestin, CD133 and Sox2. Therefore, in addition to these commonalities, signaling pathways involved in stem cell maintenance may also be expected to be active and play a role in regulating cancer stem cells. Studies suggest that this is indeed the case. For example, as described earlier, the NSC mitogen FGF has been suggested to be an autocrine glioma-promoting factor involved in driving proliferation and angiogenesis in brain tumors141,142. Like normal NSCs, cancer cells also seem to be regulated by developmental signaling pathways like Notch, Shh and Wnt that play vital roles in brain development. For example, over-expression of the Notch receptors and their ligands Delta-like 1 (DLK1) and Jagged 1 (JAG1), molecules that promote self-renewal and an expansion of NSC pools, correlate with an increased proliferative index in glioma

! "*! cells143. Like Notch, deregulated self-renewal through mutations in the developmental signaling of the Shh and Wnt pathways also promote medulloblastoma growth in mice144- 147. Although the inactivation of the tumor suppressor PTEN is well recognized to occur in many tumors including those of the brain148,149, insights into its role in NSC self- renewal and proliferation provides a new appreciation for the biological mechanisms that loss of tumor suppressors may play in cancers derived from cancer stem-like cells. By expanding the precancerous precursor pools through increased cell proliferation, it may allow for the accumulation of mutations in a normally relatively quiescent population of cells. Furthermore, the expansion of the very cells known to maintain and feed tumor growth may also lead to the progression of a low-grade tumor to a more aggressive phenotype. As previously noted, the index of stemness, as assessed by expression of stem cell markers like CD133, seems to correlate with tumor grade in CNS malignancies24,105. This association between increased self-renewal and carcinogenesis in the brain is further supported by the observation that the polycomb transcription factor Bmi1, a gene initially characterized to be involved in stem cell self-renewal, is found to be over-expressed in a number of human medulloblastomas126. Unlike PTEN, Bmi1 represents an example where a previously uncharacterized oncogene was first studied for its role in NSC biology and only then hypothesized to be involved in cancer based on the intimate relationship neural developmental pathways and tumor formation. Similarly, evaluation of the literature suggests that many of the pathways being investigated as “cancer stem cell specific” were implicated in cancer long before an appreciation of both cancer stem cell and NSCs in general. For example, although there is mounting evidence that suggests a role for developmental signaling pathways in understanding cancer, genetic and epidemiological data implicated many of these pathways in brain cancer long before they were even being studied in normal NSCs (Table 1.1). Albeit with a few exceptions (Bmi1 and Notch) the majority of developmental signaling pathways were characterized as being important to cancer development prior to wide-spread acceptance of the cancer stem cell hypothesis. For example, the role of FGF and EGF in promoting brain tumor growth was known before the initial identification of NSCs37,51,150,151. Similarly, the oncogenic role of the Wnt and Shh pathways were derived from the study of the hereditary cancers in Turcot and Gorlin syndromes respectively and

! #+! were only later studied for their involvement in NSC biology117,121,152,153. Hereditary brain tumors seen in Cowden’s syndrome, a genetic disorder involving an inherited PTEN mutation, is yet another example where a gene’s role in cancer was well established before it was recognized to be a regulator of stem cell self-renewal. Although it is important to acknowledge that the study of these pathways in stem cell biology has helped us better understand their related role in tumor biology, it must also be realized that to date very few concepts from stem cell biology have produced fundamentally new avenues for research or treatment. In summary, the “cancer stem cell hypothesis” has highlighted remarkable similarities between normal neural stem cell and cancer cell biology. A more precise appreciation of the mechanisms governing the decisions made by normal NSCs, and the translation of knowledge to cancer may lead to the identification of novel and druggable regulatory networks in cancer stem cells.

! #"!

Table 1.1 | Selected pathways implicated in the regulation of both NSCs and brain cancer

Report in Report in Pathway NSCs! Cancer! Epidermal Growth Factor (EGF) [1992]51 [1988]150 Fibroblast Growth Factor (FGF) [1992]51 [1990]151 BMI-1 [2003]128 [2004]126 Sonic Hedgehog (SHH) [2005]116 [1996]152 PTEN [2001]125 [1997]148 Wingless (Wnt) [2003]121 [1998]153 Notch [2002]111 [2005]143 Bone Morphogenic Protein (BMP) [2000]154 [2006]155

! See referenced literature for study details

! ##!

PART III: CHEMICAL BIOLOGY 1.3.1 CHEMICAL BIOLOGY As its name implies, chemical biology is an emerging multidisciplinary field that involves the study and perturbation of biological systems with chemical probes. Although the premise; observing the phenotypic effects that result from the manipulation of biological pathways, is similar to that found in other disciplines such as genetics and molecular biology, chemical biology offers distinct advantages to these more commonly used techniques. For example, unlike mutagenic approaches of altering specific gene and protein function, chemical ligands often allow for reversible, temporal and more transient effects. This is a huge advantage when deletion of a specific gene is often lethal during development or when working with primary tissue. In addition to this, titrating various concentrations of chemical modifiers allows for partial inhibition or activation of particular pathways rather than the complete knockout or non-physiological overexpression strategies commonly seen in genetic studies. Furthermore, the ability of specific chemicals to simultaneously modulate multiple isoforms or proteins belonging to the same family (or processes), allows the scientist to overcome the difficulty of pathway redundancy when probing specific processes. Similarly, combination of chemical agents allows for the design of complex experiments where a number of different unrelated pathways can be easily studied in parallel. Another advantage of chemical biology over systems utilizing genetic manipulation is that probes can be developed into therapeutics for human use after appropriate safety testing.

These examples display some of the distinct advantages that have made chemical biology a useful discipline that can address biological questions previously deemed impractical by conventional molecular biology approaches. The application of chemical biology and its benefits to primary cultures of stem cells will undoubtedly allow scientist to address challenging questions in this emerging field.

1.3.2 APPLICATIONS OF CHEMICAL BIOLOGY IN STEM CELL BIOLOGY The application of chemical biology to stem cell biology is a relatively young field, but it has already proven useful at dissecting the role of specific pathways in a variety of stem cell processes including self-renewal, proliferation and

! #$! differentiation156,157. Perhaps the most striking improvement in our understanding of stem cell self-renewal has come from the application of chemical biology to embryonic stem (ES) cells. Traditionally, ES cells have been maintained in a self-renewing and pluripotent state through co-cultures with feeder cells in media containing serum, and exogenous factors such a leukemia inhibitory factor (LIF) (for mouse) or basic fibroblast growth factor (bFGF or FGF2) (for human ES cells)158. The specific factors within these supplements that are directly responsible for ES cell maintenance are unknown, vary between laboratories and between serum/feeder batches making the manipulation and study of these cells difficult. To further complicate matters, these cocktails has been shown to alter the genetic and epigenetic state of these cells; a process hypothesized to be the route behind the various inconsistent findings reported by different laboratories in the ES cell field159. Small molecule screens for chemicals that can maintain “stemness” in ES cells (i.e. expression of Octamer-4; OCT4), have helped elucidate the specific pathways that maintain pluripotency and have made the conditions used for ES cell culture more defined and reproducible. One such screen uncovered a compound (pluripotin) that can maintain ES cell in the absence of any serum or supportive feeder cell layers. Importantly, this agent also maintained the ability of ES cells to undergo germline transmission in vivo160. Similarly, independent groups have also shown that a cocktail comprised of inhibitors of the kinase FGF receptor (FGFR), mitogen-activated protein kinase kinase (MEK) and glycogen-synthase kinase 3 (GSK3) also supports the long term self-renewal capacity of ES cells in the absence of other factors161. Taken together these studies have helped define the particular pathways involved in the maintenance of pluripotency and proliferation. These agents work together to activate proliferative pathways and inhibit pro-differentiation pathways. Growing ES cells by selectively stimulating chemically defined pathways not only makes the conditions more controlled between labs, but it removes additional factors that may have prevented manipulation of these cells for therapeutic purposes. By maintaining the cells in a pluripotent state with as few signals as possible, the precise control of self-renewal and differentiation down particular lineages may also become a more achievable goal. Systematic chemical and genetic screens that probe for the activation of reporter genes specific for differentiation down particular lineages would likely yield interesting and novel findings in these

! #%! minimal culture conditions.

Small molecules that regulate the specific differentiation of ES cells down particular lineages have also helped elucidate the cues and signals dictating cell fate decisions in pluripotent cells. A chemical genetic screen in ES cultures looking for agents that promote the induction of neuronal differentiation identified a chemical probe that interacts with GSK-3! and promotes neuronal differentiation. Although it remains to be determined if this is the only biological target involved in the lineage specific differentiation, it exemplifies the utility of small molecules in quickly assaying pathways involved in specific aspects of stem cell biology162.

A separate small molecule screen has also shown that the conversion of ES cells to cardiomyocytes can also be efficiently carried out by application of vitamin C163. Although the exact mechanism has yet to be revealed, such screens further exemplify how unanticipated actions of particular natural products can reveal new insights into biology and generate of unique hypotheses that were previously not possible with traditional candidate gene approaches.

Therapeutically, the adaptation of chemical biology reagents across culture systems and organisms allows for the efficient translation of in vitro discoveries to real clinical problems. Single molecule inhibitors of the phospho-inositide 3-kinase (PI3-K) have been shown to induce differentiation of ES cells into functional insulin secreting pancreatic !-cells. Not only do these cells display a pancreatic !-cell phenotype in culture, but they are also capable of rescuing a diabetes mellitus phenotype in mice164. In the CNS, a number of small molecules that activate signaling in pathways including FGF and Shh are known to regulate neural stem cell processes165,166. It is not surprising, given the intimate relationship between cancer and stem cell biology, that chemical inhibitors of these pathways have shown utility when translated to brain cancer therapy166. For example, the characterization of the glial promoting effects of bone morphogenic protein (BMP) in neural stem cell biology has also led to the application of these agents as differentiation therapies for human brain tumor stem cell in murine in vivo xenotransplantation cancer models132,155. The conservation of these pathways across species suggests that clinical trials on these agents could begin immediately.

! #&!

Although these examples are only a few that demonstrate unique roles of small molecules in neural stem cell biology, systematic screens to identify previously unappreciated pathways in normal and cancer neural precursors in an unbiased manner have yet to be conducted. For example, many of the small molecules screens designed to identify chemical regulators of stem cell biology have been biased to particular pathways (i.e. Wnt) already known to be involved in stem cell biology162. Although these screens have helped discover new agents that regulate NSC biology, the screen design precludes identification of novel pathways regulating neural stem cell function. Although investigation into these pathways will undoubtedly lead to new insights and potential therapies in brain cancer, parallel work into indentifying novel normal and cancer stem cell pathways and targets is also necessary to move the field forward.

“Non-hypothesis” driven screening of novel pathways regulating normal neural stem cell properties will help further our understanding of the neural stem cell “ground state”; that is our appreciation of the complete collection of signals regulating their self- renewal and mutlilineage differentiation potential. Large collections of commercially available small molecules libraries and culture methods that allow for growing neural stem cells in minimal growth factor conditions have now made such chemical genetic screens possible. Based on the aforementioned similarities between NSCs and CSCs of the brain, such unbiased approaches that aim to uncover novel pathways in normal stem cell function (and agents that can regulate them) will likely lead to translational research in the fields of both regenerative and cancer therapy in the CNS.

! #'!

PART IV: RESEARCH OBJECTIVES 1.4.1 SCOPE AND IMPACT OF MY THESIS RESEARCH The aim of my PhD research is to identify novel regulatory pathways involved in neural stem cells biology and translate this knowledge by assessing if these pathways are also maintained in cancer stem cells. My efforts thus entailed focusing on 3 objectives that are summarized in the chapters describing my results.

1.4.2 OBJECTIVES Objective 1: Develop an in vitro system that allows for the rapid screening of chemicals (and their presumed molecular pathways) that regulate the expansion of normal neural stem cells. Objective 2: Investigate the functionality of the identified pathways in various populations of brain tumor stem cells Objective 3: Characterize and gain mechanistic insight into the phenotypic and functional distribution of the identified pathways in neural stem cells.

1.4.3 SUMMARY OF PHD RESEARCH During the course of my PhD research, these 3 objectives were met and resulted in the following manuscripts: Diamandis, P., Wildenhain, J., Clarke, I. D., Sacher, A. G., Graham, J., Bellows, D. S., Ling, E. K., Ward, R. J., Jamieson, L. G., Tyers, M. & Dirks, P. B. Chemical genetics reveals a complex functional ground state of neural stem cells. Nat Chem Biol 3, 268-73 (2007).

Diamandis, P., Sacher, A.G., Tyers, M. & Dirks, P.B. New drugs for brain tumors? Insights from chemical probing of neural stem cells. Med Hypotheses 72, 683-7 (2009). ! "#$%$&'#()! *+"!#$%&'("!)*"!Cusimano M., #$"! +*"!,-./.0"!+*"!12.-34"!5*6*"!,-./.0"!7*"! 89"!7*"!!:44"!:*"!;<4-%"!=*"!>!6'-3%"!?*@*"!Neural stem cell populations phase-vary through functionally heterogeneous and equilibrating states of neurotransmitter pathway gene expression. (Submitted, 2009) !

! #(!

Taken together, these studies have yielded novel insights into the role of neurotransmitter pathways in regulating both cancer and normal neural stem cells. My work expands on previous work implicating a number of neurotransmitter pathways in NSC biology167 and demonstrates the simultaneous activity of different neurotransmitter pathways in a single NSC population168. It is also the first to implicate many of these pathways in cancer stem cell biology168,169. Furthermore, it is the first to imply that NSCs are lineage primed168 and have phase varying through gene expression profile that allows them to enter distinct states where they are uniquely position to respond to distinct differentiation cues. This work thus sheds new light into our understanding of stem cell biology of the brain and has various therapeutic applications in the fields of regenerative medicine and oncology.

! #)!

-CHAPTER 2- CHEMICAL GENETIC INTERROGATION OF NEURAL STEM CELLS: CHEMICAL GENETICS REVEAL A COMPLEX FUNCTIONAL GROUND

STATE OF NEURAL STEM CELLS

2.1 PROLOGUE Following the design and validation of a small molecule screening platform for neural precursor cells, I assayed a library of 1,267 pharmacologically active compounds for agents that inhibited the expansion of mouse embryonic day 14.5 neural stem cell cultures (E14.5 NSCs). Interestingly, although the literature implicated many of the identified hits as regulators of neurotransmitter signaling, I showed the recovered agents had a remarkable selectivity towards neural stem cell cultures when compared to CNS cells displaying a more differentiated phenotype. I also show that incubation with a variety of these agents affected the neurosphere-forming replating potential of neural stem cells and led to decreased expression of the neural precursor marker nestin. Furthermore, I demonstrate that the inhibitory effects I observe with these neuromodulatory agents are likely not due to non-specific off-target effects. Consistent with the idea that stem-like cells drive cancer, I also demonstrate that the inhibitory effects seen are conserved in both mouse and human derived cultures of brain tumor stem cells. Taken together this chemical genetic approach is suggestive that neural stem cells may be functionally primed; expressing and responding to a number of genes and pathways commonly associated with a variety of different neuronal subtypes. Furthermore, it suggests that manipulation of these pathways may have clinical implications for psychiatric diseases and cancers of the CNS. Lastly, this paper demonstrates that chemical screens of primary mNSCs provide a convenient method for the identification of agents with anti-CSC activity. The majority of this work described in this chapter was published in: Diamandis, P., Wildenhain, J., Clarke, I. D., Sacher, A. G., Graham, J., Bellows, D. S., Ling, E. K., Ward, R. J., Jamieson, L. G., Tyers, M. & Dirks, P. B. Chemical genetics reveals a complex functional ground state of neural stem cells. Nat Chem Biol 3, 268-73 (2007).

! #*!

2.2 TITLE AND CONTRIBUTORS

CHEMICAL GENETICS REVEAL A COMPLEX FUNCTIONAL GROUND

STATE OF NEURAL STEM CELLS

!

Phedias Diamandis1,2,3,4, Jan Wildenhain4, Ian D. Clarke1,2, Adrian G. Sacher1,2, Jeremy

Graham1,2, David S. Bellows3, Erick K. M. Ling1,2,5 Ryan J. Ward1,2,5, Leanne G.

Jamieson1,2,5, Mike Tyers3,4, and Peter B. Dirks1,2,5,6

1The Arthur and Sonia Labatt Brain Tumor Research Centre, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto, M5G 1X8, Canada

2Program in Developmental and Stem Cell Biology, The Hospital for Sick Children and University of Toronto, Canada

3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, M5G 1X5, Canada

4Department of Medical Genetics and Microbiology, University of Toronto, 1 Kings College Circle, Toronto, M5S 1A8, Canada

5Department of Laboratory Medicine & Pathobiology, Faculty of Medicine, University of Toronto, Banting Institute, 100 College Street, Toronto, M5G 1L5, Canada.

6Division of Neurosurgery, The Hospital for Sick Children and University of Toronto, Canada

Published in Nat Chem Biol 3, 268-73 (2007)

Reproduced with permission from the Nature Publishing Group

! $+!

Figure 2.1 | Graphic abstract 1/40'A.2!B4(4&'A%!-4C4.2!.!A90D24E!F$(A&'9(.2!B-9$(G!%&.&4!9F!(4$-.2!%&40!A422%*!

! $"!

2.3 INTRODUCTION AND SUMMARY The identification of self-renewing and multipotent neural stem cells (NSCs) in the mammalian brain holds promise for the treatment of neurological diseases and has yielded new insight into brain cancer24,99,100. However, the complete repertoire of signaling pathways that governs the proliferation and self-renewal of NSCs, which I refer to as the “ground state”, remains largely uncharacterized. Although the candidate gene approach has uncovered vital pathways in NSC biology111,123,125,128,170, to date only a few highly studied pathways have been interrogated. Based on the intimate relationship between NSC self-renewal and neurosphere proliferation128 (Fig. 2.2a), I undertook a chemical genetic screen for inhibitors of neurosphere proliferation in order to probe the operational circuitry of the NSC. The screen recovered small molecules known to affect neurotransmission pathways previously thought to operate primarily in the mature central nervous system (CNS), and also had potent inhibitory effects on cultures enriched for brain cancer stem cells. These results suggest that clinically approved neuromodulators may remodel the mature CNS and find application in the treatment of brain cancer.

! $#!

Figure 2.2 | HTS of neural precursor cells (a) Neurospheres are derived from self-renewing multi-potent NSCs and contain a heterogeneous mixture of stem, progenitor, and a very small number of differentiated cells. (b) Scatter plot of all 1,267 compounds of the LOPAC library screened against neural precursor cells. 160 compounds (!) were identified as inhibitors of neurosphere proliferation (P < 0.01), 19 compounds (") were identified as activators (P < 0.01) and the remaining agents (#) screened did not have any significant effects on proliferation (P > 0.01). (c) Examples of phenotypic variation observed in response to particular agents. Scale bars, 250!µm.!!!

! $$!

2.4 RESULTS To profile the signaling network of primary cultures of neural precursor cells (NPCs), I screened 1,267 compounds in the Library of Pharmacologically Active Compounds (LOPAC) for inhibitors of neurosphere proliferation, as measured by incorporation of the vital dye Thiazolyl Blue Tetrazolium Bromide (MTT) (Fig 2.2b and Supplementary Table 2.1). A Z’-factor171 of 0.78 and a Pearson correlation coefficient of 0.981 for replicate screens indicated that the assay was reliable (Fig. 2.3; See methods for details). 160 compounds that significantly inhibited neurosphere proliferation (P < 0.01) were clustered into groups of known pharmacologic action (Table 2.1 and Table 2.2). Known cytotoxic compounds that target essential cellular processes predictably scored as hits in the screen. Unexpectedly, however, many agents that modulate neurotransmission in the dopamine, serotonin, opioid, glutamate, vanilloid, and other pathways potently inhibited growth of NPCs. Many of these agents are used in the clinical treatment of neurological disorders and are traditionally thought to act on mature central nervous system (CNS) cell populations. These compounds induced a variety of neurosphere phenotypes, including changes in sphere number, sphere size, and cell-cell or cell-surface adhesion properties suggesting that an elaborate balance of these signaling pathways dictates NPC fate (Fig. 2.2c).

! $%!

!

Figure 2.3 | Validation of a HTS assay for neurosphere cultures (a) Scatter plot of positive (3!µM cycloheximide) and negative controls (vehicle: 0.03% DMSO) demonstrate the dynamic range of the HTS neurosphere assay. Z’-factor analysis confirmed the suitability of the assay for screening. (b) Pearson R correlation of pilot experiments performed in replicate demonstrated reproducibility and accuracy of values over the dynamic range of the screen.

! $&!

Table 2.1 | HTS bioactive pharmacological classes ! Active Total % Active Class† Agents‡ Agents in Class Cytotoxic§ 38 114 33 Biochemistry 6 46 13 Cannabinoid 1 6 17 Cholinergic 8 77 10 Cyclic Nucleotides 4 31 13 Dopamine 22 113 20 Glutamate 9 88 10 Intracellular Ca2+ 2 7 29 Ion Pump 3 16 19 Lipid 1 9 11 Na+ Channel 3 17 18 Nitric Oxide 5 37 14 Opioid 6 27 22 P2 Receptor 2 14 14 Phosphorylation 9 93 10 Serotonin 12 83 14 Vanilloid 2 5 40 Entire Screen 160 1267 13¥ ! ,5(A2$G4%!.22!A2.%%4%!H'&/!.!IJ!#A&'C4!'(!12.%%K!9F!.&!! 24.%&!LMJ! -N(2

! $'!

Table 2.2 | HTS bioactive pharmacological classes ! Total active Class in class, Total in class % Active in class Adenosine 4 56 7.1 Adrenceptor 5 103 4.9 Antibiotics 8 29 27.6 Anticonvulsant 1 12 8.3 Apoptosis 6 11 54.5 Biochemistry 6 46 13.0 Calcium Channels 2 26 7.7 Cannabinoid 1 6 16.7 Cell Cycle 5 15 33.3 Cell Stress 3 20 15.0 Cholinergic 8 77 10.4 Cyclic Neucleotides 4 31 12.9 Cytoskeleton and ECM 6 10 60.0 DNA 10 29 34.5 Dopamine 22 113 19.5 GABA 2 41 4.9 Glutamate 9 88 10.2 Histamine 2 32 6.3 Hormone 3 33 9.1 Imidazoline 1 11 9.1 Immune System 1 11 9.1 Intracellular Calcium 2 7 28.6 Ion Pump 3 16 18.8 K + Channel 1 17 5.9 Lipid 1 9 11.1 Na+ Channel 3 17 17.6 Neurotransmission 4 45 8.9 Nitric Oxide 5 37 13.5 Opioid 6 27 22.2 P2 Receptor 2 14 14.3 Phosphorylation 9 93 9.7 Prostaglandin 1 24 4.2 Serotonin 12 83 14.5 Vanilloid 2 5 40.0 Screen Total 160 1267 12.6‡ Screen Total 2§ 122 1154 10.6 †Only includes inhibitors of MTT readings. ‡Frequency of whole screen §Total calculated without cytotoxic compounds: antibiotics, apoptosis, cell cycle, cell stress, cytoskeleton, and DNA

! $(!

To verify hits from the primary screen, 43 representative candidates were retested at the original screen concentration of 3! µM; of these 40 (93%) exhibited significant activity (Table 2.3 and Fig 2.4). Because other neural cell types express and signal through a number of neurotransmitter receptors172, I assessed the selectivity and potency of each agent for a normal mouse astrocyte cell line versus NPCs. Dose-response curves were generated for 28 compounds in both neurosphere and astrocyte cultures and used to determine the effective concentration needed to decrease proliferation by 50% (EC50) (Fig. 2.5a-f and Table 2.3). A neurosphere selectivity ratio, defined as

EC50(astrocytes)/EC50(neurospheres), for each compound was determined and compared to that of known non-specific inhibitors of proliferation (Fig. 2.5a-c). Compounds that exhibited a neurosphere selectivity ratio greater than that observed in these control agents (> 5.08) were defined as NPC-specific agents (Fig. 2.5d-f and Table 2.4); 12 of the compounds tested exhibited high selectivity for NPCs. Notably, the serotonin agonist p- aminophenethyl-m-trifluoromethylphenyl pierazine (PAPP, 14) and the vanilloid receptor ligand dihydrocapsaicin were respectively 702 and 192 fold more selective for normal NPCs than astrocyte cultures. Neurospheres are comprised of a heterogeneous population of neural stem cells and lineage restricted progenitor cells. To determine if the inhibitors affected NSC self- renewal, as opposed to proliferation of more committed precursor populations, I analyzed the higher order colony forming efficiency of treated neurosphere cultures. With the exception of dihydrocapsaicin, representative compounds from the major neurotransmission classes significantly reduced higher order neurosphere formation upon re-culture in the absence of drug (Fig 2.6). The various inhibitors thus appear to target the neural precursor pool that is predominantly responsible for sphere formation.

! $)!

Table 2.3 | Assessment of potency and specificity of selected agents

Neurosphere Astrocyte Neurosphere Name Action Target Selectivity EC50 (µM) EC50(µM) Selectivity Controls Cycloheximide (1) Inhibitor Protein Synthesis 60S Ribosome 0.142 0.071 0.50 Etoposide (2) Inhibitor Topoisomerase Topo II 0.340 0.433 1.28 Carboplatin (3) Intercalator DNA - 0.489 2.453 5.08 Sobuzoxane (19) Inhibitor Topoisomerase Topo II 10.19 n.t. n/a Mevastatin (20) Inhibitor n/a Ras, Rho 0.142 n.t. n/a Taxol (21) Inhibitor Tubulin - 0.010 n.t. n/a Vinblastine (22) Inhibitor Tubulin - 0.028 n.t. n/a Vincristine (23) Inhibitor Tubulin - > 0.001 n.t. n/a

Adrenergic )VWXY64(9D.0'(4!(24)! Agonist Adrenoceptor Beta 1 39.2 n.t. n/a Rauwolscine (25) Antagonist Adrenoceptor Alpha 2 66.37 n.d. >1.43

Dopamine Perphenazine (26) Antagonist Dopamine Receptor D2 23.12 n.d. >4 (±)-Butaclamol (4) Antagonist Dopamine D2>D1 0.785 12.34 15.7 R(-)-Propylnorapomorphine (5) Agonist Dopamine D2 0.3512 8.23 23.4 R(-)-Apomorphine (6) Agonist Dopamine - 0.3499 10.19 29.1 cis-(Z)-Flupenthixol (7) Antagonist Dopamine Receptor - 0.1993 1.182 5.93 (+)-Bromocriptine (18) Agonist Dopamine Receptor D2 1.187 n.t. n/a

Cannaboid Indomethacin (27) Agonist Cannabinoid receptor CB2 n.d. n.t. n/a

Ion Channels Bepridil (28) Blocker Ca2+ Channel - 2.70 4.724 1.75 Dequalinium (29) Blocker K+ channels Apamin-sensitive 1.474 1.418 0.962

MAO Quinacrine (30) Inhibitor MAO-A/B - 0.936 n.t. n/a

Muscarinic p-F-HHSiD (8) Antagonist Acetylcholine Receptor M3>M1>M2 0.441 5.815 13.2 Methoctramine (31) Antagonist Acetylcholine Receptor M2 1.053 0.0845 0.802

NMDA Ifenprodil (9) Blocker NMDA Polyamine site 0.616 11.06 17.9 Pentamidine (32) Antagonist NMDA Receptor - 0.822 1.995 2.43

Nitric Oxide Diphenyleneiodonium (33) Inhibitor Nitric Oxide Synthase eNOS 0.011 0.0209 1.88 7-Nitroindazole (34) Inhibitor Nitric Oxide Synthase nNOS 76.3 282.6 3.71

Opioid Metaphit (35) Antagonist Opioid sigma 10.04 3.624 0.361 Carbetapentane (10) Ligand Opioid sigma 1 0.756 28.16 37.3

Phosphorylation Chelerythrine (36) Inhibitor PKC - 0.396 1.531 3.87 Fenretinide (11) Vitamin A RAR - 0.334 2.399 7.18 acid analog WHI-P131 (12) Inhibitor JAK3 - 2.346 - > 10 SB 202190 (13) Inhibitor p38 MAPK - 8.063 64.8 8.04 Tyrphostin AG 34 (37) inhibitor Tyrosine Kinase - 9.917 n.t. n/a.

Serotonin Methiothepin (38) Antagonist Serotonin 5-HT1E/F, 5-HT6 2.663 3.698 1.39 Metergoline (39) Antagonist Serotonin 5-HT2/ 5-HT1D 1.624 3.285 2.02 PAPP (14) Agonist Serotonin 5-HT1A 0.031 21.82 702 CGS-12066A (40) Agonist Serotonin 5-HT1B 2.007 14.4 7.17

Vanilloid Dihydrocapsaicin (15) Agonist Vanilloid Receptor VR1 0.218 41.83 192

Other 5-Bromo-2'-deoxyuridine (41) Inhibitor DNA - 2.045 n.t. n/a 7,7-Dimethyl-(5Z,8Z)-eicosadienoic Inhibitor Phospholipase A2 / - 5.170 n.t. n/a acid (42) Lipoxygenase n.d. = not determined at highest tested dose (30-95 mM) n.t. = not tested

! $*!

!

Figure 2.4 | Confirmation of HTS hits Normalized MTT values of a representative sample of hits taken from different pharmacological classes in the LOPAC collection. Compounds are annotated by Sigma- Aldrich catalog number. All values represent the mean of triplicate values of three independent experiments and error bars represent s.e.m.. Of the 43 compounds retested, 40 (93%) (blue) were confirmed as significant (P<0.05) when compared to control wells (green) using Student’s t-test.

! %+!

Figure 2.5 | Identification of potent neural precursor cell-specific compounds Dose-response curves and chemical structures of controls: (a) cycloheximide, (b) etoposide, and (c) carboplatin, and selected newly identified compounds: (d) dihydrocapsaicin, (e) apomorphine, and (f) PAPP. Each plot displays the fitted sigmoidal logistic curve to MTT proliferation assay readings of both astrocytes (- -$- -) and neurosphere cultures (%#%). Values represent the mean and standard error of the mean (s.e.m.) of three independent experiments.

! %"!

Table 2.4 | Highly potent and highly selective compounds identified by HTS of E14.5 mouse neurospheres

ptc1+/! ptc1+/!p53!/! Neurosphere Astrocyte Neurosphere Neurosphere Neurosphere § § Name Action Target Selectivity EC50 (µM) EC50(µM) Selectivity EC50(µM) EC50(µM) Controls Cycloheximide (1) Inhibitor Protein Synthesis 60S Ribosome 0.142 0.071 0.50 0.042 0.054 Etoposide (2) Inhibitor Topoisomerase Topo II 0.340 0.433 1.28 0.208 n.t. Carboplatin (3) Intercalator DNA - 0.489 2.453 5.08 0.196 n.t.

Selected Hits† (±) Butaclamol (4) Antagonist Dopamine Receptor D2>D1 0.785 12.34 15.7 0.751 2.533 R(-)-Propylnorapomorphine (5) Agonist Dopamine Receptor D2 0.351 8.230 23.4 0.199 n.t. R(-)-Apomorphine (6) Agonist Dopamine Receptor - 0.350 10.19 29.1 0.168 0.683 cis-(Z) Flupenthixol (7) Antagonist Dopamine Receptor - 0.199 1.182 5.93 0.187 n.t. p-F-HHSiD (8) Antagonist Acetylcholine Receptor M3>M1>M2 0.441 5.815 13.2 1.125 1.373 Ifenprodil (9) Antagonist NMDA Receptor Polyamine site 0.616 11.06 17.9 0.451 0.807 Carbetapentane (10) Agonist Opioid Receptor sigma 1 0.756 28.16 37.3 2.083 2.040 Fenretinide (11) Agonist Retinoic Acid Receptor - 0.334 2.399 7.18 0.204 n.t. WHI-P131 (12) Antagonist JAK3 - 2.346 n.d. > 10 1.525 n.t. SB 202190 (13) Antagonist p38 MAPK - 8.063 64.8 8.04 3.006 n.t. PAPP (14) Agonist Serotonin Receptor 5-HT1A 0.031 21.82 702 0.169 0.097 Dihydrocapsaicin (15) Agonist Vanilloid Receptor VR1 0.218 41.83 192 0.020 0.651

Cyclopamine (16) Antagonist Smoothened n.t. n.t. n/a 1.00 13.44

†Compounds listed represent confirmed hits with high selectivity for neural precursor cells (neurosphere selectivity > 5) §ptc1+/! and ptc1+/!p53!/! neurosphere cultures were derived from mouse cerebellar tumors samples. n.d. = not determined at highest tested dose (30 µM) n.t. = not tested

! %#!

Figure 2.6 | Identification of potent neural precursor cell-specific compounds Replating colony forming efficiency of pretreated neurosphere cultures. Values represent the number of progeny neurospheres arising from 2,000 or 1,000 cells plated in fresh media following a seven day pretreatment of NPCs with the indicated inhibitor at the estimated EC75 value. As the EC75 of apomorphine did not allow the recovery of sufficient cells, an EC50 pretreatment was used for this agent. Sphere counts for vehicle treated cells represents the mean and standard deviation (s.d.) of six separate replicates conducted during two independent experiments. All other values represent the mean of two independent experiments. Asterisks indicate a reproduced statistically significant (P < 0.05) reduction in replating efficiency compared to that observed agent’s vehicle. The larger P-value (of the two experiments) is reported. These differences (at both 2,000 and

1,000 cells per well) were confirmed (2 tailed paired t-test) for PAPP (P2000,1000 =

0.02,0.008) and apomorphine (P2000,1000 = 0.01,0.02) treated cultures in three independent trials.

! %$!

To further delineate the mechanism through which neuromodulatory agents impede expansion of NPCs in culture, I performed time course analyses for both cell viability and apoptosis. Unlike etoposide (Fig. 2.7a) and cycloheximide (data not shown), which have immediate effects on cell proliferation and viability, the neurotransmission modulators PAPP and ifenprodil decreased viable cell numbers only after 2 days post-treatment (Fig. 2.7a). Similar delayed onset-effects were observed for butaclamol, p-fluoro-hexahydrosila-difenidol (p-F-HHSiD, 8) and carbetapentane (data not shown). Consistently, caspase-3 and 7 levels were unchanged following 12 hours of PAPP and ifenprodil treatment, but increased significantly after 2 days of drug treatment (Fig. 2.7b). This increase in the apoptotic response of treated cells occurred at concentrations of drugs that did not abolish the initial proliferation or viability of these cells (Fig. 2.7c). Finally, expression of the immature NPC marker nestin was substantially decreased after treatment for 2 days with ifenprodil (Fig. 2.7d) and PAPP (data not shown). These results suggest that appropriate neurotransmission signaling is required to maintain NSC proliferation, survival and identity.

! %%!

Figure 2.7 | Temporal effects of selected neuromodulators on NPC viability and apoptotic response (a) Proliferation dynamics of PAPP, ifenprodil and etoposide treated NPCs. (b) Normalized caspase-3 and 7 activity in NPCs following 12 and 48 hour drug treatment. Asterisk indicates a significant change (2 tailed t-test) from the corresponding vehicle treated data point. (c) Corresponding MTT values taken at 12 hours and 2 days for the caspase-3 and 7 experiments shown in (b). All values represent the mean and s.d. of one representative experiment (from 3 independent trials) of NPCs treated with PAPP (1 µM), ifenprodil (3!µM), etoposide (3!µM) or vehicle. (d) Flow cytometric analysis of the neural precursor marker nestin in NPCs following 2 day treatment with ifenprodil (5!µM) or vehicle. Representative histograms of vehicle treated (20% nestin negative) and ifenprodil treated (63% nestin negative) cells compared to the isotype control (100% nestin negative) are shown from two independent experiments.

! %&!

As gene expression profiles of brain tumors resemble those of normal and embryonic NPCs25,106,146,173, agents that inhibit normal neural precursor growth may also inhibit cultures of brain tumors that are enriched for cancer stem cells99,100,106. I therefore assessed the activity of a subset of NPC-specific inhibitors against low passage (< 4) neurosphere cultures derived from spontaneously formed medulloblastomas in heterozygous patched (ptc1+/!) and ptc1+/!-p53!/! mice174. Like their normal counterparts, cancerous NPCs from these tumors grow as spheres in serum-free culture and express the neural precursor marker prominin1 (CD133) (Fig. 2.8a,b). The NPC- specific agents also potently suppressed the proliferation of both ptc1+/! and ptc1+/!p53!/! medulloblastoma precursor cell populations (Fig. 2.8c and Table 2.4). Notably, some of these agents were an order of an magnitude more effective in the inhibition of tumor cell growth in vitro than the hedgehog signaling inhibitor cyclopamine166. The expansion of normal human NPCs and human glioblastoma cells was also inhibited by neuromodulators (Table 2.5). For example, PAPP and ifenprodil, had EC50 values comparable to those of commonly used non-specific brain-tumor chemotherapeutic drugs, such as carboplatin and etoposide. Redeployment of well tolerated pharmacologically active agents may thus afford a new generation of chemotherapeutic agents specific for brain tumor stem cells.

! %'!

Figure 2.8 | Neuromodulator drug sensitivity in cancer-derived NPC (a) ptc1+/& tumors contain cells with self-renewing neurosphere-forming potential in vitro. Scale Bar, 125 mm. (b) Ptc1+/& tumors cells stain positive (M1) for the early precursor marker rominin1 (CD133 homolog) at comparable levels to primary human medulloblastomas (11.6%)103. Unstained (black) and stained (red) specimens are shown. +/& &/& (c) EC50 values (mean and s.d.) for inhibition of ptc1 p53 tumor sphere MTT proliferation by various neuromodulators. Compound identity indicated in Table 2.

! %(!

Table 2.5 | Normal and cancerous human NPCs show sensitivity to a myriad of neurotransmission modulators

hFetal hGBM1 hGBM2 precursors precursors precursors Z ![ P Name Action Target Selectivity EC50 (µM)! EC50 (µM) EC50 (µM) Controls Etoposide Inhibitor Topoisomerase Topo II 0.16 0.62 0.27 Carboplatin Intercalator DNA - 0.43 2.04 3.30

Selected Hits, Apomorphine Agonist Dopamine Receptor - 5.26 14.58 0.31 p-F-HHSiD Antagonist Acetylcholine Receptor M3>M1>M2 10.58 12.63 1.23 Ifenprodil Antagonist NMDA Receptor Polyamine site 0.42 1.99 0.206 Carbetapentane Agonist Opioid Receptor sigma 1 6.12 5.44 1.73 PAPP Agonist Serotonin Receptor 5-HT1A 0.22 1.87 0.31 Dihydrocapsaicin Agonist Vanilloid Receptor VR1 3.28 88.63 21.46 ! †Only a selected array of the identified mouse neural precursor selective agents were tested in human cells. All agents tested are displayed in this table. ¶Values against neural precursors derived from human fetal CNS tissue. 'Pathological diagnosis of hGBM1 is WHO grade IV GMB §Pathological diagnosis of hGBM2 is WHO grade IV GBM (giant cell variant).

! %)!

As even well characterized agents may exert biological effects through off-target pathways175, I verified that a number of the known receptors for various agents were indeed expressed in both normal and tumor NPCs. The dopamine (DRD2), acetylcholine (M3), NMDA (NR1), and serotonin (5HT-1A) receptors were present in primary and secondary normal mouse neurosphere cultures and ptc1+/& tumor neurosphere cultures, as determined by RT-PCR (Fig. 2.9a). In addition, I was able to use pharmacological inhibitors as a means to assess whether the growth inhibition caused by the dopamine class of neuromodulators depends on transmission through a known receptor. In one example, (+)-sulpride (17), a D2 dopamine receptor antagonist, was able to competitively rescue the inhibitory effects of the D2/3 dopamine receptor agonist, bromocriptine (18), as judged by both colony formation (Fig. 2.9a,b and Fig. 2.10) and MTT values (data not shown). To further assess the potential for off-target effects of neuromodulators in other classes, I clustered the 160 bioactive agents from the primary screen based on their chemical structures (Supplementary Table 2.1). This analysis demonstrated substantial chemical structural diversity within each of the different neuromodulator classes. For example, the 22 bioactive dopamine agents identified in the screen spanned 10 different structural motif clusters; similarly, the 12 active serotonergic agents covered 10 different chemical clusters (Fig. 2.11 and Fig. 2.12). The observed sensitivity of NPCs to these structurally diverse agents is thus likely to arise through effects on known neurotransmission receptors, as opposed to some unknown coincident target.

! %*!

Figure 2.9 | Neuromodulator drug sensitivity in NSCs represent on-target effects (a) RT-PCR gene expression profiles of a selection of neurotransmitter receptors in different precursor populations. mRNA from serum differentiated neurospheres and mouse erythroid leukemia (MEL) cells were used as positive and negative controls respectively. Vertical black line indicates non-contiguous lanes from the same experiment. (b) Inhibition of colony formation by bromocriptine in cultures with and without (+)-sulpiride supplementation. Normalized mean and s.e.m. values of a three independent triplicate cultures are shown. Sulpiride challenge significantly shifted the

EC50 of bromocriptine from 1.2!µM (without sulpiride) to 2.5!µM (with sulpiride) (P < 0.05).!indicating a rescue effect. (c) Representative micrographs of the inhibitory effects of bromocriptine on NPC expansion when challenged with a competitive antagonist. Scale bars, 500!µm.

! &+!

!

Figure 2.10 | Pharmacological competition at the D2 dopamine receptor blunts the inhibitor effects of bromocriptine (a)-(b) Micrographs from 2 independent experiments depicting that the DRD2 antagonist (+)-sulpride can rescue the NPC inhibitory effects of the dopamine agonist bromocriptine on neurosphere formation. Scale bars, 500 mm (c) Quantification of colony formation ability in bromocriptine treated cultures with and without (+)-sulpride inoculation. Values represent the mean and s.d. of a representative experiment. Asterisk indicates significant difference (p=0.011, 2-tailed t-test). This trend were reproducible in

! &"! two independent experiments (d) MTT dose response analysis of (+)-sulpride and/or bromocriptine treated cultures. Values represent the mean and s.e.m. of three independent experiments. Asterisk indicates significant different (t[5] = 5.3, p<0.01, 2-tailed paired t- test).

! &#!

!

!

Figure 2.11 | Bioactive dopamine modulators display a rich intra-class chemical diversity 8 of the 22 bioactive agents identified as dopamine signaling regulators. All 8 molecules are found in unique clusters when grouped based on 2D chemical fingerprint. A total of 10 different clusters were identified within the 22 agents known to act on the dopamine pathways. P-value represents the original significance testing preformed from the screening data. Reported drug targets displayed in this figure represents curated data published in Drugbank176 and inhouse.

! &$!

Figure 2.12 | Bioactive serotonin modulators display a rich intra-class chemical diversity 8 of the 12 bioactive agents identified as serotonin signaling regulators. All 8 molecules are found in unique clusters when grouped based on 2D chemical fingerprint. A total of 10 different clusters were identified within the 12 agents known to act on the serotonin pathway. P-value represents the original significance testing preformed from the screening data. Reported drug targets displayed in this figure represents curated data published in Drugbank176 and inhouse.

! &%!

2.5 DISCUSSION

The ex vivo and in situ manipulation of NSCs for treating neurological disorders, including brain cancer, will require an understanding of the global signaling network that regulates NSC behavior. Through a chemical genetic approach I have uncovered the existence of a complex functional “ground state”, whereby NSC proliferation and self- renewal is regulated by a plethora of signaling pathways (Fig. 2.13a,b). Significantly, this cohort includes many neurotransmission pathways previously thought to function only in mature cells of the CNS. I infer that NSC proliferation and self-renewal thus requires an appropriate local environment of neurotransmitter activities, which may be provided by a basal level of autocrine feedback from more committed cells within the neurosphere or even the NSC itself. Indeed, recent studies on individual pathways support the notion that proliferation of different progenitor subpopulations in vivo may respond to dopamine, serotonin, acetylcholine and glutamate167. Importantly, my chemical genetic profile demonstrates the simultaneous operation of these pathways in NPCs cultured under uniform experimental conditions. This sensitivity of NPC cultures to a spectrum of neuroactive compounds also supports the notion of lineage-priming in the NSC compartment, similar to that seen in hematopoietic stem cells177.

! &&!

Figure 2.13 | Functional ground state of NSCs (a) Current models of the NSC hierarchy focus on developmental signaling pathways such as Wnt, Notch, and Sonic Hedgehog. (b) Compounds identified in the HTS approach reveal that the NSC ground state and cell fate decision-making depends on a complex circuitry that includes many neurotransmission signaling pathways.

! &'!

While I have yet to definitively identify the precursor subpopulation(s) targeted by each inhibitor identified in the screen, the strong selectivity of many agents for NPCs and primary brain tumor cells suggests that the affected pathways lie high in the hierarchical organization of the neuronal lineage. Indeed, the often complete inhibition of neurosphere proliferation and the effects on secondary replating suggests that stem cells and/or very early progenitor components of the population are affected by these agents. The finding that both inhibitors and activators of specific pathways inhibit neurosphere proliferation (e.g. dopamine receptor agonists and antagonists) suggests that a complex signaling landscape dictates NSC fate178. I note that the pro-proliferative culture conditions used in the neurosphere assay may have biased the assay against identification of significant numbers of small molecules that stimulate proliferation. A small molecule activator of embryonic stem cell proliferation has recently been identified160, suggesting that analogous screens may succeed in identifying activators of NPC proliferation. The unanticipated actions of well-characterized clinical agents on NPCs may account in whole or in part for the observed clinical benefits of these agents and/or the adverse side-effects that arise after prolonged therapy. Effective in vivo concentrations of the anti-Parkinsonian drug apomorphine reach 6-7!µM179, which is substantially higher than doses that affect NPCs in vitro. The regulation of NSC proliferation by neurotransmitters may thus also dictate how the CNS is wired both during development and in the adult brain180. Recent evidence suggests that appropriate GABA stimulation of NPCs is required for the proper integration of neurons in the adult hippocampus181. Through structure activity analysis, I also identified specific chemical substitutions important for the bioactivity of these agents in our in vitro system (Fig 2.14 and Fig. 2.15). Such modifications to the core chemical structure of many clinically used agents may thus afford a way to regulate the potentially therapeutic or harmful effects these drugs exhibit on NPCs.!! In light of the evidence that CNS tumors are maintained by cancer stem cells24,99 which have similarities to normal NSCs106, the potent and selective anti-proliferative agents identified in this study may presage a next generation of therapeutic agents in brain cancer, although further in vivo testing is required24. Intriguingly, a retrospective analysis of cancer incidence in Parkinson’s patients revealed a significant reduction in the

! &(! prevalence of brain tumors182; this correlation may derive from the effect of anti- Parkinsonian drugs on the NPCs from which brain tumors are thought to arise. As the complex NSC ground state I propose is likely to at least in part define the identity of brain tumor stem cells, redeployment of pharmacologically-approved neuroactive agents may provide an immediate and non-toxic means to treat often intractable CNS tumors.

!

!

! &)!

Figure 2.14 | SAR analysis of adenosine and serotonin agonists (a) SAR analysis an array of structurally similar adenosine agonist (cluster 132). i. CGS- 21680 hydrochloride ii. 5`-N-Ethylcarboxamidoadenosine iii. HE-NECA iv. 2- Phenylaminoadenosine v. 5`-N-Methyl carboxamidoadenosine vi. N6- Cyclohexyladenosine. (b) SAR analysis an array of structurally similar serotonin agonists (cluster 127 ). i. R(+)-UH-301 (A) ii. R(+)-UH-301 (B) iii. R-(+)-8-Hydroxy-DPAT iv.

! &*!

(±)-8-Hydroxy-DPAT vi. (±)-PPHT (DRD2 agonist). “active” agents and represents agents found to significantly suppress the number of viable cells (MTT scores) in the initial screen. “inactive” agents represent agents that were predicted to have activity (but did not) based on their structural similarities; suggesting important structural changes. Functional substitutions presumed to have positive (green) and negative (red) effects on activity are circled. !

!

! '+!

Figure 2.15 | SAR analysis of dopamine agonists and antagonist (a) SAR present in structurally related dopamine antagonists (cluster 130). i. Perphenazine ii. Fluphenazine iii. Trifluoperazine iv. cis-(Z)-Flupenthixol v. Triflupromazine vi. Propionylpromazine vii. Prochlorperazine (b) SAR presnt in

! '"! structurally related dopamine agonist (cluster 84). i. Apomorphine ii. R(-)-Apocodeine iii. R(-)-Propylnorapomorphine iv. R(-)-N-Allylnorapomorphine v. R(-)-2,10,11- Trihydroxyaporphine vi. R(-)-2,10,11-Trihydroxy-N-propylnoraporphine. All agents (unless otherwise stated) were active in the initial screen. Agents are arranged in descending order with respect to their observed biological response. Functional substitutions presumed to have positive (green) and negative (red) effects on activity are circled. !

!!

! '#!

2.6 METHODS

Animals All animals used for these experiments were housed, maintained and manipulated at the Hospital for Sick Children (HSC, Toronto) in a humane manner that was in accordance with guidelines established by the Animal Care Committee. FVB and Ptch1 mice were obtained from The Jackson Laboratories (Bar Harbor, Maine) and were maintained and bred at the HSC Laboratory Animal Services (LAS) facility. For the isolation of murine embryonic tissue, FVB male and female paired each night to undergo time mating. Successful mating (and potential pregnancies) were assessed the following morning, by resistance to vaginal probing on physical examination due to the presence of male- derived seminal coagulates produced by fluids from both the vesicular and coagulating gland (plug). Female mice positive for this test (plug) were marked by ear notches and noted to their fetuses were noted to be at the E0.5 stage of development. Pregnant females were maintain for another 2 weeks, upon which they were sacrificed by cervical dislocation and E14.5 embryos were harvested from the abdominal cavities.

Isolation and culture of primary embryonic murine neural stem cell (mNSCs) from E14.5 embryos Isolation and culture of primary embryonic (e14.5) mNSC was performed as previously described in chemically-defined neural stem cell media91 containing 20 ng/mL human recombinant epidermal growth factor (EGF) (Sigma), 20 ng/mL basic fibroblast growth factor (bFGF) (Upstate) and 2 µg/mL heparin (Sigma) and fed every 2-3 days52. More specifically, following removal from the abdominal cavities of pregnant mothers, the yolk sac containing E14.5 embryos was transferred to a Petri dish containing phosphate buffered saline (PBS, Wisent Laboratories). Embryos were separated from their yolk sac and placental membrane and transferred to a fresh Petri dish where they were maintained in artificial cerebral spinal fluid (ACSF: NaCl (2mM), KCl (1 mM),

MgCl2 (1M), NaHCO3 (155 mM), Glucose (1M), CaCl2 (108 mM), Antibiotic/Antimycotic (1%; Wisent Laboratories). Under the aid of a dissection microscope (Lecia, Wetzlar, Germany), the telencephalic regions (containing both the

! '$! medial and ganglionic eminences) were excised from the surrounding dura and vasculature and placed in another Petri dish containing ACSF until all telencephalons from the remaining embryos were collected. Using a flame-narrowed pasture pipette, telencephalons were mechanically dissociated into a single cell suspension by carefully pipetting the tissue up and down 5-10 times. Following centrifugation (3 min, 1200 rpm), the ACSF supernatant was vacuumed and the cellular pellet was resuspended in complete mouse neural stem cell media (mNSC media; basal media (DMED:F12 50:50 (Gibco), Glucose (0.6%), HEPES (5 mM), Antibiotic/Antimycotic (1%)), supplemented with EGF (human recombinant; 20 ng/ml; Sigma-Aldrich Canada), bFGF (20 ng/ml; Stem Cell Technologies), Transferrin (100 ug/ml), Insulin (25 ug/ml), Putrecine (60 uM), Sodium Selenite (30 nM), Progesterone (20 nM) and Heparin (2ug/ml)) and any remaining undissociated tissue aggregates were removed by filtered the cells through a 40 uM cell strainer (BD, Falcon). A hemocytomer and the exclusion of trypan blue (Gibco) was used to assess the number of viable cells isolated from the dissection. For expansion as floating cellular clusters (Neurospheres), viable cells were plated at a density of 106/ml of complete mNSC media in non-coated 10 cm Petri dishes and maintained at 37oC, 5%

CO2 and atmospheric oxygen (20%) in incubators. Conditioned cellular media was partially replenished (50:50) with fresh mNSC media every 2-3 days of culture. Following 7 days of expansion as primary neurospheres, cellular aggregates were dissociated by first centrifugating the aggregates into a pellet and resuspended it in 2 ml of Accutase (Sigma) for 5 minutes at 37oC. The generation of a single cell suspension was further aided by mechanical dissociation by repeatedly pipetting the cell with a 1 ml filtered plastic pipette (5-10 times). The Accutase suspension was then diluted by adding an additional 10 mL of basal media. Cells were spun down and resuspending in fresh mNSC media, filtered, counted and finally plated once again at the desired density.

Secondary mNSC neurosphere culture and chemical screens Prior to chemical screens and other manipulations, the NSC fraction in culture was expanded by growing freshly dissected cells as primary neurospheres107 in bulk culture (106 cells/mL). After 7 days, primary neurospheres were collected and enzymatically digested for 3 minutes at 37oC using Accutase (Sigma), mechanically dissociated with a 1

! '%! mL pipette and passed through a cell strainer (Falcon). Viable cells were plated at low cell densities (20 cells/µL) in 96-well plates (Falcon) in a final volume of 100 µL and screened in singlets against the LOPAC library (Sigma) at a concentration of 3 µM (0.03% DMSO). On day 4, each well in the screen was supplemented with an additional 50 µL of fresh mNSC media and another aliquot of the LOPAC library (maintaining the final concentration of each compound at 3 µM). Secondary neurosphere cultures were then incubated for an additional 3 days (day 7) at which point the effect of each compound was assessed by quantifying the total proliferation of each well using the incorporation of the vital dye Thiazolyl Blue Tetrazolium Bromide (MTT) (Sigma) as previously described103.

Statistical analysis for chemical screen Background plate effects (Fig. 2.16) occurring from the evaporation of media over the course of the experiment were estimated by:

1 N b x' i = h ! i, j N " Ni j=1

' where xi, j is the value at well i of plate j, h is the number of excluded hits that were 2 standard deviations below the mean and bi represents the estimated background at each well position183.

The respective background was then subtracted from the raw MTT value measured for each point (Fig. 2.16 and Fig. 2.17). To calculate significance (z-score and P-value), the theoretical probability density function [N(1.0,0.11)] was fitted to the empirical normalized distribution obtained from the screen (Fig. 2.17). Compounds that caused optical density readings to significantly deviated from this predicted underlying distribution function (P < 0.01) were designated as bioactive184.

! '&!

Figure 2.16 | Assessment of plate edge effects Optical density scatter and mesh plot of MTT values as a function of well row and column are shown. Due to the long incubation time of plates at 37oC, a row and column- dependant edge effect emerged due to differential evaporation over the course of the screen. Systematic noise was removed as described in Supp. Fig. 2.17. !

! ''!

!

!

Figure 2.17 | Correction for plate edge effects in the screen (a) Density function of the raw data obtained from the HTS of neurospheres prior to the correction for edge-effects. (b) Density function of the normalized (black line) and the fitted theoretical distribution (red line) used to calculate significance. (c) Box plot representation of the raw data of each plate in the screen. (d) Re-plot of data after the removal of the evaporation induced systematic error helped to reduce the number of both false positives and negatives. Box length for each plot represents the interquartile range (IQR) (Q3 – Q1). The solid black line represents the median value for each plate and cutoffs represent values 1.5*(IQR) from Q3 and Q1. !

!

! '(!

Dose-Response Curves and EC50 Calculations Potency of confirmed bioactive compounds was quantified by generating dose-response curves for mNSC under the same cell density and culture conditions described for the initial screen. Starting from initial concentrations between 300-30 µM, each compound was titrated across a series of 10 half-log dilutions. Each agent was tested in triplicate in at least three independent experiments. EC50 values for each agent was calculated by fitting the data points to the four-parameter logistic sigmoidal dose-response curve:

EC ! EC Y = EC + 0 100 100 log(EC50 !X )(Hill Slope) 1+10 ! where X is the logarithm of concentration and Y is the predicted response. Curve fitting was performed with GraphPad Prism Software. ! Assessment of the NSC specific effects of selected inhibitory agents To assess effects on the NSC fraction of precursor cultures, the number of neurospheres generated from a single cell suspension of either 1000 or 2000 cells was determined after chemical treatment. Primary neurospheres were dissociated into a single cell suspension and subjected to the estimated EC75 of selected agents from different neurotransmission classes for seven days. The cultures were re-dissociated and plated in fresh media in the absence of compound; neurospheres that grew to >50!µm in diameter after a further seven days were used as an index of the number of NSC present in the original treated culture. Data shown represents the average of two independent experiments each containing 6 replicates. ! Astrocyte screen and neurosphere selectivity assessment Selectivity of each compound for mNSC was assessed by constructing dose-response curves and EC50 calculations for the normal astrocytic GFAP expressing cell line C8- D1A (American Type Culture Collection, ATCC), which is derived from cells from the cerebellum of an 8 day old mouse. Cell densities and feeding schedules for these cells were identical to those described in the mNSC cultures. Astrocytes were grown in

! ')!

DMEM media (GIBCO) supplemented with 10% fetal bovine serum; all screens were performed on astrocytes cultured for five to ten passages.

Brain Tumor Generation and Culture Mouse tumor cells were isolated from the cerebellum of 16 week old patched heterozygous (ptc1+/-) mice or ptc1+/-p53-/- mice displaying ataxia. Tumors were separated from normal tissue, dissociated and resuspended in serum-free chemically defined medium. Subsequent culture and HTS of these cells was preformed as described above for normal E14.5 NSC; data shown represents the average and standard deviation of a single experiment preformed in triplicate. Human brain tumors and neural precursors were isolated and expanded as previously described103. A detailed description of methods used for the isolation, culture, and confirmation of multipotency of all human stem cell culture is provided below. Dose response curves and EC50 calculations for the effects of tested drugs on both normal and cancerous human precursor cells were performed in serum-free conditions containing EGF and FGF at 20 cells/µL, as described above.!!! ! Time course analysis and apoptosis assay Primary neurospheres were dissociated and plated at a density of 15,000 cells/well. Cell numbers were quantified at 0, 24, 48, and 96 hours post plating by the MTT assay as previously described in triplicate and as two independent trials. The effects of PAPP, ifenprodil and etoposide on caspase-3/7 activity were assessed using the Apo-One homogeneous caspase-3/7 assay kit as described by the manufacturer (Promega). Briefly, cells were plated at 15,000 cells/well and treated with drug or vehicle. Casapse-3/7 acitvity was assessed at 12 hr and after 2 days by adding caspase substrate and incubating the sample for 4 hours. Readings were collected using an excitation wavelength of 499 nm and an emission wavelength of 521 on a florescent plate reader, background florescence was removed and values were then normalized to number of viable cells (assessed by MTT). !

! '*!

Flow cytometry Intracellular staining was performed as described elsewhere185. Primary spheres were dissociated and plated as a single cell suspension in 10 cm2 dishes at a density 0.5X106/mL and treated with either vehicle (ethanol), PAPP (3!µM) or Ifenprodil (5!µM) for 2 days. Cells were then harvested, dissociated and washed with PBS once. Cell were then fixed at room temperature in a 1.6% paraformaldehyde (Sigma) solution. Fixed cells were washed with PBS and permeabilized by adding 1mL of Methanol (-20oC) and incubating on ice for 30 minutes. Cells were then blocked with 1 mL of staining buffer (0.5% Normal Goat Serum with 4mM EDTA) for 30 minutes. Primary staining was preformed using Mouse anti-Rat Nestin (BD Pharmingen) and isotype control antibodies (Dako Cytomation) diluted to a final concentration of 1mg/mL in staining buffer. After a 30 minute incubation at 4oC, cells were washed in 1mL of staining buffer and resuspended in staining buffer containing 4 µg/mL of the secondary Alexa Fluor® conjugated goat anti-mouse IgG1 (Molecular Probes) antibody. Following another 30 minute incubation at 4oC, cells were washed and held at 4oC for 1 hour prior to analysis on a FACScan (Becton Dickenson). Gates were established using the isotype control as the negative population. Analysis of prominin1 expression in mouse tumors was preformed as previously described24. Briefly, primary spheres were dissociated into a single cell suspension and resuspended in 1X PBS with 0.5% BSA and 2 mM EDTA. 4 \:!CD133-PE (eBioscience) was added to 100 µL of cell suspension and incubated for 30min in the dark at 4°C; 4 µg/mL Propidium Iodide was added to mark dead cells. Prominin1 expression was assessed by the proportion of cells that were positive for expression above the levels see in the unstained control.

RNA isolation and RT-PCR Total RNA was isolated from cultured cells using the RNeasy extraction kit (Qiagen) and DNA contamination removed with the RNeasy MinElute Clean-up Kit (Qiagen). Approximately 100 ng of each total RNA sample was used to generate a corresponding

! (+! cDNA sample using an oligo (dT)5 primer and the Transcriptor reverse transcriptase kit (Roche). Primer pairs used were: dopamine D2 receptor (sense, 5’- TTCTTGGTGTGTTCATCATCTG-3’; antisense, 5’- CAGACTCAGCAGTGCAGGAT- 3’), serotonin receptor 5HT1A (sense, 5’-TGAGTTGTTGGGTGCCATAA-3’; antisense, 5’-CCTTCTCCATCACCACCACT-3’), acetylcholine receptor M3 (sense, 5’- GAAAAGGATGTCGCTCATCAA-3’; antisense, 5’- TCTTGTTGCACAGGGCATAG- 3’), and NMDA receptor NR1 (sense, 5’- TTCTGCAAGCGAGGACGA-3’; antisense, 5’-ACTCGTTCTTGCCGTTGATT-3’).

Structure activity relationship (SAR) analysis and chemical clustering To perform SAR analysis, I used the assumption that the experimental results obtained from the screen represented biologically true positive (TP) and true negative (TN) activities. The bioactive agents (TP) identified, were used as a positive training set using a Bayesian classifier. The leave-one-out cross validation method was then used to generate likelihood scores for each molecule screen from the LOPAC Library (Supplementary Table 2.1). The predictive power and distributions of the positive and negative training set are displayed in Figure 2.18. Compounds that had positive likelihood scores but were found to have inactivity (p>0.01) in the actual screen, were identified as interesting compounds for SAR. The inactivity of these agents could thus be attributed to important chemical substitutions (assuming these changes did not compromise the compound stability in culture). Although not discussed in the paper, examples of SAR analysis for small molecules with strong core structures can be found in Figures 2.14 and 2.15. For cluster-analysis, the 2D chemical structural fingerprints of all agents were generated and used to calculate the dissimilarity co-efficient (1-(Tanimoto)186. A maximal partitioning algorithm was then used to test different average cluster sizes (ranging from 10 to 2 molecules/cluster)187. Clustering was validated by confirming concordance in clustered chemical structures and discordance of chemicals in different clusters. Cluster assignments are displayed in Supplemental Table 2.1 in descending order of size.

! ("!

Figure 2.18 | Development and validation of training set for SAR analysis (a) Roc curve displaying the predictive power of the structural features (extended connectivity fingerprints, hydrogen donor and acceptors, rotational bounds, weight and alogP) used in training the Bayesian classifier to distinguish between active from non active compounds. (b) Density distribution of these structural learned features. The

! (#! predicted locations of the true positive (TP), true negative (TN) false positive (FP) and false negative (FN) are displayed on the density function. The joined subset of FN + FP + TP structures (total 271 molecules) were then used to identify common pharmacophores that could be used for SAR analysis. FP (11% using a likelihood score > 0) identified in this analysis represent substitutions with “unpredicted” and potential biologically important biological consequences.

Isolation of NSC from human foetal CNS tissue With the ethical approval of the Research Ethics Boards at The Hospital for Sick Children and Mount Sinai Hospital (Toronto), tissue from human foetal CNS was collected from first trimester (8-13 weeks of gestation) aborted fetuses. When applicable, subcorital forebrain tissue was separated from the surrounding spinal cord tissue in ACSF, and mechanically dissociated with the aid of a 1 ml filtered plastic pipette. Following dissociation, cells were pelleted by centrifugation and resuspended in 6 mL of human NSC media (hNSC media: basal media (DMED:F12 50:50 (Gibco), Glucose (0.6%), HEPES (5 mM), Antibiotic/Antimycotic (1%)), supplemented with EGF (human recombinant; 20 ng/ml; Sigma-Aldrich Canada), bFGF (20 ng/ml; Stem Cell Technologies), Transferrin (100 ug/ml), Insulin (25 ug/ml), Putrecine (60 uM), Sodium Selenite (30 nM), Progesterone (20 nM), Heparin (2ug/ml), NSF (1x; Clonetics) and Leukemia Inhibitory Factor (LIF, 10 ng/ml; Chemicon International). Following filtration through a 40 um cell strainer, cells were plated in a 6 cm non-coated Petri dish and o maintained as neurosphere cultures in an incubator at 37 C, 5% CO2 and atmospheric oxygen (20%). The conditioned media of the cells was partially exchanged (50:50) with fresh media every 2-3 days and sphere cultures were passage every 5-10 days required. For dissociation, cells were collected and incubated in Accutase at 37 oC for 10-20 minutes, mechanically disrupted, passed through a 40 um cell strainer and resuspended in an equal mixture of fresh and conditioned hNSC media.

Primary human brain tumor isolation and primary culture Under the ethical guidelines approved by the Research Ethics Boards at The Hospital for Sick Children, St. Michael’s Hospital, Toronro Western Hospital and the Sunnybrooke

! ($!

Health Sciences Centre (Toronto), brain tumor samples were obtained from consenting patients scheduled for therapeutic surgery to remove a malignant intracranial tumor mass. Following retrieval from the operating room on ice, specimens were first rinsed with ACSF and contaminating vasculature, calcifications and/or white matter were excised from the tumor mass using surgical scissors and forceps. The remaining tumor tissue was then manually minced with the surgical tools to a fine pulp and then further enzymatically digested for 30-60 minutes in ACSF containing Trypsin (1.33 mg/ml), Hyaluronidase (0.67 mg/ml) and Kynurenic Acid (0.1 mg/ml) to further weaken the cell- cell interactions. To remove the emzymatic cocktail, cells were washed in basal media, pelleted and resuspended and incubated for 5 min at 37oC in 10 ml of PBS containing 0.7 mg/ml of Trypsin inhibitor (Ovalbumin). This was followed by another two washes and centrifugation steps in basal media upon which the final product was repeated pipetted to further mechanically dissociate the cells into a single cell suspension. Any remaining non-dissociated tissues was removed by straining cells through a 70 um filter (BD, San Jose, CA). Contaminating red blood cells were then removed through the use of a sucrose density gradient (Lympholyte-M, Cedarlane Laboratories, Ltd). The layer containing non-red blood cells was carefully removed from the gradient mixture, washed another two times in basal media, and plated for expansion as neurospheres in non-coated 10 cm Petri dish at a density of 2x106 cells/ml. The conditioned media of the cells was partially exchanged (50:50) with fresh media every 2-3 days and sphere cultures were passage every 5-10 days required. For dissociation, cells were collected and incubated in Accutase at 37 oC for 20-30 minutes, mechanically disrupted, passed through a 40 um cell strainer and resuspended in an equal mixture of fresh and conditioned hNSC media.

Adherent culturing of human foetal and glioma neural stem (NS) cells Initial NS cultures were generated by plating early passage serum-free suspension cultures of both human foetal and glioma on 10 cm Pitre dishes precoated with poly-L- ornithine (Sigma) & laminin (Sigma). This matrix was prepared by first pre-coating the culture plates with 0.01% poly-L-ornithine for 30 minutes at room temperature; washing twice with PBS; followed by the addition of laminin (10ug/ml of PBS) for at least 6 hours at 37oC. All NS cells were maintained in complete NS media (NS Media: Neurocult NS-

! (%!

A basal media containing L-Glutamine (2mM), BSA (75 ug/ml), Antibiotic/Antimycotic (1%), and supplemented with fresh EGF (human recombinant; 10 ng/ml; Sigma-Aldrich Canada), bFGF (10 ng/ml; Stem Cell Technologies), Transferrin (100 ug/ml), Insulin (25 ug/ml), Putrecine (60 uM), Sodium Selenite (30 nM), Progesterone (20 nM), Heparin (2ug/ml), and 1X B27). The conditioned media of the cells was partially exchanged (50:50) with fresh NS media every 2-3 days and adherent cultures were passage every 5- 10 days as required. To passage adherent cultures, their media was removed, and cells were first incubated in Accutase at 37 oC for 5 minutes followed by mechanically disruption with a 1 ml filtered plastic pipette tip. The Accutase was then quenched by the addition of 10 ml of basal media and removed by pelleting the cells via centrifugation. Viable cell numbers were counted by trypan blue exclusion and desired cell densities were then resuspended in fresh NS media and plated on recently poly-L-ornithine (Sigma) & laminin coated plated.

Induction of multipotent differentiation of adherent neural precursor cultures Once NS cultures were ready for passage, cells were detached and cell number assessed as previously described. Approximately 10-50 X 105 cells were plated on pre-coated poly-L-ornithine & laminin coverslips submerged in 24 wells plates or 4-well chamber slides in complete NS media. The next day, the media was removed, cells were washed ones with basal NS media and grown in NS media whose growth factors were partially withdrawn (NS-GFW1 Media: Neurocult NS-A basal media containing L-Glutamine (2mM), BSA (75 ug/ml), Antibiotic/Antimycotic (1%), and supplemented with fresh bFGF (5 ng/ml; Stem Cell Technologies), Transferrin (100 ug/ml), Insulin (25 ug/ml), Putrecine (60 uM), Sodium Selenite (30 nM), Progesterone (20 nM), Heparin (2ug/ml), and 1X B27). Following one full week under these conditions, cells were once again washed with basal NS media and grown for an additional two week in media whose growth factors were complete withdrawn (NS-GFW2 Media: 50:50 mixture of Neurocult NS-A basal media containing L-Glutamine (2mM), BSA (75 ug/ml), Antibiotic/Antimycotic (1%) and Neural Basal Media (Gibco) supplemented with Transferrin (25 ug/ml), Insulin (6.25 ug/ml), Putrecine (15 uM), Sodium Selenite (7.5 nM), Progesterone (5 nM), Heparin (2 ug/ml), and 1X B27). During this three week

! (&! differentiation process, the conditioned media of the cells was partially exchanged (50:50) with the appropriate type of fresh media every 2-3 days. At the end of this protocol, cells were preserved by fixation and multipotent differentiation was assessed as outlined in section entitled “Immunocytochemical analysis of NS cultures”.

Immunocytochemical analysis of NS cultures Adherent NS cells were prepared for immuncytochemical analysis by growing them on poly-L-ornithine & laminin coated coverslips submerged in 24-well plates or 4-well chamber slides (as previously described). On the day the cultured cells were desired to be analyzed, the cells were directly fixed onto the cultured slides by adding an equal amount of 4% paraformaldehyde (PFA) was added to the cell culture media for 30-60 minutes. PFA was removed by first aspirating the media/PFA mixture followed by two 5 minute washes with PBS. The cells were then stored at 4oC for 1-2 weeks or until staining commenced. Staining began by permeablizing the cell membranes with the addition of 0.03% TritonX (Sigma) in PBS for five minutes followed by two additional 5 min washes with PBS. To prevent non-specific binding, permeablized cells were first incubated in a 10% normal goat serum (NGS) blocking solution of PBS that was passed through a 40 um filter for 1 hour at room temperature. To distinguish the presence of astrocytes and neurons, cells were then incubated for another 1 hour at room temperature in 10% NGS containing primary antibodies raised against either GFAP (1:1000; Dako) or !-Tubulin type III (1:500, Chemicon) respectively. The precursor phenotype of the cultured cells was characterized by incubating the cells for 1 hr at room temperature with primary antibodies against the neural precursor markers Nestin (1:1000) and Sox2 (1:1000). The excess unbound primary antibody was then removed through three successive 5 minutes washes with PBS and cells were then subjected to various fluorescently labeled secondary antibodies for 1 hr at room temperature in the dark. For the detection of bound primary antibodies raised in mouse (!-tubulin, Sox2), the anti-mouse IG-A568 antibody was used. Detection of bound primary antibodies raised in rabbits (GFAP, Nestin), the anti-rabbit IG-A488 antibody was used. Excess secondary antibody was then removed by three successive 5 minute washes with PBS. Finally, the coverslips were then mountedon to microscopic slides with 6 uL of DAKO mounting media supplemented with the

! ('! nuclear staining dye DAPI (1:500). Specimen visualization was performed on a SD 200 Sprectral Bioimaging System (ASI Ltd., Isreal) attached to a Zeiss Axioplan 2 Microscope (Carl Zeiss, Canada). Pictures were obtained and analyzed post-hoc using the AxioVision Software package.

! ((!

Supplemental Table 2.1 | HTS of Neural Precursor Cells

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID U-62066 -0.303 3.8109E-01 no Opioid Agonist kappa 1.9970 32 (±) trans-U-50488 methanesulfonate -2.833 2.3049E-03 yes Opioid Agonist kappa 7.8149 32 (-)-cis-(1S,2R)-U-50488 tartrate 0.171 4.3208E-01 no Neurotransmission Ligand Sigma receptor 5.0720 32 (-)-trans-(1S,2S)-U-50488 hydrochloride -0.297 3.8321E-01 no Opioid Agonist kappa 7.5748 32 (+)-trans-(1R,2R)-U-50488 hydrochloride -0.473 3.1802E-01 no Opioid Agonist kappa 7.5748 32 BRL 52537 hydrochloride -2.761 2.8849E-03 yes Neurotransmission Agonist kappa/mu opioid 8.7509 32 GR-89696 fumarate -0.779 2.1806E-01 no Opioid Agonist kappa 0.0475 32 AC 915 oxalate -0.890 1.8665E-01 no Opioid Ligand sigma1 3.3295 32 Fluphenazine dihydrochloride -10.197 1.0242E-24 yes Dopamine Antagonist D1/D2 20.6718 130 Trifluoperazine dihydrochloride -9.305 6.6778E-21 yes Dopamine Antagonist D1/D2 17.0927 130 Triflupromazine hydrochloride -1.820 3.4349E-02 no Dopamine Antagonist DRD2 12.3307 130 Perphenazine -11.065 9.3020E-29 yes Dopamine Antagonist DRD2 18.2918 130 Prochlorperazine dimaleate -0.860 1.9485E-01 no Dopamine Antagonist 11.6765 130 Propionylpromazine hydrochloride -1.584 5.6571E-02 no Dopamine Antagonist DRD2 6.2928 130 cis-(Z)-Flupenthixol dihydrochloride -8.822 5.5976E-19 yes Dopamine Antagonist DRD1/DRD2/A2a/ADRA1A 19.2865 130 Pheniramine maleate -0.206 4.1831E-01 no Histamine Antagonist HRH1 1.4913 109 (±)-Brompheniramine maleate -3.530 2.0756E-04 yes Histamine Antagonist HRH1 3.9173 109 (+)-Chlorpheniramine maleate -1.272 1.0172E-01 no Histamine Antagonist HRH1 3.9173 109 (+)-Brompheniramine maleate -3.976 3.5087E-05 yes Histamine Antagonist HRH1 3.9173 109 (±)-Chlorpheniramine maleate -2.116 1.7156E-02 no Histamine Antagonist HRH1 3.9173 109 Disopyramide -4.179 1.4650E-05 yes Na+ Channel Blocker 6.0562 109 Disopyramide phosphate -0.072 4.7135E-01 no K+ Channel Modulator 5.7949 109 CGS-21680 hydrochloride -1.225 1.1027E-01 no Adenosine Agonist A2a 4.7849 132 5`-N-Ethylcarboxamidoadenosine -2.587 4.8437E-03 yes Adenosine Agonist A1/A2 1.5341 132 HE-NECA -2.504 6.1439E-03 yes Adenosine Agonist A2 12.7320 132 5`-N-Methyl carboxamidoadenosine -2.582 4.9052E-03 yes Adenosine Agonist A2 > A1 -0.1474 132 2-Phenylaminoadenosine 0.573 2.8340E-01 no Adenosine Agonist A2 > A1 1.7469 132 N6-Cyclohexyladenosine -2.883 1.9695E-03 yes Adenosine Agonist A1 -2.8521 132 R(-)-2,10,11-Trihydroxyaporphine -3.309 4.6827E-04 yes Dopamine Agonist DRD2 29.2738 84 R(-)-2,10,11-Trihydroxy-N-propylnoraporphine -2.744 3.0324E-03 yes Dopamine Agonist DRD2 30.9879 84 Apomorphine hydrochloride hemihydrate -11.274 8.8620E-30 yes Dopamine Agonist 29.7434 84 R(-)-N-Allylnorapomorphine hydrobromide -7.839 2.2697E-15 yes Dopamine Agonist 30.6967 84 R(-)-Propylnorapomorphine hydrochloride -8.274 6.4584E-17 yes Dopamine Agonist DRD2 31.4575 84 R(-)-Apocodeine hydrochloride -9.865 2.9486E-23 yes Dopamine Agonist 28.9096 84 Metoclopramide hydrochloride -1.992 2.3176E-02 no Dopamine Antagonist DRD2 7.2623 83 Tiapride hydrochloride 0.689 2.4543E-01 no Dopamine Antagonist D2/D3 4.0505 83 SDZ-205,557 hydrochloride -2.681 3.6740E-03 yes Serotonin Antagonist 5-HT4 10.9325 83 Procainamide hydrochloride 0.654 2.5671E-01 no Na+ Channel Antagonist 3.7651 83 N-Acetylprocainamide hydrochloride -4.557 2.5891E-06 yes Na+ Channel Blocker 6.2381 83 N-(2-[4-(4-Chlorophenyl)piperazin-1-yl]ethyl)-3-methoxybenzamide 0.383 3.5092E-01 no Dopamine Agonist D4 6.0821 83 Naloxone benzoylhydrazone 1.915 2.7727E-02 no Opioid Agonist kappa 2.0832 59 Naloxone hydrochloride -10.137 1.8971E-24 yes Opioid Antagonist 6.8882 59 Naloxonazine dihydrochloride -0.606 2.7223E-01 no Opioid Antagonist mu1 5.4107 59

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Naltrexone hydrochloride 0.389 3.4860E-01 no Opioid Antagonist 3.5271 59 Nalbuphine hydrochloride 0.070 4.7197E-01 no Opioid Antagonist 0.2990 59 (-)-3-Methoxynaltrexone hydrochloride 0.596 2.7547E-01 no Opioid Antagonist 4.8978 59 Cephapirin sodium 0.622 2.6699E-01 no Antibiotic Cell wall synthesis 1.8833 54 Cephalothin sodium 2.264 1.1788E-02 no Antibiotic Cell wall synthesis 4.6902 54 Cephalosporin C zinc salt -3.309 4.6797E-04 yes Antibiotic Cell wall synthesis 12.0888 54 Cefotaxime sodium -1.096 1.3654E-01 no Antibiotic Cell wall synthesis 3.9918 54 Cefazolin sodium -2.580 4.9470E-03 yes Antibiotic Cell wall synthesis 19.8926 54 Ceftriaxone sodium 2.230 1.2886E-02 no Antibiotic Cell wall synthesis 1.1330 54 R(+)-UH-301 hydrochloride -2.517 5.9202E-03 yes Serotonin Agonist 5-HT1A 16.3801 127 S(-)-UH-301 hydrochloride -2.706 3.4078E-03 yes Serotonin Antagonist 5-HT1A 16.3801 127 R-(+)-8-Hydroxy-DPAT hydrobromide -0.692 2.4460E-01 no Serotonin Agonist 5-HT1A 9.7698 127 (±)-8-Hydroxy-DPAT hydrobromide 0.656 2.5587E-01 no Serotonin Agonist 5-HT1A 9.7698 127 (±)-PPHT hydrochloride -1.915 2.7715E-02 no Dopamine Agonist DRD2 9.6773 127 S(+)-Raclopride L-tartrate 1.901 2.8633E-02 no Dopamine Antagonist DRD2 4.1295 126 S(-)-IBZM -0.117 4.5360E-01 no Dopamine Ligand DRD2 5.4292 126 (-)-Sulpiride -2.647 4.0550E-03 yes Dopamine Antagonist DRD2 7.5872 126 (±)-Sulpiride -0.284 3.8816E-01 no Dopamine Antagonist DRD2 7.5872 126 S-(-)-Eticlopride hydrochloride -5.158 1.2497E-07 yes Dopamine Antagonist DRD2 14.6402 126 3-Tropanyl-3,5-dichlorobenzoate -9.272 9.0927E-21 yes Serotonin Antagonist 5-HT3 10.0230 29 4`-Chloro-3-alpha-(diphenylmethoxy)tropane hydrochloride -9.672 1.9822E-22 yes Dopamine Blocker Reuptake 13.5550 29 DL-Homatropine hydrobromide 0.744 2.2844E-01 no Cholinergic Antagonist Muscarinic 3.3069 29 Benztropine mesylate -3.749 8.8704E-05 yes Cholinergic Antagonist Muscarinic 12.5609 29 Aminobenztropine -1.455 7.2819E-02 no Cholinergic Ligand Muscarinic 9.8555 29 SKF-525A hydrochloride -1.928 2.6951E-02 no Multi-Drug Resistance Inhibitor Microsomal oxidation 3.1837 20 PRE-084 1.311 9.4859E-02 no Opioid Agonist sigma1 4.7579 20 Carbetapentane citrate -8.021 5.2461E-16 yes Opioid Ligand sigma1 12.5359 20 Procaine hydrochloride 0.550 2.9111E-01 no Na+ Channel Blocker 1.0800 20 (±)-threo-1-Phenyl-2-decanoylamino-3-morpholino-1-propanol hydrochloride -1.109 1.3382E-01 no Sphingolipid Inhibitor Glucosylceramide synthase 1.4587 20 GBR-12909 dihydrochloride -10.899 5.8437E-28 yes Dopamine Inhibitor Reuptake 17.7542 119 GBR-12935 dihydrochloride -11.274 8.8620E-30 yes Dopamine Inhibitor Reuptake 17.4032 119 Diphenhydramine hydrochloride 0.845 1.9907E-01 no Histamine Antagonist HRH1 6.6928 119 Orphenadrine hydrochloride 0.455 3.2463E-01 no Cholinergic Antagonist Muscarinic 4.7871 119 Doxepin hydrochloride -1.012 1.5566E-01 no Adrenoceptor Inhibitor Uptake 0.7541 67 Amitriptyline hydrochloride 0.707 2.3990E-01 no Adrenoceptor Inhibitor Uptake 2.9998 67 Chlorprothixene hydrochloride -5.008 2.7566E-07 yes Dopamine Antagonist DRD2 12.4473 67 Thiothixene hydrochloride -1.899 2.8770E-02 no Dopamine Antagonist D1/D2 12.4318 67 (-)-Bicuculline methbromide, 1(S), 9(R) -1.842 3.2731E-02 no GABA Antagonist GABA-A 3.4384 58 (+)-Bicuculline -0.199 4.2119E-01 no GABA Antagonist GABA-A 15.5316 58 (+)-Hydrastine -0.836 2.0152E-01 no GABA Antagonist GABA-A 17.7536 58 Noscapine hydrchloride -2.455 7.0361E-03 yes Opioid Ligand 21.4022 58 Ritodrine hydrochloride -2.332 9.8562E-03 yes Adrenoceptor Agonist beta2 1.5332 53 R(-)-Denopamine -4.486 3.6321E-06 yes Adrenoceptor Agonist beta1 2.2356 53

! ()!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID DL-alpha-Methyl-p-tyrosine -3.908 4.6497E-05 yes Neurotransmission Inhibitor Tyrosine hydroxylase -0.1777 53 alpha-Methyl-DL-tyrosine methyl ester hydrochloride 0.370 3.5559E-01 no Neurotransmission Inhibitor Tyrosine hydroxylase 0.5909 53 LY-53,857 maleate -0.338 3.6761E-01 no Serotonin Antagonist 5-HT2/5-HT1C 0.3126 51 Mesulergine hydrochloride -2.160 1.5386E-02 no Dopamine Agonist 4.5314 51 Metergoline -4.342 7.0622E-06 yes Serotonin Antagonist 5-HT2/5-HT1D 14.2794 51 Pergolide methanesulfonate -2.657 3.9457E-03 yes Dopamine Agonist D2/D1 1.4244 51 Retinoic acid -0.866 1.9333E-01 no Apoptosis Activator 1.3311 22 13-cis-retinoic acid -1.469 7.0905E-02 no Transcription Regulator RAR-alpha, beta 1.3311 22 Retinoic acid p-hydroxyanilide -6.013 9.0821E-10 yes Cell Cycle Inhibitor 9.8493 22 Astaxanthin 1.571 5.8038E-02 no Cell Stress Inhibitor Antioxidant 1.2987 22 Promethazine hydrochloride 0.475 3.1754E-01 no Histamine Antagonist HRH1 3.5263 16 Promazine hydrochloride -0.888 1.8723E-01 no Dopamine Antagonist DRD2 6.8657 16 10-(alpha-Diethylaminopropionyl)-phenothiazine -1.545 6.1222E-02 no Biochemistry Inhibitor Butyrylcholinesterase 4.2861 16 Thioridazine hydrochloride 0.236 4.0687E-01 no Dopamine Antagonist D1/D2 7.4168 16 Clomipramine hydrochloride -0.844 1.9938E-01 no Serotonin Inhibitor Reuptake 0.4609 108 Chlorpromazine hydrochloride 0.751 2.2645E-01 no Dopamine Antagonist 9.8583 108 Clorgyline hydrochloride 1.520 6.4194E-02 no Neurotransmission Inhibitor MAO-A 0.3833 108 NG-Hydroxy-L-arginine acetate -10.958 3.0541E-28 yes Nitric Oxide Metabolite NOS -0.6344 104 L-2-aminoadipic acid -2.334 9.8098E-03 yes Glutamate Inhibitor Glutamine synthetase -1.4913 104 alpha-Guanidinoglutaric acid -9.041 7.7797E-20 yes Nitric Oxide Inhibitor NOS 3.7998 104 ML-9 -9.322 5.7099E-21 yes Phosphorylation Inhibitor MLCK 17.0244 78 ML-7 -11.274 8.8620E-30 yes Phosphorylation Inhibitor MLCK 16.9036 78 HA-100 -0.689 2.4534E-01 no Phosphorylation Inhibitor PKA / PKC / MLCK 2.0402 78 Decamethonium dibromide -1.572 5.7997E-02 no Cholinergic Agonist Nicotinic 1.3227 76 Hexamethonium bromide -0.820 2.0614E-01 no Cholinergic Antagonist Nicotinic 0.8360 76 Hexamethonium dichloride -10.634 1.0320E-26 yes Cholinergic Antagonist Nicotinic 0.4196 76 NS 521 oxalate -11.274 8.8620E-30 yes Glutamate Modulator Benzimidazolone 2.9504 55 Pimozide -0.232 4.0814E-01 no Dopamine Antagonist DRD2 2.0490 55 Domperidone -1.525 6.3676E-02 no Dopamine Antagonist DRD2 2.2927 55 Purvalanol A -11.234 1.3858E-29 yes Phosphorylation Inhibitor CDK 16.1452 50 CGP-74514A hydrochloride -9.117 3.8610E-20 yes Phosphorylation Inhibitor Cdk1 19.7353 50 Tyrphostin AG 1478 1.870 3.0751E-02 no Phosphorylation Inhibitor EGFR 3.1243 50 Pirenperone 0.228 4.0987E-01 no Serotonin Antagonist 5-HT2 1.7303 48 LY-310,762 hydrochloride 0.051 4.7962E-01 no Serotonin Antagonist 5-HT1D 3.6520 48 Ketanserin tartrate -3.325 4.4160E-04 yes Serotonin Antagonist 5-HT2 6.5571 48 Methiothepin mesylate 1.868 3.0896E-02 no Serotonin Antagonist 5-HT1E, 5-HT1F, 5-HT6 6.2477 33 (±)-Octoclothepin maleate 1.473 7.0371E-02 no Dopamine Antagonist DRD2 1.6480 33 Methoxamine hydrochloride -3.255 5.6606E-04 yes Adrenoceptor Agonist alpha1 4.8466 33 Phenamil methanesulfonate 1.485 6.8800E-02 no Na+ Channel Inhibitor Amiloride sensitive 3.3719 14 Amiloride hydrochloride -0.799 2.1207E-01 no Na+ Channel Blocker Epithelial 2.6796 14 3`,4`-Dichlorobenzamil -2.581 4.9275E-03 yes Ion Pump Inhibitor Na+/Ca2+ exchanger 11.4518 14 Methotrexate -10.203 9.6328E-25 yes DNA Metabolism Inhibitor 31.7569 13 (-)Amethopterin -8.704 1.5969E-18 yes DNA Metabolism Inhibitor 31.7569 13

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Aminopterin -9.783 6.6618E-23 yes Antibiotic Inhibitor Dihydrofolate reductase 28.1012 13 cis-(±)-8-OH-PBZI hydrobromide -2.297 1.0800E-02 no Dopamine Agonist D3 1.3911 8 (±)-7-Hydroxy-DPAT hydrobromide -11.102 6.1185E-29 yes Dopamine Agonist D3 12.3042 8 R-(+)-7-Hydroxy-DPAT hydrobromide -2.590 4.7938E-03 yes Dopamine Agonist D3 12.3042 8 1-(m-Chlorophenyl)-biguanide hydrochloride 1.795 3.6304E-02 no Serotonin Agonist 5-HT3 3.4618 136 1-Phenylbiguanide -2.689 3.5800E-03 yes Serotonin Agonist 5-HT3 2.6508 136 U-99194A maleate 1.255 1.0468E-01 no Dopamine Antagonist D3 2.9061 133 YS-035 hydrochloride 0.043 4.8281E-01 no Ca2+ Channel Blocker L-type 3.7365 133 Ifenprodil tartrate -8.904 2.6930E-19 yes Glutamate Blocker Polyamine site NMDA 11.1544 128 Ro 25-6981 hydrochloride -10.524 3.3394E-26 yes Glutamate Antagonist NMDA-NR2B 12.4160 128 Trazodone hydrochloride -2.644 4.0994E-03 yes Serotonin Inhibitor Reuptake 17.8168 122 BRL 15572 0.312 3.7764E-01 no Serotonin Antagonist 5-HT1D 1.9532 122 2`,3`-dideoxycytidine -1.367 8.5880E-02 no Immune System Inhibitor Reverse Transcriptase 1.8676 115 Cytosine-1-beta-D-arabinofuranoside hydrochloride -11.274 8.8620E-30 yes DNA Metabolism Inhibitor 4.9565 115 Efaroxan hydrochloride -2.338 9.6932E-03 yes Imidazoline Antagonist I1 2.0987 113 Methoctramine tetrahydrochloride -7.012 1.1764E-12 yes Cholinergic Antagonist M2 7.2873 113 3-Amino-1-propanesulfonic acid sodium -6.691 1.1076E-11 yes GABA Agonist GABA-A -2.1148 111 3-Aminopropylphosphonic acid -2.331 9.8664E-03 yes GABA Agonist GABA-B -2.8464 111 Indomethacin morpholinylamide -3.398 3.3995E-04 yes Cannabinoid Ligand CB2 6.9161 106 Indomethacin -0.356 3.6081E-01 no Prostaglandin Inhibitor COX 0.4544 106 2,3-Dimethoxy-1,4-naphthoquinone -10.398 1.2595E-25 yes Cell Stress Modulator 3.2447 103 NSC 95397 -1.098 1.3600E-01 no Phosphorylation Inhibitor Cdc25 1.5944 103 Cefaclor -3.009 1.3115E-03 yes Antibiotic Cell wall synthesis 14.9709 100 Cephalexin hydrate -3.966 3.6478E-05 yes Antibiotic Cell wall synthesis 13.0448 100 Ethylene glycol-bis(2-aminoethylether)-N,N,N`,N`-tetraacetic acid 1.507 6.5918E-02 no Biochemistry Inhibitor Carboxypeptidase B 2.1283 91 Diethylenetriaminepentaacetic acid -0.458 3.2363E-01 no Biochemistry Inhibitor Zn2+-dependent protease 1.7783 91 7,7-Dimethyl-(5Z,8Z)-eicosadienoic acid -3.045 1.1618E-03 yes Lipid Inhibitor PLA2 / Lipoxygenase 4.7938 86 Oleic Acid -0.345 3.6507E-01 no Phosphorylation Activator PKC 0.1829 86 Daidzein -2.931 1.6903E-03 yes Cell Cycle Inhibitor Aldehyde dehydrogenase 7.0109 80 Genistein -0.469 3.1957E-01 no Phosphorylation Inhibitor Tyrosine kinase 1.5385 80 3-Methoxy-morphanin hydrochloride -7.714 6.0807E-15 yes Glutamate Antagonist 8.7457 79 Dextromethorphan hydrobromide monohydrate -0.793 2.1384E-01 no Glutamate Antagonist NMDA 2.8232 79 Dequalinium analog, C-14 linker -11.274 8.8620E-30 yes Phosphorylation Inhibitor PKC-alpha 23.1039 74 Dequalinium dichloride -9.540 7.1408E-22 yes K+ Channel Blocker 22.8573 74 Chloroquine diphosphate 0.499 3.0880E-01 no DNA Intercalator DNA 5.2087 72 Quinacrine dihydrochloride -9.712 1.3360E-22 yes Neurotransmission Inhibitor MAO 16.3953 72 (+)-cis-Dioxolane iodide -3.402 3.3444E-04 yes Cholinergic Agonist Muscarinic 3.6333 70 OXA-22 iodide -1.666 4.7849E-02 no Cholinergic Agonist Muscarinic 3.6333 70 (±)-Butaclamol hydrochloride -5.934 1.4790E-09 yes Dopamine Antagonist D2>D1 23.5780 68 (+)-Butaclamol hydrochloride -11.274 8.8620E-30 yes Dopamine Antagonist 23.5780 68 Sanguinarine chloride -10.008 7.0621E-24 yes Ion Pump Inhibitor Na+/K+ ATPase 29.6105 66 Chelerythrine chloride -10.671 6.9758E-27 yes Phosphorylation Inhibitor PKC 29.3180 66 Chlorambucil 0.081 4.6781E-01 no DNA Intercalator 2.6857 64

! (*!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Melphalan -3.907 4.6645E-05 yes DNA Metabolism Intercalator GCC 0.0225 64 L-765,314 -0.831 2.0305E-01 no Adrenoceptor Antagonist alpha-1B 0.7451 57 Prazosin hydrochloride -2.902 1.8555E-03 yes Adrenoceptor Antagonist alpha1 14.2207 57 L-745,870 hydrochloride -11.274 8.8620E-30 yes Dopamine Antagonist D4 6.9747 44 L-750,667 trihydrochloride -1.184 1.1822E-01 no Dopamine Antagonist D4 6.9180 44 N-Vanillylnonanamide -9.029 8.6770E-20 yes Vanilloid Ligand 9.1987 42 Dihydrocapsaicin -10.003 7.4015E-24 yes Vanilloid Agonist 9.3231 42 Acyclovir -1.215 1.1227E-01 no Immune System Inhibitor Viral DNA synthesis 1.5146 40 Ganciclovir -8.939 1.9604E-19 yes Cell Cycle Inhibitor G2-M checkpoint 5.7853 40 MDL 28170 -1.139 1.2743E-01 no Cell Cycle Inhibitor Calpain I / II 1.2360 38 Z-L-Phe chloromethyl ketone -2.901 1.8594E-03 yes Biochemistry Inhibitor Chymotrypsin A-gamma 6.0322 38 Nimesulide -2.395 8.3067E-03 yes Prostaglandin Inhibitor COX-2 5.3913 36 Niclosamide -2.784 2.6841E-03 yes Antibiotic Protonophore 5.6937 36 Vincristine sulfate -10.792 1.8712E-27 yes Cytoskeleton and ECM Inhibitor Tubulin 54.7161 35 Vinblastine sulfate salt -9.396 2.8241E-21 yes Cytoskeleton and ECM Inhibitor Tubulin 53.7145 35 U-74389G maleate -8.942 1.9194E-19 yes Cell Stress Inhibitor 13.4525 31 U-83836 dihydrochloride 0.954 1.7004E-01 no Cell Stress Inhibitor 3.3297 31 Podophyllotoxin -11.147 3.7067E-29 yes Cytoskeleton and ECM Inhibitor 18.7107 15 Etoposide -11.102 6.1554E-29 yes Apoptosis Inhibitor Topo II 32.8884 15 GR 127935 hydrochloride -1.081 1.3994E-01 no Serotonin Antagonist 5-HT1B/1D 6.6535 11 SB 224289 hydrochloride -4.786 8.5192E-07 yes Serotonin Antagonist 5-HT1B 22.2519 11 MG 624 -10.660 7.8645E-27 yes Cholinergic Antagonist Nicotinic 6.1540 4 N,N,N-trimethyl-1-(4-trans-stilbenoxy)-2-propylammonium -11.127 4.6237E-29 yes Cholinergic Antagonist Nicotinic 7.1135 4 Arecoline hydrobromide -0.111 4.5581E-01 no Cholinergic Agonist 3.3496 2 Arecaidine propargyl ester hydrobromide -9.931 1.5191E-23 yes Cholinergic Agonist M2 5.4180 2 GR 113808 0.992 1.6061E-01 no Serotonin Antagonist 5-HT4 5.3510 135 DSP-4 hydrochloride 1.518 6.4499E-02 no Adrenoceptor Neurotoxin 0.6039 134 Cephradine 0.711 2.3849E-01 no Antibiotic Cell wall synthesis 6.0389 131 Cefsulodin sodium salt hydrate -2.277 1.1404E-02 no Antibiotic Cell wall synthesis 0.2849 129 (R,R)-cis-Diethyl tetrahydro-2,8-chrysenediol -1.053 1.4619E-01 no Hormone Antagonist ER-beta 0.7913 125 Metaphit methanesulfonate -4.967 3.3952E-07 yes Opioid Antagonist sigma 13.5640 124 Betaine hydrochloride -2.668 3.8119E-03 yes Biochemistry Metabolite -2.2762 123 Pentamidine isethionate -11.274 8.8620E-30 yes Glutamate Antagonist NMDA 7.4941 121 Iodoacetamide -8.706 1.5709E-18 yes Biochemistry Inhibitor -0.8463 120 Piribedil maleate -0.334 3.6902E-01 no Dopamine Agonist D3 0.4062 118 Isoliquiritigenin -2.817 2.4232E-03 yes Cyclic Nucleotides Activator Guanylyl cyclase 0.2233 117 SB 204070 hydrochloride -5.886 1.9782E-09 yes Serotonin Antagonist 5-HT4 16.2381 116 Dubinidine -2.437 7.4034E-03 yes Anticonvulsant 11.4546 114 Tamoxifen citrate -10.643 9.3778E-27 yes Phosphorylation Inhibitor PKC 12.2426 112 XK469 -9.355 4.1753E-21 yes Apoptosis Inhibitor TopoII beta 15.1840 110 beta-Lapachone -10.428 9.2210E-26 yes Apoptosis Activator 8.9996 107 m-Iodobenzylguanidine hemisulfate -2.564 5.1721E-03 yes Apoptosis Activator 6.9672 105 L-687,384 hydrochloride -9.156 2.7015E-20 yes Opioid Agonist sigma1 9.5091 102

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Ropinirole hydrochloride -0.255 3.9930E-01 no Dopamine Agonist DRD2 0.9738 101 1-(2-Chlorophenyl)-1-(4-chlorophenyl)-2,2-dichloroethane -2.508 6.0751E-03 yes Hormone Inhibitor Corticosteroid 6.0772 99 Hydrocortisone -3.049 1.1479E-03 yes Hormone Cortisol -4.7854 98 Bepridil hydrochloride -2.655 3.9608E-03 yes Ca2+ Channel Blocker 14.2861 97 GYKI 52466 hydrochloride -1.864 3.1193E-02 no Glutamate Antagonist AMPA/kainate 1.4609 96 Idarubicin -11.274 8.8620E-30 yes DNA Metabolism Inhibitor 17.7420 95 PAPP -10.728 3.7791E-27 yes Serotonin Agonist 5-HT1A 9.3713 94 Emetine dihydrochloride hydrate -10.633 1.0516E-26 yes Apoptosis Activator 22.1870 93 5-Fluorouracil -7.749 4.6212E-15 yes Cell Cycle Inhibitor Thymidylate synthetase 2.4326 92 Ellipticine -8.321 4.3538E-17 yes Cell Cycle Inhibitor CYP1A1 / TopoII 3.0215 90 SB 415286 -2.870 2.0517E-03 yes Phosphorylation Inhibitor GSK-3 13.0577 89 TMB-8 hydrochloride 1.707 4.3949E-02 no Intracellular Calcium Antagonist 0.5698 88 Guanidinoethyl disulfide dihydrobromide -4.521 3.0805E-06 yes Nitric Oxide Inhibitor iNOS 3.2504 87 Triamterene -4.642 1.7211E-06 yes Na+ Channel Blocker 13.7962 85 Calmidazolium chloride -8.582 4.6481E-18 yes Intracellular Calcium Inhibitor Ca2+ATPase 23.6956 82 (S)-(+)-Camptothecin -9.670 2.0264E-22 yes Apoptosis Inhibitor TopoI 14.3223 81 Brefeldin A from Penicillium brefeldianum -11.274 8.8620E-30 yes Cytoskeleton and ECM Inhibitor Golgi apparatus 14.2446 77 Diphenyleneiodonium chloride -10.068 3.8434E-24 yes Nitric Oxide Inhibitor eNOS 7.7864 75 2-Chloro-2-deoxy-D-glucose -2.364 9.0301E-03 yes Biochemistry Analog Glucose 5.2008 73 CGS-12066A maleate -2.575 5.0073E-03 yes Serotonin Agonist 5-HT1B 10.1700 71 (±)-AMT hydrochloride -2.366 8.9866E-03 yes Nitric Oxide Inhibitor iNOS 2.7281 69 Reserpine -1.023 1.5310E-01 no Serotonin Inhibitor Uptake 1.1851 65 Moxisylyte hydrochloride -1.018 1.5439E-01 no Adrenoceptor Antagonist alpha1 0.1163 63 Chlorothiazide -3.165 7.7549E-04 yes Biochemistry Inhibitor Carbonic anhydrase 8.3425 62 Amsacrine hydrochloride -11.274 8.8620E-30 yes DNA Repair Inhibitor TopoII 13.9612 61 CNS-1102 -3.782 7.7642E-05 yes Glutamate Antagonist NMDA 12.3150 60 Colchicine -10.753 2.8583E-27 yes Cytoskeleton and ECM Inhibitor Tubulin 16.7429 56 BTCP hydrochloride 0.836 2.0154E-01 no Dopamine Blocker Reuptake 2.4581 52 Benzamidine hydrochloride 1.509 6.5593E-02 no Biochemistry Inhibitor Peptidase 0.3080 49 Zaprinast -2.506 6.1062E-03 yes Cyclic Nucleotides Inhibitor PDE V 11.2791 47 Carboplatin -6.204 2.7600E-10 yes DNA Intercalator 3.9150 46 2-(alpha-Naphthoyl)ethyltrimethylammonium iodide -6.762 6.7869E-12 yes Cholinergic Inhibitor Choline Acetyltransferase 4.9215 45 SKF 96365 -4.497 3.4502E-06 yes Ca2+ Channel Inhibitor 10.6091 43 Ancitabine hydrochloride -9.951 1.2427E-23 yes DNA Metabolism Inhibitor 10.7570 41 Azathioprine -7.913 1.2594E-15 yes P2 Receptor Inhibitor Purine synthesis 8.9517 39 3`-Azido-3`-deoxythymidine -5.023 2.5440E-07 yes Immune System Inhibitor Reverse transcriptase 10.8585 37 Mevastatin -6.278 1.7126E-10 yes Antibiotic Inhibitor Ras, Rho 17.6321 34 Thapsigargin -9.559 5.9618E-22 yes Intracellular Calcium Releaser 22.8812 30 Taxol -11.274 8.8620E-30 yes Cytoskeleton and ECM Inhibitor Tubulin 28.0486 28 Oxymetazoline hydrochloride -2.720 3.2662E-03 yes Adrenoceptor Agonist alpha2A 2.7662 27 Quinolinic acid -2.419 7.7776E-03 yes Glutamate Antagonist NMDA -1.2408 26 Sobuzoxane -9.600 3.9970E-22 yes Gene Regulation Inhibitor Topo II 14.1486 25 TCPOBOP -2.008 2.2335E-02 no Transcription Agonist CAR 0.2981 24

! )+!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Raloxifene hydrochloride -9.222 1.4569E-20 yes Hormone Modulator ER 17.4412 23 Rotenone -11.274 8.8620E-30 yes Cell Stress Modulator Mitochondria 17.3197 21 6-Methoxy-1,2,3,4-tetrahydro-9H-pyrido[3,4b] indole -2.661 3.8951E-03 yes Neurotransmission Inhibitor MAO 3.6170 19 Metolazone -8.113 2.4793E-16 yes Ion Pump Inhibitor Na+/Cl- transporter 12.9560 18 SKF 86466 0.882 1.8883E-01 no Adrenoceptor Antagonist alpha2 2.7306 17 1-Aminocyclopropanecarboxylic acid hydrochloride -2.401 8.1676E-03 yes Glutamate Agonist NMDA-Glycine -0.7432 12 SB 216763 -2.542 5.5106E-03 yes Phosphorylation Inhibitor GSK-3 11.5423 10 Actinonin -3.501 2.3141E-04 yes Biochemistry Inhibitor Leucine aminopeptidase 10.0832 9 2-Methylthioadenosine diphosphate trisodium -10.359 1.8958E-25 yes P2 Receptor Agonist P2Y 2.5127 7 NS 2028 -2.684 3.6413E-03 yes Cyclic Nucleotides Inhibitor Guanylate cyclase 8.6194 6 Protoporphyrin IX disodium -3.441 2.9003E-04 yes Cyclic Nucleotides Activator Guanylyl cyclase 24.5736 5 Oligomycin A -9.731 1.1127E-22 yes Antibiotic Inhibitor 32.1070 3 Mitoxantrone -10.205 9.3930E-25 yes DNA Metabolism Inhibitor 8.1117 1 MRS 1754 0.927 1.7702E-01 no Adenosine Antagonist A2B -18.0505 2-methoxyestradiol -0.983 1.6273E-01 no Hormone Metabolite Estrogen -11.8404 Cysteamine hydrochloride -0.599 2.7463E-01 no Somatostatin Depleter -2.3036 alpha,beta-Methylene adenosine 5`-triphosphate dilithium -1.290 9.8601E-02 no P2 Receptor Agonist P2X > P2Y -3.6150 O-Methylserotonin hydrochloride -0.172 4.3173E-01 no Serotonin Agonist -13.8908 Se-(methyl)selenocysteine hydrochloride 0.412 3.4032E-01 no Cell Cycle Inhibitor -3.8644 Myricetin -0.911 1.8106E-01 no Phosphorylation Inhibitor Casein Kinase II -13.2584 NG-Monomethyl-L-arginine acetate -1.386 8.2897E-02 no Nitric Oxide Inhibitor NOS -5.3300 MK-912 -0.582 2.8028E-01 no Adrenoceptor Agonist alpha2A -2.8536 (±)-3-(3,4-dihydroxyphenyl)-2-methyl-DL-alanine -1.894 2.9132E-02 no Neurotransmission Inhibitor L-aromatic amino acid dec -6.9387 MRS 2159 -0.267 3.9488E-01 no P2 Receptor Antagonist P2X1 -14.3488 2,6-Difluoro-4-[2-(phenylsulfonylamino)ethylthio]phenoxyacetamide -0.498 3.0915E-01 no Glutamate Agonist AMPA -7.6987 Lorglumide sodium -2.092 1.8225E-02 no Cholecystokinin Antagonist CCK-A -4.4235 LY-278,584 maleate -1.074 1.4143E-01 no Serotonin Antagonist 5-HT3 -5.9712 R(+)-Lisuride hydrogen maleate -1.041 1.4905E-01 no Dopamine Agonist DRD2 -23.9479 L-703,606 oxalate -0.507 3.0612E-01 no Tachykinin Antagonist NK1 -1.7788 Levallorphan tartrate -0.346 3.6450E-01 no Opioid Antagonist -0.6160 S-(-)-Lisuride -0.147 4.4144E-01 no Dopamine Agonist DRD2 -21.7059 Linopirdine -0.553 2.9029E-01 no Cholinergic Releaser -7.9848 L-741,626 -1.009 1.5639E-01 no Dopamine Antagonist DRD2 -11.7936 L-733,060 hydrochloride -1.465 7.1425E-02 no Tachykinin Antagonist NK1 -1.2037 R(-)-Me5 -1.414 7.8639E-02 no Na+ Channel Antagonist -3.4341 (-)-Naproxen sodium 0.602 2.7345E-01 no Prostaglandin Inhibitor COX -7.5011 4-Methylpyrazole hydrochloride 1.042 1.4861E-01 no Biochemistry Inhibitor Alcohol dehydrogenase -5.4206 Nocodazole -0.189 4.2502E-01 no Cytoskeleton and ECM Inhibitor beta-tubulin -8.4905 N-omega-Methyl-5-hydroxytryptamine oxalate salt 0.656 2.5593E-01 no Serotonin Ligand -21.3103 Moxonidine hydrochloride 0.217 4.1407E-01 no Adrenoceptor Agonist alpha2A -9.0361 MRS 1845 -0.006 4.9753E-01 no Ca2+ Channel Inhibitor SOC -8.9172 N-Methyl-1-deoxynojirimycin -1.335 9.0866E-02 no Biochemistry Inhibitor Glucosidase -4.4000 MRS 1523 -0.001 4.9948E-01 no Adenosine Antagonist A3 -6.3328

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Metaproterenol hemisulfate -1.022 1.5346E-01 no Adrenoceptor Agonist beta2 -10.9972 Mianserin hydrochloride 0.975 1.6489E-01 no Serotonin Antagonist -3.8848 8-Methoxymethyl-3-isobutyl-1-methylxanthine -0.340 3.6683E-01 no Cyclic Nucleotides Inhibitor PDE I -11.1680 MK-886 -1.549 6.0662E-02 no Leukotriene Inhibitor -12.4571 Mexiletene hydrochloride -1.259 1.0406E-01 no Na+ Channel Blocker -4.9652 Methylergonovine maleate -0.785 2.1635E-01 no Dopamine Antagonist -21.8872 Molsidomine 1.567 5.8608E-02 no Nitric Oxide Donor -7.5890 3-Methyl-6-(3-[trifluoromethyl]phenyl)-1,2,4-triazolo[4,3-b]pyridazine 0.091 4.6377E-01 no Benzodiazepine Agonist BZ1 -3.2042 Mizoribine 1.642 5.0274E-02 no DNA Metabolism Inhibitor IMP dehydrogenase -5.7895 S-Methylisothiourea hemisulfate -0.211 4.1640E-01 no Nitric Oxide Inhibitor iNOS -1.3890 N-Methyl-D-aspartic acid -1.860 3.1415E-02 no Glutamate Agonist NMDA -3.3399 MJ33 1.364 8.6259E-02 no Lipid Inhibitor PLA2 -10.8297 MRS 2179 -1.246 1.0638E-01 no P2 Receptor Antagonist P2Y1 -10.4540 Meloxicam sodium -1.140 1.2723E-01 no Prostaglandin Inhibitor COX-2 -4.9935 Morin -2.216 1.3352E-02 no Cell Stress Inhibitor Antioxidant -8.1816 Minoxidil -1.189 1.1715E-01 no K+ Channel Activator ATP sensitive -4.7999 Meclofenamic acid sodium 2.233 1.2790E-02 no Prostaglandin Inhibitor COX / 5-Lipoxygenase -6.5636 Milrinone 0.055 4.7805E-01 no Cyclic Nucleotides Inhibitor PDE III -13.5090 (±)-alpha-Methyl-4-carboxyphenylglycine 0.727 2.3362E-01 no Glutamate Antagonist Metabotropic -7.9747 1-Methylhistamine dihydrochloride 0.307 3.7950E-01 no Histamine Metabolite -5.4372 S-Methyl-L-thiocitrulline acetate -1.751 3.9966E-02 no Nitric Oxide Inhibitor NOS -5.2887 Melatonin -0.241 4.0489E-01 no Melatonin Agonist -15.8002 L-Methionine sulfoximine 0.087 4.6545E-01 no Glutamate Inhibitor Glutamine synthase -4.8953 (±)-Metoprolol (+)-tartrate -0.269 3.9400E-01 no Adrenoceptor Antagonist beta1 -15.9164 6-Methyl-2-(phenylethynyl)pyridine hydrochloride 0.094 4.6269E-01 no Glutamate Antagonist mGluR5 -4.7043 Mibefradil dihydrochloride -0.027 4.8926E-01 no Ca2+ Channel Blocker T-type -8.5973 N6-Methyladenosine 0.742 2.2892E-01 no Adenosine Agonist -6.9910 (S)-MAP4 hydrochloride -0.553 2.9029E-01 no Glutamate Antagonist mGluR4,6,7 -4.5665 (±)-Methoxyverapamil hydrochloride -1.407 7.9775E-02 no Ca2+ Channel Antagonist L-type -1.2039 Metrazoline oxalate 0.107 4.5726E-01 no Imidazoline Ligand -8.0272 GW9662 1.475 7.0156E-02 no Transcription Inhibitor PPAR-gamma -6.3233 Sodium Taurocholate -0.760 2.2352E-01 no Multi-Drug Resistance Modulator Conjugate Pathway -13.7869 Amifostine 0.359 3.5974E-01 no Cell Stress Inhibitor Cytoprotectant -5.7141 Acetazolamide 1.111 1.3326E-01 no Biochemistry Inhibitor Carbonic anhydrase -3.7190 A-315456 0.363 3.5836E-01 no Adrenoceptor Antagonist alpha1D -8.5278 GR 4661 0.102 4.5957E-01 no Serotonin Agonist 5-HT1D -11.8529 2-Hydroxysaclofen -1.624 5.2176E-02 no GABA Antagonist GABA-B -3.8336 Nicardipine hydrochloride 1.869 3.0816E-02 no Ca2+ Channel Antagonist L-type -17.3875 Nifedipine -0.294 3.8448E-01 no Ca2+ Channel Antagonist L-type -12.3093 7-Nitroindazole -1.349 8.8727E-02 no Nitric Oxide Inhibitor nNOS -9.4168 6-Nitroso-1,2-benzopyrone 0.011 4.9549E-01 no Transcription Inhibitor PARP -6.2052 Nilutamide 0.119 4.5270E-01 no Hormone Inhibitor Androgen -5.4365 NF 023 1.115 1.3250E-01 no P2 Receptor Antagonist P2X1 -15.4323

! )"!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Nimustine hydrochloride -0.652 2.5707E-01 no DNA Intercalator -6.1040 Norcantharidin -1.251 1.0549E-01 no Phosphorylation Inhibitor PP2A -2.1578 (+)-Nicotine (+)-di-p-toluoyl tartrate -2.233 1.2776E-02 no Cholinergic Agonist Nicotinic -5.5323 Naltrindole hydrochloride -0.314 3.7675E-01 no Opioid Antagonist delta -8.3412 N-(p-Isothiocyanatophenethyl)spiperone hydrochloride -0.415 3.3899E-01 no Dopamine Antagonist DRD2 -3.6574 NO-711 hydrochloride -0.462 3.2213E-01 no GABA Inhibitor Uptake -2.5809 Atropine methyl bromide -0.872 1.9172E-01 no Cholinergic Antagonist Muscarinic -16.1459 Amperozide hydrochloride -0.413 3.3993E-01 no Serotonin Ligand -2.3932 Aminoguanidine hemisulfate -1.219 1.1135E-01 no Nitric Oxide Inhibitor NOS -3.7892 Agmatine sulfate 0.588 2.7823E-01 no Imidazoline Agonist -3.1654 4-Aminobenzamidine dihydrochloride -1.287 9.9097E-02 no Biochemistry Inhibitor Trypsin -1.1605 Mifepristone -1.970 2.4404E-02 no Hormone Antagonist Progesterone -8.2079 L-alpha-Methyl-p-tyrosine -1.800 3.5926E-02 no Neurotransmission Inhibitor Tyrosine hydroxylase -0.1777 Monastrol -1.162 1.2260E-01 no Cell Cycle Inhibitor EgG5 -10.4422 1-Methylimidazole -0.847 1.9852E-01 no Prostaglandin Inhibitor COX -2.1232 Mecamylamine hydrochloride -2.108 1.7531E-02 no Cholinergic Antagonist Nicotinic -6.6672 Methapyrilene hydrochloride 0.628 2.6514E-01 no Histamine Antagonist HRH1 -6.6557 Memantine hydrochloride -0.141 4.4392E-01 no Glutamate Antagonist NMDA -2.7403 Me-3,4-dephostatin 0.583 2.7980E-01 no Phosphorylation Inhibitor PP1B / SHPTP-1 -7.1731 Minocycline hydrochloride 0.442 3.2929E-01 no Cell Cycle Inhibitor -11.6737 Maprotiline hydrochloride -0.361 3.5888E-01 no Adrenoceptor Inhibitor Reuptake -5.6469 H-8 dihydrochloride -1.583 5.6734E-02 no Phosphorylation Inhibitor PKA, PKG -12.8645 Proglumide -1.907 2.8236E-02 no Cholecystokinin Antagonist -0.7173 (±)-Muscarine chloride -1.905 2.8377E-02 no Cholinergic Agonist Muscarinic -4.2822 (+)-MK-801 hydrogen maleate -0.098 4.6095E-01 no Glutamate Antagonist NMDA -7.4893 (-)-MK-801 hydrogen maleate 1.133 1.2862E-01 no Glutamate Antagonist NMDA -7.4893 2-Methyl-5-hydroxytryptamine maleate 0.572 2.8372E-01 no Serotonin Agonist 5-HT3 -11.8280 alpha-Methyl-5-hydroxytryptamine maleate -0.780 2.1782E-01 no Serotonin Agonist 5-HT2 -20.4376 L-alpha-Methyl DOPA -0.897 1.8486E-01 no Biochemistry Inhibitor Aromatic amino acid decar -6.9387 Methysergide maleate -0.020 4.9209E-01 no Serotonin Antagonist -8.8270 Methylcarbamylcholine chloride 1.913 2.7879E-02 no Cholinergic Agonist Nicotinic -8.1124 MDL 26,630 trihydrochloride 0.327 3.7170E-01 no Glutamate Agonist NMDA-Polyamine -4.5631 ZM 39923 hydrochloride 0.746 2.2793E-01 no Phosphorylation Inhibitor JNK-3 -2.6434 3-Morpholinosydnonimine hydrochloride -1.324 9.2801E-02 no Nitric Oxide Donor -5.4749 p-MPPI hydrochloride 0.261 3.9720E-01 no Serotonin Antagonist 5-HT1A -9.5404 MDL 105,519 2.565 5.1579E-03 no Glutamate Antagonist NMDA-Glycine -6.9319 Metrifudil -0.212 4.1594E-01 no Adenosine Agonist A2 -8.6255 p-MPPF dihydrochloride -2.160 1.5404E-02 no Serotonin Antagonist 5-HT1A -9.6044 Niflumic acid -1.958 2.5127E-02 no Prostaglandin Inhibitor COX-2 -5.2950 Nialamide -0.029 4.8829E-01 no Neurotransmission Inhibitor MAO -7.2169 Nomifensine maleate 0.669 2.5166E-01 no Dopamine Inhibitor Reuptake -6.9261 nor-Binaltorphimine dihydrochloride 0.388 3.4898E-01 no Opioid Antagonist kappa -6.3339 Neostigmine bromide 1.228 1.0966E-01 no Cholinergic Inhibitor Acetylcholinesterase -5.7277

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID CR 2249 -1.002 1.5820E-01 no Glutamate Agonist NMDA-Glycine -6.4049 S-(4-Nitrobenzyl)-6-thioinosine -0.427 3.3464E-01 no Adenosine Inhibitor Uptake -10.0990 S-Nitroso-N-acetylpenicillamine -1.645 4.9980E-02 no Nitric Oxide Donor -5.3835 NAN-190 hydrobromide -0.619 2.6809E-01 no Serotonin Antagonist 5-HT1A -8.8603 NCS-356 2.833 2.3079E-03 no GABA Agonist gamma-Hydroxybutyrate -5.1409 S-Nitrosoglutathione 0.698 2.4266E-01 no Nitric Oxide Donor -9.5229 NCS-382 -1.733 4.1521E-02 no GABA Antagonist gamma-Hydroxybutyrate -4.9225 Nalidixic acid sodium -0.314 3.7659E-01 no Antibiotic Inhibitor DNA Gyrase -12.6120 5-Nitro-2-(3-phenylpropylamino)benzoic acid -0.734 2.3139E-01 no Cl- Channel Blocker -7.9444 NF449 octasodium salt 1.851 3.2049E-02 no G protein Antagonist Gs-alpha -13.0359 Nordihydroguaiaretic acid from Larrea divaricata (creosote bush) 2.630 4.2730E-03 no Leukotriene Inhibitor Lipoxygenase -6.6401 (-)-Nicotine hydrogen tartrate salt 0.045 4.8206E-01 no Cholinergic Agonist Nicotinic -3.2819 NG-Nitro-L-arginine 0.139 4.4492E-01 no Nitric Oxide Inhibitor NOS -7.8438 Naphazoline hydrochloride -1.724 4.2333E-02 no Adrenoceptor Agonist alpha -4.9453 3-Nitropropionic acid -0.563 2.8667E-01 no Cell Stress Toxin -4.6291 NG-Nitro-L-arginine methyl ester hydrochloride -1.411 7.9176E-02 no Nitric Oxide Inhibitor NOS -5.3340 (±)-Normetanephrine hydrochloride -1.796 3.6219E-02 no Adrenoceptor Metabolite Norepinephrine -2.4055 Nortriptyline hydrochloride -1.095 1.3669E-01 no Adrenoceptor Inhibitor Uptake -4.0728 NADPH tetrasodium 0.302 3.8128E-01 no Nitric Oxide Cofactor -10.4977 Valproic acid sodium -0.095 4.6218E-01 no Anticonvulsant -4.3774 Praziquantel -2.296 1.0839E-02 no Antibiotic Ca2+ Ionophore -6.6721 Propafenone hydrochloride -0.967 1.6670E-01 no K+ Channel Blocker hKv1.5 -12.0390 5alpha-Pregnan-3alpha-ol-11,20-dione -0.808 2.0960E-01 no GABA Modulator GABA-A -14.4850 Pempidine tartrate -1.719 4.2825E-02 no Cholinergic Antagonist Nicotinic -2.9296 Piracetam -0.376 3.5353E-01 no Glutamate Modulator AMPA -7.7089 Phosphomycin disodium -1.402 8.0513E-02 no Antibiotic Cell wall synthesis -4.3038 Pyrilamine maleate 0.989 1.6127E-01 no Histamine Antagonist HRH1 -6.9749 Piroxicam 0.253 4.0025E-01 no Prostaglandin Inhibitor COX -7.4543 3-n-Propylxanthine -1.094 1.3705E-01 no Adenosine Antagonist A1 > A2 -11.5600 Phenylephrine hydrochloride -1.257 1.0442E-01 no Adrenoceptor Agonist alpha1 -11.1360 Pentylenetetrazole -0.927 1.7686E-01 no Neurotransmission Modulator CNS -3.7535 (+)-Pilocarpine hydrochloride -1.548 6.0832E-02 no Cholinergic Agonist Muscarinic -3.6587 Pilocarpine nitrate -1.213 1.1256E-01 no Cholinergic Agonist Muscarinic -5.4850 Nitrendipine -1.610 5.3738E-02 no Ca2+ Channel Antagonist L-type -18.7148 Nimodipine -0.854 1.9650E-01 no Ca2+ Channel Antagonist L-type -16.0786 Nisoxetine hydrochloride 0.834 2.0226E-01 no Adrenoceptor Blocker Reuptake -11.0438 Nylidrin hydrochloride -1.375 8.4499E-02 no Adrenoceptor Agonist beta -0.0022 N6-Cyclopentyl-9-methyladenine -2.112 1.7347E-02 no Adenosine Antagonist A1 -3.0271 Naltriben methanesulfonate -0.361 3.5920E-01 no Opioid Antagonist delta2 -6.1337 Naftopidil dihydrochloride -0.117 4.5334E-01 no Adrenoceptor Antagonist alpha1 -12.5115 BW 245C 1.036 1.5017E-01 no Prostanoids Agonist DP -6.5950 NS-1619 -0.488 3.1285E-01 no K+ Channel Activator Ca2+ activated -0.7044 NBQX disodium -1.675 4.6988E-02 no Glutamate Antagonist AMPA/kainate -8.4171

! )#!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID (±)-Octopamine hydrochloride -1.567 5.8506E-02 no Adrenoceptor Agonist alpha -2.8789 N-Oleoylethanolamine -0.412 3.4030E-01 no Sphingolipid Inhibitor Ceramidase -2.3589 Oxolinic acid 0.918 1.7933E-01 no Antibiotic Inhibitor DNA Gyrase -2.8871 Olomoucine 1.553 6.0230E-02 no Phosphorylation Inhibitor PK -3.6670 Sodium Oxamate 0.745 2.2817E-01 no Biochemistry Inhibitor Lactate Dehydrogenase -4.3313 Oxybutynin Chloride 0.708 2.3948E-01 no Cholinergic Antagonist Muscarinic -4.4857 Oxiracetam -0.613 2.6992E-01 no Nootropic -5.2602 Ouabain -1.568 5.8446E-02 no Ion Pump Inhibitor Na+/K+ ATPase -5.3647 ODQ -0.592 2.7686E-01 no Cyclic Nucleotides Inhibitor NO-sensitive guanylyl cyc -3.1837 Ofloxacin -0.489 3.1247E-01 no Antibiotic DNA Synthesis -6.7858 Oxotremorine sesquifumarate salt 0.855 1.9623E-01 no Cholinergic Agonist M2 -8.1581 Oxatomide 0.596 2.7568E-01 no Immune System Modulator -1.1963 Oxaprozin 0.066 4.7363E-01 no Prostaglandin Inhibitor -4.2022 Oxotremorine methiodide -0.700 2.4191E-01 no Cholinergic Agonist Muscarinic -8.3627 Progesterone 0.425 3.3535E-01 no Hormone Progesterone -12.7869 Palmitoylethanolamide -0.913 1.8051E-01 no Cannabinoid Agonist CB2 -2.3737 Piceatannol 0.596 2.7567E-01 no Phosphorylation Inhibitor Syk / Lck -10.2033 Parthenolide -0.724 2.3459E-01 no Serotonin Inhibitor -4.0799 Pindolol -0.079 4.6853E-01 no Adrenoceptor Antagonist beta -25.5497 O-Phospho-L-serine 0.547 2.9212E-01 no Glutamate Antagonist NMDA -5.3290 (±)-Propranolol hydrochloride 0.357 3.6065E-01 no Adrenoceptor Antagonist beta -17.1649 Picrotoxin 0.777 2.1855E-01 no GABA Antagonist GABA-C -3.0384 4-Phenyl-3-furoxancarbonitrile 0.931 1.7584E-01 no Nitric Oxide Donor -6.6936 Pentoxifylline -1.637 5.0810E-02 no Cyclic Nucleotides Inhibitor PDE -16.7236 L-Glutamic acid, N-phthaloyl- -0.589 2.7788E-01 no Glutamate Agonist NMDA -4.9887 Pancuronium bromide 0.408 3.4176E-01 no Cholinergic Antagonist -12.6977 3-alpha,21-Dihydroxy-5-alpha-pregnan-20-one 0.444 3.2849E-01 no GABA Modulator GABA-A -15.4033 Pirfenidone -1.482 6.9175E-02 no Immune System Inhibitor -1.4000 1,3-Dimethyl-8-phenylxanthine 0.752 2.2609E-01 no Adenosine Antagonist A1 -19.9006 PPNDS tetrasodium -2.067 1.9369E-02 no P2 Receptor Antagonist P2X1 -17.6537 PD 404,182 0.091 4.6392E-01 no Biochemistry Inhibitor KDO-8-P synthase -4.8559 Papaverine hydrochloride -0.299 3.8240E-01 no Cyclic Nucleotides Inhibitor PDE -1.5223 Pentolinium di[L(+)-tartrate] 0.166 4.3402E-01 no Cholinergic Antagonist Nicotinic -4.8241 1-Phenyl-3-(2-thiazolyl)-2-thiourea 1.208 1.1348E-01 no Dopamine Inhibitor beta-Hydroxylase -7.0804 Thiolactomycin -1.300 9.6869E-02 no Antibiotic Inhibitor Myristate synthesis -4.4790 Cisplatin 1.523 6.3908E-02 no DNA Intercalator -1.2264 Palmitoyl-DL-Carnitine chloride 0.200 4.2092E-01 no Phosphorylation Modulator PKC -2.9027 R(-)-N6-(2-Phenylisopropyl)adenosine -2.023 2.1549E-02 no Adenosine Agonist A1 -5.5885 N-Phenylanthranilic acid -1.610 5.3659E-02 no Cl- Channel Blocker -6.7826 S(-)-p-Bromotetramisole oxalate -1.295 9.7578E-02 no Phosphorylation Inhibitor Alkaline phosphatase -8.7258 5-Aminovaleric acid hydrochloride -1.670 4.7454E-02 no GABA Antagonist GABA-B -2.9370 (±)-Nipecotic acid -0.722 2.3529E-01 no GABA Inhibitor Uptake -4.7970 Azelaic acid 0.106 4.5792E-01 no DNA Metabolism Inhibitor -1.4943

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Tryptamine hydrochloride 1.245 1.0658E-01 no Serotonin Ligand -15.7942 5-Fluoroindole-2-carboxylic acid -0.008 4.9699E-01 no Glutamate Antagonist NMDA-Glycine -8.5717 PD 168,077 maleate -1.516 6.4777E-02 no Dopamine Agonist D4 -9.9614 SU 6656 0.316 3.7583E-01 no Phosphorylation Inhibitor Src family kinase -9.8898 Quercetin dihydrate -2.302 1.0654E-02 no Cyclic Nucleotides Inhibitor PDE -14.2835 Quinidine sulfate -1.930 2.6809E-02 no Na+ Channel Antagonist -0.1268 Quipazine dimaleate -1.444 7.4350E-02 no Serotonin Agonist -10.3024 Quinine sulfate -0.269 3.9413E-01 no K+ Channel Antagonist -0.1268 (+)-Quisqualic acid 0.436 3.3137E-01 no Glutamate Agonist AMPA -5.8739 Quazinone 0.205 4.1894E-01 no Cyclic Nucleotides Inhibitor PDE III -4.2156 (-)-Quinpirole hydrochloride -1.565 5.8745E-02 no Dopamine Agonist D2/D3 -8.0415 Quipazine, N-methyl-, dimaleate -0.325 3.7242E-01 no Serotonin Agonist 5-HT3 -5.2567 Quipazine, 6-nitro-, maleate 0.978 1.6400E-01 no Serotonin Inhibitor Reuptake -14.7627 Quinelorane dihydrochloride -1.567 5.8557E-02 no Dopamine Agonist DRD2 -5.7572 (±)-Quinpirole dihydrochloride -0.451 3.2590E-01 no Dopamine Agonist D2 > D3 -6.9743 Cortexolone -0.609 2.7135E-01 no Hormone Precursor Cortisol -10.9265 Phenelzine sulfate -1.536 6.2233E-02 no Neurotransmission Inhibitor MAO-A/B -4.6157 Phosphonoacetic acid -0.888 1.8714E-01 no DNA Inhibitor DNA Polymerase -3.1115 (-)-Perillic acid -1.321 9.3323E-02 no G protein Inhibitor p21 Ras -5.2521 Pyrazinecarboxamide -1.617 5.2920E-02 no Antibiotic -4.8113 Primidone 0.619 2.6807E-01 no Anticonvulsant -4.0977 Pirenzepine dihydrochloride -0.545 2.9289E-01 no Cholinergic Antagonist M1 -3.8093 Putrescine dihydrochloride -0.769 2.2096E-01 no Glutamate Agonist NMDA-Polyamine -2.3789 Phentolamine mesylate -0.319 3.7483E-01 no Adrenoceptor Antagonist alpha -7.3473 Phloretin 0.916 1.7971E-01 no Ca2+ Channel Blocker L-Type -2.5179 Pargyline hydrochloride -0.279 3.8998E-01 no Neurotransmission Inhibitor MAO-B -2.0187 Phorbol 12-myristate 13-acetate -0.265 3.9551E-01 no Phosphorylation Activator PKC -2.8929 1,3-PBIT dihydrobromide 2.221 1.3159E-02 no Nitric Oxide Inhibitor NOS -4.5252 1,4-PBIT dihydrobromide -1.179 1.1914E-01 no Nitric Oxide Inhibitor NOS -5.7763 Phenylbutazone 0.790 2.1470E-01 no Prostaglandin Substrate Prostaglandin peroxidase -3.3633 Picotamide 1.736 4.1318E-02 no Thromboxane Antagonist TXA2 -1.0308 Tranylcypromine hydrochloride -1.441 7.4837E-02 no Neurotransmission Inhibitor MAO -6.3559 (S)-Propranolol hydrochloride -0.535 2.9645E-01 no Adrenoceptor Blocker beta -17.1649 Ammonium pyrrolidinedithiocarbamate -0.417 3.3843E-01 no Nitric Oxide Modulator NOS -5.9065 (±)-cis-Piperidine-2,3-dicarboxylic acid 0.155 4.3829E-01 no Glutamate Agonist NMDA -7.4174 Protriptyline hydrochloride 0.760 2.2355E-01 no Adrenoceptor Blocker Reuptake -6.9557 6(5H)-Phenanthridinone 0.677 2.4936E-01 no Transcription Inhibitor PARP -1.7136 5alpha-Pregnan-3alpha-ol-20-one 0.934 1.7527E-01 no GABA Modulator GABA-A -15.2536 Propantheline bromide -1.534 6.2542E-02 no Cholinergic Antagonist Muscarinic -4.4811 Piperidine-4-sulphonic acid -0.424 3.3586E-01 no GABA Agonist GABA-A -2.9963 Paromomycin sulfate 0.821 2.0582E-01 no Antibiotic Protein synthesis -4.4506 1,10-Phenanthroline monohydrate -0.886 1.8793E-01 no Biochemistry Inhibitor Metalloprotease -3.3402 Prilocaine hydrochloride 1.509 6.5665E-02 no Na+ Channel Blocker -8.1733

! )$!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Propentofylline 0.156 4.3790E-01 no Adenosine Inhibitor Transporter -15.1399 (S)-(-)-propafenone hydrochloride -0.752 2.2591E-01 no Adrenoceptor Blocker beta -12.0390 Pyridostigmine bromide 0.231 4.0862E-01 no Cholinergic Inhibitor Cholinesterase -1.7195 R(+)-3PPP hydrochloride -2.213 1.3452E-02 no Dopamine Agonist DRD2 -10.0895 S(-)-3PPP hydrochloride -0.421 3.3705E-01 no Dopamine Agonist DRD2 -10.0895 3-Phenylpropargylamine hydrochloride -2.043 2.0514E-02 no Dopamine Inhibitor Dopamine beta-hydroxylase -3.1720 N6-2-Phenylethyladenosine 0.531 2.9763E-01 no Adenosine Agonist A1 -7.1641 N6-Phenyladenosine 0.728 2.3320E-01 no Adenosine Agonist A1 -1.7086 Phaclofen 0.970 1.6600E-01 no GABA Antagonist GABA-B -7.2387 (±)-Pindobind 0.257 3.9841E-01 no Adrenoceptors Ligand beta -24.1169 SKF 94836 0.246 4.0292E-01 no Calcium Signaling Inhibitor PDE III -11.6790 IC 261 -1.630 5.1502E-02 no Phosphorylation Inhibitor CK-1delta/epsilon -7.3987 S(-)-Pindolol -1.122 1.3093E-01 no Serotonin Agonist 5-HT1A -25.5497 Pinacidil -2.117 1.7149E-02 no K+ Channel Activator -11.3913 Pregnenolone sulfate sodium 2.653 3.9869E-03 no GABA Antagonist GABA-A -11.1447 PPADS 1.106 1.3440E-01 no P2 Receptor Antagonist P2 -14.7027 S(+)-PD 128,907 hydrochloride -1.174 1.2027E-01 no Dopamine Agonist D3 -2.1728 Phenylbenzene-omega-phosphono-alpha-amino acid -0.014 4.9432E-01 no Glycine Antagonist -0.7417 Phthalamoyl-L-glutamic acid trisodium 0.251 4.0096E-01 no Glutamate Agonist NMDA -5.5159 PD 98,059 0.903 1.8334E-01 no Phosphorylation Inhibitor MEK2 -3.9645 (±)-PD 128,907 hydrochloride 1.164 1.2221E-01 no Dopamine Agonist D3 -2.1728 S-(4-Nitrobenzyl)-6-thioguanosine 0.102 4.5951E-01 no Adenosine Inhibitor -7.6238 4-Aminopyridine -1.516 6.4813E-02 no K+ Channel Blocker A-type -4.8117 Atropine sulfate 1.244 1.0672E-01 no Cholinergic Antagonist Muscarinic -4.7393 Atropine methyl nitrate 1.292 9.8265E-02 no Cholinergic Antagonist Muscarinic -17.9825 Arcaine sulfate -0.330 3.7075E-01 no Glutamate Antagonist NMDA-Polyamine -1.0159 Sphingosine 1.404 8.0132E-02 no Phosphorylation Inhibitor PKC -1.5940 SB 269970 hydrochloride -1.100 1.3570E-01 no Serotonin Antagonist 5-HT7 -1.6559 Spiperone hydrochloride -0.517 3.0266E-01 no Dopamine Antagonist DRD2 -7.9131 SR 2640 -1.548 6.0825E-02 no Leukotriene Antagonist CysLT1 -9.7171 (-)-Scopolamine,n-Butyl-, bromide 0.649 2.5831E-01 no Cholinergic Antagonist Muscarinic -16.9904 SB 205384 0.797 2.1277E-01 no GABA Modulator GABA-A -8.6705 CV-3988 -0.437 3.3102E-01 no Cytokines & Growth Fa Antagonist PAF -6.2993 Sulindac -0.194 4.2302E-01 no Prostaglandin Inhibitor COX -8.2302 Succinylcholine chloride -1.930 2.6777E-02 no Cholinergic Antagonist Nicotinic -4.6390 Salbutamol -1.678 4.6677E-02 no Adrenoceptor Agonist beta2 -8.6603 Salmeterol -0.490 3.1222E-01 no Adrenoceptor Agonist beta2 -8.8943 SU 5416 -0.662 2.5394E-01 no Phosphorylation Inhibitor VEGFR PTK -10.8964 (-)-Scopolamine methyl bromide -0.847 1.9854E-01 no Cholinergic Antagonist Muscarinic -16.9163 Ruthenium red -0.074 4.7055E-01 no Ion Pump Inhibitor Mitochondrial uniporter -0.9840 Rutaecarpine 0.686 2.4645E-01 no K+ Channel Blocker -7.8454 Resveratrol -0.230 4.0924E-01 no Prostaglandin Inhibitor COX -3.8617 REV 5901 -1.184 1.1813E-01 no Leukotriene Antagonist LTD4 -7.0078

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Rottlerin 0.304 3.8047E-01 no Phosphorylation Inhibitor PKC / CaM Kinase III -6.0230 Ranolazine dihydrochloride -2.104 1.7711E-02 no Lipid Inhibitor pFOX -8.2305 Rolipram -2.091 1.8254E-02 no Cyclic Nucleotides Inhibitor PDE IV -5.1219 Phosphoramidon disodium -0.510 3.0512E-01 no Biochemistry Inhibitor Endopeptidase -20.5364 Roscovitine 0.003 4.9891E-01 no Phosphorylation Inhibitor CDK -0.0626 Ro 8-4304 -2.188 1.4331E-02 no Glutamate Antagonist NMDA-NR2B -3.3769 RX 821002 hydrochloride 0.249 4.0167E-01 no Adrenoceptor Antagonist alpha2 -5.5331 Ribavirin -1.194 1.1620E-01 no Cell Cycle Inhibitor IMP dehydrogenase -5.4344 Ranitidine hydrochloride 0.023 4.9093E-01 no Histamine Antagonist H2 -8.8616 Ritanserin -0.696 2.4332E-01 no Serotonin Antagonist 5-HT2/5-HT1C -7.7433 Rauwolscine hydrochloride -1.747 4.0288E-02 no Adrenoceptor Antagonist alpha2 -2.0332 Ro 16-6491 hydrochloride -2.181 1.4590E-02 no Neurotransmission Inhibitor MAO-B -1.0523 Ro 41-1049 hydrochloride 1.328 9.2137E-02 no Neurotransmission Inhibitor MAO-A -4.7684 Ro 41-0960 -0.360 3.5942E-01 no Neurotransmission Inhibitor COMT -7.6263 Reactive Blue 2 2.003 2.2573E-02 no P2 Receptor Antagonist P2Y -10.1270 Riluzole -1.033 1.5090E-01 no Glutamate Antagonist Release -1.7431 Risperidone 1.694 4.5097E-02 no Dopamine Antagonist DRD2 -10.2931 Rilmenidine hemifumarate 0.526 2.9930E-01 no Imidazoline Agonist I1 -5.7097 Ro 04-6790 dihydrochloride 0.343 3.6587E-01 no Serotonin Antagonist 5-HT6 -5.2757 (±)-Sotalol hydrochloride -0.813 2.0811E-01 no Adrenoceptor Antagonist beta -4.9715 SB-366791 -1.419 7.7915E-02 no Vanilloid Antagonist VR1 -3.0156 Sodium nitroprusside dihydrate 2.404 8.1010E-03 no Nitric Oxide Releaser -1.0698 (±)-Synephrine -0.149 4.4085E-01 no Adrenoceptor Agonist alpha -5.2875 Sulfaphenazole -1.362 8.6671E-02 no Multi-Drug Resistance Inhibitor Cytochrome P4502C -9.1951 Seglitide -0.662 2.5398E-01 no Somatostatin Agonist sst2 -18.2993 Sulindac sulfone 0.592 2.7680E-01 no Prostaglandin Inhibitor -7.3445 Cortexolone maleate -1.434 7.5773E-02 no Dopamine Antagonist DRD2 -14.1896 SR 57227A -1.400 8.0749E-02 no Serotonin Agonist 5-HT3 -7.7427 (-)-Scopolamine hydrobromide -0.742 2.2919E-01 no Cholinergic Antagonist Muscarinic -10.2442 SC-560 1.113 1.3287E-01 no Prostaglandin Inhibitor COX-1 -4.6793 Semicarbazide hydrochloride -1.405 8.0029E-02 no Neurotransmission Inhibitor MAO -5.0341 (-)-Scopolamine methyl nitrate -0.451 3.2606E-01 no Cholinergic Antagonist Muscarinic -18.9174 DL-Stearoylcarnitine chloride 0.638 2.6181E-01 no Phosphorylation Inhibitor PKC -2.3326 Spermidine trihydrochloride 1.259 1.0410E-01 no Glutamate Ligand NMDA-Polyamine -4.9703 SNC80 -0.457 3.2366E-01 no Opioid Agonist delta -4.2921 SKF 83959 hydrobromide -1.357 8.7380E-02 no Dopamine Agonist D1 -10.4451 Spermine tetrahydrochloride 0.394 3.4672E-01 no Glutamate Antagonist NMDA-Polyamine -3.3021 SKF 75670 hydrobromide -0.142 4.4350E-01 no Dopamine Agonist D1 -6.1094 SC 19220 -1.640 5.0479E-02 no Prostaglandin Antagonist EP1 -1.8074 SKF 89626 -0.044 4.8258E-01 no Dopamine Agonist D1 -10.7228 SKF 83565 hydrobromide 0.789 2.1492E-01 no Dopamine Agonist D1 -7.5982 N-Oleoyldopamine 0.463 3.2158E-01 no Neurotransmission Ligand CB1 -9.5690 Spironolactone -0.526 2.9950E-01 no Hormone Antagonist Mineralocorticoid -7.4931

! )%!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID SCH-202676 hydrobromide 0.360 3.5948E-01 no G protein Modulator GPCR -3.7669 D-Serine -0.143 4.4320E-01 no Glutamate Agonist NMDA-Glycine -4.1465 Albuterol hemisulfate -0.949 1.7130E-01 no Adrenoceptor Agonist beta2 -7.1289 N-Succinyl-L-proline 0.725 2.3436E-01 no Neurotransmission Inhibitor ACE -5.6487 Acetamide -1.940 2.6178E-02 no Biochemistry Inhibitor Carbonic anhydrase -2.0047 N-(4-Aminobutyl)-5-chloro-2-naphthalenesulfonamide -0.923 1.7804E-01 no Intracellular Calcium Antagonist Calmodulin -1.6370 L-azetidine-2-carboxylic acid 0.984 1.6263E-01 no Biochemistry Inhibitor Collagen -3.9080 p-Aminoclonidine hydrochloride -0.048 4.8095E-01 no Adrenoceptor Agonist alpha2 -8.3850 3-aminobenzamide -1.034 1.5048E-01 no Apoptosis Inhibitor PARS -4.4936 (±)-Norepinephrine (+)bitartrate 0.099 4.6056E-01 no Adrenoceptor Agonist -10.1650 4-Amino-1,8-naphthalimide 1.875 3.0420E-02 no Apoptosis Inhibitor PARP -1.0499 (±)-alpha-Lipoic Acid 0.342 3.6624E-01 no Cell Stress Coenzyme Pyruvate dehydrogenase -1.9717 DL-Thiorphan -0.312 3.7768E-01 no Neurotransmission Inhibitor Enkephalinase -6.4858 Tulobuterol hydrochloride -1.767 3.8635E-02 no Adrenoceptor Agonist beta -5.3849 Tyrphostin AG 34 -2.319 1.0192E-02 no Phosphorylation Inhibitor Tyrosine kinase -12.9224 Triamcinolone -0.184 4.2701E-01 no Hormone Agonist Glucocorticoid -7.5110 S(-)-Timolol maleate -0.767 2.2140E-01 no Adrenoceptor Antagonist beta -14.8252 Triprolidine hydrochloride -0.406 3.4234E-01 no Histamine Antagonist HRH1 -3.7367 Tyrphostin AG 112 0.333 3.6947E-01 no Phosphorylation Inhibitor Tyrosine kinase -9.5119 Tyrphostin 1 -0.642 2.6031E-01 no Phosphorylation Inhibitor EGFR -9.6061 Tyrphostin 23 -0.840 2.0056E-01 no Phosphorylation Inhibitor EGFR -17.6797 TFPI hydrochloride -1.156 1.2388E-01 no Nitric Oxide Inhibitor nNOS -1.7477 Na-p-Tosyl-L-lysine chloromethyl ketone hydrochloride 0.005 4.9808E-01 no Cyclic Nucleotides Inhibitor Adenylyl cyclase -9.0928 Tyrphostin 25 -0.259 3.9777E-01 no Phosphorylation Inhibitor EGFR -14.8744 1-[2-(Trifluoromethyl)phenyl]imidazole -0.158 4.3738E-01 no Nitric Oxide Inhibitor NOS -6.3332 SU 4312 0.269 3.9390E-01 no Phosphorylation Inhibitor KDR -8.4051 SR 59230A oxalate -1.222 1.1095E-01 no Adrenoceptor Antagonist beta3 -11.1051 SKF 89976A hydrochloride -0.233 4.0797E-01 no GABA Inhibitor GAT-1 -0.1813 SIB 1757 -1.752 3.9859E-02 no Glutamate Antagonist mGluR5 -8.2872 SIB 1893 -0.848 1.9822E-01 no Glutamate Antagonist mGluR5 -4.4486 1-(1-Naphthyl)piperazine hydrochloride 0.230 4.0901E-01 no Serotonin Antagonist 5-HT2 -7.5485 1-(2-Methoxyphenyl)piperazine hydrochloride -0.639 2.6148E-01 no Serotonin Agonist 5-HT1 > 5-HT2 -9.6082 Spiroxatrine 0.122 4.5148E-01 no Serotonin Agonist 5-HT1A -7.0795 SR-95531 -0.957 1.6917E-01 no GABA Antagonist GABA-A -4.7649 (±)-6-Chloro-PB hydrobromide -2.267 1.1691E-02 no Dopamine Agonist D1 -10.5959 SKF 91488 dihydrochloride -0.766 2.2184E-01 no Histamine Inhibitor Histamine N-methyltransfe -4.4216 Suramin hexasodium 2.284 1.1174E-02 no P2 Receptor Antagonist P2X, P2Y -16.9942 SQ 22536 0.562 2.8721E-01 no Cyclic Nucleotides Inhibitor Adenylyl cyclase -4.4435 Sepiapterin 1.795 3.6295E-02 no Nitric Oxide Cofactor NOS -8.8178 R(-)-SCH-12679 maleate 1.609 5.3786E-02 no Dopamine Antagonist D1 -1.3407 (±)-SKF 38393, N-allyl-, hydrobromide -1.590 5.5966E-02 no Dopamine Agonist D1 -2.9637 SB 206553 hydrochloride -1.365 8.6107E-02 no Serotonin Antagonist 5-HT2C/5-HT2B -9.5886 L-Tryptophan 0.224 4.1134E-01 no Serotonin Precursor -17.6188

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Tranilast 0.174 4.3108E-01 no Leukotriene Inhibitor LTC4 -6.2394 Taurine 0.533 2.9710E-01 no Glycine Agonist -2.4249 Tolbutamide -0.303 3.8087E-01 no Hormone Releaser Insulin -13.2463 Tetraethylthiuram disulfide 2.347 9.4627E-03 no Biochemistry Inhibitor Alcohol Dehydrogenase -2.7989 Tetraisopropyl pyrophosphoramide -1.114 1.3262E-01 no Biochemistry Inhibitor Butyrylcholinesterase -3.4763 Tetramisole hydrochloride 2.649 4.0391E-03 no Phosphorylation Inhibitor Phosphatase -9.6745 Trihexyphenidyl hydrochloride 1.215 1.1219E-01 no Cholinergic Antagonist Muscarinic -3.5988 Theophylline 0.370 3.5573E-01 no Adenosine Antagonist A1 > A2 -11.3611 (E)-4-amino-2-butenoic acid -2.149 1.5824E-02 no GABA Agonist GABA-C -5.0086 Tetradecylthioacetic acid -0.324 3.7306E-01 no Transcription Agonist PPAR-alpha -5.5493 Trequinsin hydrochloride -0.971 1.6581E-01 no Cyclic Nucleotides Inhibitor PDE III -4.8392 Tyrphostin AG 879 -1.607 5.4042E-02 no Phosphorylation Inhibitor TrkA -11.2234 Tetraethylammonium chloride 0.123 4.5107E-01 no Cholinergic Antagonist Nicotinic -0.9450 Tolazamide 1.431 7.6219E-02 no Hormone Releaser Insulin -13.2030 Terbutaline hemisulfate 0.641 2.6079E-01 no Adrenoceptor Agonist beta -8.9158 4-Hydroxyphenethylamine hydrochloride 1.030 1.5154E-01 no Dopamine Agonist -2.6218 Trimipramine maleate 0.568 2.8512E-01 no Serotonin Inhibitor Reuptake -5.4862 Tyrphostin AG 490 1.802 3.5809E-02 no Phosphorylation Inhibitor JAK2 -17.3461 TTNPB -2.158 1.5472E-02 no Transcription Ligand RAR-alpha, beta, gamma -6.8131 Tetrahydrozoline hydrochloride -0.469 3.1944E-01 no Adrenoceptor Agonist alpha -6.7607 Tyrphostin AG 494 -2.275 1.1443E-02 no Phosphorylation Inhibitor EGFR -18.5744 N-p-Tosyl-L-phenylalanine chloromethyl ketone 0.337 3.6819E-01 no Biochemistry Inhibitor Chymotrypsin alpha -6.2244 (6R)-5,6,7,8-Tetrahydro-L-biopterin hydrochloride 0.976 1.6460E-01 no Neurotransmission Cofactor Tyrosine -10.0438 Tyrphostin AG 527 -1.607 5.3975E-02 no Phosphorylation Inhibitor EGFR -20.3163 Theobromine -1.392 8.1993E-02 no Adenosine Antagonist A1 > A2 -8.7094 (±)-Taxifolin 1.321 9.3301E-02 no Cell Stress Inhibitor Antioxidant -12.8430 Tyrphostin AG 528 -0.148 4.4112E-01 no Phosphorylation Inhibitor EGFR -13.4954 Terazosin hydrochloride -0.057 4.7719E-01 no Adrenoceptor Antagonist alpha1 -1.6037 Tyrphostin AG 537 -1.249 1.0589E-01 no Phosphorylation Inhibitor EGFR -20.6782 Tyrphostin AG 555 -0.681 2.4787E-01 no Phosphorylation Inhibitor EGFR -20.1332 Tyrphostin AG 698 1.510 6.5504E-02 no Phosphorylation Inhibitor EGFR -21.0985 Tyrphostin AG 808 -1.315 9.4172E-02 no Phosphorylation Inhibitor Tyrosine kinase -25.9684 Thio-NADP sodium 0.366 3.5715E-01 no Intracellular Calcium Blocker NAADP-induced -12.7713 Tyrphostin AG 835 2.230 1.2874E-02 no Phosphorylation Inhibitor Tyrosine kinase -20.3163 Amantadine hydrochloride -0.653 2.5684E-01 no Dopamine Releaser -3.1592 Aminophylline ethylenediamine -1.537 6.2146E-02 no Adenosine Antagonist A1/A2 -16.7085 S-(p-Azidophenacyl)glutathione -0.003 4.9890E-01 no Multi-Drug Resistance Modulator Glutathione S-transferase -10.4568 N-Acetyl-5-hydroxytryptamine -1.060 1.4447E-01 no Melatonin Precursor -21.5479 Aurintricarboxylic acid 0.571 2.8403E-01 no Apoptosis Inhibitor TopoII -7.1275 (±)-2-Amino-4-phosphonobutyric acid 1.893 2.9164E-02 no Glutamate Antagonist NMDA -4.4713 N-arachidonylglycine 0.387 3.4935E-01 no Cannabinoid Inhibitor FAAH -4.2508 WIN 62,577 2.351 9.3727E-03 no Tachykinin Antagonist NK1 -17.1837 S(-)-Willardiine 1.251 1.0541E-01 no Glutamate Agonist AMPA/kainate -7.1410

! )&!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID WAY-100635 maleate -0.497 3.0976E-01 no Serotonin Antagonist 5-HT1A -14.1470 S-5-Iodowillardiine -0.355 3.6125E-01 no Glutamate Agonist AMPA -6.9856 Xylazine hydrochloride 0.375 3.5365E-01 no Adrenoceptor Agonist alpha2 -8.1210 Xamoterol hemifumarate -0.124 4.5053E-01 no Adrenoceptor Agonist beta1 -15.3573 Xylometazoline hydrochloride 0.296 3.8361E-01 no Adrenoceptor Agonist alpha -1.2434 Xanthine amine congener 0.001 4.9954E-01 no Adenosine Antagonist A1 -24.1320 Yohimbine hydrochloride -1.656 4.8820E-02 no Adrenoceptor Antagonist alpha2 -2.0332 YC-1 0.433 3.3233E-01 no Cyclic Nucleotides Activator Guanylyl cyclase -10.1316 Zonisamide sodium -0.917 1.7959E-01 no Anticonvulsant -8.1550 Zardaverine -0.341 3.6646E-01 no Cyclic Nucleotides Inhibitor PDE III/ PDE IV -5.7003 Zimelidine dihydrochloride -0.649 2.5817E-01 no Serotonin Inhibitor Reuptake -4.1985 Tetracaine hydrochloride -0.020 4.9221E-01 no Na+ Channel Modulator -1.2589 Tyrphostin 47 0.178 4.2956E-01 no Phosphorylation Inhibitor EGFR -19.4361 Tyrphostin 51 -0.918 1.7939E-01 no Phosphorylation Inhibitor EGFR -10.5293 T-1032 -2.020 2.1672E-02 no Cyclic Nucleotides Inhibitor PDE V -5.7604 I-OMe-Tyrphostin AG 538 -0.737 2.3059E-01 no Phosphorylation Inhibitor IGF-1 RTK -15.0767 Tyrphostin AG 538 -0.395 3.4624E-01 no Phosphorylation Inhibitor IGF-1 RTK -19.5274 Trimethoprim -1.230 1.0940E-01 no Antibiotic Inhibitor Dihydrofolate reductase -2.7962 Tomoxetine 0.167 4.3359E-01 no Adrenoceptor Inhibitor Reuptake -10.8802 T-0156 0.077 4.6928E-01 no Cyclic Nucleotides Inhibitor PDE V -10.0449 D-609 potassium -0.896 1.8503E-01 no Lipid Inhibitor PIPLC -6.4571 Tyrphostin AG 126 -0.410 3.4082E-01 no Phosphorylation Inhibitor TNFalpha -16.9960 Terfenadine 0.755 2.2503E-01 no Histamine Antagonist HRH1 -3.2974 Tropicamide 0.823 2.0533E-01 no Cholinergic Antagonist M4 -15.2981 THIP hydrochloride 0.361 3.5894E-01 no GABA Agonist GABA-A -6.8890 Trifluperidol hydrochloride -1.621 5.2465E-02 no Dopamine Antagonist D1/D2 -5.7072 3-Tropanyl-indole-3-carboxylate hydrochloride -1.528 6.3309E-02 no Serotonin Antagonist 5-HT3 -5.4495 Tracazolate -0.315 3.7633E-01 no GABA Modulator -10.9470 3-Tropanylindole-3-carboxylate methiodide -0.344 3.6556E-01 no Serotonin Antagonist 5-HT3 -15.8039 Telenzepine dihydrochloride 0.381 3.5148E-01 no Cholinergic Antagonist M1 -3.9471 Thioperamide maleate -0.154 4.3894E-01 no Histamine Antagonist H3 -18.2620 (±)-Thalidomide 0.521 3.0117E-01 no Cytoskeleton and ECM Inhibitor TNFalpha -5.6805 R(+)-Terguride -2.004 2.2512E-02 no Dopamine Agonist -15.7008 Thiocitrulline -0.147 4.4139E-01 no Nitric Oxide Inhibitor nNOS, eNOS -7.7510 Tyrphostin A9 -1.746 4.0363E-02 no Phosphorylation Inhibitor PDGFR -9.9304 TPMPA 0.874 1.9101E-01 no GABA Antagonist GABA-C -5.8605 U-75302 -0.172 4.3163E-01 no Leukotriene Agonist BLT1 -8.3563 Uridine 5`-diphosphate sodium 0.707 2.3966E-01 no P2 Receptor Agonist P2Y -7.6702 U-73122 -1.194 1.1617E-01 no Lipid Inhibitor PLC, A2 -15.5197 SKF 95282 dimaleate 0.424 3.3576E-01 no Histamine Antagonist H2 -2.9694 4-Imidazoleacrylic acid -0.264 3.9608E-01 no Histamine Inhibitor Histidine ammonia-lyase/ -12.7023 Urapidil hydrochloride 0.157 4.3754E-01 no Adrenoceptor Antagonist alpha1 -8.1990 Urapidil, 5-Methyl- -0.783 2.1677E-01 no Adrenoceptor Antagonist alpha1A -7.7682

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID U-69593 -0.728 2.3336E-01 no Opioid Agonist kappa -1.0973 UK 14,304 -0.210 4.1672E-01 no Adrenoceptor Agonist alpha2 -7.3309 U-101958 maleate -2.065 1.9472E-02 no Dopamine Antagonist D4 -4.7843 U0126 0.711 2.3855E-01 no Phosphorylation Inhibitor MEK1/MEK2 -7.7131 (±)-Verapamil hydrochloride 1.008 1.5666E-01 no Ca2+ Channel Modulator L-type -1.4780 VUF 5574 0.581 2.8057E-01 no Adenosine Antagonist A3 -7.2944 Vinpocetine 0.433 3.3256E-01 no Cyclic Nucleotides Inhibitor PDE I -0.0438 Vancomycin hydrochloride from Streptomyces orientalis -0.064 4.7468E-01 no Antibiotic Cell wall synthesis -0.4820 (±)-gamma-Vinyl GABA -0.701 2.4166E-01 no GABA Inhibitor Transaminase -1.7994 (±)-Vesamicol hydrochloride -1.967 2.4563E-02 no Cholinergic Inhibitor ACh storage -4.9336 Wortmannin from Penicillium funiculosum 0.408 3.4169E-01 no Phosphorylation Inhibitor PI3K -8.7892 1400W dihydrochloride -0.275 3.9170E-01 no Nitric Oxide Inhibitor iNOS -2.6662 WB 64 -1.683 4.6151E-02 no Cholinergic Ligand M2 -4.4336 ( R)-(+)-WIN 55,212-2 mesylate -0.681 2.4803E-01 no Cannabinoid Agonist -1.2169 GABA -0.439 3.3035E-01 no GABA Agonist -2.7784 Acetyl-beta-methylcholine chloride -0.338 3.6762E-01 no Cholinergic Agonist M1 -1.7344 5-azacytidine -1.398 8.1104E-02 no DNA Metabolism Inhibitor DNA methyltransferase -1.7173 5-(N-Ethyl-N-isopropyl)amiloride -0.289 3.8646E-01 no Ion Pump Blocker Na+/H+ Antiporter -11.6381 3-Aminopropionitrile fumarate 1.234 1.0853E-01 no Multi-Drug Resistance Substrate CYP450 -9.0810 Apigenin 1.824 3.4069E-02 no Cell Cycle Inhibitor -7.0447 Gabaculine hydrochloride -2.256 1.2023E-02 no GABA Inhibitor GABA transaminase -5.5637 AA-861 1.517 6.4696E-02 no Leukotriene Inhibitor 5-lipoxygenase -3.1329 9-Amino-1,2,3,4-tetrahydroacridine hydrochloride -1.021 1.5361E-01 no Cholinergic Inhibitor Cholinesterase -4.2209 AL-8810 -1.806 3.5496E-02 no Prostaglandin Antagonist FP Receptor -1.0281 1-Aminobenzotriazole 0.524 3.0028E-01 no Multi-Drug Resistance Inhibitor CYP450, chloroperoxidase -3.6565 O-(Carboxymethyl)hydroxylamine hemihydrochloride -2.058 1.9775E-02 no Biochemistry Inhibitor Aminotransferase -3.5668 5-(N,N-Dimethyl)amiloride hydrochloride 2.396 8.2873E-03 no Ion Pump Blocker Na+/H+ Antiporter -10.8889 Amiprilose hydrochloride -0.778 2.1821E-01 no Immune System Modulator -3.5961 Sandoz 58-035 0.332 3.6977E-01 no Lipid Inhibitor ACAT -7.3180 (±)-2-Amino-3-phosphonopropionic acid 1.250 1.0569E-01 no Glutamate Antagonist NMDA -3.8618 L-Arginine 0.559 2.8802E-01 no Nitric Oxide Precursor -3.9612 (±)-2-Amino-7-phosphonoheptanoic acid -1.020 1.5391E-01 no Glutamate Antagonist NMDA -4.2353 (±)-2-Amino-5-phosphonopentanoic acid -1.045 1.4797E-01 no Glutamate Antagonist NMDA -4.3519 L-732,138 0.503 3.0758E-01 no Tachykinin Antagonist NK1 > NK2, NK3 -13.3105 Acetylsalicylic acid -0.481 3.1520E-01 no Prostaglandin Inhibitor COX-3 > COX-1 > COX-2 -4.6918 5-(N-Methyl-N-isobutyl)amiloride 1.871 3.0654E-02 no Ion Pump Blocker Na+/H+ Antiporter -11.0245 Acetylthiocholine chloride 0.004 4.9832E-01 no Cholinergic Agonist Nicotinic -4.7778 4-Androsten-4-ol-3,17-dione 0.194 4.2293E-01 no Hormone Inhibitor Aromatase -14.8147 2-(2-Aminoethyl)isothiourea dihydrobromide -1.251 1.0541E-01 no Nitric Oxide Inhibitor NOS -2.9427 cis-Azetidine-2,4-dicarboxylic acid -2.251 1.2198E-02 no Glutamate Modulator NMDA -3.5649 trans-Azetidine-2,4-dicarboxylic acid -0.349 3.6350E-01 no Glutamate Agonist mGluR1, mGluR5 -3.5649 AGN 192403 hydrochloride 0.703 2.4098E-01 no Imidazoline Ligand I1 -4.7455 AIDA 0.154 4.3880E-01 no Glutamate Antagonist mGluR1 -7.5362

! )'!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID A-77636 hydrochloride 1.440 7.4902E-02 no Dopamine Agonist D1 -5.7649 ATPA 0.283 3.8853E-01 no Glutamate Agonist Kainate -10.2981 ARL 67156 trisodium salt -0.168 4.3329E-01 no P2 Receptor Inhibitor ecto-ATPase -9.5979 Beclomethasone 0.267 3.9475E-01 no Hormone Glucocorticoid -9.5218 2,3-Butanedione monoxime -1.810 3.5125E-02 no K+ Channel Blocker ATP-sensitive -3.0410 SB 222200 -1.695 4.5077E-02 no Tachykinin Antagonist NK3 -0.3519 1-benzoyl-5-methoxy-2-methylindole-3-acetic acid -0.955 1.6978E-01 no Multi-Drug Resistance Inhibitor MRP1 -1.2294 p-Benzoquinone -1.571 5.8049E-02 no DNA Repair Inhibitor G:C site -1.6650 8-Bromo-cGMP sodium -1.005 1.5751E-01 no Cyclic Nucleotides Activator -12.4831 Bromoenol lactone 0.248 4.0190E-01 no Lipid Inhibitor PLA2 -2.8194 Benzamide 0.550 2.9122E-01 no Apoptosis Inhibitor PARS -2.6316 N-Acetyl-L-Cysteine -1.621 5.2499E-02 no Glutamate Antagonist -4.6646 N-Acetyltryptamine 1.020 1.5384E-01 no Melatonin Agonist - An -17.6160 (±)-Atenolol -0.658 2.5518E-01 no Adrenoceptor Antagonist beta1 -14.7808 5alpha-Androstane-3alpha,17beta-diol -0.381 3.5156E-01 no Hormone Metabolite Androgen -15.2440 L-allylglycine -1.155 1.2402E-01 no Biochemistry Inhibitor -3.9370 H-9 dihydrochloride -0.190 4.2469E-01 no Phosphorylation Inhibitor cAMP- and cGMP-dependent -10.7677 6-Aminohexanoic acid -2.179 1.4649E-02 no Immune System Inhibitor Blood Clotting -3.2682 ATPO -1.680 4.6478E-02 no Glutamate Antagonist GluR1-4 -9.5467 Allopurinol 0.147 4.4145E-01 no Cell Stress Inhibitor Xanthine oxidase -6.5941 Amiodarone hydrochloride 0.929 1.7639E-01 no Adrenoceptor Agonist alpha/beta -0.5073 4-(2-Aminoethyl)benzenesulfonyl fluoride hydrochloride 0.363 3.5814E-01 no Biochemistry Inhibitor Serine Protease -6.0757 Alprenolol hydrochloride -2.143 1.6057E-02 no Adrenoceptor Antagonist beta -15.2992 Altretamine -1.322 9.3085E-02 no DNA Metabolism Inhibitor -1.0683 N-Acetyldopamine monohydrate 0.467 3.2034E-01 no Dopamine Precursor -11.2009 Aminoguanidine hydrochloride -0.001 4.9971E-01 no Nitric Oxide Inhibitor NOS -2.9994 BW 284c51 -0.402 3.4396E-01 no Cholinergic Inhibitor Acetylcholinesterase -4.8061 Adenosine -0.209 4.1740E-01 no Adenosine Agonist -2.9711 L-Aspartic acid 0.163 4.3519E-01 no Glutamate Agonist -3.3129 N-(4-Amino-2-chlorophenyl)phthalimide -0.602 2.7356E-01 no Anticonvulsant -4.4471 Adenosine 3`,5`-cyclic monophosphate -1.188 1.1734E-01 no Phosphorylation Activator PKA -7.9942 L(-)-Norepinephrine bitartrate -1.418 7.8127E-02 no Adrenoceptor Agonist alpha, beta1 -10.1650 5-(N,N-hexamethylene)amiloride 0.039 4.8462E-01 no Ion Pump Inhibitor Na+/H+ Antiporter -13.0169 4-Androstene-3,17-dione -0.692 2.4456E-01 no Hormone Precursor Androgen -13.5513 (±)-p-Aminoglutethimide -0.444 3.2849E-01 no Biochemistry Inhibitor P450-dependendent hydroxy -3.6754 (±)-HA-966 1.157 1.2368E-01 no Glutamate Antagonist NMDA-glycine -4.5024 Androsterone 1.947 2.5753E-02 no Hormone Androgen -17.5390 Antozoline hydrochloride 0.485 3.1371E-01 no Imidazoline Agonist -4.1995 Aniracetam -1.355 8.7788E-02 no Glutamate Agonist AMPA -6.1620 1,3-Diethyl-8-phenylxanthine 2.005 2.2458E-02 no Adenosine Antagonist A1 -20.0593 8-(p-Sulfophenyl)theophylline 1.047 1.4759E-01 no Adenosine Antagonist A1 > A2 -19.8946 1,3-Dipropyl-8-p-sulfophenylxanthine 4.743 1.0546E-06 no Adenosine Antagonist A1 > A2 -22.1456 2-Methylthioadenosine triphosphate tetrasodium -0.776 2.1877E-01 no P2 Receptor Agonist P2Y -0.6760

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Adenosine amine congener -0.334 3.6901E-01 no Adenosine Agonist A1 -4.4147 Amoxapine -1.041 1.4885E-01 no Adrenoceptor Inhibitor Uptake -6.0297 R(+)-Atenolol 3.764 8.3501E-05 no Adrenoceptor Antagonist beta1 -14.7808 S(-)-Atenolol 0.617 2.6861E-01 no Adrenoceptor Antagonist beta1 -14.7808 1-Allyl-3,7-dimethyl-8-p-sulfophenylxanthine -2.178 1.4691E-02 no Adenosine Antagonist A2 -13.2461 trans-(±)-ACPD 0.315 3.7642E-01 no Glutamate Agonist Metabotropic -2.0313 1-Amino-1-cyclohexanecarboxylic acid hydrochloride -0.749 2.2678E-01 no Neurotransmission Substrate -1.3026 Alaproclate hydrochloride -0.172 4.3153E-01 no Serotonin Inhibitor Reuptake -1.0468 Rp-cAMPS triethylamine 1.186 1.1788E-01 no Phosphorylation Inhibitor PKA -12.7973 SB 200646 hydrochloride 2.437 7.3995E-03 no Serotonin Antagonist 5-HT2C/2B -11.7331 D(-)-2-Amino-7-phosphonoheptanoic acid 0.985 1.6220E-01 no Glutamate Antagonist NMDA -4.2353 Acetohexamide -2.240 1.2547E-02 no Hormone Releaser Insulin -12.1480 SKF 97541 hydrochloride -0.736 2.3101E-01 no GABA Agonist GABA-B -3.2485 cis-4-Aminocrotonic acid -1.088 1.3840E-01 no GABA Agonist GABA-C -5.0086 N6-2-(4-Aminophenyl)ethyladenosine -1.323 9.2944E-02 no Adenosine Agonist A3 -7.3150 Agroclavine -0.911 1.8116E-01 no Dopamine Agonist -10.9663 gamma-Acetylinic GABA 1.285 9.9459E-02 no GABA Inhibitor GABA transaminase -2.0503 AB-MECA 0.760 2.2351E-01 no Adenosine Agonist A3 -6.1580 Alloxazine -0.442 3.2909E-01 no Adenosine Antagonist A2b -5.0147 CGP-7930 -0.411 3.4052E-01 no GABA Modulator GABA-B -2.6698 CGP-13501 -1.096 1.3653E-01 no GABA Modulator GABA-B -2.8414 CP55940 1.396 8.1367E-02 no Cannabinoid Agonist -6.5601 L-Cycloserine -0.155 4.3834E-01 no Sphingolipid Inhibitor Ketosphinganine synthetas -5.7186 (+)-Catechin Hydrate 0.511 3.0484E-01 no Cell Stress Inhibitor Antioxidant -11.1574 Chlorpropamide -0.190 4.2463E-01 no Hormone Releaser Insulin -9.0400 1-(4-Chlorobenzyl)-5-methoxy-2-methylindole-3-acetic acid -1.584 5.6573E-02 no Multi-Drug Resistance Inhibitor MRP1 -9.0898 Choline bromide -0.730 2.3268E-01 no Cholinergic Substrate Choline acetyltransferase -2.6283 Ceramide -0.827 2.0421E-01 no Phosphorylation Inhibitor Diacylglycerol kinase -0.7919 CB 1954 -1.070 1.4227E-01 no DNA Intercalator -7.7981 Carcinine dihydrochloride -1.023 1.5311E-01 no Cell Stress Inhibitor Antioxidant -10.2686 Corticosterone -2.285 1.1148E-02 no Hormone Glucocorticoid -8.9885 Cortisone -0.581 2.8073E-01 no Hormone Corticosteroid -8.5526 3-Bromo-7-nitroindazole -1.680 4.6484E-02 no Nitric Oxide Inhibitor NOS -9.6580 (+)-Bromocriptine methanesulfonate 0.028 4.8886E-01 no Dopamine Agonist DRD2 -21.8304 O6-benzylguanine -1.177 1.1967E-01 no DNA Repair Inhibitor -6.8215 N-Bromoacetamide -1.362 8.6669E-02 no Na+ Channel Modulator -1.7090 Benzamil hydrochloride -1.054 1.4591E-01 no Ion Pump Blocker Na+/H+, Na+/Ca2+ Pump -5.3837 L-Buthionine-sulfoximine 0.513 3.0383E-01 no Multi-Drug Resistance Inhibitor -5.7085 DL-Buthionine-[S,R]-sulfoximine -1.022 1.5341E-01 no Multi-Drug Resistance Inhibitor -5.7085 Bumetanide -2.182 1.4545E-02 no Ion Pump Inhibitor Na+-K+-2Cl- cotransporter -5.3951 Betaine aldehyde chloride -0.070 4.7207E-01 no Cholinergic Metabolite Choline dehydrogenase -2.4729 Benazoline oxalate -0.335 3.6890E-01 no Imidazoline Agonist I2 -5.9911 BWB70C -0.893 1.8586E-01 no Leukotriene Inhibitor 5-lipoxygenase -2.7024

! )(!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID 5-Bromo-2`-deoxyuridine 2.580 4.9400E-03 no DNA Metabolism Inhibitor -3.6110 (±)-Baclofen -1.069 1.4250E-01 no GABA Agonist GABA-B -7.7439 SB 202190 -1.124 1.3042E-01 no Phosphorylation Inhibitor p38 MAPK -8.6028 Bay 11-7085 2.180 1.4646E-02 no Cell Cycle Inhibitor IkB-alpha -8.0406 Betaxolol hydrochloride 0.984 1.6260E-01 no Adrenoceptor Antagonist beta1 -14.7476 Betamethasone 0.881 1.8913E-01 no Hormone Glucocorticoid -9.5218 Buspirone hydrochloride -1.775 3.7910E-02 no Serotonin Agonist 5-HT1A -8.0068 Benserazide hydrochloride 0.007 4.9702E-01 no Biochemistry Inhibitor Decarboxylase -7.5822 Budesonide -1.994 2.3069E-02 no Hormone Cortisol -4.0628 8-Bromo-cAMP sodium 2.309 1.0469E-02 no Cyclic Nucleotides Activator -13.0857 Ro 20-1724 0.587 2.7848E-01 no Cyclic Nucleotides Inhibitor cAMP phosphodiesterase -4.4839 Bestatin hydrochloride 0.643 2.6004E-01 no Biochemistry Inhibitor Aminopeptidase -6.6215 Bretylium tosylate -0.002 4.9932E-01 no Adrenoceptor Blocker -8.8373 BP 897 -1.715 4.3173E-02 no Dopamine Agonist D3 -6.2599 (E)-5-(2-Bromovinyl)-2`-deoxyuridine -2.123 1.6863E-02 no Immune System Inhibitor HSV1 -2.5745 Chloroethylclonidine dihydrochloride 0.861 1.9473E-01 no Adrenoceptor Antagonist alpha1B -7.8626 6-Fluoronorepinephrine hydrochloride -1.168 1.2140E-01 no Adrenoceptor Agonist alpha -6.6614 Bromoacetyl alprenolol menthane 1.442 7.4623E-02 no Adrenoceptor Antagonist beta -14.5755 Benoxathian hydrochloride 0.265 3.9540E-01 no Adrenoceptor Antagonist alpha1 -7.3187 Phenoxybenzamine hydrochloride -1.512 6.5240E-02 no Adrenoceptor Blocker alpha -2.5903 Bupropion hydrochloride 0.267 3.9469E-01 no Dopamine Blocker Reuptake -3.1372 (±)-Bay K 8644 0.853 1.9691E-01 no Ca2+ Channel Agonist L-type -9.8317 Bromoacetylcholine bromide -0.355 3.6139E-01 no Cholinergic Ligand -4.7430 BMY 7378 dihydrochloride -0.659 2.5500E-01 no Serotonin Agonist 5-HT1A -8.6878 R(+)-6-Bromo-APB hydrobromide 1.532 6.2732E-02 no Dopamine Agonist D1/D5 -6.4948 N6-Benzyl-5`-N-ethylcarboxamidoadenosine -0.798 2.1254E-01 no Adenosine Agonist A3 -4.0973 BU224 hydrochloride 0.404 3.4294E-01 no Imidazoline Antagonist I2 -6.6354 B-HT 933 dihydrochloride -1.280 1.0022E-01 no Adrenoceptor Agonist alpha2 -0.5572 BRL 37344 sodium -1.725 4.2299E-02 no Adrenoceptor Agonist beta3 -6.1686 BRL 54443 maleate 0.338 3.6775E-01 no Serotonin Agonist 5-HT1E/1F -17.4858 BW 723C86 0.725 2.3431E-01 no Serotonin Agonist 5-HT2B -17.1873 Citicoline sodium -0.133 4.4704E-01 no Lipid Inhibitor PLA2 -1.5676 Ciprofibrate -1.326 9.2430E-02 no Transcription Ligand PPAR-alpha -7.0194 6-Chloromelatonin -1.659 4.8549E-02 no Melatonin Agonist -16.1502 Carmustine 2.191 1.4231E-02 no DNA Intercalator -5.6758 PK 11195 -0.468 3.2005E-01 no GABA Antagonist Benzodiazepine -5.6993 Caffeic Acid 0.320 3.7452E-01 no Cell Stress Inhibitor Antioxidant -12.2125 Cilostazol 0.822 2.0558E-01 no Cyclic Nucleotides Inhibitor PDE III -6.1607 Caffeine 0.411 3.4041E-01 no Adenosine Inhibitor Phosphodiesterase -14.5759 Cyclophosphamide monohydrate -0.976 1.6459E-01 no DNA Intercalator -5.5771 Caffeic acid phenethyl ester 1.306 9.5730E-02 no Cell Cycle Inhibitor NFkB -11.5965 Cinoxacin -1.095 1.3684E-01 no Antibiotic Inhibitor -0.0911 Carisoprodol -1.795 3.6351E-02 no Neurotransmission Skeletal muscle -6.5522

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Centrophenoxine hydrochloride -0.330 3.7061E-01 no Nootropic -0.6260 Clemastine fumarate -2.023 2.1547E-02 no Histamine Antagonist HRH1 -5.4913 beta-Chloro-L-alanine hydrochloride -1.390 8.2254E-02 no Biochemistry Inhibitor Alanine aminotransferase -3.1222 Pyrocatechol -2.053 2.0033E-02 no Cell Cycle Inhibitor -2.1080 CPCCOEt -0.807 2.0990E-01 no Glutamate Antagonist mGluR1 -6.4393 L-Canavanine sulfate -0.902 1.8343E-01 no Nitric Oxide Inhibitor iNOS -5.4287 Cortisone 21-acetate -2.218 1.3282E-02 no Hormone Cortisol -9.8430 Cyproterone acetate -1.803 3.5658E-02 no Hormone Antagonist Androgen -7.0698 DL-p-Chlorophenylalanine methyl ester hydrochloride 0.128 4.4922E-01 no Neurotransmission Inhibitor Tryptophan hydroxylase -3.4208 Ciclosporin -0.186 4.2628E-01 no Phosphorylation Inhibitor Calcineurin phosphatase -6.2107 D-Cycloserine -1.240 1.0748E-01 no Glutamate Agonist NMDA-Glycine -5.7186 8-(4-Chlorophenylthio)-cAMP sodium 1.217 1.1188E-01 no Cyclic Nucleotides Activator -10.7572 Carbamazepine 1.114 1.3255E-01 no Anticonvulsant -5.5460 Captopril -1.285 9.9379E-02 no Neurotransmission Inhibitor ACE -6.0164 Carbachol -0.034 4.8651E-01 no Cholinergic Agonist -7.3897 Chlorzoxazone -1.496 6.7306E-02 no Nitric Oxide Inhibitor iNOS -2.0853 L-Cysteinesulfinic Acid 1.458 7.2457E-02 no Glutamate Ligand -4.0432 9-cyclopentyladenine 0.499 3.0895E-01 no Cyclic Nucleotides Inhibitor Adenylate cyclase -1.7579 Cimetidine 1.665 4.7922E-02 no Histamine Antagonist H2 -15.6553 Cyclobenzaprine hydrochloride 0.072 4.7136E-01 no Serotonin Antagonist 5-HT2 -0.3834 Clemizole hydrochloride 2.788 2.6501E-03 no Histamine Antagonist HRH1 -4.3715 2-Chloroadenosine -0.396 3.4604E-01 no Adenosine Agonist A1 > A2 -4.4821 Bethanechol chloride -0.649 2.5807E-01 no Cholinergic Agonist Muscarinic -3.3990 Cinnarizine -1.375 8.4610E-02 no Ca2+ Channel Blocker -2.7921 1-(3-Chlorophenyl)piperazine dihydrochloride 0.952 1.7066E-01 no Serotonin Agonist 5-HT1 -0.9409 SB 204741 -2.299 1.0744E-02 no Serotonin Antagonist 5-HT2B -10.7182 4-Chloromercuribenzoic acid -0.032 4.8726E-01 no Biochemistry Inhibitor -4.9331 (-)-Cotinine -0.510 3.0507E-01 no Cholinergic Metabolite Nicotinic -9.2586 CL 316,243 0.621 2.6722E-01 no Adrenoceptor Agonist beta3 -9.7686 7-Chloro-4-hydroxy-2-phenyl-1,8-naphthyridine -0.401 3.4422E-01 no Adenosine Antagonist A1 -6.1894 Clotrimazole -0.716 2.3705E-01 no K+ Channel Inhibitor Ca2+-activated K+ channel -1.7138 Cyproheptadine hydrochloride 1.232 1.0898E-01 no Serotonin Antagonist 5-HT2 -5.8370 5`-(N-Cyclopropyl)carboxamidoadenosine -1.701 4.4454E-02 no Adenosine Agonist A2 -0.6480 Cefmetazole sodium 2.503 6.1504E-03 no Antibiotic Cell wall synthesis -3.7838 Clozapine 0.547 2.9208E-01 no Dopamine Antagonist D4 > D2,D3 -3.3166 (±)-p-Chlorophenylalanine -0.247 4.0241E-01 no Neurotransmission Inhibitor Tryptophan hydroxylase -4.5617 Clofibrate -1.738 4.1149E-02 no Lipid Modulator Lipoprotein lipase -4.1199 CB34 -1.081 1.3979E-01 no Benzodiazepine Ligand -5.9593 DL-Cycloserine -0.299 3.8263E-01 no Sphingolipid Inhibitor Ketosphinganine synthase, -5.7186 McN-A-343 -0.089 4.6455E-01 no Cholinergic Agonist M1 -4.9137 Cystamine dihydrochloride 2.230 1.2874E-02 no Glutamate Inhibitor Transglutaminase -0.9068 Calcimycin 2.172 1.4942E-02 no Intracellular Calcium Ca2+ -11.3039 Cantharidin -0.578 2.8158E-01 no Phosphorylation Inhibitor PP2A -3.0587

! ))!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Citalopram hydrobromide -1.858 3.1616E-02 no Serotonin Inhibitor Reuptake -4.9204 Clonidine hydrochloride 1.041 1.4898E-01 no Adrenoceptor Agonist alpha2 -8.6158 Cilostamide 0.176 4.3029E-01 no Cyclic Nucleotides Inhibitor PDE III -7.2538 Chelidamic acid 0.726 2.3390E-01 no Glutamate Inhibitor L-glutamic decarboxylase -5.4374 N6-Cyclopentyladenosine -0.784 2.1655E-01 no Adenosine Agonist A1 -3.3957 Cantharidic Acid -0.553 2.9016E-01 no Phosphorylation Inhibitor PP1 / PP2A -4.0610 Phenytoin sodium -0.217 4.1428E-01 no Anticonvulsant -7.1884 S(-)-Pindolol -0.965 1.6722E-01 no Adrenergic Antagonist beta -25.5497 (-)-alpha-Methylnorepinephrine 1.193 1.1634E-01 no Adrenoceptor Agonist -6.3305 Dilazep hydrochloride 0.245 4.0329E-01 no Adenosine Inhibitor Uptake -4.5746 1,7-Dimethylxanthine -0.662 2.5390E-01 no Adenosine Antagonist A1 > A2 -7.6830 Daphnetin -0.980 1.6359E-01 no Phosphorylation Inhibitor PK -7.3553 DM 235 -2.019 2.1731E-02 no Nootropic -6.9226 5,5-Dimethyl-1-pyrroline-N-oxide -1.666 4.7874E-02 no Cell Stress Inhibitor Antioxidant -3.0181 Diacylglycerol Kinase Inhibitor II -1.390 8.2282E-02 no Phosphorylation Inhibitor Diacylglycerol kinase -3.7809 Dihydrexidine hydrochloride -0.415 3.3896E-01 no Dopamine Agonist D1 -5.3869 N-Methyldopamine hydrochloride -1.219 1.1143E-01 no Dopamine Agonist -11.1570 1,1-Dimethyl-4-phenyl-piperazinium iodide -1.057 1.4522E-01 no Cholinergic Agonist -2.2053 Cyclothiazide -0.907 1.8221E-01 no Glutamate Agonist AMPA -2.7150 8-Cyclopentyl-1,3-dipropylxanthine -0.070 4.7201E-01 no Adenosine Antagonist A1 -21.9125 8-Cyclopentyl-1,3-dimethylxanthine -0.125 4.5035E-01 no Adenosine Antagonist A1 -20.3794 (±)-CPP 0.321 3.7398E-01 no Glutamate Antagonist NMDA -4.3608 2-Cyclooctyl-2-hydroxyethylamine hydrochloride -0.575 2.8276E-01 no Neurotransmission Inhibitor PNMT -3.8568 5-Carboxamidotryptamine maleate 1.667 4.7711E-02 no Serotonin Agonist 5-HT7 -19.4950 7-Chlorokynurenic acid 1.241 1.0733E-01 no Glutamate Antagonist NMDA -4.9223 (±)-CGP-12177A hydrochloride -1.611 5.3544E-02 no Adrenoceptor Agonist beta -15.3877 S-(-)-Carbidopa -2.314 1.0324E-02 no Biochemistry Inhibitor Aromatic amino acid decar -8.9751 (±)-Chloro-APB hydrobromide -0.111 4.5581E-01 no Dopamine Agonist D1 -6.3740 Y-27632 dihydrochloride -0.125 4.5007E-01 no Phosphorylation Inhibitor ROCK -7.9142 2-Chloroadenosine triphosphate tetrasodium 1.149 1.2531E-01 no P2 Receptor Agonist P2Y -8.8552 (+)-Cyclazocine 0.612 2.7040E-01 no Opioid Antagonist -1.0736 Capsazepine 1.260 1.0386E-01 no Vanilloid Agonist -7.7040 Chlormezanone -0.947 1.7194E-01 no Neurotransmission Modulator Muscle relaxant -6.2721 8-(3-Chlorostyryl)caffeine -1.214 1.1231E-01 no Adenosine Antagonist A2A -12.4931 CGS-15943 0.178 4.2922E-01 no Adenosine Antagonist A1 -6.3372 Cirazoline hydrochloride 0.424 3.3580E-01 no Adrenoceptor Agonist alpha1A -6.6318 CGP 20712A methanesulfonate 0.649 2.5815E-01 no Adrenoceptor Antagonist beta1 -13.7874 (2S,1`S,2`S)-2-(carboxycyclopropyl)glycine 1.233 1.0878E-01 no Glutamate Agonist mGluR2 -5.6999 CNQX disodium -1.904 2.8471E-02 no Glutamate Antagonist AMPA/Kainate -8.1220 CX 546 -1.050 1.4686E-01 no Glutamate Modulator AMPA -5.5188 Chloro-IB-MECA -0.847 1.9846E-01 no Adenosine Agonist A3 -5.2500 WB-4101 hydrochloride -2.247 1.2321E-02 no Adrenoceptor Antagonist alpha1A -7.3187 DNQX 0.931 1.7599E-01 no Glutamate Antagonist Kainate/quisqualate -6.4394

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Dihydroouabain 0.203 4.1942E-01 no Ion Pump Inhibitor Na+/K+ Pump -5.4302 Dobutamine hydrochloride 0.319 3.7505E-01 no Adrenoceptor Agonist beta1 -7.2941 Dihydrokainic acid 1.329 9.1976E-02 no Glutamate Blocker Kainate -5.6560 P1,P4-Di(adenosine-5`)tetraphosphate triammonium -0.287 3.8718E-01 no Biochemistry Inhibitor -9.7230 Debrisoquin sulfate -0.303 3.8078E-01 no Neurotransmission Antihyperten -5.4151 2`,3`-didehydro-3`-deoxythymidine -1.878 3.0177E-02 no Immune System Inhibitor Reverse Transcriptase -1.1860 Droperidol 0.645 2.5958E-01 no Dopamine Antagonist D1/D2 -5.2369 L-3,4-Dihydroxyphenylalanine methyl ester hydrochloride 0.277 3.9093E-01 no Dopamine Precursor -9.1222 1,4-Dideoxy-1,4-imino-D-arabinitol -0.009 4.9656E-01 no Phosphorylation Inhibitor Glycogen phosphorylase -6.9439 2,4-Dinitrophenyl 2-fluoro-2-deoxy-beta-D-glucopyranoside 0.772 2.2005E-01 no Biochemistry Inhibitor exo-beta-(1,3)-Glucanase -4.0687 D-ribofuranosylbenzimidazole -2.144 1.6019E-02 no Transcription Inhibitor -6.9546 Diltiazem hydrochloride 0.540 2.9476E-01 no Ca2+ Channel Antagonist L-type -4.7662 SB 203186 0.937 1.7450E-01 no Serotonin Antagonist 5-HT4 -6.9689 Dihydroergotamine methanesulfonate -0.779 2.1785E-01 no Serotonin Antagonist -17.9041 2,3-Butanedione -1.642 5.0346E-02 no Cytoskeleton and ECM Inhibitor Myosin ATPase -1.4471 N,N,N`,N`-Tetramethylazodicarboxamide -0.639 2.6132E-01 no Cell Stress Modulator Thiols -2.7413 (S)-3,5-Dihydroxyphenylglycine -0.336 3.6828E-01 no Glutamate Agonist mGluR1 -7.1241 Doxylamine succinate 1.736 4.1270E-02 no Histamine Antagonist HRH1 -3.7735 Desipramine hydrochloride 0.123 4.5119E-01 no Adrenoceptor Inhibitor Uptake -7.4669 5,5-Diphenylhydantoin 1.383 8.3398E-02 no Anticonvulsant -6.5440 N',N'-Dimethylarginine hydrochloride -1.373 8.4933E-02 no Nitric Oxide Inhibitor NOS -6.7125 Clodronic acid -0.184 4.2694E-01 no Cytoskeleton and ECM Inhibitor MMP1 / collagenase -3.7954 Dihydroergocristine methanesulfonate -0.773 2.1977E-01 no Dopamine Agonist -18.9281 2,6-Diamino-4-pyrimidinone -1.072 1.4188E-01 no Phosphorylation Inhibitor GTP cyclohydrolase I -1.3883 DL-alpha-Difluoromethylornithine hydrochloride -0.424 3.3565E-01 no Angiogenesis Inhibitor ODC -2.9978 SCH-28080 2.107 1.7547E-02 no Ion Channels Inhibitor H+/K+-ATPase -8.4419 S(-)-DS 121 hydrochloride -0.100 4.6025E-01 no Dopamine Antagonist Autoreceptor -10.3580 Vanillic acid diethylamide -0.595 2.7599E-01 no Vanilloid Agonist -3.4750 Epibestatin hydrochloride -0.787 2.1568E-01 no Biochemistry Inhibitor Metalloprotease -6.6215 Etodolac -0.558 2.8836E-01 no Prostaglandin Inhibitor COX -5.9520 Enoximone -0.624 2.6642E-01 no Cyclic Nucleotides Inhibitor PDE III -5.2654 ET-18-OCH3 -0.626 2.6570E-01 no Lipid Inhibitor PIPLC -13.2721 Etazolate hydrochloride -0.247 4.0263E-01 no Adenosine Inhibitor Phosphodiesterase -8.8336 7-Cyclopentyl-5-(4-phenoxy)phenyl-7H-pyrrolo[2,3-d]pyrimidin-4-ylamine -1.412 7.8972E-02 no Phosphorylation Inhibitor Ick -3.2409 E-64 -1.709 4.3748E-02 no Biochemistry Inhibitor Cysteine protease -7.4853 Diacylglycerol kinase inhibitor I -1.782 3.7390E-02 no Phosphorylation Inhibitor Diacylglycerol kinase -7.8178 Demeclocycline hydrochloride -0.065 4.7428E-01 no Antibiotic Protein synthesis -8.0592 Diclofenac sodium 0.299 3.8255E-01 no Prostaglandin Inhibitor COX -3.6420 DL-erythro-Dihydrosphingosine -0.079 4.6871E-01 no Phosphorylation Inhibitor PKC / PLA2 / PLD -0.1690 R-(-)-Desmethyldeprenyl hydrochloride -0.506 3.0657E-01 no Neurotransmission Inhibitor MAO-B -5.6089 2,2`-Bipyridyl 1.455 7.2862E-02 no Biochemistry Inhibitor Metalloprotease -3.7379 Dicyclomine hydrochloride 1.414 7.8624E-02 no Cholinergic Antagonist Muscarinic -0.8242 3,4-Dichloroisocoumarin -1.120 1.3141E-01 no Biochemistry Inhibitor Serine Protease -4.9206

! )*!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID DBO-83 -1.279 1.0038E-01 no Cholinergic Agonist Nicotinic -7.5674 Dephostatin 0.621 2.6732E-01 no Phosphorylation Inhibitor CD45 Tyrosine Kinase -3.9458 3-deazaadenosine 1.770 3.8360E-02 no Immune System Inhibitor -6.0324 (Z)-Gugglesterone 3.205 6.7567E-04 no Lipid Signaling Antagonist FRX -10.5383 Danazol 2.152 1.5713E-02 no Hormone Inhibitor -15.4776 N,N-Dihexyl-2-(4-fluorophenyl)indole-3-acetamide 1.916 2.7694E-02 no Benzodiazepine Ligand Mitochondria -7.3781 SP600125 -1.986 2.3536E-02 no Phosphorylation Inhibitor c-JNK -8.4226 Diazoxide 0.045 4.8201E-01 no K+ Channel Activator ATP-sensitive -3.3923 3,4-Dihydroxyphenylacetic acid -0.583 2.7988E-01 no Dopamine Metabolite -10.2325 Dantrolene sodium 0.798 2.1230E-01 no Intracellular Calcium Inhibitor Release -8.4822 DCEBIO 0.810 2.0900E-01 no K+ Channel Activator hlK1 -3.8055 1-Deoxynojirimycin hydrochloride 1.242 1.0713E-01 no Biochemistry Inhibitor alpha-glucosidase -6.6269 L-3,4-Dihydroxyphenylalanine 2.440 7.3488E-03 no Dopamine Precursor -9.9117 Dipyridamole -0.727 2.3372E-01 no Adenosine Inhibitor -1.9619 Doxazosin mesylate -1.861 3.1346E-02 no Adrenoceptor Blocker alpha1 -2.6413 Doxycycline hydrochloride -0.558 2.8849E-01 no Antibiotic Protein synthesis -14.0343 6,7-ADTN hydrobromide 0.357 3.6056E-01 no Dopamine Agonist -6.3543 Dipropyldopamine hydrobromide -1.346 8.9197E-02 no Dopamine Agonist -3.0223 Amfonelic acid 0.488 3.1277E-01 no Dopamine Modulator -13.3956 Icilin 1.643 5.0157E-02 no Neurotransmission Agonist CMR1 -6.3622 (±)-SKF-38393 hydrochloride 1.280 1.0030E-01 no Dopamine Antagonist D1 -8.8933 R(+)-SCH-23390 hydrochloride 1.375 8.4525E-02 no Dopamine Antagonist D1 -2.8442 (±)-DOI hydrochloride -1.345 8.9370E-02 no Serotonin Agonist 5-HT2/5-HT1C -1.8081 (±)-2,3-Dichloro-alpha-methylbenzylamine hydrochloride -2.251 1.2197E-02 no Neurotransmission Inhibitor PNMT -3.2749 4-DAMP methiodide 0.331 3.7027E-01 no Cholinergic Antagonist M3 -4.8290 1,3-Dipropyl-7-methylxanthine -0.089 4.6456E-01 no Adenosine Antagonist A2 -17.0207 Propofol 0.377 3.5315E-01 no Cholinergic Inhibitor Muscarinic -2.8649 Dextrorphan D-tartrate 1.569 5.8358E-02 no Glutamate Antagonist NMDA -3.2343 R(+)-Butylindazone 0.029 4.8834E-01 no Ion Pump Inhibitor K+/Cl- transport -9.4197 DPMA 0.036 4.8553E-01 no Adenosine Agonist A2 -4.6563 3,5-Dinitrocatechol -0.529 2.9828E-01 no Neurotransmission Inhibitor COMT -6.6346 N,N-Dipropyl-5-carboxamidotryptamine maleate -1.730 4.1815E-02 no Serotonin Agonist 5-HT1A -13.3021 6,7-Dichloroquinoxaline-2,3-dione 0.127 4.4932E-01 no Glutamate Antagonist NMDA-glycine -3.3370 3,7-Dimethyl-I-propargylxanthine 1.454 7.3015E-02 no Adenosine Antagonist A2 -15.2349 5,7-Dichlorokynurenic acid 0.901 1.8378E-01 no Glutamate Antagonist NMDA-glycine -4.6989 4-Diphenylacetoxy-N-(2-chloroethyl)piperidine hydrochloride 0.864 1.9366E-01 no Cholinergic Antagonist Muscarinic -1.3361 1,10-Diaminodecane 0.139 4.4460E-01 no Glutamate Agonist (inv NMDA-polyamine -2.0129 Dihydro-beta-erythroidine hydrobromide -1.788 3.6854E-02 no Cholinergic Antagonist nAch -2.8674 N-(3,3-Diphenylpropyl)glycinamide -1.243 1.0691E-01 no Glutamate Blocker NMDA -5.7246 Glibenclamide -0.360 3.5941E-01 no K+ Channel Blocker ATP-dependent -8.7301 GW2974 -0.516 3.0279E-01 no Phosphorylation Inhibitor EGFR / ErbB-2 -12.2358 Guanfacine hydrochloride 0.927 1.7695E-01 no Adrenoceptor Agonist alpha2 -2.0297 L-Glutamic acid hydrochloride -0.473 3.1822E-01 no Glutamate Agonist -2.3031

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID L-Glutamine 0.828 2.0389E-01 no Glutamate Agonist -5.0432 Guanidinyl-naltrindole di-trifluoroacetate -0.204 4.1929E-01 no Opioid Antagonist kappa -9.3192 GW1929 0.015 4.9398E-01 no Transcription Agonist PPAR-gamma -9.5700 GW5074 -0.162 4.3570E-01 no Phosphorylation Inhibitor Raf1 kinase -8.2139 GW7647 -0.992 1.6052E-01 no Transcription Agonist PPAR-alpha -12.1570 Gallamine triethiodide -0.916 1.7995E-01 no Cholinergic Antagonist M2 -4.7485 S-Ethylisothiourea hydrobromide -1.599 5.4874E-02 no Nitric Oxide Inhibitor NOS -4.2580 Edrophonium chloride -0.806 2.1001E-01 no Cholinergic Inhibitor Acetylcholinesterase -5.3039 Ebselen -0.251 4.0072E-01 no Leukotriene Inhibitor -2.2253 rac-2-Ethoxy-3-hexadecanamido-1-propylphosphocholine -1.269 1.0223E-01 no Phosphorylation Inhibitor PKC -13.3506 rac-2-Ethoxy-3-octadecanamido-1-propylphosphocholine -1.112 1.3298E-01 no Phosphorylation Inhibitor PKC -12.3289 N-Ethylmaleimide -0.028 4.8877E-01 no Biochemistry Inhibitor Isocitrate dehydrogenase -3.3843 (-)-Epinephrine bitartrate -0.530 2.9791E-01 no Adrenoceptor Agonist -13.0202 (±)-Epinephrine hydrochloride -0.328 3.7148E-01 no Adrenoceptor Agonist -12.7284 Ethosuximide -0.612 2.7014E-01 no Anticonvulsant -1.4347 Endothall -1.143 1.2644E-01 no Phosphorylation Inhibitor PP2A -4.0878 Emodin -0.483 3.1438E-01 no Phosphorylation Inhibitor p56lck TK -5.2573 (-)-Physostigmine -0.218 4.1377E-01 no Cholinergic Inhibitor Cholinesterase -3.2224 NBI 27914 0.134 4.4651E-01 no Neurotransmission Antagonist CRF1 -3.9632 beta-Estradiol -0.562 2.8696E-01 no Hormone Estrogen -12.0545 Estrone 0.041 4.8372E-01 no Hormone Estrogen -13.0120 Methyl beta-carboline-3-carboxylate 0.060 4.7599E-01 no Benzodiazepine Agonist -9.2192 N-Methyl-beta-carboline-3-carboxamide 0.699 2.4235E-01 no GABA Antagonist GABA-A -9.5740 Methyl 6,7-dimethoxy-4-ethyl-beta-carboline-3-carboxylate -1.326 9.2453E-02 no Benzodiazepine Agonist -5.1499 (-)-Eseroline fumarate 0.634 2.6308E-01 no Cholinergic Inhibitor Cholinesterase -5.6206 (S)-ENBA 0.208 4.1767E-01 no Adenosine Agonist A1 -8.4799 erythro-9-(2-Hydroxy-3-nonyl)adenine hydrochloride -0.766 2.2197E-01 no Adenosine Inhibitor Adenosine deaminase -1.3832 Ergocristine 0.408 3.4172E-01 no Dopamine Agonist -27.2658 Felbamate -0.459 3.2327E-01 no Glutamate Antagonist -8.5534 Fusidic acid sodium 2.045 2.0427E-02 no Cell Cycle Inhibitor -9.8579 Fenoterol hydrobromide -1.841 3.2821E-02 no Adrenoceptor Agonist beta2 -10.3422 S-(+)-Fluoxetine hydrochloride -0.809 2.0932E-01 no Serotonin Inhibitor Reuptake -9.1211 R-(-)-Fluoxetine hydrochloride -0.128 4.4905E-01 no Serotonin Inhibitor Reuptake -9.1211 Fluvoxamine maleate 0.132 4.4737E-01 no Serotonin Inhibitor Reuptake -9.4524 1-(4-Fluorobenzyl)-5-methoxy-2-methylindole-3-acetic acid -0.538 2.9543E-01 no Multi-Drug Resistance Inhibitor MRP1 -9.0898 Furegrelate sodium 0.364 3.5800E-01 no Phosphorylation Inhibitor Thromboxane synthase -12.3958 Fiduxosin hydrochloride 0.200 4.2056E-01 no Adrenoceptor Antagonist alpha1 -5.1869 Furosemide 1.889 2.9456E-02 no Ion Pump Inhibitor Na+,K+,Cl- cotransport -5.9592 p-Fluoro-L-phenylalanine -0.702 2.4126E-01 no Neurotransmission Substrate Tyrosine Hydroxylase -4.5617 Fenofibrate -2.116 1.7190E-02 no Transcription Agonist PPAR-alpha -0.2642 Fenspiride hydrochloride -0.576 2.8223E-01 no Adrenoceptor Antagonist alpha -0.7723 Flumazenil -0.366 3.5703E-01 no Benzodiazepine Antagonist -10.4543 Foliosidine -0.097 4.6140E-01 no Anticonvulsant -6.0396

! *+!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Fusaric acid -0.141 4.4382E-01 no Dopamine Inhibitor Dopamine beta-hydroxylase -7.0076 Flecainide acetate -1.498 6.7005E-02 no Na+ Channel Blocker -5.4913 Fenoldopam bromide 0.052 4.7919E-01 no Dopamine Agonist D1 -7.5224 Forskolin -1.331 9.1668E-02 no Cyclic Nucleotides Activator Adenylate cyclase -4.1972 Famotidine 0.055 4.7811E-01 no Histamine Antagonist H2 -7.5810 FSCPX 0.782 2.1697E-01 no Adenosine Antagonist A1 -25.0121 Farnesylthiosalicylic acid 0.906 1.8244E-01 no G protein Antagonist Ras -4.1651 Flunarizine dihydrochloride 0.395 3.4627E-01 no Ion Pump Blocker Na+/Ca2+ channel -1.9671 5-fluoro-5`-deoxyuridine 0.379 3.5246E-01 no DNA Metabolism Inhibitor -2.9280 Flupirtine maleate -0.484 3.1420E-01 no Glutamate Antagonist NMDA -9.8407 Flutamide -0.113 4.5490E-01 no Hormone Inhibitor Androgen -7.3424 Fexofenadine hydrochloride -0.757 2.2445E-01 no Histamine Antagonist HRH1 -5.6445 Formoterol -0.838 2.0104E-01 no Adrenoceptor Agonist beta2 -16.0812 Felodipine -0.966 1.6707E-01 no Ca2+ Channel Blocker L-type -10.2014 Fluspirilene 1.589 5.6001E-02 no Dopamine Antagonist D2/D1 -5.8124 Furafylline 1.025 1.5277E-01 no Biochemistry Inhibitor P450IA2 -16.0242 FPL 64176 -1.099 1.3591E-01 no Ca2+ Channel Activator L-type -6.8541 Fluoxetine hydrochloride -1.311 9.4858E-02 no Serotonin Inhibitor Reuptake -9.1211 GR 125487 sulfamate salt -0.194 4.2300E-01 no Serotonin Antagonist 5-HT4 -4.2655 IEM-1460 0.685 2.4674E-01 no Glutamate Inhibitor AMPA -5.2560 Ibudilast 0.379 3.5244E-01 no Cyclic Nucleotides Inhibitor PDE IV -6.2032 Imidazole-4-acetic acid hydrochloride -0.922 1.7820E-01 no GABA Antagonist GABA-C -12.6089 Indirubin-3`-oxime -0.397 3.4587E-01 no Phosphorylation Inhibitor CDK -8.4387 Imazodan 0.462 3.2217E-01 no Cyclic Nucleotides Inhibitor PDE II -12.4077 Ipratropium bromide -0.308 3.7904E-01 no Cholinergic Antagonist Muscarinic -8.9698 2-Iodomelatonin -0.135 4.4637E-01 no Melatonin Agonist -7.1283 SB 228357 0.034 4.8634E-01 no Serotonin Antagonist 5-HT2B/2C -4.6876 IMID-4F hydrochloride -1.906 2.8337E-02 no K+ Channel Blocker -3.6024 R(-)-Isoproterenol (+)-bitartrate -1.485 6.8743E-02 no Adrenoceptor Agonist beta -13.6004 Isoguvacine hydrochloride -2.220 1.3206E-02 no GABA Agonist GABA-A, GABA-C -5.7435 Guvacine hydrochloride -1.345 8.9234E-02 no GABA Inhibitor Uptake -5.2169 (±)-AMPA hydrobromide -0.427 3.3461E-01 no Glutamate Agonist AMPA/kainate -9.1686 Muscimol hydrobromide 0.186 4.2612E-01 no GABA Agonist GABA-A, GABA-C -6.3913 Guanabenz acetate -1.906 2.8344E-02 no Adrenoceptor Agonist alpha2 -5.3611 gamma-D-Glutamylaminomethylsulfonic acid -0.046 4.8146E-01 no Glutamate Antagonist Kainate -6.9425 Glipizide -0.830 2.0341E-01 no K+ Channel Blocker ATP-sensitive -13.3300 GYKI 52895 -1.258 1.0419E-01 no Dopamine Inhibitor Reuptake -0.3735 Gabapentin 0.884 1.8841E-01 no Anticonvulsant -4.8782 (±)-Vanillylmandelic acid -1.811 3.5052E-02 no Adrenoceptor Metabolite -1.2797 6-Hydroxymelatonin 0.377 3.5318E-01 no Melatonin Metabolite -15.2014 4-Hydroxy-3-methoxyphenylacetic acid -0.567 2.8526E-01 no Dopamine Metabolite -2.3077 MHPG piperazine 0.053 4.7896E-01 no Adrenoceptor Metabolite -2.0185 Hypotaurine 0.275 3.9158E-01 no Cell Stress Inhibitor Antioxidant -3.4031

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Haloperidol -0.740 2.2967E-01 no Dopamine Antagonist D2/D1 -7.7150 Hydralazine hydrochloride 0.259 3.9772E-01 no Neurotransmission Inhibitor MAO-A/B -6.5236 4-Imidazolemethanol hydrochloride -2.077 1.8919E-02 no Histamine Inhibitor Histinol Dehydrogenase -9.7374 Hydrocortisone 21-hemisuccinate sodium -0.040 4.8392E-01 no Hormone Cortisol -10.6646 6-Hydroxy-DL-DOPA 1.204 1.1431E-01 no Adrenoceptor Neurotoxin -6.8703 DL-threo-beta-hydroxyaspartic acid -0.522 3.0100E-01 no Glutamate Inhibitor Transport -2.7459 Hydroxytacrine maleate -0.204 4.1905E-01 no Cholinergic Inhibitor Cholinesterase -8.2465 Lithium Chloride -0.911 1.8108E-01 no Neurotransmission Inhibitor Inositol monophosphatase -0.4563 Hydrochlorothiazide -0.212 4.1605E-01 no Biochemistry Inhibitor Carbonic anhydrase -2.9759 SB 218795 1.359 8.7060E-02 no Neurotransmission Antagonist NK3 -2.6055 Hispidin 0.890 1.8684E-01 no Phosphorylation Inhibitor PKC-beta -12.0505 17alpha-hydroxyprogesterone -0.578 2.8155E-01 no Hormone Metabolite Progesterone -10.7469 1,3,5-tris(4-hydroxyphenyl)-4-propyl-1H-pyrazole 0.772 2.2016E-01 no Hormone Agonist ER-alpha -3.3002 1-(4-Hydroxybenzyl)imidazole-2-thiol 0.763 2.2280E-01 no Dopamine Inhibitor Dopamine beta-hydroxylase -5.7482 Histamine dihydrochloride -1.476 6.9936E-02 no Histamine Agonist -12.2180 Harmane -1.403 8.0334E-02 no Imidazoline Agonist I1 -6.4705 L-Histidine hydrochloride -1.421 7.7629E-02 no Histamine Precursor -13.7537 Dopamine hydrochloride 1.352 8.8231E-02 no Dopamine Agonist -10.1142 Hydroxyurea 1.131 1.2897E-01 no DNA Metabolism Inhibitor Ribonucleoside reductase -2.1618 MHPG sulfate potassium -0.556 2.8923E-01 no Adrenoceptor Metabolite -5.0540 5-Hydroxyindolacetic acid -1.301 9.6595E-02 no Serotonin Metabolite -19.9055 L-Hyoscyamine -0.661 2.5429E-01 no Cholinergic Antagonist -6.4660 Hydroquinone 0.428 3.3434E-01 no Leukotriene Inhibitor -0.3944 BU99006 1.182 1.1870E-01 no Imidazoline Ligand I2 -8.1496 3-Hydroxybenzylhydrazine dihydrochloride -0.487 3.1326E-01 no Biochemistry Inhibitor Amino acid decarboxylase -6.6452 Serotonin hydrochloride -2.197 1.4025E-02 no Serotonin Agonist -19.9324 L-165,041 0.306 3.7973E-01 no Lipid Signaling Agonist PPAR-gamma -1.9292 5-Hydroxy-L-tryptophan 0.030 4.8799E-01 no Serotonin Precursor -20.5215 Hydroxylamine hydrochloride 1.635 5.1057E-02 no Neurotransmission Inhibitor MAO -2.3474 4-Hydroxybenzhydrazide -0.789 2.1512E-01 no Biochemistry Inhibitor -3.0491 Hemicholinium-3 0.787 2.1562E-01 no Cholinergic Blocker Uptake -5.1174 HA-1004 hydrochloride 0.316 3.7596E-01 no Phosphorylation Inhibitor PK -7.3536 H-7 dihydrochloride 0.681 2.4785E-01 no Phosphorylation Inhibitor PKC -3.0203 Hexahydro-sila-difenidol hydrochloride, p-fluoro analog -1.492 6.7900E-02 no Cholinergic Antagonist M3>M1>M2 -1.5458 Histamine, R(-)-alpha-methyl-, dihydrochloride -1.469 7.0855E-02 no Histamine Agonist H3 -11.9901 5-hydroxydecanoic acid sodium 0.233 4.0782E-01 no K+ Channel Blocker -5.4081 Leflunomide 4.103 2.0382E-05 no Immune System Inhibitor -5.4344 VER-3323 hemifumarate salt -0.761 2.2321E-01 no Serotonin Agonist 5-HT2C/5-HT2B -4.5451 Lidocaine hydrochloride -0.111 4.5578E-01 no Na+ Channel Modulator -5.0834 Lidocaine N-ethyl bromide quaternary salt 0.265 3.9538E-01 no Na+ Channel Antagonist -7.1021 L-Leucinethiol, oxidized dihydrochloride -0.584 2.7961E-01 no Biochemistry Inhibitor Aminopeptidase -1.6871 LE 300 -1.070 1.4234E-01 no Dopamine Antagonist D1 -1.8116 Lansoprazole -0.034 4.8660E-01 no Ion Pump Inhibitor H+ pump -11.4889

! *"!

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID LFM-A13 4.500 3.4002E-06 no Phosphorylation Inhibitor BTK -7.6465 Luteolin -1.323 9.2973E-02 no Cell Stress Inhibitor Antioxidant -12.5139 L-655,240 -0.121 4.5183E-01 no Thromboxane Antagonist TXA2 -9.9163 Loratadine -1.823 3.4160E-02 no Histamine Antagonist HRH1 -7.9079 (-)-Tetramisole hydrochloride -1.866 3.1005E-02 no Phosphorylation Inhibitor -9.6745 L-655,708 -0.613 2.6988E-01 no Benzodiazepine Ligand GABA-A -11.3947 LY-294,002 hydrochloride -1.217 1.1184E-01 no Phosphorylation Inhibitor PI3K -3.1040 Loxapine succinate -0.748 2.2732E-01 no Dopamine Antagonist -4.0117 (±)-Ibotenic acid -1.650 4.9495E-02 no Glutamate Agonist NMDA -7.9415 Isotharine mesylate -0.639 2.6136E-01 no Adrenoceptor Agonist beta -8.8817 (±)-Ibuprofen -0.262 3.9653E-01 no Prostaglandin Inhibitor COX -8.3439 IIK7 -0.354 3.6152E-01 no Melatonin Agonist -8.6271 (±)-Isoproterenol hydrochloride -0.572 2.8376E-01 no Adrenoceptor Agonist beta -14.1845 3-Isobutyl-1-methylxanthine -0.774 2.1947E-01 no Adenosine Inhibitor Phosphodiesterase -16.0203 Idazoxan hydrochloride -1.448 7.3866E-02 no Imidazoline Ligand I1 / I2 -6.8373 1-(5-Isoquinolinylsulfonyl)-3-methylpiperazine dihydrochloride 0.477 3.1672E-01 no Phosphorylation Inhibitor PKC -2.8972 (-)-Isoproterenol hydrochloride 0.237 4.0646E-01 no Adrenoceptor Agonist beta -14.1845 1-(5-Isoquinolinylsulfonyl)-2-methylpiperazine dihydrochloride 0.335 3.6874E-01 no Phosphorylation Inhibitor PKA / PKC -3.0203 Imipramine hydrochloride -0.527 2.9903E-01 no Serotonin Blocker Reuptake -1.0708 Isoxanthopterin 0.234 4.0752E-01 no Cell Stress Metabolite -7.9514 Iproniazid phosphate -0.249 4.0158E-01 no Neurotransmission Inhibitor MAO -8.4760 S(+)-Isoproterenol (+)-bitartrate -0.296 3.8350E-01 no Adrenoceptor beta -13.6004 L-N6-(1-Iminoethyl)lysine hydrochloride -0.381 3.5164E-01 no Nitric Oxide Inhibitor iNOS -6.5483 3-Iodo-L-tyrosine -0.048 4.8068E-01 no Neurotransmission Inhibitor Tyrosine hydroxylase -3.7497 L-N5-(1-Iminoethyl)ornithine hydrochloride -0.934 1.7519E-01 no Nitric Oxide Inhibitor NOS -6.6284 Ivermectin 1.507 6.5855E-02 no Cholinergic Modulator alpha7 nACh -2.3255 Imiloxan hydrochloride -0.988 1.6160E-01 no Adrenoceptor Antagonist alpha2B -5.9673 CR 2945 0.545 2.9285E-01 no Cholecystokinin Antagonist CCK-B -13.6579 S(+)-Ibuprofen -0.733 2.3164E-01 no Prostaglandin Inhibitor COX -8.3439 p-Iodoclonidine hydrochloride -0.161 4.3619E-01 no Adrenoceptor Agonist alpha2 -7.1078 R(+)-IAA-94 -0.010 4.9595E-01 no Cl- Channel Inhibitor -8.5026 Indatraline hydrochloride -0.724 2.3465E-01 no Dopamine Inhibitor Reuptake -0.1381 Iofetamine hydrochloride -0.759 2.2398E-01 no Neurotransmission Analog -4.5840 ICI 204,448 hydrochloride -0.364 3.5809E-01 no Opioid Agonist kappa -2.3593 ICI 118,551 hydrochloride 1.101 1.3555E-01 no Adrenoceptor Antagonist beta2 -6.5568 Imetit dihydrobromide -0.245 4.0313E-01 no Histamine Agonist H3 -13.0801 1,5-Isoquinolinediol 0.027 4.8934E-01 no Apoptosis Inhibitor PARS -5.7715 IB-MECA -1.089 1.3806E-01 no Adenosine Agonist A3 -3.8357 3-(1H-Imidazol-4-yl)propyl di(p-fluorophenyl)methyl ether -0.015 4.9415E-01 no Histamine Antagonist H3 -5.6768 Isonipecotic acid -0.457 3.2387E-01 no GABA Agonist GABA-A -4.8582 JWH-015 0.622 2.6688E-01 no Cannabinoid Agonist CB2 -4.4913 JL-18 -0.286 3.8739E-01 no Dopamine Antagonist D4>D2 -8.5500 Kainic acid -0.317 3.7567E-01 no Glutamate Agonist Kainate -5.9803

Supplementary Table 1 | HTS of Neural Precursor Cells

Product Name Z score p value Activity Class Action Selectivity Likelihood score Cluster ID Ketoconazole 0.586 2.7892E-01 no Multi-Drug Resistance Inhibitor Cytochrome P450c17 -0.0583 Ketorolac tris salt -0.726 2.3405E-01 no Prostaglandin Inhibitor COX -6.3587 Ketoprofen -1.164 1.2226E-01 no Prostaglandin Inhibitor COX-1 -9.3283 K 185 0.567 2.8521E-01 no Melatonin Antagonist -5.3276 Ketotifen fumarate 0.957 1.6921E-01 no Histamine Antagonist HRH1 -10.4823 Kynurenic acid 0.702 2.4140E-01 no Glutamate Antagonist NMDA-Glycine -6.8297 Kenpaullone -1.555 5.9978E-02 no Phosphorylation Inhibitor CDK1, CDK2, CDK5 -10.0974 Karakoline -1.625 5.2047E-02 no Cholinergic Antagonist Nicotinic -3.9500 L-701,324 0.451 3.2590E-01 no Glutamate Antagonist NMDA-Glycine -2.3937 loxoprofen -0.883 1.8853E-01 no Prostaglandin Inhibitor COX -10.0239 Labetalol hydrochloride -1.097 1.3628E-01 no Adrenoceptor Antagonist beta -7.9308 L-162,313 0.792 2.1410E-01 no Neurotransmission Agonist AT1 -14.2550 Lidocaine N-methyl hydrochloride -0.683 2.4738E-01 no Na+ Channel Blocker -7.4131 LY-367,265 -0.407 3.4202E-01 no Serotonin Antagonist Reuptake -10.9577 L-368,899 0.291 3.8555E-01 no Neurotransmission Antagonist Oxytocin receptor -7.6946 Lomefloxacin hydrochloride 0.137 4.4536E-01 no Antibiotic Inhibitor DNA Gyrase -12.1961 Lamotrigine -1.095 1.3681E-01 no Anticonvulsant -2.3009 alpha-Lobeline hydrochloride 0.550 2.9121E-01 no Cholinergic Agonist Nicotinic -2.4879 Loperamide hydrochloride -0.498 3.0938E-01 no Opioid Ligand -4.0032 Lonidamine 0.436 3.3148E-01 no Cell Stress Inhibitor Mitochondrial hexokinase -5.7024

! *#!

-CHAPTER 3- NEUROCHEMICAL ATTENUATION OF BRAIN TUMOR GROWTH IN VIVO

3.1 PROLOGUE

In the previous chapter I identified that a wide variety of neurotransmitter agents are involved in regulating the expansion of embryonic mouse derived NSC cultures. The sensitivity to these agents was also maintained in populations of neural precursors derived from both mouse and human brain cancers. Prompted by these findings, I investigate if repeated administration of neuromodulatory drugs can attenuate brain tumor growth in different mouse models of brain tumors. Following the induction of tumorigenesis in various clinically relevant models of brain cancer, I randomized mice into different groups and treated them daily with either a dopamine or serotonin agonist that were recovered from my original HTS screen168. I then assayed the effects of these agents on survival and tumor growth. My results suggest that the efficacy of the drugs on tumor progression depended on the model used. In a relatively aggressive mouse model of brain tumors (induced by irradiation), neither drug was able to attenuate in vivo brain tumor growth. In contrast to this, in a slower growing model (induced by a chemical carcinogen), substantial differences between treated and untreated mice were noted when the mice were scarified and the tumor size was compared. Although the numbers of mice used in the second mouse model were too small to assess the statistical significance of these results, they optimistically suggest that neuromodulation may effectively decease brain tumor load. In addition to strengthening the potential use of clinically approved neurotransmission agents as novel and non-toxic chemotherapeutics, this data highlights the utility of in vitro HTS of normal NSCs for cancer drug discovery. The majority of the work described in this chapter was presented orally at the Stem Cell Network Annual General Meeting 2008 and is ongoing work being conducted in Dirks laboratory.

! *$!

3.2 TITLE AND CONTRIBUTORS

NEUROCHEMICAL ATTENUATION OF BRAIN TUMOR GROWTH IN VIVO

Phedias Diamandis1-4, Ryan J Ward1,2,5, Lilian Lee1,2, Kevin Graham1,2, Adrian Satcher1,2,

Mike Tyers3,4,7 & Peter B. Dirks1,2,5.6

1The Arthur and Sonia Labatt Brain Tumour Research Centre, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

2Program in Developmental and Stem Cell Biology, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto M5G 1X5, Canada.

4Department of Molecular Genetics, University of Toronto, 1 Kings College Circle, Toronto M5S 1A8, Canada.

5Department of Laboratory Medicine and Pathobiology, Faculty of Medicine, University of Toronto, Banting Institute, 100 College Street, Toronto M5G 1L5, Canada.

6Division of Neurosurgery, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

7Present address: Wellcome Trust Centre for Cell Biology and Institute of Cell Biology, School of Biological Sciences, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, United Kingdom

! *%!

3.3 SUMMARY Brain tumors are thought to be maintained by rare cancer stem cell-like cells (CSC) found within the tumor bulk24,188. In addition to exhibiting the cardinal properties of stem cells; self-renewal and multi-potential differentiation, these immature cancer cells share similar signaling signatures with normal neural stem cells (NSCs)25. Understanding the pathways regulating normal precursors may thus lead to the development of novel chemotherapeutics aimed at depleting CSCs. Using this premise, in Chapter 2 I performed a high throughput screen (HTS) on normal NSCs and recovered a large array of neuromodulatory agents that inhibited the in vitro expansion of both normal and cancer derived Nestin+Sox2+ precursors derived from human and mouse CNS tissue168. Interestingly, in Chapter 4 of this thesis, I also note through a compilation of retrospective data that patient populations chronically on neuromodulatory therapy experience a decrease in brain tumor incidence (p=0.01)169. Interestingly, the reduction of cancer incidence in patients chronically on neuromodulatory therapy seemed to be specific for the brain and not other types of tumors189. To further substantiate their potential use as chemotherapeutics, I investigated the in vivo efficacy of apomorphine (a dopamine agonist) and PAPP (a serotonin agonist) in a number of brain tumor animal models. My results indicate that consistent with my in vitro data, in situ brain tumor growth can be modulated via neuropharmacological perturbations. Chronic injections of both dopamine and serotonin agonists reduced brain tumor load in Ptc1+/- +ENU mice. Contradictory to this, administration of these agents to brain tumors derived in irradiated Ptc1+/- mice did not improve survival. Potential reasons for the failure of these agents to attenuate brain tumor growth in the Ptc1+/- + irradiation model are discussed. In addition to providing support for the potential use of clinically approved neurotransmission agents as novel and non-toxic chemotherapeutics, this data highlights the utility of in vitro HTS in cancer drug discovery.

3.4 INTRODUCTION Medulloblastoma is the most common malignant brain tumour in children and remains one of the leading causes of mortality in this age group. In addition to this, it remains the leading cause of morbidity in the pediatric population with survivors of brain

! *&! tumours often suffering serious developmental and intellectual impairment as a consequence of the non-specific toxicities of anti-cancer therapies149. I have previously demonstrated that brain tumors are maintained by a rare subpopulation of cancer cells with stem-like properties24,103. Unlike the majority of tumor cells, these cancer stem-like cells, that can be identified by their CD133 expression, retain the unique capacity to regrow a phenocopy of the original tumour when injected into brains of NOD-SCID mice. In great contrast, the non-cancer stem cell portion of brain tumours (CD133-), which usually makes up the majority of the tumour mass, do not retain the capacity to regenerate any tumour growth24. These results suggest that only therapies that target and efficiently ablate the cancer stem cell fraction of these tumors will prevent recurrence and lead to the development of novel and effective therapies. Molecularly, perturbations in hedgehog (Hh) signaling are found in approximately 30% of human medulloblastomas. Furthermore, Gorlin’s syndrome patients, who carry a germline mutations in the Hh receptor Patched-1 (Ptc1) have been found to be at increased risk for developing medulloblastomas149. This genetic link between the Ptc1 allele and brain cancer is maintained in mice heterozygous for Ptc1. These mice carry an increased risk of medulloblastomas of 10-30%, in comparison to the wildtype background in which medulloblastoma development is essentially absent144. In addition to carrying this increased baseline incidence of medulloblastoma growth, tumorigenesis in these mice can be further increased upon exposure to a number of environmental carcinogens. For example, when exposed to early postnatal irradiation, the frequency of meduloblastomas in these mice is augmented to nearly 100% incidence by 8-12 weeks, with a notable increase in the severity of disease166,190. Unpublished data from our laboratory also supports that N-ethyl-N-nitrosourea (ENU)-induced DNA damage during embryogenesis also increases (albeit to a lesser extent) the frequency and severity of tumors in these mice. Recently, Ward and colleagues188, along with others191 have shown that like the human disease, medulloblastomas formed in Ptc1 deficient mice contains subpopulations of cells that express the stem cell marker CD15. These cells display a neural precursors phenotype, are clonogenic, maintain a multi-lineage differentiation capacity, have elevated Hh signaling and have an enriched tumorigenic potential when orthotopically transplanted into mice. The low tumor-forming ability of cells found

! *'! within the tumor bulk also supports the idea that, like the human disease, these tumors are organized in a functional hierarchy and thus represent clinically relevant mouse models of cancer188. Given their cancer stem cell-like properties and the high tumor incidence rates that can be generated in Ptc1 mice when coupled with irradiation or ENU, in vivo assessment of the effects of various chemotherapeutic agents on medulloblastoma growth can be performed with relatively few mice. In Chapter 2, the results of an in vitro screen demonstrated that large arrays of neuromodulatory agents spanning various neurotransmitter classes have potent inhibitory effects on cultures enriched for brain cancer stem cells derived from both mouse and human samples168. These results suggest that some of these agents, which are already clinically approved for the management of various (4$-929B'A.2 diseases, can be rapidly redeployed as non-toxic chemotherapeutic agents. Here I attempt to address this prospect by investigating the anti-cancer effects of neuromodulatory agents among various in vivo mouse models of brain tumors. Specifically, I test to see if highly specific and potent agents identified to inhibit normal neural precursors in my initial HTS chemical screen, also affect the in vivo brain tumor growth in these mouse models.

3.5 RESULTS I first assessed the effects of various neuromodulatory agents on the survival of Ptc1+/- mice that were irradiated at day 0. Mice were treated at 4 weeks onwards and signs of brain tumors (i.e. ataxia and domed heads) began to appear as early as 55 days (~ 8 weeks) in some mice. Overall, 80% of Ptc1+/- irradiated mice developed lethal brain tumors, with the remaining living past 110 days and never developing CNS tumors. Between the different control and neuromodulation-treated groups, no differences in survival or brain cancer incidence were observed (Fig 3.1). All deaths in each group were confirmed to be tumor related, based on symptoms and histological assessment of the brains of mice at time of sacrifice. This suggested that pharmacological treatment of brain tumor derived in irradiated Ptc1+/- mice with neurotransmitter modulatory agents is not an effective therapy.

! *(!

Figure 3.1 | Effects of neuromodulators on irradiated Ptc1+/- mice Apomorphine and PAPP do not improve the survival of medulloblastomas related deaths in irradiated Ptc1+/- mice. Daily treatment with apomorphine or vehicle of Ptc1+/- (w 3 Gy of irradiation at P0) commenced after 4 postnatal weeks. At the first consistent signs of medulloblastoma, mice were sacrificed and age at time of death was used for survival analysis. No significant differences in survival were observed between the control and treatment groups.

! *)!

I then assessed the effectiveness of neuromodulators in the Ptc1+/- +ENU brain tumor model. The incidence of brain tumors in this model was observed to be lower than the irradiated model with approximately 65% of ENU-treated mice forming tumors. Initial signs of brain tumors in this model were also delayed and did not appear until approximately 154 post-natal days (22 weeks). At this time, all mice in this experiment were sacrificed (5 vehicle, 3 PAPP and 2 apomorphine treated mice). At time of sacrifice, four of the five control mice exhibited large tumors that were comparable in size to all previous attempts to form tumors using this method. Although the incidence of tumors in the treated groups was approximately the same (3/5 total: 2 PAPP and 1 apomorphine treated mice had visible tumors), all neuromodulatory treated mice showed substantially smaller tumors when compared to the controls in this experiment and others from previous untreated experimental groups (Fig 3.2 and Fig 3.3). Although the number of mice used in this experiment was too small to achieve statistical significance, these experiments suggests that in contrast to our previous attempts in irradiated Ptc1+/- mice, neuromodualtory treatment of Ptc1+/- +ENU induced brain tumors, substantially decreases the size and symptoms of these tumors.

! **!

Figure 3.2 | Effects of apomorphine on in vivo medulloblastoma growth The dopamine agonist apomorphine substantially decreases the size of medulloblastomas generated in Ptc1+/- +ENU mice. Mice were sacrificed at first sign of symptoms in either group. In this experiment, this was after 22 postnatal weeks. The brains of mice were dissected, fixed in PFA, and sectioned into 6 µm slices. Tumors were histologically identified using hematoxylin and eosin staining. Sections used for head-to-head comparisons between treated and control groups are representative sections taken from the approximate center off each tumor.

! "++!

Figure 3.3 | Effects of PAPP on in vivo medulloblastoma growth The serotonin agonist apomorphine substantially decreases the size of medulloblastomas generated in Ptc1+/- +ENU mice. Mice were sacrificed at first sign of symptoms in either group. In this experiment, this was after 18 postnatal weeks. The brains of mice were dissected, fixed in PFA, and sectioned into 6 µm slices. Tumors were histologically resolved using hematoxylin and eosin staining. Sections used for head-to-head comparisons between treated and control groups are representative sections taken from the approximate center of each tumor.

! "+"!

3.6 DISCUSSION Although these results are preliminary, they suggest contrasting conclusions regarding the effectiveness of neuromodulatroy agents at controlling the in vivo growth of brain tumors. My results indicate that daily treatment of four-week old irradiated Ptc1+/- mice does not alter the incidence or the progression of brain tumors. Interestingly, others have had remarkable success at attenuating brain tumors in this model with glutamate agonists192. These contrasting results may be derived from the different neuromodulatory agents used or in their drug treatment schedule that significantly deviated from ours. In contrast to commencing treatment at four weeks, the authors of the other study began treatment at postnatal day 0. Although this treatment regimen may not be clinically relevant, these earlier treatment points may have allowed for the agents to effectively target the tumor before it becomes too large to control with single neuromodulatory agents. In support of this interpretation, unpublished CT scanning of irradiated Ptc1+/- mouse brains shows that the tumors formed in the CNS can become quite large, even after 5 weeks. It is thus possible that in my experimental design, beginning treatment at 4 weeks may consequently result in treatment of a tumor that is too large to adequately control with my selected neuromodulatory agents. Although the alternate study can be criticized for commencing treatment too early to have any clinical relevance, it does support the utility of neuromodulatory drugs as novel anticancer agents. To theirs’ and my defense, such a mouse model that leads to the rapid generation of tumors in such a large proportion of mice may not be reflective of the human disease. Therefore, the failure of neurmodulatory agents in attenuating brain tumors during later time points (as in my experiment) may be something not shared in the slower growing tumors seen in humans. Further work to identify the time point which best represents the human disease and what (if any) human relevance this mouse model has is required before proper interpretations regarding the utility of these agents as anti-cancer therapeutics can be made. Furthermore, it is important to note differences in the signaling and regulation of the neurotransmitter receptors targeted by the agents used in this Chapter and those by Iacovelli and colleagues. As suggested in data found in Chapter 5, the down-stream effects of dopamine/serotonin may be dramatically different that those of glutamate

! "+#! signaling. Other reasons that could account for the discrepancy in the efficacy of these agents to regulate cancer growth may relate to how the expression of these pathways is regulated. For example, glutamate receptor activation are though to potentiate further glutamate receptor expression and activity in a positive feedback loop through Src- dependant pathways193. In contrast to this, may other G-protein dependant neurotransmitter pathways are though to have arrestin-mediated negative feedback loops194,195 that desensitize cells to these signals following chronic stimulation. This divergence in mechanisms may account for why repeated glutamate agonism may constitute a more effective treatment for regulating cancer growth when compared the stimulation of other G-protein dependant neurotransmitter pathways in the long run. In contrast to this, my results in the Ptc1+/- +ENU brain tumor model are preliminary but promising. All three mice with tumors in the neuromodulatory treated groups displayed a reduction in brain tumor size when compared to control mice. Additional support for a reduction in tumor growth stems from observations of tumor sizes in previous experiment with Ptc1+/- +ENU mice. All untreated mice in our records present with exceptionally large tumors after 20 weeks of in vivo growth. This suggests that the substantially smaller tumors seen in all of the neuromodulatory treated mice are likely a consequence of treatment rather than random variation. Unfortunately, due to the smaller than expected litter sizes and tumor incidence in this experiment, the analysis did not have the appropriate statistical power to allow for significance testing. Additional experiments with larger sample sizes are needed to properly address this hypothesis with statistically confidence. It is interesting to hypothesize why the two different models of brain tumors yielded such contrasting results. One explanation for these differences may be due to the differences in growth dynamics. Brain tumors in the Ptc1+/- +ENU model, where I show a possible reduction in tumor size following neuromodulatory treatment, seem to have a longer latency period (~20 wks) compared to that seen in the Ptc1+/- +irradiation model (~12 wks). Thus, there may be a minimal length of time for which treatments must be administered in order to be effective. Alternatively, treatments may need to be initiated when tumors are still within certain size constraints. This idea is supported by the fact

! "+$! that even in the irradiated model, effective reduction in both tumor incidence and size have been reported when treatment is initiated immediately after birth192. Even though these mouse models display some of the hallmarks of human derived cancers, the model (if any) that is most representative of the dynamics seen in the human disease needs to be determined before further conclusions can be made. Driving such a high incidence of cancer that appear relatively early in mice may discredit both models as accurately depicting the true progression of brain cancer in humans. These models should be tested against drugs in current clinical use to ensure that they yield similar (albeit non- curative results) to those seen in humans. This is needed before the evaluation of other agents can be made. In addition, since surgical removal of the tumor bulk is commonly performed in the clinical setting, mouse models in which treatments can be initiated when tumors are quite small192, analogous to post-surgery residual tumor load, may be a more accurate representation of the clinical situation. Regardless of these aforementioned model limitations, I feel that the preliminary in vivo data in the Ptc1+/- +ENU, when coupled with the in vitro data presented in the previous chapters, strengthens a neuromodulatory hypothesis of brain cancer. Currently, the generation of additional in vivo data to further address the statistical significance of these findings is currently an ongoing effort in the Dirks laboratory. Positive results from these studies will bring further support for the development of neurotransmitter-based anti-brain tumor therapeutics. Given the lack of effective treatments in brain tumor therapy, further positive results from these experiments will support the design of clinical trials to properly assess the efficacy of these treatments in the human disease.

3.7 METHODS Drug Preparation This study focused on the in vivo brain tumor effects of the pan-dopamine receptor agonist apomorphine and the serotonin 5-HT1A agonist PAPP. Apomorphine and PAPP were ordered from Sigma (USA) and dissolved into compatible solvents. Apomorphine was dissolved in sterile water. PAPP was dissolved in a mixture of 50% ethanol and 50% 0.1 M HCl. The in vivo concentrations used were 10 mg/kg for PAPP and 20 mg/kg apomorphine. These concentrations are known to be well tolerated by both mice and

! "+%! rats196,197. Based on the literature, these dosages do not have any known severe side effects. The volumes administered to achieve these concentrations were calculated weekly based on the weight of each mouse. An electronic bench top scale was used to measure the mass of each mouse and the injected volumes to achieve the desired drug concentrations ranged approximately between 75-150 µL depending on the mouse’s mass. All drug aliquots were kept at -20oC until they were ready for use and were stored for no longer than two months after they were made. Drug fidelity was also assessed by observing previously reported stereotypical behavior in injected mice196,197. Mice injected with apomorphine always displayed a momentary hyperactive state while PAPP administration caused a delayed but consistent reduction in locomotion.

Animals Ptc1+/- +Irradiation mouse model Ethical approval for all experiments was obtained from the Hospital for Sick Children’s Animal Care Committee. Ptc1+/- male and female mice (The Jackson Laboratories) were paired each night to undergo time mating. Successful mating (and potential pregnancies) were assessed the following morning, by resistance to vaginal probing on physical examination. This indicates the presence of male-derived seminal coagulates produced by fluids from both the vesicular and coagulating gland (plug) and is indicative of recent sexual activity. Female mice positive for this test (plug) were monitored daily for delivery of litters. To augment brain tumor incidence in this model, litters were subjected to irradiation (3 grays) immediately after birth. My previous data suggests that the irradiated Ptc1+/- mice yields brain tumors in almost all mice by 8 to 12 weeks. These frequencies and time of onset is comparable to previously published reports190. Following 4 weeks, irradiated mice were genotyped and mice that were confirmed to be heterozygous at the Ptc1 locus began daily treatment with either apomorphine or PAPP. Treatment continued for 16 weeks via intraperitoneal injections or until mice showed symptoms of tumors during daily scheduled monitoring. Upon signs of tumor-related or non-tumor related distress, mice were sacrificed in a CO2 chamber. Ataxia, domed head, paralysis, dehydration, hunched posture, weight loss, lethargy are all symptoms of brain tumors in these mouse models.

! "+&!

For this model, drug efficacy was determined using Kaplan-Meier survival analysis. Overall survival for this experiment was recorded using an “intention to treat” experimental design (Fig 3.4).

Ptc1+/- +ENU mouse model Ptc1+/- males and females were paired each night to undergo time mating. Successful matings (and potential pregnancies) were assessed the following morning, by resistance to vaginal probing on physical examination. Female mice positive for this test (plug) were marked by ear notches and noted to have fetuses at the E0.5 stage of development. Pregnant females were maintained for another 2 weeks (wk), upon which the mothers (harboring E14.5 fetuses) were injected (intraperitoneal) with 100 µL of RY4&/<2YRY ('&-9%9$-4. (ENU) (Sigma Aldrich, USA) dissolved in saline. Following delivery and 3 post-natal weeks, the resulting litters was genotyped and Ptc1+/- pups were followed until early adulthood (4 wk). At this point, mice were randomized to either a control (vehicle) or treatment (serotonin or dopamine agonist) group and treated at the doses described above. Drugs were administered daily for an additional 14 to 18 weeks. At this point, mice were sacrificed and tumors within the different groups were histologically assessed by size. Due to the smaller sample size of this experiment, I did not have sufficient statistical power to detect differences in survival. I thus opted to assess drug effects by comparing the size of tumors with and without treatment. Drug administration continued for 16 weeks via intraperitoneal injections or until first signs of mice showing any symptoms of tumours during daily monitoring of mice. Upon any signs of tumor-related or non-tumor related distress, all mice in the experiment were sacrificed in a CO2 chamber. Drug efficacy in these experiments was histologically assessed based on the size of the tumor growth. This was done by surgical removal of the brain, followed by fixation in 4% PFA and storage at 4OC (Fig 3.4). Histological analysis was then carried out as outlined in the “immunohistochemistry” portion of the methods section.

! "+'!

Figure 3.4 | In vivo experimental outline Outline of neuromodulatory treatment plan for medulloblastomas generated in Ptc1+/- +ENU mice. Pregnant Ptc1+/- females were injected with ENU at E14.5. The resulting litters were genotyped and Ptc1+/- pups were followed until early adulthood (4 wk). At this point, the mice were randomized to either a control (vehicle) or treatment (serotonin or dopamine agonist) group. Mice were then treated daily for an additional 14 to 18 weeks, at which point mouse tumors in the different groups were assessed by size. A similar strategy was used for the Ptc1+/- + Irradiated model, but overall survival rather than tumor size was used to evaluate the efficacy of each drug.

! "+(!

Immunohistochemistry Following surgical removal, and fixation in paraformaldehyde (4%), mouse CNS tissue was dehydrated through increasing gradients of alcohol. Following this step, brains were embedded in paraffin. Tissues was then sectioned to generate 6 µm slices and immobilized on slides. Tissues were allowed to dry and processed using a standard protocol. These were finally stained using eosin and hematoxylin and large unorganized hypernucleated regions observed with a light microscope were used to assess tumor size. To allow for fair comparisons, sections from each tumor were selected that represented the maximal surface area seen on all of the sectioned slides.

! "+)!

-CHAPTER 4- NEW DRUGS FOR BRAIN TUMORS? INSIGHTS FROM CHEMICAL PROBING OF NEURAL STEM CELLS 4.1 PROLOGUE

In the previous chapter, I tested if agents affecting neurotransmitter signaling could regulate the in vivo growth of brain cancer in various clinically-inspired model models. The results from these experiments varied between the different mouse models used and require further testing. Given the caveats that accompany mouse models of human disease (discussed in the previous chapter), and the wide use of many of these agents for the treatment of a variety of psychiatric disorders, I decided to look at cancer patterns within different human populations to determine if neuromodulators have anti- cancer effects. I conducted a retrospective epidemiological meta-analysis to address the hypothesis that patients receiving pharmacological therapy for psychiatric disorders would have altered brain tumor incidence rates. The results of this study are presented in this chapter. This analysis demonstrates that the reported brain cancer incidence rates in patients with a variety of mental disorders is in fact significantly reduced. In addition to this analysis, I provide explanations why this correlation may have been previously overlooked. The discussion also provides a review of the literature to speculate to a possible mechanism and additional evidence advocating a role for these agents in brain cancer stem cell biology. This study builds on ideas derived from the previous chapter and continues to lend support to the idea that neurotransmitter systems are involved in regulating brain cancer. The work described in this chapter was published in:

Diamandis, P., Sacher, A.G., Tyers, M. & Dirks, P.B. New drugs for brain tumors? Insights from chemical probing of neural stem cells. Med Hypotheses 72, 683-7 (2009). ! ! ! ! ! ! !

! "+*!

4.2 TITLE AND CONTRIBUTORS

! NEW DRUGS FOR BRAIN TUMORS? INSIGHTS FROM CHEMICAL PROBING OF NEURAL STEM CELLS

Phedias Diamandis1,2,3,4, Adrian G. Sacher1,2, Mike Tyers3,4,7* & Peter B. Dirks1,2,5,6*

1The Arthur and Sonia Labatt Brain Tumour Research Centre, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

2Program in Developmental and Stem Cell Biology, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto M5G 1X5, Canada.

4Department of Molecular Genetics, University of Toronto, 1 Kings College Circle, Toronto M5S 1A8, Canada.

5Department of Laboratory Medicine and Pathobiology, Faculty of Medicine, University of Toronto, Banting Institute, 100 College Street, Toronto M5G 1L5, Canada.

6Division of Neurosurgery, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto M5G 1X8, Canada.

7Present address: Wellcome Trust Centre for Cell Biology and Institute of Cell Biology, School of Biological Sciences, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, United Kingdom !

Published in Med Hypotheses 72, 683-7 (2009)

Reproduced with permission from the Elsevier

! ""+!

4.3 SUMMARY The cancer stem cell hypothesis posits a direct relationship between normal neural stem cells (NSCs) and brain tumour stem cells (BTSCs). New insights into human brain tumour biology and treatment should thus emerge from the study of normal NSCs. These parallels have recently been exploited in a chemical genetic screen that identified a broad repertoire of neurotransmission modulators as inhibitors of both NSC and BTSC expansion in vitro. Prompted by these findings, I sought epidemiological support for effects of neuromodulation of brain tumours in vivo. I present observations from data collected from retrospective clinical studies suggesting that patients with a wide variety of neuropsychiatric disorders have decreased brain tumour incidence. I speculate that this reduction may derive from the use of drugs that collaterally affect the normal neural precursor compartment, and thereby limit a population that is suspected to give rise to brain tumours. Standard chronic neuropharmacological interventions in clinical neuropsychiatric care are thus candidates for redeployment as brain cancer therapeutics.

4.4 INTRODUCTION High-grade gliomas represent at least one third of all primary brain tumours diagnosed in the adult population. Unfortunately, even with the progress made in radiation and chemotherapeutic regimens used following standard surgical resection, the median survival of these patients remains at 9-12 months41,42. In fact, in the last 40 years, the most major clinically significant milestone in glioma management, the DNA alkylating agent temozolomide, prolongs the median survival time from 12.1 to 14.6 months47. Given these unfavorable outcomes, and the so few if any documented examples of complete remission48, brain tumour treatment strategies must apparently shift away from traditional anti-neoplastic drug classes. Recent evidence suggests that brain tumours are maintained by rare cancer cells with stem cell-like properties24,105,132. Moreover, the discovery of stem cells in the postnatal brain suggests that normal neural precursors, that is stem cells (NSCs) or their close downstream progenitors not only may direct neuronal regeneration but that such cells may be the root cause of brain cancers. The inability of traditional therapeutics to eliminate rare brain tumour stem cells (BTSCs) may account for the frequent therapeutic

! """! failure and uniform clinical relapse8. It follows that the development of agents that act on BTSCs should afford more effective means to treat brain cancer. To date, few drug discovery assays have been developed that test targeting of stem cells, especially cancer stem cells. Indeed, my recent chemical genetic survey suggests that diverse neurotransmission pathways, traditionally thought to be only operational in differentiated neural cells, inhibit both normal and cancerous neural precursor cell expansion in vitro and thus opening up the possibility of their use as antineoplastic agents for human brain tumors168. To gain a clinical perspective on this hypothesis, I sought to determine whether patients diagnosed with a variety of neuropsychiatric disorders (and hence presumed to be on chronic neuromodulatory medication) exhibit different brain tumour incidence rates when compared to the general population.

4.5 METHODS AND RESULTS The analysis of historical cohorts has made it possible to identify correlations between many cancers and human behaviour; however, the relative rarity of brain cancer and the typically late-stage diagnosis hampers statistical analysis. Brain cancer is thus a disease with few known risk and preventative factors, including the potential association with extrinsic environmental modifiers, such as use of neuromodulatory drugs in neurologic or psychiatric treatment. Fortunately, a differential incidence of more prevalent cancers (such as breast, skin, and lung) among neuropsychiatric patients has prompted investigation of the relationships of these co-morbidities. Reports of brain tumour incidence in some of these studies allowed us to retrospectively assess correlations between neuropsychiatric diagnosis and brain cancer risk. I looked for recently published studies (year 2000 onwards) that reported patients with co-morbid neurologic or psychiatric conditions and cancer. From these, I found nine studies182,189,198- 204 that reported brain cancer incidence rates following an initial neurologic or psychiatric diagnosis. In my analysis of these studies, I make the important but reasonable assumption that the patient population is exposed to some form of chronic neuromodulatory pharmacologic therapy. This supposition is almost certainly valid for recent studies on Parkinson’s disease, schizophrenia and major depressive illness.

! ""#!

In perhaps the most notable study, the incidence of breast and other cancer types was examined among 144,364 subjects previously diagnosed with Parkinson’s disease (PD)182. Within this study was the unremarked-upon correlation that PD patients experienced a 5-fold reduction (~0.625% vs. ~0.125%; P<0.01) in the incidence of brain tumours, as compared to a control population. The continuous administration of anti- Parkinsonian drugs in this cohort might have decreased neural precursor and/or BTSC proliferation, thereby reducing the risk of brain cancer. Other studies that have followed brain tumour incidence in patient populations presumed to be treated with neuromodulatory drugs revealed less striking correlations (See “Reported SIR” in Table 4.1). For example, a number of studies have reported non-significant reductions in standardized incidence ratio (SIR) of brain tumours in schizophrenia patients198,200-203.

! ""$!

Table 4.1 | Standardized incidence ratios (SIR) of brain tumours in cohorts previously diagnosed with a variety of mental disorders

“Reported SIR” “Revised SIR” Study Year Disease (95% CI) (95% CI) † Lichtermann et al.203 2001 Schizophrenia 0.88 (0.62&1.20) 0.88 (0.62&1.20) Dalton et al.189 2002 Bipolar psychosis 0.82 (0.53&1.20) 0.64 (0.31&1.17) † Dalton et al.189 2002 Unipolar psychosis 1.19 (0.99&1.43) 0.99 (0.70&1.34) † Dalton et al.189 2002 Reactive Depression 1.20 (0.92&1.55) 0.73 (0.42&1.18) † Dalton et al.189 2002 Dysthymia 1.34 (1.05&1.68) 0.82 (0.52&1.23) † Lalonde et al.182 2003 Parkinson’s 0.20 (0.17 -0.23) 0.20 (0.17 -0.23) Carney et al.199 2004 Any mental (Male) 2.09 (1.22&3.59) - Carney et al.199 2004 Any mental (Female) 2.12 (1.40 &3.21) - Goldacre et al.201 2005 Schizophrenia 0.74 (0.29&1.53) 0.74 (0.29&1.53) Olsen et al.204 2005 Parkinson’s 1.32 (0.90&1.90) 0.85 (0.31&2.34) † Grinshpoon et al.202 2005 Schizophrenia (Male) 0.56 (0.32&0.81) 0.56 (0.32&0.81) Grinshpoon et al.202 2005 Schizophrenia (Female) 0.94 (0.62&1.27) 0.94 (0.62&1.27) Barak et al.198 2005 Schizophrenia 0.20 (0.00&1.09) 0.20 (0.00&1.09) Dalton et al.200 2005 Schizophrenia (Male) 0.74 (0.42&1.20) 0.74 (0.42&1.20) Dalton et al.200 2005 Schizophrenia (Female) 0.78 (0.44&1.26) 0.78 (0.44&1.26)

Diamandis et al. 2007 Combined 1.15( (1.01-1.30) * 0.80(¥ (0.67&0.95) ** ! †Qualifying brain cancer cases have been modified as noted by authors to only include cases more than 2 years after mental disorder diagnosis. (Excludes data from Lalonde et al (2003). Although supportive, the large sample size in this study would significantly skew the results of the analysis. ¥ Excludes data from Carney et al (2004) as authors did not reanalyze their data following the observed temporal discrepancy in their brain tumour SIR. *p = 0.04; **p = 0.01

! ""%!

In contrast, several studies report a positive correlation between neuropsychiatric illness and brain cancer incidence189,199,204. All of these studies, however, reveal a characteristic bimodal temporal pattern of brain tumour incidence. For example, in a cohort of patients with depression, a remarkably high SIR of 3.27 for brain tumour incidence was observed within the first year following diagnosis, but this value steadily decreased to 0.84 after 10 or more years from initial diagnosis189. A similar bimodal phenomenon has been reported in another study of Parkinson’s disease patients: despite an initial increase in brain tumour incidence (SIR=1.32), an inverse association was observed (SIR = 0.85) five or more years after diagnosis204. A further study uncovered an elevated overall brain tumour incidence rate (SIR>2.00) in patients with mental illnesses, with a median time to subsequent brain tumour diagnosis of only 18 months199. This apparent initial increase in brain tumour incidence in psychiatric patients might, however, derive from occult pre-existing tumours that are responsible for the presenting psychiatric symptoms. An early disproportionate increase in brain tumour incidence in such patients, therefore, would be entirely expected; this increase may obscure any potential underlying relationship between neuropsychiatric disease, its treatment, and brain tumour incidence. To account for this possible masking effect, I re-examined previous studies after exclusion of brain tumour that closely followed on psychiatric diagnosis (I excluded cancers diagnosed <2 years from initial hospitalization). This analysis uncovered a consistent negative correlation between psychiatric diagnosis and cancer incidence (See “Revised SIR” in Table 4.1). Although the low number of patients in each individual study precludes statistical significance, the combined data reveals a significant decrease in brain tumour incidence in patients with psychiatric disorders (SIR=0.8, P=0.01, computed using Comprehensive Meta-Analysis software). Interestingly, the reduction of cancer incidence reported in some of these patient populations that are chronically on neuromodulatory therapy seems to be specific for the brain and not other organ sites189. Although I do not attempt to account for any potential cofounding variables such as differences in the surveillance or autopsy rates in the psychiatric population, I feel that these observed correlations, when coupled with the observed effects of psychiatric drugs on NSC/BTSC populations, intriguingly suggest that neuromodulators in common clinical practice may have protective effects against brain cancer through suppression of

! ""&!

NSC/BTSC proliferation168. Another important but unavoidable limitation of the data reviewed in this essay was the need to consolidate the SIRs of potentially different brain tumor subtypes from patients diagnosed with, and treated, for a variety of different neurological disorders. Although stratification of patients into different cohorts that better define these epidemiological differences may help refine which specific brain tumor subtypes and drug classes are most amenable to this therapeutic strategy, the rarity of CNS cancers (226 over the 9 studies I present) makes such further detailed insight not readily forthcoming. Therefore, although intriguing, these observations should be seen as hypothesis-generating for future epidemiological work rather an effort to lobby for a change in the current standard of care for primary brain tumors. Well-controlled prospective epidemiological studies that correlate the incidence of brain tumour sub-types with specifically defined drug regimes of patients suffering from diverse neuropsychiatric disorders will be required to rigorously address and confirm this hypothesis. Obtaining sufficient sample sizes for such detailed analyses presents a large practical obstacle and will likely require committed long term group efforts to properly elucidate even a strong association between brain tumor incidence and specific neuropharmacological medication.

4.6 DISCUSSION The current understanding of cancer, stem cells and cancer stem cell biology provides some additional mechanistic insight into the above epidemiological observations. Brain cancers are typically heterogeneous in nature and consist of cells from a variety of committed neuronal and glial cell lineages in conjunction with the more rare primitive cells that appear to drive tumor growth. Cancer initiation may thus occur in immature multipotent cells of the central nervous system (CNS)24. A number of molecular and functional similarities between BTSCs and NSCs support this “neural precursor” oncogenesis model105,132. Classic carcinogenic134, oncogenic virus135 and genetic mouse models suggest that the subventricular stem/early precursor cell compartment is the likely site of disease initiation. Notably, intraventricular infusion of the malignant glioma associated ligand epidermal growth factor (EGF)205,206 leads to the rapid expansion of the adjacent subventricular zone (SVZ) precursor pool, aberrant migration of the cells away from stereotypical pathways in the cerebral cortex, and

! ""'! formation of hyperplastic lesions that protrude into the ventricles207. Furthermore, mice with deletions in the p53208and p16INK4a129 genes, which normally restrain cell proliferation and are commonly lost in gliomas209, have larger and more dense SVZ precursor pools. Lastly, glioblastomas arising in mice deficient for p53 and NF1 in neural precursors also have all early lesions present in the SVZ210. Chemical agents that reduce or limit the expansion of neural precursor regions may therefore also attenuate the initiation and/or expansion of the brain malignancies (Fig. 4.1).

! ""(!

Figure 4.1 | Neural precursor compartments and brain cancer The mature adult brain contains remnant populations of undifferentiated NSCs in the dentate gyrus and subventricular zone (center panel). i. Agents leading to the expansion of this compartment (e.g., EGF, oncogenic viruses) cause cancer-like lesions in the brain. ii. Factors reducing the size of this compartment may thus reduce the incidence of brain cancer.

! "")!

The concept that stem cells underlie neurogenesis in the adult brain marked a paradigm shift in developmental neurobiology51,211. In response to a wide variety of molecular, behavioral and environmental cues, quiescent NSCs of the adult SVZ and hippocampal dentate gyrus are stimulated to divide, migrate and become newly integrated and functional post-mitotic neurons, illustrating the dynamic responsiveness of the brain’s stem cell compartments167. Interestingly, these anatomical compartments overlap with regions collaterally innervated by multiple neurotransmitter systems, further suggesting that neuromodulatory signals may play a vital role in the governance of stem cell fate in the adult brain167. For example, uncommitted hippocampal NSCs express functional L-type calcium channels and respond to excitatory NMDA signals by increasing intracellular calcium levels that promote the expression of neuronal differentiation genes and subsequent neurogenesis212 (Fig. 4.2a). A wide variety of other neurmodulatory signals are also implicated in the regulation of neurogenesis, extending this phenomenon to include the dopaminergic213, cholinergic214, GABAergic215,216, serotonergic217,218 and glutamineric212 neurotransmitter systems (Fig 4.2b). Ectopic changes in these neurotransmitter levels have been shown to stimulate the quiescent NSC compartment, followed by subsequent differentiation of these cells towards neuronal lineages. These neurogenic events occur at clinically relevant dosages of neuroactive agents213 and, moreover, changes at the cellular level precede the known clinical benefits of antidepressants218. Analysis of human Alzheimer’s disease specimens also support this neurogenic model of drug action219, although the primary effects on NSC pools in neurodegenerative diseases may be independent of drug administration220,221. All told, these studies demonstrate that neuromodulatory agents can regulate NSC-containing compartments (Fig. 4.2c). The appropriate maintenance of these compartments is thought to rely on the correct balance of symmetric and asymmetric stem cell divisions, which may be altered in favour of excess symmetric stem cell divisions in cancer222-224 (Fig. 4.3i). Alternatively, chronic administration of neuromodulatory agents may deplete NSC pools in the neuropsychiatric patient and thereby protecting against cancer initiation (Fig. 4.3ii). In support of this, in other experimental cancer systems the size of the stem cell pool is directly related to cancer incidence225.

! ""*!

Figure 4.2 | Neuromodulatory effects on neural precursors (a) Like mature glutaminergic neurons, NSCs have been shown to express and respond to NMDA excitatory stimuli, resulting in the expression of differentiation programs and the generation of mature functional neurons. (b) Neurotransmitter-induced neurogenesis also occurs following dopaminergic, GABAergic, serotonergic and cholinergic stimuli. (c) The initiation of neuronal-specific differentiation of neural precursors occurs at clinically relevant dosages of neuromodulators (e.g., haloperidol) in mouse models. (d) Like their normal NSC counterparts, the promiscuous expression and responsiveness to neurotransmission are also conserved in cancer-derived neural precursor populations (BTSC, brain tumor stem cells)

! "#+!

Figure 4.3 | NSC division symmetry and brain cancer NSC compartments in the adult brain are maintained by symmetric and asymmetric cell divisions. i. a bias towards symmetric divisions (e.g., increased EGF signalling) may expand these pools and lead to cancer. ii. Signals favouring precursor differentiation (e.g., bone morphogenic protein, neuromodulators), may lead to the depletion of the self- renewing population and suppression of brain cancer initiation.

! "#"!

Interestingly, mutations in neurotransmission pathways including sodium, potassium, and calcium channels have recently been shown to be a common occurrence in gliomas226. These include mutations in all five of the major neurotransmission classes, serotonin, dopamine, acetylcholine, glutamate and GABA. Intriguingly, this analysis suggest that neuromodulatory mutations could represent driver mutations (that confer selective advantage) rather than just representing passenger genetic alteration (that do not effect net tumor growth)226. Furthermore, unlike other mutations (i.e. PTEN, TP53, RB1) discovered in the analysis commonly found mutated in other cancers (i.e. pancreases, breast, colon) mutations in neurotransmission pathways appeared to be a feature unique to gliomas226. These mutually exclusive cancer-driving mutations may thus be a consequence of the unique mechanism these pathways play in regulating neural precursor pool size and tendency to differentiate. In addition to this study, chromosomal regions like 17p13.3 commonly lost in malignant astrocytomas227, whose pathophysiological mechanism continues to remains elusive, intriguingly contain genes for the receptors of neurotransmission receptors including (TRPV1 and TRPV3 in the case of 17p13.3). Such perturbations in basal neurotransmission inputs from the surrounding neural stem cell niche may be enough to change the dynamics of the NSC compartment in favour of a more hyper-proliferative and hypo-neurgenic state. Activation of normal lineage-specific signaling pathways in neural precursor derived cancer cells may provide a means of depleting the undifferentiated pool of cells that initiate and maintain brain tumor growth (Fig. 4.2c,d). Indeed, it has recently been demonstrated that that pro-differentiating agents such as bone morphogenic proteins (BMPs)154 can promote the differentiation of BTSCs and prolong the life of mice xenotransplanted with human tumors155. Moreover, in vivo neuromodulation of the metabotropic glutamate receptor 4 (mGlu4); a candidate target for the treatment of both Parkinson’s disease228,229 and generalized anxiety disorders230, by the mGlu4 potentiator PHCCC has also been shown to attenuate medulloblastoma formation in mice192. Similarly, mGlu2/3 antagonism also suppresses both the in vitro and in vivo growth of human glioma cells231,232. The generality of this phenomenon is suggested by the finding that a broad spectrum of drugs that act on diverse neurotransmission pathways inhibit both normal and cancerous neural precursor cell proliferation in vitro168 (Fig. 4.2d).

! "##!

Notably, the most potent anti-BTSC agents identified in these in vitro screens included the clinically prescribed dopamine agonist apomorphine233 and the glutamate antagonist ifenprodil234. The cohort of well-tolerated neuropharmacological agents used in standard clinical practice offer the prospect of rapid redeployment following clinical trials in patients with brain cancer.

! "#$!

-CHAPTER 5- NEURAL STEM CELL POPULATIONS PHASE VARY THROUGH EQUILIBRATING STATES OF DISTINCT NEUROTRANSMITTER PATHWAY GENE EXPRESSION

5.1 PROLOGUE

In this chapter, I looked to better phenotypically and functionally characterize the neurotransmitter pathway gene expression landscape of purified populations of human NSCs. I show that hNSCs devoid of mature neuronal marker expression express a variety of different neurotransmitter pathway genes at the mRNA and protein level. When examining the mRNA expression at the single cell level, I find that neurotransmitter pathway genes are heterogeneous expressed throughout the culture. I show that this heterogeneity can be re-derived from single cell colonies at equilibrating frequencies to those seen to those seen in the parental population. Prospectively, I show that this reversibility is independent on the neurotransmitter status of the cells. Lastly, I demonstrate that this heterogeneity limits the response of the population to exogenous cues and may act as a protective mechanism to maintain the NSC compartment throughout development and adult life. The majority of this work described in this chapter is currently in an unpublished manuscript format:

"#$%$&'#()! *+"!#$"! +*"!#$%&'("!)*"!,-./.0"!+*"!12.-34"!5*6*"!,-./.0"!7*"!89"!7*"!!:44"!:*"! ;<4-%"! =*"! >! 6'-3%"! ?*@*"! Neural stem cell populations phase vary through functionally heterogeneous and equilibrating states of neurotransmitter pathway gene expression. (Under Review 2009) !

! "#%!

5.2 TITLE AND CONTRIBUTORS

NEURAL STEM CELL POPULATIONS PHASE VARY THROUGH EQUILIBRATING STATES OF DISTINCT NEUROTRANSMITTER PATHWAY GENE EXPRESSION

Phedias Diamandis1,2,3,4, Ryan Austin1,2, Maria Cusimano1,2, Kelvin Au1,2 Kevin

Graham1,2, Ian D. Clarke1,2, Jeremy Graham1,2, Jenny Ho1,2, Lilian Lee1,2, Mike

Tyers3,4,5, and Peter B. Dirks1,2,6,7

1The Arthur and Sonia Labatt Brain Tumor Research Centre, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto, M5G 1X8, Canada

2Program in Developmental and Stem Cell Biology, The Hospital for Sick Children and University of Toronto, Canada

3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, M5G 1X5, Canada

4Department of Molecular Genetics, University of Toronto, 1 Kings College Circle, Toronto, M5S 1A8, Canada

5Wellcome Trust Centre for Cell Biology, School of Biological Sciences, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, United Kingdom

6Department of Laboratory Medicine & Pathobiology, Faculty of Medicine, University of Toronto, Banting Institute, 100 College Street, Toronto, M5G 1L5, Canada.

7Division of Neurosurgery, The Hospital for Sick Children and University of Toronto, Canada

! "#&!

5.3 SUMMARY & INTRODUCTION The brain is bathed in neurochemicals that govern neuronal function and communication between neurons and supporting glia. Recent evidence suggests that neurotransmitters regulate multipotent neural stem cell (NSC) function during development215, adulthood167,213,216 and multiple disease states218,221. Here, I show that phenotypically uniform populations of human multipotent NSCs express low levels of genes spanning the major neurotransmitter (NT) systems. Single cell RT-PCR analysis revealed heterogeneous NT patterns throughout the culture and that these patterns could be re- established from single cells. Prospectively isolated a7nAChR+ and a7nAChR- hNSC populations recreated heterogeneous NT-positive populations, suggesting that hNSCs demonstrate reversible and phase varying patterns of NT expression that are intrinsically encoded in the lineage. Neurotransmitter simulation of hNSC pools demonstrated that only restricted subpopulations are capable of responding to specific neurochemcial cues. These results suggest that stochastic sampling of different neurochemical states in NSCs may temporally and spatially control fate decisions in response to extrinsic cues, consistent with the general concept of “lineage priming” in hNSCs. Thus, the brain may exert electrochemical control of its own neural precursor compartment; heterogeneity at the single cell level may impose limitations on stem cell functional responses to local environmental cues or therapeutic manipulation, or may enable a readiness of specific subpopulations to simultaneously respond to a diversity of extrinsic signals.

5.4 RESULTS I profiled the expression of different NT genes in multipotent human neural precursors, derived as adherent NSC cultures from primary human fetal CNS tissue235 (Fig 5.1a). These cultures represent homogeneous symmetrically self-renewing multipotent populations of nestin+ and Sox2+ cells that lack detectable expression of both early and mature neuronal markers !III-tubulin and NeuN, and are thus devoid of differentiated cells (Fig 5.1b-d). RT-PCR analysis of these bulk cultures demonstrated expression of a wide variety of neurotransmitter pathway genes typically associated with many differentiated neuronal subtypes (Fig 5.2a). In vivo, these pathways are expressed

! "#'! in neurogenic regions of the brain, receive inputs from multiple neuronal afferents, and have been implicated in regulating neurogenesis167. To demonstrate that the detected NT transcripts produced expected full-length protein products, as opposed to abortive expression of 5' regions236,237, I used western blot analysis to detect proteins encoded by the NT genes DRD2 and GluR2 (Fig 5.2b). Consistent with low-level expression, I was unable to detect these neuronal markers by immunofluorescence analysis with available antibodies (data not shown). These results suggest that, albeit at low levels, lineage specific NT-related genes are translated into protein products in NSC pools. ! !

! "#(!

!

! !

Figure 5.1 | Phenotypically uniform populations of hNSC express functional neurotransmitter pathways (a) Different neurotransmitter pathways implicated in NSC biology. (b) The innate phenotypic heterogeneity of primary neurosphere cultures (neuronal marker TujIII expression is shown in red, GFAP expression is shown green) (c,d) Adherent cultures of fetal hNSCs homogenously express the neural stem cell markers Nestin and Sox2 and are completely devoid of the mature neuronal markers NeuN and TujIII.

! "#)!

Figure 5.2 | Phenotypically uniform populations of hNSC express functional neurotransmitter pathways (a) Phenotypically uniform populations of undifferentiated hNSC express genes associated with mature CNS NT pathways. (DRD2: Dopamine receptor D2; TH: Tyrosine hydroxylase; TPH1: Tryptophan hydroxylase 1; CHRM3: Muscarinic acetylcholine receptor M3; CHRNA7: Cholinergic receptor, nicotinic, alpha 7; CHRNA9: Cholinergic receptor, nicotinic, alpha 9; GRIN2B: Glutamate receptor, ionotropic, N-methyl D-aspartate 2B; GRIA1: Glutamate receptor, ionotropic, AMPA 1; GABRB1: Gamma-aminobutyric acid (GABA) A receptor, beta 1; GABRB2: Gamma- aminobutyric acid (GABA) A receptor, beta 2; GAD67: Glutamate decarboxylase 67).

Housekeeping genes used for positive control: B2M:! ]2 microglobulin; PBGD: hydroxymethylbilane synthase. (b) Western blot analysis of DRD2 and GluR2 (Glutamate receptor, metabotropic 2) protein expression in hNSC cultures. The sizes of the prominent bands shown in these blots is consistent with those seen in positive controls (whole mouse brain extracts, not shown in this blot) and have been confirmed by others to represent the DRD2238,239/GluR2240,241 proteins using these antibodies.

! "#*!

I have previously shown that growth of bulk populations of mouse neural precursor-derived neurospheres cultures can be modulated by perturbations in NT pathways168. I thus investigated if the low levels of NT proteins present in more highly purified adherent hNSC populations were functional. I monitored the induction of early response genes by acetylcholine as a surrogate readout for the activation of cholinergic receptors and their downstream GPCRs. Brief acetylcholine stimulation of hNSCs led to activation of specific early response genes known to be involved in the co-ordination of differentiation and maturation of neuronal cells242-245 (Fig 5.3a,b). Importantly, this activation could be completely blocked by pretreatment with the muscarinic cholinergic antagonist atropine, suggesting this induction was a direct consequence of cholinergic (muscarinic) receptor activation. Given the intracellular convergence of many of these neurotransmitter pathways246, I tested and confirmed this activation of early response genes is mimicked by other neurotransmitters (Fig 5.3b). I find that this activation could be efficiently blocked chelerythrine chloride, suggesting that many of these receptor signal through a PKC dependant process (Fig 5.3c). I also observe activation of the neurogenic factor BNDF shortly following the induction of these early response genes and suggest how these pathways may orchestrate cell fate decisions in these cells (Fig 5.3d). Lastly, I also detected a time-dependant increase in g-aminobutyric acid (GABA) in conditioned media of hNSC cells by mass spectrometry, consistent with the presence of functional glutamate decarboxylase (GAD67) protein (Fig. 5.4). Taken together, these results suggest that functional neurotransmitter machinery resides within ostensibly homogeneous hNSC cultures.

! "$+!

!

!""# $""# #!!!!$ ;!$ %37#$ #!!!$ #:$ %37;$ #!!$ #!$ %37C$ #!$ :$ %&'()*+,$$-./,$01)23+$ !$ #$

<=>?$ 9@1$ 9@1$A$ %&'()*+,$-./,$01)23+$42$5677#$$$ 9*7.B42+$ !"#$ %""# &""# #F$ #!!!!$

#;$ #!!!$

#!!$ E$

#!$ G<5H$(859$ D$ 5677#$(859$ %&'()*+,$01)23+$42$ #$ !$

%&'()*+,$-./,$01)23+$42$ <=>?$ 901$ !"#$

Figure 5.3 | Phenotypically uniform populations of hNSC express functional neurotransmitter pathways (a,b) Neurotransmitter stimulation of hNSC cultures leads to activation of (a) early growth response (Egr) and (b) Nuclear receptor related 1 (Nurr1) genes as assessed by quantitative RT-PCR. (a) Brief acetylcholine stimulation (30 mM, 45 min) of hNSC cultures leads to increased Egr gene expression as assessed by quantitative RT-PCR. This acetylcholine-dependant gene expression was completely abolished by pre-incubation (30 mM, 45 min) with the muscarinic acetylcholine receptor antagonist atropine. (b) The activation of Nurr1 by a variety of different neurotransmitters. Although dopamine was unable to activate Nurr1 levels, the more stable DRD2 agonist R(-)- Propylnorapomorphine (NPA) did. (c) Neurotranmitter induction of Nurr1 levels can be blocked by the PKC inhibitor chelerythrine chloride. (d) Acetylcholine stimulation (30 mM, 3 hr) of hNSC cultures leads to activation of BDNF as assessed by quantitative RT- PCR. Standard deviations for a representative experiment are shown.

! "$"!

Figure 5.4 | Functional correlates of GAD67 protein expression in hNSC Time dependant detection of GABA in condition media of hNSC cultures as assessed by mass spectrometry. Each of the two lines represents replicates preformed in independent experiments.

! "$#!

Other phenotypically homogenous populations of stem cells have been reported to express genes and their protein products heterogeneously throughout the culture247-249. To assess the prospect of heterogeneity in the distribution patterns of neurotransmitter pathway genes in NSCs, I next assessed the presence of NT mRNAs in single cells (see Fig 5.5 for an overview of the method)250. Analysis of live sorted single Nestin+Sox2+TujIII-NeuN- hNSCs revealed that neurotransmitter-related transcript expression is heterogeneous (Fig. 5.6a-c). For example, the expression of the GABA-A receptor, beta 2 (GABRB2) was expressed at detectable levels in 58% of cells (Fig. 5.6b); in contrast, approximately only 1% of cells expressed DRD2 (Fig. 5.6c). To ensure that the heterogeneous detection of transcripts was not due to inadequate sensitivity, I applied RT-PCR, to a limiting dilution of NSCs and estimated the frequency of DRD2+ cells across different cell densities by fitting to a Poisson distribution (Fig. 5.6d and Fig 5.7). This analysis predicted a similar frequency (1%) of DRD2 positive cells in culture as compared to single cell analysis, suggesting that the frequency estimates were independent of both input cell number and mRNA abundance. These findings demonstrate that different NT genes are expressed in NSC cultures in a heterogeneous cell-by-cell fashion.

! "$$!

Figure 5.5 | Multiplex RT-PCR assay for assessing mRNA expression in single hNSCs Single live hNSCs were deposited into 96-cell PCR plates using FACS. Cells were then lysed and gene specific primers were used to transcribe mRNA species of interest in a reverse transcriptase reaction. The completed cDNA product was then taken through a multiplex PCR amplification step. To resolve expression of each specific mRNA species, replica plates were made and each subjected through a unique gene-specific nested PCR step. The presence and absence of transcript was assessed by running the product on a 2% agarose gel. See methods for further details.

! "$%!

!

Figure 5.6 | hNSCs heterogeneously express NT pathways genes in an equilibrium state that can be regenerated from single cell clones (a-c) Single cell RT-PCR analysis of hNSC for the housekeeping gene (B2M), and the NT genes GABRB2 and DRD2. As a negative control, wells A1-A4 contain 0 cells; as a positive control, wells H9-H12 each contained 100 cells. All other wells contained a single sorted hNSC. (a) B2M is homogenously expressed in cultures of hNSC. (b,c) GABRB1 and DRD2 exhibit heterogeneous single cell expression in hNSC cultures. (c) Limiting dilution analysis of DRD2 expression in 128 hNSC cells down to a single sorted hNSC cell (12 replicates). (d,e) Bulk hNSC culture heterogeneity may be derived from (d) heterogeneous cultures of restricted self-renewing progenitors or (e) an intrinsic single cell-dependant mechanisms.

! "$&!

Figure 5.7 | Limiting dilution RT-PCR assay In a similar manner to the outlined single cell RT-PCR approach, FACS was used to deposit 12 replicates of diluting cell numbers (1-128 cell/row) on a 96-well PCR plate. Cells were then lysed and ran through the same nested RT-PCR steps previously described. Once resolved on a gel, the percentage of negative reactions at each cell density was plotted and fitted to a Poisson distribution to estimate the frequency of cells

! "$'! expressing particular transcripts in the population. This assay was used to estimate frequencies of cell expressing very rare transcripts (~1% of cells) since it samples more cells per plate when compared to the single cell approach shown above (2816 cell/plate vs. the 96 cells/plate).

! "$(!

Recent evidence suggests that some previously presumed sources of adult mouse NSCs actually represent a mosaic collection of spatially restricted lineage restricted self- renewing progenitors251 (Fig. 5.6e,f). To assess if such mosaicism could potentially explain the observed heterogeneity in NT gene expression, I sorted live single cells from bulk cultures and expanded these cells into clones. I then assessed the mRNA and protein expression patterns of different clones across a battery of different lineage-specific NT pathways (Fig. 5.7). Given the low prevalence of cells expressing DRD2 mRNA transcripts in the parental population, progenitor heterogeneity should yield only a rare proportion of colonies that express DRD2 at the mRNA and protein level (Fig 5.7e,f). In striking contrast to this prediction, all tested colonies had detectable DRD2 protein expression (Fig. 5.7a). Furthermore, the colonies isolated from these cultures recreated the diversity of NT genes expressed in bulk NSC cultures and recapitulated the relative expression frequency of each NT gene at the single cell level when approximated by tandem 25 cells multiplex RT-PCR reactions (Fig. 5.7b and Methods). These findings indicate that the observed heterogeneity in NT expression is not a consequence of mixed lineage-restricted progenitor cultures and instead is intrinsic to all self-renewing NSCs found in these cultures. The reproducibility in the relative expression frequencies I observe also suggests that NSCs are able to re-equilibrate to a stable frequency pattern at the population level.

! "$)!

Figure 5.8 | hNSCs heterogeneously express NT pathways genes in an equilibrium state that can be regenerated from single cell clones (a) Expanded single cell clones exhibit expression of DRD2 protein by western blot analysis. (b) Determination of heterogenous mRNA expression by tandem RT-PCR analysis on small hNSC cell populations; colonies derived from single cells regenerated a similar pattern of mRNA heterogeneity as the originating parental bulk populations. !

! "$*!

I then addressed whether the maintenance of heterogeneous NT-related gene and protein expression in neural precursors arises from either a classical hierarchical cascade of gene expression or stochastic fluctuation between different metastable states (Fig 5.9). As I was unable to identify sufficiently sensitive antibodies to extracellular domains of NT receptors (data not shown), I sorted hNSC populations using fluorescently labeled (FITC)-)-Bungarotoxin (BTX), a snake venom toxin that specifically binds with high affinity to the a7 subunit of the nicotinic acetylcholine receptor (nAChR)252. Individual hNSC cells that expressed a7nAChR were identified by FACS as those with signals above background levels observed in non-neuronal cell lines; this FITC-BTX reactivity was outcompeted by known chemical inhibitors of the BTX-a7nAChR interaction (Fig 5.10 and Fig 5.11)252. Consistent with the above observations (Fig 5.8), only a rare population of cells within the culture specifically stained for FITC-BTX. Following isolation and a 14-day expansion period in vitro, both positive and negative a7nAChR subpopulations were able to regenerate the original heterogeneous pattern observed in the parental NSC population (Fig 5.12). In line with my previous findings168, I observed a slight reduction in proliferation rate of a7nAChR positive populations; however, this population readily recovered and expanded to yield cultures that were morphologically indistinguishable from the other populations (data not shown). These results indicate that a NSC subpopulation defined by expression of a specific NT pathway can reversibly and spontaneously transit between different states. Importantly, the observed dynamics of NT gene distribution cannot be explained by a hierarchical model of heterogeneity.

! "%+!

Figure 5.9 | Heterogeneous patterns of NT protein expression are reversible and re- equilibrate (a,b) Clonally-derived pattern of heterogeneous gene expression can be explained by (a) classic hierarchical models of stem cell heterogeneity or (b) equilibration through random phase variation.

! "%"!

Figure 5.10 | FITC-"-Bungarotoxin selectively labels a small subpopulation of hNSCs (a) Titration of FITC-"-Bungarotoxin in hNSC culture. (b) Titration of FITC-"- Bungarotoxin in a non-neuronal fibroblast cells line.

! "%#!

Figure 5.11 | "-Bungarotoxin labels a specific subpopulation of "7nAChR positive hNSCs The binding efficiency of FITC-"-Bungarotoxin by flow cytometry in the presence and absence of chemicals known to compete with "-Bungarotoxin. The specific "7nAChR antagonist MG-624 that maps to the same binding site as "-Bungarotoxin substantially reduced the subpopulation of cells positive for FITC-"-Bungarotoxin. A substantial reduction in the positive subpopulation was also noted by pre-incubation with both nicotine and acetylcholine.

! "%$!

Figure 5.12 | Heterogeneous patterns of NT protein expression are reversible and re- equilibrate hNSC populations with distinct! "7nAChR status were prospectively sorted using the nicotinic acetylcholine receptor antagonist (FITC)-^-Bungarotoxin and allowed to re- equilibrate over ~5-7 doubling periods. All sorted populations had a high degree of purity and the ability to self-renew for extended periods of time. !

! "%%!

The variable NT gene expression pattern among hNSCs suggests that these populations might exhibit a heterogeneous response to NT stimulation. I tested if the observed induction of early response genes in NSCs following acetylcholine stimulation (Fig 5.3) was restricted to only a discrete fraction of the population using limiting dilution RT-PCR. Tandem analysis of small NSC populations revealed that the induction of Nurr1, which responds to acetylcholine receptor activation, occurred in only ~1% of the cell population of cells upon acetylcholine stimulation (Fig 5.13 and Fig 5.14). This heterogeneous response was not due a technical limitation of my detection method, as shown uniform detection of Nurr1 expression after mixing of different mRNA fractions (Fig 5.15a,b). Despite the fact that Nurr1 responded in only a small subpopulation of cells, all expanded single cell-derived colonies acquired the capacity to respond to acetylcholine treatment (Fig 5.16), consistent with the above results (Fig. 5.8). Also consistent with individual cells sampling through differentially responsive neurotransmitter cell states is the observation of a relatively large differentiation response (>1%) following chronic (10 day) acetylcholine treatment (Fig 5.17). Taken together, these findings suggest that the heterogeneous gene expression states I observe in hNSC pools are functional and vary between different states in time, such that overall pool distributions can be readily regenerated from any specific sub-pool. !

! "%&! Figure 4

!"#

$"#

Nurr1 0.03%

DMSO64 DRD2 0.83% 2hr PBGD N/A

Nurr1 1.18%# ACH 64 * (30 uM) DRD2 0.82% 2hr PBGD N/A *N.S. ; # p=0.008

Figure 5.13 | hNSCs populations heterogeneously activate gene expression following NT receptor stimulation (a) Limiting cell dilution RT-PCR assay of Nurr1 expression in hNSC cultures following acetylcholine stimulation. (b) Expression analysis of Nurr1, and the loading controls DRD2 and PBGD, in the 64-cell data point (12 replicates) from the dilution curve shown in (a).

! "%'!

Figure 5.14 | Time course of early response gene induction by acetylcholine in hNSCs (a-d) Fold change of different early response genes in the presence and absence of acetylcholine. Changes in gene expression were monitored over 24 hours of acetylcholine stimulation. The induction in all genes seen at 1-2 hours could be abolished by atropine when assessed at 45 minutes (not shown). It is noteworthy that for all genes assayed,

! "%(!

Nurr1 levels increased most substantially following acetylcholine stimulation. The induction of Nurr1 after 2 hours was thus used in future experiments to assess if cells respond heterogeneously/homogenously to acetylcholine (Figure 3) as it represents my most robust conditions of neurotransmitter induced gene activation.

! "%)!

Figure 5.15 | hNSCs populations heterogeneously activate gene expression following NT receptor stimulation (a) Schematic of mixing control for RT-PCR assay sensitivity. (b) Expression analysis of Nurr1 by RT-PCR in unmixed and mixed samples from acetylcholine treated cultures.

! "%*!

Figure 5.16 | Muscarinic Acetylcholine agonists activate Egr1 in NSCs pools An array of different acetylcholine agonists were used to assess the competency of both bulk and clonally formed hNSCs lines to respond to acetylcholine receptor stimulation. Acetylcholine and various muscarinic agonist could activate Egr-1 in both bulk and all clonally derived cultures tested. Similar activation patterns were observed for Egr-2, Egr- 3 and Nurr1 (data not shown). This induction was specific for muscarinic receptors as specific acetylcholine nicotinic agents ((-)-Nicotine, (+)-Epibatidine) could not elicit this response. I did not observe non-specific gene induction (i.e. Sp-1) following muscarinic activation. Bargraph represents fold induction as assessed by quantitative RT-PCR following 1 hour of chemical stimulation (30 µM). !!

! Figure 4

"&+!

!"# $"# .#/%0*#123#µ45# .#/%0*#123#µ45# .#*67896:;<#123#µ45# .#*67896:;<#123#µ45# .#=>3# .#&?@#1233#µ45#

%&'(##)*&'#+,-(((# %&'(##)*&'#+,-(((#

Figure 5.17 | Chronic acetylcholine administration augments !III-tubulin expression during differentiation (a,b) Differentiation of hNSCs in the (a) absence or (b) presence of acetylcholine. Cells were grown on laminin and growth factors were sequentially withdrawn in the presence of BDNF and forskolin. Images shown are representative fields following a 10-day differentiation process. !III-tubulin is used a surrogate readout of neuronal differentiation. !

! "&"!

5.5 DISCUSSION Mounting evidence suggests that stem cell populations exhibit substantial nongenetic heterogeneity253. Embryonic stem cells display heterogeneous expression of Nanog and Stella, and also show spontaneous transition from one state to the other247,248 reverting original population heterogeneity after cell sorting. Furthermore, these distinct states transiently retain different potentials in response to stimuli248. Similar observations have also been recently made in hematopoietic stem cells249. Stem cells have been reported to characteristically maintain a large proportion of simultaneous transcriptional activity at promoters associated with both precursor and differentiated states236. My data demonstrates that hNSC pools exhibit patterns of NT gene expression that are reversible, equilibrating, and heterogeneous. Importantly, the variable expression of NT genes is mirrored by responsiveness to relevant NT signals. The capacity of hNSCs to reversibly express different NT programs supports the notion of lineage priming in that cells are poised to effect gene expression programs that ultimately reflect downstream committed cell types177. These different metastable NSC states presumably engage with the local neurochemical environment to specify different NSC responses, and possibly specifying appropriate neuronal or differentiated cell types. This type of biological response is formally analogous to phase variation in pathogens, whereby the pathogen is primed to respond to different host defenses by random phase variation in gene expression254-256. Although such evidence of subtype specific differentiation in response to neurotransmitters exists, mechanistic interpretations has been troubled given the intracellular convergence of many of these NT pathways246,257,258. Reversible heterogeneous states with altered potentiality could also explain how cells are compartmentalized to momentary process common intracellular signals differently. To my knowledge, my results provide the first indication of such metastable states in a human-derived stem cell pool. Phase variation in the hNSC compartment states may intrinsically limit the NSC responses to signals that promote self-renewal or differentiation259. Although this nongenetic heterogeneity may involve random processes, the simultaneous equilibration of multiple neurotransmitter pathways back to a reproducible ground state suggests that additional non-stochastic mechanisms may intrinsically control the frequencies of each

! "&#!

NT in hNSCs. Heterogeneity in NSC populations would serve to preclude an excess response to any given signal and thereby mitigate potential depletion of the NSC pool. In addition, the superposition of different anticipatory states may enable simultaneous multiple responses in a complex signaling environment. As I have shown in vitro, variation would also allow recovery from depletion of any given subpopulation, thereby maintaining a fully competent multi-potential stem cell compartment over time. The roles of different neurochemical niches in NSC commitment and self-renewal requires further definition in vivo. My findings have implications for efforts to purify populations of stem cells for applications in regenerative medicine. If the intrinsic heterogeneity I observe in hNSCs applies to other stem cell types, the potential for generation of specific desired cell types may be limited even in the presence of proper stimuli. As noted, even ES cells appear to exhibit phase variation in the expression of key fate determinants such as Nanog260. Artificial stabilization of specific NT receptor expression patterns may be needed to efficiently produce neuronal subtypes in the treatment of Alzheimer’s, Parkinson’s and other neurodegenerative diseases. Finally, I speculate that brain tumor stem cells may exhibit similar phase variation patterns as normal NSC cells; if so, single agent therapies may prove less than effective in the face of cancer cell phenotypic diversity261.

5.6 METHODS Adherent culturing of human foetal neural stem (hNSC) cells. With the approval of the Research Ethics Boards at The Hospital for Sick Children and Mount Sinai Hospital (Toronto), I obtained tissue from human foetal CNS (8-13 weeks of gestation). Subcortical forebrain tissue was mechanically dissociated with the aid of a 1 ml filtered plastic pipette. Some primary samples were initially grown on non-coated plates to allow aggregates to form, but most were plated adherently in dishes precoated with 0.01% poly-L-ornithine (Sigma) & laminin (10ug/ml in PBS, Sigma). All hNSCs were maintained in complete NS media (NS Media: Neurocult NS-A basal media containing L-Glutamine (2mM), BSA (75 ug/ml), Antibiotic/Antimycotic (1%), and supplemented with fresh EGF (human recombinant; 10 ng/ml; Sigma-Aldrich Canada), bFGF (10 ng/ml; Stem Cell Technologies), Transferrin (100 ug/ml), Insulin (25 ug/ml),

! "&$!

Putrecine (60 µM), Sodium Selenite (30 nM), Progesterone (20 nM), Heparin (2!µg/ml), and 1X B27). Cells were grown and maintained as adherent cultures in an incubator at

o 235 37 C, 5% CO2 and atmospheric oxygen (20%) . The conditioned media of the cells was partially exchanged (50:50) with fresh media every 2-3 days and cultures were passaged every 5-10 days required. For dissociation, cells were collected and incubated in Accutase at 37 oC for 10-20 minutes, and then replated in an equal mixture of fresh and conditioned hNSC media. ! Immunocytochemical analysis of hNSC cultures. Adherent NS cells were prepared for immuncytochemical analysis by growing them on poly-L-ornithine & laminin coated coverslips. On the day the cultured cells were analyzed, the cells were directly fixed with 4% paraformaldehyde (PFA) for 30-60 minutes. PFA was removed by first aspirating the media/PFA mixture followed by two 5 minute washes with PBS. Permeablizing was performed with 0.3% Triton-X-100 (Sigma) in PBS for five minutes followed by two additional 5 min washes with PBS. Non-specific binding was blocked by incubating slides in 10% normal goat serum (NGS) for 1 hour at room temperature. Astrocytes and neurons were identified using antibodies raised GFAP (1:1000; Dako) or!!-Tubulin type III (1:500, Chemicon) respectively. NeuN (1:500) was also used as a neuronal marker. The neural precursor markers Nestin (1:1000) and Sox2 (1:1000) were used to identify the more primitive cells in culture. Antibodies reactivity was probed using the secondary antibodies anti-mouse IG-A568 and anti-rabbit IG-A488. antibody was used. Excess secondary antibody was then removed by three successive 5 minute washes with PBS. DAKO mounting media supplemented with the nuclear staining dye DAPI (1:500) was used to identify cells. Specimen visualization was performed on a SD 200 Sprectral Bioimaging System (ASI Ltd., Israel) attached to a Zeiss Axioplan 2 Microscope (Carl Zeiss, Canada). Pictures were obtained and analyzed post-hoc using the AxioVision Software package. ! Western blot analysis Cells were lysed for 30 min on ice with RIPA Buffer (150 mM NaCl, 10mM Tris pH 7.2, 0.1% SDS, 1% Triton X-100, 1% deoxycholate, and a protease inhibitor tablet (Roche).

! "&%!

Cells were then sonicated for 3x 10 seconds and the supernatant was re-suspended in 5X SDS sample buffer (250mM Tris pH 6.8, 10% SDS, 50% glycerol, 0.02% bromophenol blue, 10%!!-mercaptoethanol) and boiled for 5 minutes. 300 µg of protein were separated using SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to Immobolin-P PVDF membrane (Millipore). Membrane was blocked in milk powder dissolved in tris buffered saline-tween (0.137M NaCl, 2.7 mM KCl, 25 mM Tris-HCl pH 7.5, 0.1% Tween-20). The membrane was probed using a 1:100 D2DR (B-10) and 1:100 GluR-2 (N-19) (Santa Cruz Biotechnology) overnight and washed three times in TBST for 5 minutes at room temperature. The membrane was probed with an anti-mouse-HRP secondary antibody in milk-TBST for 1 hour at room temperature and washed three times in TBST for 5 minutes. Detection was accomplished using ECL Western blotting Detection System (GE Healthcare) and visualized on a film developer.

RNA isolation and RT-PCR analysis of bulk hNSC cultures Total RNA from bulk cultured cells was isolated using the RNeasy extraction kit (Qiagen) and DNA contamination removed with the RNAse-Free DNAse Kit as outlined in their product insert (Qiagen). RNA was quantified using a NanoDrop instrument (Thermo scientific) and 1! µg of total RNA from each sample was used for cDNA generation. Reverse transcription was done using Transcriptor Reverse Transcriptase

(Roche) and random hexamer primers (p[dN]6). The presence of specific genes was then assessed by PCR. To avoid amplification of contaminating genomic DNA, all primer pairs used throughout the paper were designed to span at least 1 intron.

Quantitative RT-PCR Isolation of RNA and subsequent synthesis of cDNA from various samples was carried out as described above. Relative abundance of specific mRNA species was then assessed using a lightcycler quantitative PCR instrument and iQ SYBR Green Supermix (Biorad). The relative abundance of each gene was assess using the delta-delta-Ct method with all values normalized to their respective GAPDH levels.

! "&&!

Generation of clonally derived NSC clones For the isolation of single cells, cells were prepared into a single cell suspension, washed, counted and resuspended at a final density of 0.5 x 106/ 100!µL of staining buffer (SB: 1 X PBS, 0.5% BSA, 4 mM EDTA, 10 mM HEPES) into 5 ml polysrene tubes (BD Falcon) and stained with propidium iodide (PI: Molecular Probes). With the aid of a FACSAria flow cytometer (Becton Dickinson, 100 um nozzle, 20 psi), live cells (negative for the incorporation of PI) were sorted into 96-well PLO & laminin precoated tissue culture plates. Gating parameters to identify live cells were established by Forward Scatter, Side Scatter, and PI florescence to ensure that extracellular debris, dead cells, or cell doublets were discarded from the analysis. Cell doublets were identified and excluded based on their pulse width and height of both FSC and SSC light diffraction patterns. Single cells were accurately deposited into the center of tissue culture plate with the aid of a motorized X-Y stage that repositions the plate to the next well following completion of each event.

Single cell gene expression profiling by RT-PCR Multiplex single cell RT-PCR was adapted from others250 and carried out to the approach outlined in Figure 5.10. Briefly, cells were accutased and live cells were identified with propidium iodide (PI: Molecular Probes) staining. Using a flow cytometer (BD FACSAria Cell Sorting System, BD, USA), single cells were then sorted into 96-well

PCR plates containing 13!µL of sterile RNAase/DNAse-free H20 (GIBCO) supplemented with 20 mg/mL of MS2 bacterophage RNA (Roche). As controls, zero cells were deposited into the first four wells of the plate (no template control) and 100 PI-negative cells were deposited into the last four wells during the sort. Single PI-negative viable cells (live gate) were deposited into the remaining wells of the plate. Immediately following the sort, plates were sealed tightly with optical tape and contents were flash- frozen by submersion in -70oC ethanol. This process also allowed for efficient cell lysis. ! To assess the expression of a specific array of genes, a sorted plate was first incubated in a 95oC pre-heated therocycler (Biorad) for 5 minutes. Once completed, 2 mL of a cocktail of gene specific primers (ones desired to be analyzed; at a final concentration of 1! µM each) was manually added to each well and the reverse

! "&'! transcription process was carried out as outlined in the Roche manual. I find that 1 U/ well of the Transcriptor Reverse Transcriptase enzyme (Roche) is sufficient for this. Following the synthesis of cDNA, the desired genes underwent a first-round multi-plex amplification PCR step for 36 cycles at 65oC by adding 20!µL of pre-mixed “PCR master mix” containing: 4 µL 10X PCR Buffer, 4 µL dNTPs (2 mM stock), 0.33 µL of each reverse primer (10 µM stock), 0.33!µL of each forward primer (10!µM stock), 0.2 µL Taq polymerase/reaction and X µL H2O for a total of 40!µL. First-round product was then replica plated into new PCR plates and a nested gene specific second-round PCR reaction was carried out for 45 cycle at 58oC. Detectable levels of gene expression for each cell was then determined by assessing the presence or absence of the complete PCR product by gel electrophoresis of a 2% agarose ethidum bromide stained gel. !!! Assessment of heterogeneity by tandem analysis of small hNSC populations To simplify the assessment of heterogeneous gene expression and estimated gene expression prevalence (proportion of cells expressing detectable levels of specific mRNA transcript) of rarely expressed genes (<5% of cells), a variation of the outlined single cell protocol was developed. This involved analysis of small hNSC populations (~25-50 cells) rather than single cells and assumes that if the heterogeneous expression of a gene is rare enough, a positive band detected from small population of cells (~25-50) is likely to be derived from the mRNA of a single cellular entity found within that sample of cells. This thus allows the screening of large number of cells and a more representative frequency of gene expression to be estimated for the cell population. Briefly, cells are collected and prepared for cell sorting as described in the section entitled “Single cells gene expression profiling by RT-PCR”. Cells were sorted as previously described, but as populations (10-12 replicates) of defined sizes (typically between ~20-50 cells for genes detectable in only ~0.1-10% of cells at the single cell level). The gene specific RT, first round multi- plex amplification PCR, and gene-specific nested PCR steps were then repeated as previously described for the desired genes. Detectable levels of gene expression for each subpopulation were then determined by assessing the presence or absence of PCR product by gel electrophoresis of a 2% agarose ethidum bromide stained gel. The

! "&(! frequency of cells expressing a particular transcript at detectable levels was then estimated using the formula:

! Where f (x) = estimated detectable prevalence at a given density, P = number of positive populations for the detection of a specific gene, N = number of negative populations, and n is the particular cell density used for this experiment. Frequency estimations were also achieved by performing a limiting dilution RT-PCR method similar to the cell based assays used to assess colony forming units in culture and fitting it to a Poisson distribution (Fig 5.7). This method also acts as a sensitivity control by allowing visualization and confirmation of a relatively linear relationship of the data points that would suggest that the estimated frequency is well approximated and consistent at varying cell densities and thus independent on the quantity of mRNA input. ! Quantification and prospective isolation of a7nAChR+ and a7nAChR- hNSC populations To quantify and isolate sub-populations of cells expressing the nicotinic acetylcholine receptor-"7 ("7nAChR), fluorescein isothiocyanate (FITC)-conjugated a-bungarotoxin (Invitrogen) was used. Briefly, adherent NS cells were detached using Accutase (Sigma), quenched in PBS and then incubated in ice cold PBS containing either (FITC)-conjugated a-bungarotoxin (1!µM) or water (control) for one hour at 4 oC in the dark. The cells were resuspended in 300 mL of PBS containing 2 µL/mL of Propidium Iodide to permit the identification of live cells. Cells expressing "7nAChR were identified by the presence of a characteristic sub-population of cells which showed a persistent staining pattern at various concentrations of (FITC)-conjugated a-bungarotoxin (0.1-10 µM). Sorting for both positive and negative fractions based on this distinct staining pattern yielded a high degree of post-sort purity. The specificity of this reagent to!"7nAChR was confirmed by the absence of positive nicotinic receptor staining in non-neuronal tissues (1 µM)

! "&)!

(fibroblasts) and substantially reduction in number of positive cells (0.5 µM) following a brief pre-incubated (30 minutes) with the competitive agonists (Acetylcholnie and Nicotine) or a specific "7nAChR competitive antagonist (MG-624) at 125 µM (See Fig 5.10, and Fig 5.11).

Assessment of a functional heterogeneous response Due to limitations of the research tools available, I adapted the “tandem analysis of small hNSC populations” to act as a surrogate for protein function (gene induction following presumed receptor activation). To assess for a potential heterogeneous! response, I first optimized my system to be “biased” towards the detection of a homogenous response (if one was in fact present). This was done by selecting the most highly induced gene and time of maximal gene activation in my system. In bulk cultures of NSCs, a time course analysis revealed that the early response genes I identified to specifically response to acetylcholine stimulation reach maximum levels following approximately 1-2 hours of stimulation (Fig 5.14). Furthermore, of these genes, I indentified Nurr1 as the most highly induced gene under my conditions. Therefore, to avoid any potential time dependant or gene specific bias in the assessment of heterogeneous response to neurotransmitter cues, I evaluated how Nurr1 activation was distributed within a population of NPCs following 2 hours of stimulation as these factors were most favorable for the detection of a homogenous response. As an additional control, I pre-mixed and re- distributed the mRNA isolated for a separate set of small sub-populations. A uniform detection of bands in all replicates by this method would suggest that the assay had sufficient sensitivity to detect a homogenous response if one was present. ! !

! "&*!

-CHAPTER 6-

DISCUSSION: NEUROCHEMICAL CONTROL OF NEURAL STEM CELLS

6.1 PROLOGUE

In light of the findings presented in the thesis, this final chapter aims to briefly review previously published reports in support of the neurochemical control of neural stem cells. Following this brief introduction, I use knowledge gained from these studies and the results presented in this thesis to re-interpret some clinical and biological phenomena that are perhaps related to neurotransmitter effects on stem cell pools. I speculate about mechanisms by which these drugs may act intracellularly to cause the observed biological effects and comment on the role of neural stem cells in the benefits seen with psychiatric pharmacological therapy. I also address the concept of how neurotransmitters may play a role in specifying subtype specific neuronal differentiation during development, learning and repair. Lastly, I hypothesize how the delicate balance of basal neurotransmitter levels may be vital for maintaining a healthy nervous system and how perturbations in these levels may lead to dramatically different CNS diseases. This chapter concludes with a summary of the contributions this thesis has made to the field and suggests what future work in this area should focus on. This work has been formatted in a perspective piece for a potential future publication:

Diamandis, P., Cusimano MS., Tyers, M. & Dirks, P.B. Neurochemcial Control of Neural Stem cells. Manuscript in preparation (2009).!

! "'+!

6.2 TITLE AND CONTRIBUTORS

NEUROCHEMICAL CONTROL OF NEURAL STEM CELLS

Phedias Diamandis1,2,3,4, Maria Cusimano1,2, Mike Tyers3,4,5* & Peter B. Dirks1,2,6,7*

1The Arthur and Sonia Labatt Brain Tumor Research Centre, The Hospital for Sick Children and University of Toronto, 555 University Avenue, Toronto, M5G 1X8, Canada

2Program in Developmental and Stem Cell Biology, The Hospital for Sick Children and University of Toronto, Canada

3Samuel Lunenfeld Research Institute, Mount Sinai Hospital, 600 University Avenue, Toronto, M5G 1X5, Canada

4Department of Molecular Genetics, University of Toronto, 1 Kings College Circle, Toronto, M5S 1A8, Canada

5Wellcome Trust Centre for Cell Biology, School of Biological Sciences, The University of Edinburgh, Mayfield Road, Edinburgh EH9 3JR, United Kingdom

6Department of Laboratory Medicine & Pathobiology, Faculty of Medicine, University of Toronto, Banting Institute, 100 College Street, Toronto, M5G 1L5, Canada.

7Division of Neurosurgery, The Hospital for Sick Children and University of Toronto, Canada

*Correspondence or requests for materials can be addressed to M.T. ([email protected]) or P.B.D. ([email protected])

! "'"!

6.3 INTRODUCTION

Evidence suggesting that fluctuations in endogenous neurochemicals could promote remodelling and regeneration of the adult brain dates back to Hippocrates, who noted that malaria-induced convulsions dramatically alleviated manic symptoms262. Although several present-day psychiatric treatments are based on similar principles; inducing a transient increase in neurotransmitter levels, their therapeutic neurobiological mode of action is not fully understood262,263. For example, it is puzzling how immediate and brief rises in neurochemcials following electroconvulsive therapy (ECT) can have sustained symptomatic clinical benefits long after therapy has been completed263. Similarly, despite the immediate effect of SSRIs on monoamine metabolism and neurotransmitter levels, 8-12 weeks of therapy is needed before the majority of depressed patients achieve clinically significant outcomes264. These observations suggest that, in addition to temporary rises in neurotransmitter levels, more permanent downstream changes in the architecture of the CNS may occur during and following the completion of neuromodulatory therapy. Understanding these key changes may help us develop more disease-specific therapies with fewer side effects. In this final chapter, I discuss the potential role neurotransmitters play in remodelling the CNS through their effects on the neural stem cell (NSC) compartments of the brain. I introduce some emerging concepts regarding the intrinsic neurotransmitter biology of NSCs, and discuss relevant implications for CNS development, repair and their potential contribution to the pathogenesis of multiple disease states. I hope that the ideas presented here will highlight some of the questions that I feel still remained to be answered and provide suggestions on how these scientific problems can be properly addressed.

6.4 NEUROTRANSMITTER SIGNALLING PROMOTES NEUROGENESIS IN NSCS

Long-held beliefs about the static nature of the adult brain were finally put to rest following the discovery of neural stem cells (NSCs); self-renewing, multipotent cells able to generate all the major cell types of the central nervous system. Limited to largely quiescent populations of cells in the adult subventricular zone (SVZ) and the subgranular

! "'#! zone (SGZ) of the hippocampal dentate gyrus, NSCs are continually mobilized by a variety of environmental cues to divide, differentiate into newly functional neurons and integrate into the complex circuitry of the brain167,265,266. Interestingly, these naive precursor compartments are heavily innervated by dense collateral inputs from serotonergic267, dopaminergic213,221, noradrenergic267, cholinergic214 and glutaminergic268 neurons, suggesting that neurotransmitters may play a critical role in NSC maintenance, self-renewal, and differentiation within these neurogenic niches167. Though usually associated with mature differentiated neuronal subtypes, both in vitro and in vivo defined populations of naïve NSCs have been shown to express functional neurotransmitter receptors and transporters. For example, D2-dopamine- receptor staining has been observed on both NSCs and progenitors213,221. In Chapter 2 and Chapter 5, I have extended this work by demonstrating that at least in vitro, a single NSC population expresses a multitude of receptors spanning all the major neurotransmitter classes168. It is notable to mention that although there is only a handful of neurotransmitters (i.e. dopamine, serotonin, acetylcholine) in the brain, the numerous receptor subtypes (i.e. D1-D5 dopamine receptors) within each class, makes the precise phenotypic neurotransmitter characterization of these cell populations a large undertaking. For example, in the case of serotonin, evidence for the expression of a 167 number of receptors subtypes including 5-HT1A, 5-HT1B, 5-HT2 exist , but it is unclear if the remaining serotonin receptor subtypes are not expressed or simply not reported. To further compound this problem, the expression patterns of these neurotransmitter receptors within the different precursors regions found in the adult brain suggest that neurotransmitter pathway gene expression may spatiotemporally vary among NSCs and their progeny during development217. Precisely defining the neurotransmitter ground state of these cells may thus require huge efforts given the multiple precursor regions, gene expression changes occurring during development and adulthood, and the variability that exists between different organisms. Markers that allow for routine isolation of purified NSC populations, coupled with sensitive gene microarray and proteomic techniques will be required for such a systematic undertaking. Even without a complete and systematic characterization of the neurotransmitter expression patterns in these cells, it is evident from the existing literature that these

! "'$!

“naïve” multi-potential NSCs, even prior to lineage commitment, express a number of neurotransmitter pathway genes that have been traditionally associated with the more mature neuronal cells of the CNS. The presence of these receptors is not without functional importance. While potentially behaving as growth regulators269-279 during developmental periods, neurotransmitters appear to also promote activation of neuronal programs in the adult brain212,216. Similarly, cell division, survival and neurogenesis are enhanced by as much as three-fold218 in the subgranular zone (SGZ) of the dentate gyrus following increased 217,280-282 serotonin signalling at the 5-HT1A receptor . Further corroborating this result is the observation that lesions in local serotonergic neuronal inputs, or the inhibition of serotonin synthesis, leads to significant reductions in both NSC expansion and differentiation. These effects could be reversed by restoring normal neurotransmitter levels with raphe grafts containing serotonin-producing neurons273,283. Similarly, removal of the dopamine projections known to infiltrate the SVZ leads to decreased proliferation and differentiation of NSCs221,284,285. These changes are restored to normal levels when dopamine agonist specific to the D2-dopamine receptors are injected in vivo. Although the changes observed in NSCs may be derived by a number of indirect secondary pathways286, the preserved sensitivity of NSCs to dopamine amongst in vitro cultures that support only NPC growth, suggests the mechanisms are likely to be due to the direct action of this neurotransmitter on NSCs (Fig 6.1). In fact, activation of not only the serotonergic and dopamine pathways, but also the noradrenergic287, cholinergic214, and glutaminergic268 pathways above baseline levels, has been shown to contribute to neurogenesis in the dentate gyrus. It is important to note that these effects are not exclusive to neurotransmitter pathways involving G-protein coupled receptors (GPCRs). Many studies also report functional ligand-gated ionic channel receptors in neural progenitors as well277,278,288-290. For example, functional L-type calcium channels that respond to NMDA are expressed on uncommitted NSCs of the hippocampus. When stimulated, they promote activation of neuronal programs and subsequent neurogenesis212. Taken together, these studies support the idea that neurotransmitters directly signal to receptors expressed on NSCs (and likely also through

! "'%! additional indirect mechanisms), and play a role in regulating the proliferation and differentiation of these cells.

! "#$%&'!()*! !"#$#%&"'()*(#+,!! "'&!

,)! +)! Indirect regulation of NSCs Direct regulation of NSCs by neurotransmitters by neurotransmitters

NSCs Differentiated cells NT receptor

NT Growth factors, cytokines, and other signals ! Figure 6.1 | Mechanisms by which neurotransmitters may modulate NSC pools

(a) Our current appreciation of neurotransmitter pathways in the CNS suggests that neurotransmitters may regulate NSC pools indirectly through their actions on more differentiated cells. Neurotransmitter signaling in these cells may lead to the secretion of growth factors and other cytokines that regulate the growth of NSCs in the surrounding areas. (b) Given the conserved effects of these agents on isolated in vitro cultures of purified NSCs and the expression of neurotransmitter receptors on NSCs (both in vitro and in vivo), neurotransmitters may also exert their NSC effects by acting directly on these cells.

! "''!

It is interesting to note that antidepressant drugs from a variety of neurotransmitter classes including NA and 5-HT reuptake inhibitors281,282,291, mood stabilizers292-294, and atypical antidepressants295, have been shown to increase the proliferation and survival of new neurons in the dentate gyrus. Even more striking is the fact that fluoxetine treatment (Prozac, modulates serotonin levels) had no beneficial effects on irradiated mice, in which neurogenesis is abolished218. This strongly suggests that the clinical effectiveness of antidepressants in diseased states is intimately linked to the promotion of hippocampal stem cell proliferation, and not only the survival of mature CNS cells (Fig 6.2). As previously alluded to, the sheer complexity of functional effects of neurotransmitter signalling on NSCs is compounded by the vast permutations of receptor subtypes that can exist for each neurotransmitter class. For example, treatment with D2/3 receptor agonists has been shown to promote proliferation284,285, and removal of dopaminergic projections accordingly reduces proliferation in the SVZ221. However, others have contradicting data suggesting that the dopamine antagonist haloperidol increases proliferation of NSCs213. I have also observed robust but conflicting inhibitory effects with a variety of agonist and antagonist belonging to the same neurotransmitter class168. These contrasting results suggest that the way in which dopamine; and likely other neurotransmitters; regulate NSCs and their progeny is multifaceted, or may depend on different receptors found on different cells. It is also interesting to note that the neurogenic region of the SVZ is an area with remarkable overlap between afferent dopamine and serotonin neurons167. Efficient neurogenesis may thus require simultaneous activation of a number of different neurotransmitter pathways. Drugs like haloperidol that act to change the levels of multiple neurotransmitters in vivo, may thus lead to neurogenesis through different and more complex mechanisms than agents that specifically act on single neurotransmitter receptor subtypes (Fig 6.2). Differences in the environments that these animals are maintained in and their effects on the endogenous release of other neurotransmitters may further complicate the interpretation of altering a single pathway at a time.

! "'(!

6/89H1/-/I#8! 4-(1/#3/5!.6"!! 4-(1/#3/5!! 4-(1/#3/5!301:):#8;!4-(1/#3/5!,:/1#88!,0*70*;! 71,8)9/1#+,-!! 71,2/-)*,1!7,,83!! 4-(1/#3/5!5)

!!!!!!"#$%! ./01,2/-/3)3! &'()*#+,-!!

"#$%! "#$%! "#$%!

=)201/!>?$!!@!A,*/-+#8!/

! "')!

6.5 NSCS IN NEUROTRANSMITTER DEFICIENT DISEASED STATES

A number of CNS disease states are characterized by deficient neurotransmitter levels, and thus provide natural models for studying the pathological changes in neurochemical levels on the precursor compartments167,219,296,297. For instance, in addition to the hallmark loss of dopaminergic neurons in the substantia nigra, both experimental mouse models and human patients with Parkinson’s disease display reduced NSC proliferation in the subventricular zone (SVZ)221,285. Specifically, brains of post-mortem patients with Parkinson’s disease have reduced numbers of proliferative cells in the subependymal zone as well as fewer precursors in the subgranular zone and olfactory bulb221. When coupled with the data generated in mouse models, it suggests that changes in neurotransmitter levels following loss of dopaminergic neurons leads to reduced neurogenesis. Similar effects were observed in mouse models of Alzheimer’s disease, a neurodegenerative disorder characterized by deficiencies in cholinergic neurons. Rodriguez et al identified significant decreases in SVZ and SGZ cell proliferation, which both exacerbated with age and correlated with cognitive deficits298,299. Even schizophrenic patients, when assessed by Ki-67 staining of post-mortem brain, displayed significantly reduced NSC proliferation in the dentate gyrus300. A number of contrasting results only further illustrate that the effect of various neurotransmitters on neural stem cells is complex, and may depend on other factors such as the expression of different receptors on different NSCs. For example, activation of dopamine in neural progenitors has been shown by different studies to have both proliferative and inhibitory effects on NSC pools168,213,221,284,285. Similar conflicting effects of neurotransmitters on NSCs have been suggested for serotonin217,218,301, glutamate268,302 and acetylcholine168. The specific assays used as readout for NSC effects and the “specific” agents used to draw consistent conclusion regarding the effects of each neurotransmitter class need to be standardized in order to more accurately interpret the effect of each pathway on NSC biology (Fig 6.2). Given that many patients are on chronic neuromodulatory medications for their illnesses, it is hard to rule out competing NSCs effects of their therapies. The compensatory rise (or decrease) in neurotransmitter levels may thus explain some of the conflicting results regarding the effects of neurotransmitter changes in humans. As almost

! "'*! all psychiatric patients elect to participate in a combination of neuropharmacological therapies during the course of their disease, it may become increasing difficult to interpret analysis of post-mortem brains from this patient population in the future. It is thus likely that I will need to rely on developing accurate animal models of these diseases or develop sensitive non-invasive techniques to properly understand the NSC changes resulting from different CNS disorders in humans. Given the requirement of neurogenesis for the clinical benefits of antidepressant medications, identifying ways to generate specific neuronal subtypes that are lost in particular CNS disease states would be a major step forward in the field. Chemical genetic screens designed to help expand our understanding of how NSC neurogenesis and proliferation are regulated by neurochemicals that selective target only specific neurotransmitter receptor subtypes may eventually lead to more disease-specific therapies. Understanding the specific NSC and neuronal deficits that may be occurring secondary to each patient’s disease state is thus critical for designing such effective therapies. A more detailed understanding of the specific signalling similarities and difference between the different receptors subtypes may contribute to simplifying this problem. In the next sections, I review the downstream pathways that may be operational in these cells following neurochemical stimulation and hypothesize about possible mechanisms by which perturbations in these pathways may lead to different CNS diseases.

! "(+!

6.6 DOWNSTREAM EFFECTS OF NEUROTRANSMITTER SIGNALLING IN NSCS

Given the variety of different neurotransmitter signals and receptors found within the precursor compartments of the brain, it may be somewhat surprising that similar proliferative and neurogenic effects are observed following activation of these different pathways167. It is also intriguing why agents acting on similar neurotransmitter classes have been reported to have such contrasting effects. Furthermore, without an appreciation for downstream signalling effects of these different receptor classes, it may be surprising why the induction of early response genes known to be involved in dopaminergic signalling (i.e. Nurr1 induction) are also regulated by other neurotransmitters including serotonin, acetylcholine and histamine. As described in Chapter 5, my data shows that that the induction of Nurr1 may also occur through the activation of muscarinic receptors by acetylcholine. I extend this observation to show that multiple neurotransmitter pathways converge on Nurr1 induction, through a protein kinase C (PKC)-dependant pathway. This data demonstrates how similar neurogenic effects can be elicited through distinct neurotransmitter pathways. Although our understanding of the intracellular signalling cascades activated by neurotransmitter receptors stems from work done in more mature cells of the CNS246, data in neural precursor cells suggests that similar mechanisms may exist in the more immature cells of the brain. In neurons, many of these receptors are linked to common G- protein coupled receptors that generate second messengers including the conversion of AMP to cyclic AMP (cAMP) or the hydrolysis of phosphatidylinositol 4,5-bisphosphate

(PIP2) to inositol 1,4,5-trisphosphate (IP3) and diacylglycerol (DAG) (Fig 6.3). The latter pair of second messengers lead to increased intracellular calcium stores and the cooperative activation of PKC. This convergent increase in calcium and activated PKC is thought to be involved in activating gene expression vital for regulating the biological changes seen in these cells246. It appears these second messengers are also responsible for the neurogenic effects of neurotransmitters in NSCs246. This is supported by data from our lab suggesting that chemical blockade of PKC signalling following neurotransmitter stimulation completely abrogates the induction of early response genes involved in differentiation. Similar to what is seen in mature neurons, NSC-induced neurogenesis by the binding of glutamate and GABA to ion-channels also seems to depend on changes in

! "("! intracellular calcium levels212,216,303. These studies also support the activation of additional pro-differentiation genes such as NeuroD following neurotransmitter activation. It is interesting to note that receptors of the same neurotransmitter subclass can regulate different and even opposing second messenger effects based on the intracellular G-proteins present within the cell. The combination of specific receptors and their different adapter proteins, which differ between species and stages of development, may help explain why studies continually show contradictory results when assessing the effects of neurotransmitters on NSCs. Alternatively, neurotransmitter receptors can exert their biological effects through changes in cAMP levels (Fig 6.3). Increased cAMP levels are associated with protein kinase A (PKA) activity. This and calcium dependant activation of calmoduin-dependant protein kinase (CaMK) increases cAMP response element binding protein (CREB) activity and leads to other gene expression changes vital for neuronal growth, maturation and survival263. This represents yet another example of how neurotransmitter receptors coupled to different G-protein can have signalling pathways that converge intracellularly to promote similar effects in NSCs. Another important thing to note regarding neurotransmitter receptor biology is that both stimulatory or inhibitory cAMP dependant G-proteins can be activated by the same neurotransmitters (i.e. G-proteins coupled to the

5-HT1A receptor). As many of the receptors found on stem cells are thought to change during the course of development and adulthood224,304, the particular G-protein coupled to the neurotransmitter receptor can lead to dramatically different effects even in the context of the same ligand (Fig 6.3). Identifying the expression of such neurotransmitters pathways without properly noting the specific G-protein coupled to that particular receptor may be insufficient to allow predictions regarding the effects of signalling in the particular NSC populations of interest.

! "(#!

cAMP Mechansim

Gs: 5-HT1A, 5-HT4, D1 dopaminergic

A2 Adenosine !-adrenergic PLC Mechanism Gi: 5-HT1A , 5-HT1B, 5- Mascarinic, 5-HT , 5- HT1D, D2 dopaminergic, 1A HT1C, 5-HT2, "1-adrenergic, Muscarinic, GABAB D2 dopaminergic, glutamate

() '(!# !"#$ # -.-0# !"# !$%&&#

6+*!$ 6+*!$ .-/#

#'()$ %&$ -,( # ()$ #'()$ #' #'()$ ()$ #'()$ #'

)*+,# *+!"$ #&%3$ !"#$%"&''

34,5$ ,-../0%1./$ %1.(0$%1.2$

Figure 6.3 | Potential intracellular effects of neurotransmitters on NSCs ! In neurons, neurotransmitters signal through the binding of neurotransmitters to their respective receptors and activating one of the two principle GPCR pathways; cAMP or the phosphatidylionsitol signaling pathway. In the cAMP pathway, activation of the

! "($! stimulative regulative G protein (Gs) induces adenylyl cyclase (AC) to increase cAMP levels. This leads to increased PKA activity and a subsequent CREB-dependant increase in BDNF expression. Other neurotransmitter receptors conversely decrease cAMP levels and subsequent downstream effects through their association with an inhibitory regulative

G protein (Gi). The other major pathways activated by neurotransmitter receptors include the phosphatidylionsitol signaling pathway. Following activation of the Gq/11 G-protein subunit, receptor mediated signals activate membranous phosopholipase C (PLC). PLC hydrolyzes phosphatidylinositol 4,5-bisphosphate (PIP2) into the second messengers diacyglyerol (DAG) and inositol 1,4,5-triphosphate (IP3). Elevated IP3 levels increase the release of Ca2+ from the endoplasmic reticulum (ER) and mitrochondrial stores. This change in calcium levels activates calmoduin-dependant protein kinase (CaMK). CaMK then tranduces the signal through the cAMP response element binding protein (CREB) and leads to gene expression patterns similar to those seen by changes in cAMP levels. With the help of DAG, increased calcium levels also lead to the activation of protein kinase C (PKC) that goes on to activation signal transduction pathways with a variety of cellular effects including changes in gene expression associated with differentiation. Ligand gated ion channels that also increase intracellular calcium stores also lead to the activation of similar signal transduction pathways. Some examples of neurotransmitter receptors involved in activating each pathway are provided to exemplify the cross talk occurring between different neurotransmitter receptor subtypes and classes.

! "(%!

Ultimately, it appears that further to the downstream activation of these limited early response genes, neurotransmitter signalling culminates in the transcription of BDNF, a major neurotrophic factor known to play key roles in the survival, guidance, and normal functioning of neurons. Interestingly, BDNF is induced as early as one hour following ECT, and has been thought to account for clinical outcomes following antidepressant treatment by contributing to the survival of existing neurons and improving synaptic function263,305-307. However, a striking but only recently appreciated observation is that these increased BDNF levels are most prominent in the hippocampal dentate gyrus, one of the few known coordinates of the adult brain that harbours neural precusors. Within this neurogenic region, BDNF levels are reported to increase as much as 20-fold from baseline following neuromodulatory therapy308. In fact, the selective deletion of BDNF in the dentate gyrus, but not the CA1 region of the hippocampus, significantly attenuated the antidepressant effects of these treatments306. Similar observations were made in the subventricular zone (SVZ), where exposure to BDNF increased production of both olfactory bulb interneurons and striatal neurons that are not normally generated in adult brains263. Although some contrasting results exist309-312, this spatiotemporal bias suggests that, in addition to promoting survival of existing neurons in the brain, BDNF may also induce more permanent changes through effects on NSCs. The expression of BDNF in neurogenic regions following antidepressant therapy seems to be mediated by a number of different neurotransmitter receptors263. Muscarinic agonists markedly increase BDNF mRNA in the dentate gyrus of both early postnatal and adult rats, while the blockade of glutamate receptors or the stimulation of the GABAergic system reduces levels of this neurotrophin in the hippocampus313,314. Pharmacologic activation of serotonin and norepinephrine receptors increases both expression of BDNF and precursor proliferation217,218,301,315. Moreover, unpublished data from our laboratory suggests that BDNF is upregulated in purified populations of human NSCs following treatment with acetylcholine, dopamine, and serotonin. Each of these neurotransmitters seems to influence BDNF expression by modulating levels of calcium or cAMP; both secondary messengers that ultimately converge on the activation of CREB, a transcription factor implicated in regulating the BDNF promoter316-318 (Fig 6.3).

! "(&!

The regulation of BDNF is one of the many convergent pathways that may lie downstream of activated neurotransmitter receptors. Other neurotrophins like ciliary neurotrophic factor (CNTF) have also been suggested to stimulate adult neurogenesis following neuromodulation as well263. Through activation of the CNTFa receptor, CNTF have been shown to increase both self-renewal (through increasing expression of Notch1) and differentiation110,319. Although not the focus of this thesis, additional signals that may be involved are reviewed elsewhere167 and help further explain why such a diverse array of distinct neurotransmitter signals may converge to cause similar effects in NSC pools. Conversely, there may also be additional signals that are only activated by certain second messenger signals that are unique to only particular neurotransmitter receptor classes. Interestingly, others have also demonstrated that activation of CaMK leads to epigenetic changes in the cell that may be vital in the initiation of differentiation and neurogenesis. Schneider et al noted that through a calcium dependant mechanism, a novel NMDA agonist led to a decrease in nuclear histone deacetylase 5 (HDAC5) activity, de- repression of neuronal neurotransmitter pathway genes like NR-1 and an increase in neuronal protein immunoreactivity303. Other neurotransmitter induced changes in DNA- structure have also been reported in undifferentiated stem cell pools320. While these findings are of significant interest for regenerative therapies, we have yet to fully elucidate the downstream biological mechanisms underlying the effects of neurotransmitters on adult NSCs and their niches. Given the growing list of neurotransmitter receptor subtypes identified to play a role in NSCs, it is likely that the contradictory differences that are reported in the literature may also stem from multiple inter- and intra-class signalling cooperatively following administration of non-selective neuromodulatory agents (i.e. Haloperidol or dopamine). By understanding the intracellular signalling pathways activated by each neurotransmitter class and their specific subtype, explaining and foreseeing the effects of different neuromodulatory agents on NSCs function may become a more realistic goal.

! "('!

6.7 NEUROTRANSMITTER-INDUCED NEURONAL LINEAGE AND PHENOTYPE

SPECIFICATION

The mature cell phenotypes generated by NSCs following differentiation are generally described as either neurons, glia, or oligodendrocytes51. However, this simplistic definition of multipotency overlooks the potential of NSCs to produce the myriad of neuronal subtypes (i.e. dopaminergic or cholinergic neurons) in the adult brain. While there are numerous studies implicating different neurotransmitters in regulating neurogenesis, it is still uncertain as to whether these cues lead to a preferential production of particular neuronal subtypes. While the convergent intracellular signalling pathways of neurotransmitter receptors would suggest otherwise, a number of studies have demonstrated that the neuronal phenotype produced in vitro depends on the type of exogenous neurotransmitter present257,258. For example, while Studer and colleagues were able to produce 18% DAergic neurons in the absence of dopamine, Riaz et al significantly amplified this effect to 60-90% DAergic neurons by exposing the precursor cultures to dopamine257,258,321. These cells were immunoreactive for tyrosine hydroxylase, and produced endogenous dopamine that was secreted into the media322. Consistent with the neurotransmitter signalling pathways know to be functional in mature neurons, Du and colleagues demonstrated that the dopaminergic promoting effects of dopamine during differentiation requires PKC signalling323-326. This phenomenon also extends into other neurotransmitter classes as similar results were seen with the addition of serotonin during differentiation of NSCs. Using the serotonin synthesizing enzyme tryptophan hydroxylase (TPH) and the secretion of serotonin into the medium as a surrogate readout for serotoninergic neurons, addition of serotonin significantly increased this specific neuronal subtype when compared to controls322. This result has been replicated by others who consistently showed that the addition of serotonin, but not other neurotransmitters (GABA, dopamine, glutamate) could enrich the number of serotoninergic neurons to 15% of the overall culture327. Interestingly, Riaz and colleagues were unable to generate noradrenergic cell phenotypes using similar differentiation strategies322 suggesting this lineage specification mechanism is not maintain through all the neurotransmitter lineages. It is also possible

! "((! that additional/different factors are required for the genesis of certain neuronal phenotypes. For example, it is important to note that the specific generation of serotoninergic or dopamineric neurons (using their specific neurotransmitters) in these experiments required co-culture with BDNF and the cAMP potentiator forskolin258,322. It is thus possible that other neurotransmitter lineages may be specified only in the presence of other neurotrophic factors. Such co-operative cytokine mechanisms may help maintain the potentiality for so many subtypes in a system where there is substantial convergence in the intracellular signalling of individual neurotransmitters pathways. Neurotransmitter screens in the presence of other neurotrophic factors (i.e. CNTF) may help elucidate this synergistic requirement. Using a novel NMDA agonist, Schneider and colleagues showed that through a NMDA-receptor calcium-mediated mechanism, culturing neural progenitors in the presence of NDMA agonists activates gene expression of the glutamate receptors GluR2 and NR1303. Although they did not assess the induction of receptors associated with other lineages, their data suggest that addition of neurotransmitters to differentiating cultures, in addition to activating the neurotransmitter synthesizing enzymes, may also upregulate the synthesis of additional receptors that can respond to environmental ligands present in excess. Preliminary data from our laboratory also suggests that acetylcholine treatment increases the expression of particular cholinergic receptors rather than acetylcholine synthesizing enzymes, but these observations require further testing. This relatively limited body of work supports the idea of neurotransmitter-induced lineage specification. Although, the exact mechanism by which this specification occurs is still not completely understood, it is thought to require neurotransmitter receptor-ligand binding257. For example, Zhou and Iacovitti demonstrated that replacing dopamine or serotonin with their receptor agonists was as equally effective at specifying differentiation down a certain lineage as the neurotransmitters themselves. Furthermore, co-cultures with pathways specific antagonists completely blocked the phenotype- inducing effects of neurotransmitters327. Given the convergence of many of these neurotransmitter signalling pathways on PKC, it suggests that in addition to common downstream effects, such as the induction of BDNF and other neurotrophic factors, there

! "()! may be differences in collateral neurotransmitter-induced pathways or in other regulators at the DNA level accounting for these observed phenotypic differences. It is also possible that lineage specification is a result of neurotransmitters promoting the survival of certain neuronal subtypes (Fig 6.2). Although such as indirect mechanism may also explain changes in the neurotransmitter phenotype, the hypothesis that neurotransmitters can direct the generation of specific neurons provides an elegant way for the brain to appropriately respond to fluctuating neurotransmitter levels. Such a mechanism is supported by BrdU labelling experiments demonstrating the generation of newly formed neurons following neurotransmitter treatment167,218. By selectively increasing the output of only particular neurons that are capable of responding to certain neurotransmitters present in excess levels, the CNS can match the supply and demand of specific neurotransmitters is a very controlled manner. Generating serotonin-responsive neurons in a landscape of excess dopamine, for instance, would provide little if any biological benefit. Cooperative stimulation of non-neurotransmitter ligands such as FGF- 8 and SHH has already been implicated in the preferential induction of dopaminergic neurons245, and thus seems plausible that other ligands, including neurotransmitters, may also have this potential. Further work is needed to precisely understand the specific role of these neurochemicals in this neuronal lineage specification process. Experiments looking at differences in gene or protein expression among cultures differentiated in the presence of different neurotransmitters may help determine if such specification signals exist in this context. Another possibility for specifying unique lineages given the convergent signalling of these different neurotransmitters may be due the activation of specific BDNF isoforms. Metsis et al showed that different neurotransmitters activate different isoforms of BDNF313. Subtle downstream changes due to differential activation of BDNF and other neurotrophic isoforms may explain how differences in lineage specification may be derived in the presence of different neurotransmitter signals. Further work into this area is needed to properly understand the mechanisms underlying these processes. The ability to not only regulate neurogenesis, but also to have control over the neurotransmitter phenotypes of these newly formed neurons will undoubtedly be a breakthrough in regenerative medicine. Chemical and genetic screens that systematically evaluate the

! "(*! induction of particular neurotransmitter genetic programs could help provide answers to this problem. Lastly, many of the aforementioned studies have demonstrated the generation of serotinergic neurons in the presence of serotonin and the generation of dopaminergic neurons in the presence of dopamine. Although this may be the simplest demonstration of lineage specification by neurotransmitters, the production of specific neurons that produce more of a particular neurotransmitter already in “excess” (i.e. generation of dopaminergic neurons in the presence of dopamine) may be counterintuitive. Increasing the density of neurons expressing the specific receptors able to respond to this initial stimulus (as in the case of Schneider et al) may be more advantageous. A more careful understanding of the particular types of neuronal connections found within the brain may be required before we can understand and properly interpret which particular neurotransmitter-induced subtype specification programs makes biological sense. As of now, it seems evident that the differentiation suptypes induced by different neurotransmitters (even those signalling through similar intracellular pathways) are unique amongst one another. Identifying the specific expression differences that can explain these patterns may also help better understand the algorithm NSCs use to properly organize the circuitry found in the mature brain.

6.8 LINEAGE PRIMING AND NEUROTRANSMITTER HETEROGENEITY IN NSCS

The study of neurotransmitter expression, production and response has traditionally been reserved for mature neuronal and astrocytic subtypes. In contrast to this, NSCs and their immediate precursors have been most typically described by the expression of primitive, stem cell specific markers. Such a naïve description was once the predominating model in hematopoietic stem cells (HSCs) as well. However, it is now accepted that even prior to the commitment to either a myeloid or erythroid cell fate, HSCs express genes belonging to both of these mutually exclusively mature hematopoietic lineages250. Although not yet widely appreciated, similar observations have been made in neural stem cell compartments; where expression of neuronal subtype- specific receptors not only exists, but is found to be heterogeneously distributed167,168,213. For example, individual cultures of DRD2+ and DRD2- cells isolated from the SVZ of

! ")+! rats were both capable of forming stem cell derived colonies when grown in vitro213. Although the multi-potentiality of both fractions was not tested, the ability of both DRD2-positive and negative fractions to exhibit the NSCs property of self-renewal, suggests that DRD2 expression status may not distinguish the NSC population from their more mature progeny. One interpretation of this heterogeneous DRD2 expression is that the NSCs in the adult brain are regionally-restricted251. Evidence from the precise anatomical dissection of radial glial cells indicates that the lineage potential of cells isolated from different regions of the SVZ depends on their anatomical location; a characteristic that could not be reversed251. Although this study elegantly demonstrated this property, it is possible that the populations of cells dissected, although primitive, may represent restricted self- renewing progenitors rather than bonifide stem cells that lie upstream to these more committed cells. The ability to steer NSCs down particular lineages with different neurotransmitters suggests that less precise dissection of these radial glial cells for typical in vitro cultures may consequently also isolate true stem cells from surrounding structures. Another explanation of this varied neurotransmitter-receptor status in NSCs arises from observations that stem cells characteristically maintain large proportions of open chromatin and undergo simultaneous transcriptional activity at promoters associated with both precursor and differentiated cell identities236. This may result in stochastic gene expression patterns that “prime” NSCs for the rapid activation of certain lineage-specific genes during differentiation. Such mechanisms have been implicated in the heterogeneity of a number of stem cell populations253. In embryonic stem cells, the expression of the stem cell markers Nanog and Stella has been shown to spontaneously and reversibly transit between varying levels247,248 that dictate differential responses to particular stimuli248. Hematopoietic stem cells have also been shown to express stochastic levels of Sca-1 that compartmentalize cells with distinct responses to certain stimuli249. Data discussed in Chapter 5 also support this concept, as hNSC pools exhibit patterns of neurotransmitter pathway gene expression that are reversible, equilibrating, and heterogeneous (Fig 6.4). Distinct from these other reports, where the functionality of the differentially expressed genes is yet to be determined, I show that phase varying

! ")"! expression of neurotransmitter receptors directly affects the responsiveness of only specific NSC subpopulations to relevant signals. The simultaneous superposition of varying neurotransmitter pathway gene expression levels may create a complex system where subpopulations within the NSC compartment can simultaneously respond to different cues (Fig 6.5). This capacity of hNSCs to reversibly express different lineage- specific gene programs also suggests that this lineage-primed state may engage with the local neurochemical environment to specify different NSC responses, and possibly specify appropriate neuronal or differentiated cell types177. This type of biological response draws parallels to phase varying patterns seen in pathogens, whereby individual pathogens are primed to respond to different host defences by random phase variation in gene expression254-256. Furthermore, stochastic phase variation in the hNSC compartment may intrinsically limit the NSC responses to signals that promote self-renewal or differentiation259. This strategy would serve to preclude an excess response to any given signal and thereby mitigate potential depletion of the NSC pool. As I have shown in vitro, stochastic variation would also allow recovery from depletion of any given subpopulation, thereby maintaining a fully competent multi-potential stem cell compartment over time (Fig 6.6 and Fig 6.7). Interestingly, I note that the frequency of hNSCs expressing GABRB2 (~58%) compared to DRD2 (~1%) mirrors the relative abundance of their corresponding mature neuronal subtypes found in the adult brain. However, while an appealing hypothesis, the roles of different neurochemical niches in NSC commitment and self-renewal requires further definition in vivo. As discussed, the heterogeneous expression of neurotransmitter-receptors on NSCs may act to limit the self-renewing and differentiation potential of NSCs. Such a model would suggest that dramatic changes in the neurotransmitter landscape (and/or the expression of their respective receptors) may consequently alter the balance in NSC biology. Later on in the discussion, I speculate on the potential pathological effects of mutations in neurotransmitter pathway gene that may alter the basal activity of NSCs. I use two examples to illustrate this concept; (1) Cancer (decreased neurotransmitter signalling) and (2) the autism spectrum disorder Rett syndrome (increase neurotransmitter signalling).

! ")#!

Serotonin Dopamine NMDA

Serotonin Dopamine NMDA

Figure 6.4 | Stochastic variation of neurotransmitter pathway gene expression in NSCs. My data suggest that neurotransmitter pathway genes are heterogeneously expressed in NSC populations at both the mRNA and protein levels. These variations are reversible and transit between one another. This heterogeneity also limits the proportions of cells capable of responding to particular stimuli.

! ")$!

a. Dopamine Glutamate Serotonin

Serotonin Dopamine

NMDA Homogenous stem cell compartment

b. Dopamine Glutamate Serotonin

Dopamine Serotonin

NMDA Heterogeneous stem cell compartment

Figure 6.5 | Simultaneous actions of different neurotransmitters on NSC populations (a) Model of NSC pools homogenously expressing a variety of neurotransmitter receptors that are sensitive to ligands found in neighboring environments (b) Model of NSC pools heterogeneously expressing a variety of neurotransmitter receptors that are sensitive to ligands found in neighboring environments. In the first model, the signaling of multiple pathways simultaneously would allow for only a single outcome to occur at a particular instance in time. In the second model, the action of different neurotransmitters on different subpopulations of cells would facilitate a more complex and heterogeneous response to the different environmental cues.

! ")%!

Dopamine

Serotonin Dopamine

NMDA Homogenous stem cell compartment

Neurogenesis + Stem cell exhaustion

Homogenous stem cell compartment

Figure 6.6 | Consequences of neurotransmitter signaling in a homogenous stem cell compartment Homogenous expression of neurotransmitter receptors would suggest that increased chemical signals for a single neurotransmitter receptor in the environment could result in a pre-mature exhaustion in the NSC pool following neurogenesis.

! ")&!

Dopamine

Serotonin Dopamine

NMDA Heterogeneous stem cell compartment

Serotonin

NMDA Heterogeneous stem cell compartment

Neurogenesis + Stem cell maintenance

Serotonin Dopamine

NMDA Heterogeneous stem cell compartment

Figure 6.7 | Consequences of neurotransmitter signaling in a heterogeneous stem cell compartment Phase variation in the expression of neurotransmitter receptors may act to limit neurotransmitter actions in NSC populations. Following neurogenesis of a particular neurotransmitter responsive subpopulation, the remaining NSCs can re-equilibrate the full variety of receptors found in the initial population. This heterogeneous response allows for more controlled changes in the NSC population to neurotransmitters and may help prevent the complete exhaustion of these precursor pools during development and adulthood. Note: In this cartoon each cell state is only shown to express one

! ")'! neurotransmitter receptor. This was done for simplicity. It is yet to be determined if NSCs express one or more neurotransmitter receptors at a time. It is quite possible that NSCs are expressing one, two, three or even more receptors at a time, all phase varying at their own rates.

! ")(!

6.9 STOCHASTIC GENE EXPRESSION IN NSCS

In Chapter 5, I demonstrate that phenotypically homogenous undifferentiated populations of NSCs235 express functional neurotransmitter pathways in a heterogeneous, reversible and perhaps stochastic manner. Such cell-to-cell variation has also been described elsewhere. For example, expression of type 1 pili in isogenic populations of Escherichia coli phase varies with time328-332. Variations in the cell cycles of synchronized E. coli also suggests that random factors contribute to generating uncoordinated biological systems333-336. Cell-to-cell genetic variability has also been noted in many mammalian cell populations337-339. For example, the expression of cell surface markers such as CD2 has been shown to be stochastic in mammalian cells340. Many suggest that these temporal variations in the proteins levels of clonally derived cells acts as a natural mechanism of generating genotypic, phenotypic and functional diversity in isogenic populations341-345. “Stochastic events”, a term often used to describe these processes, refer to cases by which the distributions of biological characteristics seen in individual cells are a function of random events. Here I review some of the proposed mechanisms of how stochastic gene expression is generated in cells and discuss their implications to stem cell pools. There are a number of hypotheses about how and why such cell-to-cell variation exists. Some argue that loose regulation of protein synthesis aids in reducing the high energy costs associated with gene transcription. In their simplified model, the relative abundance of particular proteins within a cell is dictated by the number of transcripts made, and the resulting number of protein copies made from each mRNA transcript by cytoplasmic ribosomes328. Assuming each cell had equal numbers of ribosomes, generating large enough numbers of transcripts would saturate the ribosome activity and result in an equal distribution of protein synthesis between each cell. Creating such a bottleneck in the kinetics of translation would come at a high transcriptional energy cost to the cell. A perhaps more efficient process, would be to transcribe relatively fewer mRNA molecules and rely on ribosomes to generate multiple protein copies from each transcript. Given that this process involves the co-incidental interaction of rare molecules, the cost of this more energy efficient process is an inherent cell-to-cell noise in proteins numbers produced and their associated downstream responses. A skewed distribution of

! "))! proteins per transcripts, which occurs when there are relatively few transcripts generated following promoter activation it thought to produce large inter-cell variability328. This model argues that the balance between each cell’s energy and functional requirements dictates the degree of cell-to-cell variability of each protein within a population. Such variability in proteins levels can alter the kinetics and expression levels of other proteins and lead to multiple and additional layers of cellular variation328. This cascade of events eventually leads to the evolution of multiple subpopulations of cells with different phenotypic and functional properties. This noise-driven variability is thought to provide a protective mechanism for cells against a wide range of environmental conditions328. Although such stochastic transcriptional effects on protein numbers have also been shown to occur in clonal population of eukaryotic cells346, others argue that cell-to- cell variability is derived from perhaps different mechanisms in higher eukaryotes237. They suggest that cellular differences in mammalian cells arise from bursts of protein production occurring at variable sizes and points in time347,348 and unrelated to variable translation rates of rare mRNA transcripts349,350. These random bursts of mRNA transcription are large, random and not related to extrinsic transcriptional activators237. Large differences between cells are thus derived from the length of time a gene locus and its promoter remain in an activated state. Prolonged lengths in the intervals between promoter activation, fewer transcription factor binding sites, and levels of activator proteins associated with the gene of interest also lead to large cell-to-cell variability237. The number of bursts on the other hand appears to be random and depend on chromatin remodeling351,352. Conversely, it is thought that variation in vital proteins may be buffered at the protein level via slow degradation rates. Although the consensus about which of the above mechanisms drives heterogeneous gene expression in clonally derived stem cell populations has yet to be characterized, data suggests that stochastic gene expression also occurs in stem cells. For example, even prior to lineage commitment, multipotent stem cell co-express a variety of transcription factors and genes exclusively associated with specific mature lineages250,353. This loose regulation of gene transcription is supported by single-cell RT-PCR experiments demonstrating the co-expression of multilineage genes in hematopoietic

! ")*! stem cells250. Through chemical interrogation, I also demonstrate that homogenous populations of NSCs235 express genes and functional protein in a heterogeneous and reversible manner. Many critics have argued that stochastic patterns of gene expression as detected by sensitive methods like RT-PCR may represent artifacts and be inconsequential236. In defense to this, data I presented in Chapter 5 supports the presence of functional proteins that heterogeneously control both gene activation and differentiation. At the chromatin level, different lineage specific genes are found to be simultaneously made accessible in relatively homogenous cell popluations353. Similarly, even without any detectable transcription, chromatin at genomic regions containing developmental genes is relaxed and “primed” or “poised” to allow for transcription to occur354. Unlike mature cells, where the chromatin of lineage specific genes are made fully available, the loci of these genes in uncommitted cells is found to be only partially remodeled177,353. Similar to the models described in prokaryotes, this partial accessibility may lead to only low levels of transcripts and result in stochastic protein production in individual cells. More homogenous levels of lineage-committed genes may require full access to these chromosomal regions and only occur during or following lineage commitment. Although the nomenclature used to characterize the multiple receptors found on mature cellular entities of the brain still requires improvement, stochastic expression of these genes in NSCs may allow for deterministic outcomes following activation. For example, some suggest that differentiation is a non-specific process resulting in the generation of randomly assorted cells of different phenotypes304. When coupled with data presented by others257,258,322, this heterogeneous expression of neurotransmitter pathway genes may helps regulate not only the production of different cellular CNS fates, but also steer the production of different neuronal subtypes. McAdams and Arkin argue that the existence of lineage choices that can be precisely regulated, does not need to be separated from stochastic gene expression patterns328. In fact, it is hypothesized that these expression patterns allow for a certain amount of flexibility in cellular responses, while still maintaining probabilistic and reproducible outcomes328.

! "*+!

Interestingly, I observed large differences in the expression of particular neurotransmitter receptors among cells. For example, mRNA of the GABA receptor, GABRB2, was found in the majority of cells (~60%) profiled; while acetylcholine and dopamine receptor expression was found in fewer cells (~1%). These differences in “noise” may have evolved and be a function of chromosomal positions that allow ubiquitous CNS neurotransmitter programs (i.e. GABA) to be more frequently sampled than more specialized neurotransmitter classes (i.e. dopamine). Such an organization of genes on chromosomes is noted for cell cycle regulators and results in domain-level noise that is a function of the frequency of chromatin remodeling at the specific gene locus. Essential cell cycle genes for example, exhibit less noise than non-essential cell-cycle genes355. Genes found at chromatin that is frequently open exhibit less noise than genes found at sites where chromatin is predominantly found in an inactive conformation355. This process may have evolutionary evolved to allow for controlled numbers of cell fates while still maintaining the benefits that heterogeneity and stochasticity offer to stem cell populations. For example, these patterns may allow for the major neurotransmitter pathways to be ubiquitously expressed throughout the CNS and favor more specialized NT systems to present only in specific regions of the brain; dopaminergic cells in the subsantia nigra. As described by Enver and colleagues, the endless combinations of how the 30,000 genes in the human genome are expressed, and the existence of only one hundred or so cell types, suggests that the expression of genes is not completely random253. It is thus possible that cells are only capable of responding when stochastic events generate specific attractor states; co-expression of a system of genes that allow a cell to generate a specific response to its surrounding environment. Large single cell mRNA or protein profiling would be needed to address this prospect. No one has yet to definitively address if stochastic multi-lineage gene expression in individual cells occurs simultaneously at low levels or through bursts or fluctuations between different genetic programs250. Although my results suggest that cells fluctuate between different programs, live mRNA fluorescent imaging of cells is needed to properly address if different lineage programs are constitutively simultaneously transcribed or if they follow a more burst-like mechanism250.

! "*"!

Another hurdle with models supporting low protein expression, is understanding the mechanisms of how protein products are co-localized once made. Modeling experiments suggest that low levels of protein expression that are not co-localized with their interacting partners may never functionally interact before the proteins are degraded356. Most of the human genome (95%) is expressed at levels of 0.5-5 mRNA transcripts/cell357. It is difficult to know which mechanism applies to which genes and in what circumstances. Whichever mechanism may account for this variation in stem cells, the presence of reversible expression of neurotransmitter pathway genes in these cells implies their expression is stochastic in nature. Whatever the origin of this random process may be, these expression patterns allow for functionally distinct subpopulations to be generated and maintain. Given the importance of stem cells in the production, integrity and repair of mammalian organ, these mechanisms may act to protect stem cells against catastrophic environmental conditions328. Further work into understand these processes may lead to better protocols for manipulating these cells both ex vivo and in vivo.!

6.10 REDUCED NEUROTRANSMITTER SIGNALLING, NSCS AND CANCER

Brain tumours are maintained by rare cancer stem-like cells24,103,132 suggesting that the transformation of normal neural stem cells (NSCs) or their close downstream progenitors may be the initiating event in brain cancers. Such a hypothesis is further supported by the large number of molecular and functional similarities shared between BTSCs and NSCs105,132. Given the inhibitory and/or pro-neurogenic effects of various neurotransmitters pathways on NSCs, the endogenous release of neurotransmitters from neighbouring neurons may act as an intrinsic mechanism that limits the non-physiological expansion and carcinogenic potential of NSCs is the adult brain. Pathological changes in neurotransmitter levels may thus further limit or potentiate brain tumour initiation and/or expansion. In support of this, recent genome-wide analyses of human gliomas demonstrates that mutations in neurotransmission pathways including sodium, potassium, and calcium

! "*#! channels are common226. Carcinogenic mutations are also thought to occur in the serotonin, dopamine, acetylcholine, glutamate and GABA genes226; pathways know to regulate the self-renewal and differentiation potential of NSCs. Interestingly, unlike mutations in tumour suppressors genes (i.e. PTEN, TP53, RB1) that are commonly seen in many cancers (i.e. pancreases, breast, colon), alternations in genes associated with neurotransmission pathways are unique to gliomas226. These genetics differences suggest a vital role of neurotransmitters in regulating neural precursor pool size and promoting neurogenesis. Interestingly, chromosomal regions (i.e. 17p13.3) that are commonly lost in malignant astrocytomas227; also harbour neurotransmitter receptors (i.e. TRPV1 and TRPV3) implicated in regulating NSC expansion168. The loss of particular receptors from NSCs that render them unresponsive to particular neurotransmitters may deregulate self- renewal in NSCs and promote transformation. These results also have implications for the potential effectiveness of neuromodulators as anti-cancer agents. Given that neurotransmitter receptors are often mutated in cancer, the effectiveness of a particular agent may also be dictated by the specific mutations found in each cancer. Pre-screening for these mutations, and choosing agents that target conserved pathways could help overcome such obstacles. Furthermore, in Chapter 5 I show that many of these receptors appear to be heterogeneously expressed in NSC populations. This suggests that single agent therapy may not be effective at targeting and removing all cancer initiating cells within a tumor. A multi-drug cocktail regimen that spans a large array of neurotransmitter pathways may be required to efficiently deplete cancer stem cell to numbers needed to achieve complete and long term remission rates in brain tumor patients. The prospect that changes in the responsiveness of NSCs to neurotransmitter levels can lead to cancer growth, implies that restoring these levels though pharmacological agents may prove as an effective chemotherapeutic strategy. Activation of the metabotropic glutamate receptor 4 (mGlu4); a candidate target for the treatment of both Parkinson’s disease228,229 and generalized anxiety disorders230, limits medulloblastoma formation in mice192. Similar to this, mGlu2/3 antagonism inhibits the growth of human glioma cells both in vitro and in vivo231,232. Drugs spanning all the major neurotransmitter pathways (including the clinically prescribed dopamine agonist

! "*$! apomorphine233) have now been identified as inhibitors of cancerous neural precursor cell proliferation in vitro169. Well-tolerated neuropharmacological agents used in standard clinical practice may thus offer new avenues for brain cancer therapy. In further support of this, data collected from retrospective clinical studies suggesting that patients with a wide variety of neuropsychiatric disorders have decreased brain tumour incidence. This reduction was speculated to be derived from the use of drugs that collaterally affect the normal neural precursor compartment, and thereby limiting the population that is suspected to give rise to brain tumours. Clinically used neuropsychiatric drugs may thus be attractive candidates for redeployment as brain cancer therapeutics.

6.11 INCREASED NEUROTRANSMITTER SIGNALLING, NSCS AND RETT

SYNDROME

Given that deletions in neurotransmitter pathway genes may promote cycling and increased pools of progenitors in the brain, one would hypothesize that mechanisms that promote the expression of particular neurotransmitter pathways may lead to smaller progenitor pools as a consequence of premature differentiation during development. Here I review our current understanding of Rett syndrome (RTT), and hypothesize how elevated neurotransmitter activity may alter proper CNS maturation and lead to neurodevelopemental disorders. RTT is a severe neurodevelopemental disorder characterized by autistic-like deficits that are pathologically associated with decreased levels of excitatory glutamineric neurons and disproportionally higher levels of inhibitory GABA signals358. Although brains of RTT patients exhibit abnormal neuronal morphology, the lack of neuronal death suggest this disease represents a neurodevelopemental rather than a neurodegenerative disorder359. Molecularly, the majority of RTT is caused by de novo mutations in Methly-CpG- binding protein 2 (MeCP2)360. MeCP2 binds to specific regions of DNA, recruits histone deacetylases (HDACs) and forms multimeric protein structures at methylated CpG sites that transcriptionally repress corresponding gene targets361-367. In RTT, it is thought that

! "*%! the lack of functional MeCP2 forces altered gene expression patterns that compromises proper neuronal maturation during development. One of the most well characterized targets regulated by MeCP2 is Dlx5368 (Fig 6.8). Dlx5 belongs to the GABAergic family of genes and is specifically involved in inducing GABAergic neuronal differentiation368-370. Dlx5 has been shown to be expressed in neural progenitors found in the subventricular zone as well as newly formed neurons during post-mitotic differentiation371. The inability of MeCP2 mutants to suppress dlx5 may result in its continued expression throughout development and neuronal differentiation368. Such aberrant neurotransmitter expression patterns may alter the relative sizes of the subpopulations able to respond to particular signals in the constant neurotransmitter environment found in the adult brain may consequentially change the proportions of neuronal subtypes to those seen in RTT; increased GABAergic neurons at the expense of other subtypes (i.e. glutaminergic) (Fig 6.9).

! "*&!

a)

!"#$%"&''

Sin3A HDAC MECP2

!"#$%

b)

HDAC Sin3A #()*+$,&-''

!"#$%"&''

MECP2 !"#$%

&'% &'% &'% &'%

Figure 6.8 | Regulation of Dlx5 gene expression by MeCP2 (a) MeCP2 (methyl-CpG-binding protein 2) suppresses transcription of genes associated with promoters containing methylated CpGs. This is mediated by binding of MeCP2 to methylated DNA followed by recruitment and formation of chromatin remodeling complex containing the transcriptional repressor SIN3A and histone deacetylases (HDACs). HDAC activity promotes condensation of chromatin and decreased transcriptional activity of neighboring genes. Dlx5 is a GABAergic transcription factor associated with methyl-CpG-binding domains and is regulated by MeCP2. (b) Phosphorylation or mutations (as in the case of RTT) in MeCP2 can disrupt DNA- MeCP2 interactions and leads to increased transcription of these genes. Dlx5 is found to be overexpressed in RTT models.

! "*'!

a) Normal stem cell compartment

Dopamine Glutamate

GABA

b) MeCP2 deficient stem cell compartment

Dopamine Glutamate

GABA

Figure 6.9 | Altered neurotransmitter pathway gene expression in NSC and neural progenitors may affect proportions of mature neuronal phenotypes (a) NSC compartments heterogeneously express a variety of different neurotransmitter receptors. The proportion of cells expressing each neurotransmitter pathway gene may dictate the landscape of subtype specific neuronal lineages when NSC compartments are exposed to environmental cues. (b) MeCP2 mutation in RTT may decrease repression of MeCP2 targeted genes such as the GABAergic transcription factor dlx5. Changes in the expression of these genes can alter the proportion of NSCs expressing particular neurotransmitter transcripts (i.e. increase cells expressing GABAergic genes) and may significantly alter the proportion of neuronal subtypes induced by neurotransmitter mediated differentiation.

! "*(!

Elevated levels of GABA have been noted in similar neurodevelopmental disorders such as Angelman syndrome372,373. This inability to turn off dlx5 in RTT may account for the misdistribution of glutaminergic and GABAergic neurons during development and help explain how alterations in MeCP2 can account for the reduced cortical activity and symptoms seen in this disease358. Further support that increase GABA signalling may increase the proportions of GABAergic neurons stems from studies suggesting that specific neurotransmitters signals may promote neuronal subtypes that correlate with the particular neurotransmitter receptor activated257. The concept of lineage priming may also shed light on additional mechanisms that may contribute to alterations in the proportions of neuronal subtypes. Undifferentiated stem cells have been shown to simultaneously express a multitude of lineage specific genes at low levels well before lineage commitment. Chromatin structure from many of these sites remains relaxed and accessible for gene transcription during this undifferentiated state177,374. Lineage specification can thus be thought to require, in conjunction with an up-regulation of genes associated with the specific adopted lineage, the repression of the remaining genes not required for that particular lineage. Transcriptional repressors like MeCP2 may orchestrate such steps during differentiation and maturation of stem cells and their progeny (Fig 6.10). Neurotransmitter induced changes in the expression of NMDA receptors in NSCs has been shown to occur through changes in HDAC activity303. The specific activity of co-repressors like MeCP2 may be involved in this regulation. If in fact NSCs simultaneously express a multitude of neurotransmitter pathway genes213,218,302,375 and related transcription factors (i.e. dlx5371), silencing the expression of genes not required in particular neuronal populations may be required for proper maturation. Alterations in MeCP2 that hinder its inability to turn off dlx5 in the brains of RTT patients may explain the observed misdistribution of glutaminergic and GABAergic neurons and account for the reduced cortical activity and symptoms seen in RTT358.

! "*)!

Dopamine

PKC

GABA

Dopamine

PKC

MeCP2

Dopamine responsive mRNA MeCP2 regulated neurotransmitter gene mRNA

Figure 6.10 | MeCP2-mediated repression of gene transcription during differentiation In their undifferentiated state, NSCs express mRNA of a wide variety of neurotransmitter pathway genes and associated transcription factors. Neurotransmitter signaling has been shown to induce biased expression of lineage specific genes and induce differentiation. Induction of MeCP2 during neural differentiation may thus act to suppress the expression of certain neurotransmitter pathway genes not associated with the chosen neuronal lineage during differentiation.

! "**!

In addition to dlx5, MeCP2 has been shown to also target other neurotransmission pathway genes such as the sodium channel type II (SCN2A)376. Mutations in other members of this family (SCN1A377, SCN1B378) that are involved in neuronal voltage- gated sodium channels and that lead to phenotypic psychiatric behaviours in humans mimic those seen in MeCP2 mutant mice. The spectrum of ion channels and neurotransmitter pathway genes regulated by MeCP2 may therefore be much greater that currently appreciated. These findings generalize the importance of MECP2-mediated regulation of neurotransmitter pathway gene expression in NSCs and their more committed progeny. Interestingly, depressed BDNF levels have also been reported in brain harbouring MeCP2 mutations363. Given the repressive effects of GABAergic signalling on BDNF transcription313,314, over-representation of functional GABA-related genetic programs in MeCP2-mutant NSC and progenitors may limit the transcription of these neurotrophic factors shown to be required for proper neurogenesis. Changes in neurotransmitter pathway gene expression have been shown to be involved in a number of “stem cell” disorders. While our current understanding of these diseases and the mechanisms involved in neurochemcial regulation of NSCs are unclear, the clinical observations reported in this thesis strongly support a role of these pathways in regulating NSC biology. Further work in this area will help clarify the mechanisms involved.

6.12 FUTURE DIRECTIONS AND CONCLUDING REMARKS

This thesis has provided experimental and clinical evidence supporting a role of a large myriad of neurotransmitters in the regulation of both normal and cancerous neural precursor cells. Although others have reported similar results in normal stem cell populations, I feel this thesis has greatly expanded on the previous work and is one of the first (if not the first) to show their efficacy of these agents in regulating cancer stem cell populations. Furthermore, this thesis demonstrates and discusses the implications of the dramatic effects that endogenous or ectopic changes in neurotransmitter levels may have on normal neural development and biology. I hope this and future work into this

! #++! phenomenon will allow a better understanding of the biology of neurotransmitter pathways in NSCs and lead to the development of therapies to help treat a variety of CNS disorders. Perhaps one of the most interesting directions this project can take is understanding how specific neuronal neurotransmitter phenotypes can be generated following differentiation. Although others have already shown that the addition of dopamine and serotonin can lead to changes to the differentiation potential of these cells, the regulation of these genes is still unclear. In Chapter 5, I show that similar to these other neurotransmitters, acetylcholine also efficiently directs the differentiation of NSCs. I hope that by expanding the array of neurotransmitter pathways known to be simultaneously active in these cells, the methodologies and options for specifically and precisely altering the neurotransmitter phenotypes will be significantly improved. Breakthroughs in this field will undoubtedly advance the therapeutic options for a number of CNS diseases. The characterization of the heterogeneous neurotransmitter expression patterns described in this work is another significant advancement in this field. My findings have implications for efforts to purify populations of stem cells for applications in regenerative medicine. If the intrinsic heterogeneity I observe in hNSCs applies to other stem cell types, the potential for generation of specifically desired cell types may be limited even in the presence of proper stimuli. As noted, even ES cells appear to exhibit phase variation in the expression of key fate determinants such as Nanog260. Artificial stabilization of specific NT receptor expression patterns may be needed to efficiently produce neuronal subtypes in the treatment of Alzheimer’s, Parkinson’s and other neurodegenerative diseases. Finally, I speculate that brain tumor stem cells may exhibit similar phase variation patterns as normal NSC cells; if so, single agent therapies may prove less than effective in the face of cancer cell phenotypic diversity261. Identifying chemical or genetic ways to manipulate the proportions of cells expressing particular receptors phenotypes may help overcome this heterogeneous phase-varying pattern of expression present in these cells. Given the multitude of receptors and pathways simultaneously at play in neural stem cells, additional work is needed before a sufficient appreciation of neurotransmitter effects of NSCs can be attained. I hope that the efforts and contributions described in this

! #+"! thesis highlight some novel biological observations regarding the phenotype and function of neurotransmitter pathways in both normal neural and brain cancer-derived populations of stem cells. Furthermore, I hope that the eventual clinical application of these findings will one day help improve the live of those suffering from a variety of CNS diseases.

! #+#!

REFERENCES

! L*! 740.2"!#*!4&!.2*!1.(A4-!%&.&'%&'A%"!_MM`*!!"#!$%&'(#)#!*+%!./"!aLYbc!V_MM`X*! _*! ,.-A'."!=*!,*-.$*#!$%&'(#/$&01#2#/+34('1#5667"!V#04-'A.(!1.(A4-!d9A'4&<"! #&2.(&."!,#"!_MMaX*! e*! @-9H("!=*:*"!:'D%A90O"!7*!>!d(%3*#)#?'@!1./"!gM_YLL!V_MM`X*! g*! +($G%9("!#*,*!;H9!B4(4&'A!/'&%!V09-4!9-!24%%X!&9!A.(A4-*!=$0#8'9#!$%&'(!2"! LgaYc_!V_MMLX*! c*! 8.(./.("!6*!>!h4'(O4-B"!)*#*!;/4!/.220.-3%!9F!A.(A4-*!!'**!233"!gaYaM! V_MMMX*! a*! )4<."!;*"!=9--'%9("!d*7*"!12.-34"!=*T*!>!h4'%%0.("!5*:*!d&40!A422%"!A.(A4-"!.(G! A.(A4-!%&40!A422%*!=$04('!424"!LMgYLL!V_MMLX*! `*! @.9"!d*!4&!.2*!,2'90.!%&40!A422%!D-909&4!-.G'9-4%'%&.(A4!O!,'22'2.(G"!6*,*!1.(A4-!O'929B!,'22'2.(G"!6*,*!:4$3.40'.!%&40!A422%!.(G!&/4!4C92$&'9(!9F!A.(A4-Y %&40YA422!-4%4.-A/*!=$0#8'9#!$%&'(!."!eLLY_L!V_MMgX*! LL*! ?.-G.2"!)*"!12.-34"!=*T*!>!=9--'%9("!d*7*!#DD2<'(B!&/4!D-'(A'D24%!9F!%&40YA422! O'929B!;'22"!7*i*!1<&929B'A.2!G409(%&-.&'9(!9F!&/4!A29(.2!(.&$-4! 9F!%D244(!A929('4%!G4-'C4G!F-90!&-.(%D2.(&4G!09$%4!0.--9H!A422%*!=$04('! 256"!fg_Yf!VLbceX*! Le*! d'0'(9C'&A/"!:*"!=A1$229A/"!i*#*!>!;'22"!7*i*!;/4!6'%&-'O$&'9(!9F!1929(!=A1$229A/"!i*#*!=9$%4!0<4290.!&$09-!%&40! A422%Q!.!D-'0.-!j.(!64-!,..B"!8*!#!k$.(&'&.&'C4!#%%.!d.209("!d*i*!?-'0.-!:4C'("!#*,*!89%&!-4%'%&.(A4!&9!A.(A4-*!12'('A.2! 4ED4-'04(&%!O!6'-3%"!?*@*!1.(A4-!%&40!A422%Q!.&!&/4!/4.GH.&4-%!9F!&$09-! G4C429D04(&*!"%%4#8'9#:$0<-*!0"!LagY`b!V_MMaX*! Lb*! @9((4&"!6*!>!6'A3"!7*i*!8$0.(!.A$&4!0<429'G!24$340'.!'%!9-B.('l4G!.%!.! /'4-.-A/!12.-34"!=*T*! ?-9%D4A&'C4!'G4(&'F'A.&'9(!9F!&$09-'B4('A!O-4.%&!A.(A4-!A422%*!:(-&#=$0*#"&$@# C&+#D#C#"!233"!eb`eY`!V_MMeX*!

! #+$!

_L*! h.(B"!7*1*!4&!.2*!8'B/!24C42!4(B-.F&04(&!9F!RN6nd156!0'A4!O!;.(B"!6*,*! 8'4-.-A/'A.2!9-B.('l.&'9(!9F!D-9%&.&4!A.(A4-!A422%!'(!E4(9B-.F&!&$09-%Q!&/4! 16ffo.2D/._O4&.Lo!A422!D9D$2.&'9(!'%!4(-'A/4G!'(!&$09-Y'('&'.&'(B!A422%*! !$%&'(#8'1!76"!cabcY`Mg!V_MMaX*! _a*! 1922'(%"!#*;*"!@4--<"!?*#*"!8!=.'&2.(G"!R*7*!?-9%D4A&'C4! 'G4(&'F'A.&'9(!9F!&$09-'B4('A!D-9%&.&4!A.(A4-!%&40!A422%*!!$%&'(#8'1!7."! LMbfcYgL!V_MMgX*! _`*! )'AA'Yj'&'.('"!:*!4&!.2*!5G4(&'F'A.&'9(!.(G!4ED.(%'9(!9F!/$0.(!A929(YA.(A4-Y '('&'.&'(B!A422%*!=$04('!44."!LLLYg!V_MMaX*! _b*! Np@-'4("!1*#*"!?9224&&"!#*"!,.22'(B4-"!d*!>!6'A3"!7*i*!#!/$0.(!A929(!A.(A4-!A422! A.D.O24!9F!'('&'.&'(B!&$09$-!B-9H&/!'(!'00$(9G4F'A'4(&!0'A4*!=$04('!44."! LMcYLM!V_MMaX*! eM*! :'"!1*!4&!.2*!5G4(&'F'A.&'9(!9F!D.(A-4.&'A!A.(A4-!%&40!A422%*!!$%&'(#8'1!76"!LMeMY a!V_MMaX*! eL*! h$"!1*!4&!.2*!d'G4!D9D$2.&'9(!A422%!'%92.&4G!F-90!04%4(A/<0.2!(49D2.%0%!/.C4! &$09-!'('&'.&'(B!D9&4(&'.2*!!$%&'(#8'1!76"!`_LcY__!V_MMaX*! e_*! dA/.&&9("!;*!4&!.2*!5G4(&'F'A.&'9(!9F!A422%!'('&'.&'(B!/$0.(!042.(90.%*!=$04('! 4.2"!efgYb!V_MM`X*! ee*! q/.(B"!d*!4&!.2*!5G4(&'F'A.&'9(!.(G!A/.-.A&4-'l.&'9(!9F!9C.-'.(!A.(A4-Y'('&'.&'(B! A422%!F-90!D-'0.-FG# ;'H$0-*!0"!`eYb_!VLbafX*! ec*! T-'4G4(%&4'("!#*7*"!,9-%3.m."!7*T*!>!+$2.B'(."!R*R*!T'O-9O2.%&!D-4A$-%9-%!'(! (9-0.2!.(G!'--.G'.&4G!09$%4!/40.&9D9'4&'A!9-B.(%*!>FG#;'H$0-*!4"!_caYaf! VLbacX*! ea*! )4<(92G%"!@*#*"!;4&l2.FF"!h*!>!h4'%%"!d*!#!0$2&'D9&4(&!i,TY-4%D9(%'C4!%&-'.&.2! 40O-<9('A!D-9B4('&9-!A422!D-9G$A4%!(4$-9(%!.(G!.%&-9A<&4%*!)#='4(-1&+!20"! fgcgYaf!VLbb_X*! e`*! h4'%%"!d*!4&!.2*!=$2&'D9&4(&!1Rd!%&40!A422%!.-4!D-4%4(&!'(!&/4!.G$2&! 0.00.2'.(!%D'(.2!A9-G!.(G!C4(&-'A$2.-!(4$-9.E'%*!)#='4(-1&+!27"!agbbYcMb! VLbbcX*!

! #+%! eb*! 64#(B42'%"!:*=*!@-.'(!&$09-%*!=#>%3*#)#?'@!144"!LLfY_e!V_MMLX*! fM*! 740.2"!#*!4&!.2*!1.(A4-!%&.&'%&'A%"!_MMa*!!"#!$%&'(#)#!*+%!.6"!feYcc!V_MMaX*! fL*! @$-B4-"!?*1*"!j9B42"!T*d*"!,-44("!d*@*!>!d&-'34"!;*#*!,2'9O2.%&90.!0$2&'F9-04!.(G! .(.D2.%&'A!.%&-9A<&90.*!?.&/929B'A!A-'&4-'.!.(G!D-9B(9%&'A!'0D2'A.&'9(%*! !$%&'(!.7"!LLMcYLL!VLb`gX*! f_*! ,.2.('%"!i*!>!@$A3(4-"!7*!1/409&/4-.D!@-9A3"!1*! ;409l9290'G4Q!.!-4C'4H!9F!'&%!G'%A9C4-<"!A/40'A.2!D-9D4-&'4%"!D-4YA2'('A.2! G4C429D04(&!.(G!A2'('A.2!&-'.2%*!!$%&'(#J('$0#8'9!01"!egYcL!VLbbaX*! fa*! 19/4("!=*8*"!79/(%9("!7*)*!>!?.lG$-"!)*!T99G!.(G!6-$B!#G0'('%&-.&'9(!6-$B! .DD-9C.2!%$00.-%3*#)#?'@!1.0"!b`aYbc!V_MMgX*! fb*! d&$DD"!)*!>!84B'"!=*i*!;.-B4&'(B!O-.'(Y&$09-!%&40!A422%*!=$0#E+-0'&<%-*!0."! LbeYf!V_MMaX*! gM*! ,92G0.("!d*#*!>!R9&&4O9/0"!T*!R4$-9(.2!D-9G$A&'9("!0'B-.&'9("!.(G! G'FF4-4(&'.&'9(!'(!.!C9A.2!A9(&-92!($A24$%!9F!&/4!.G$2&!F40.24!A.(.-!h4'%%"!d*!,4(4-.&'9(!9F!(4$-9(%!.(G!.%&-9A<&4%!F-90!'%92.&4G! A422%!9F!&/4!.G$2&!0.00.2'.(!A4(&-.2!(4-C9$%!%<%&40*!C&+'%&'!0.."!LaMaYLM! VLbb_X*! g_*! ;-9D4D4"!j*!4&!.2*!6'%&'(A&!(4$-.2!%&40!A422%!D-92'F4-.&4!'(!-4%D9(%4!&9!i,T!.(G! T,T!'(!&/4!G4C429D'(B!09$%4!&424(A4D/.29(*!K'9#E+-*!03/"!LccY``!VLbbbX*! ge*! ,.B4"!T*8*!d&40!A422%!9F!&/4!A4(&-.2!(4-C9$%!%<%&40*!!4((#IG+%#='4(-.+-*!/"! caLYc!VLbb`X*! gf*! ;40D24"!d*!;/4!G4C429D04(&!9F!(4$-.2!%&40!A422%*!=$04('!424"!LL_Ya!V_MMLX*! gg*! ,.B4"!T*8*!=.00.2'.(!(4$-.2!%&40!A422%*!C&+'%&'!0/6"!LfeeY`!V_MMMX*! gc*! ;-9D4D4"!j*"!1-.'B"!1*,*"!=9-%/4.G"!1*=*!>!C.(!G4-!+99<"!6*!;-.(%F9-0'(B! B-9H&/!F.A&9-Y.2D/.!($22!.(G!%4(4%A4(&!0'A4!%/9H!G4A-4.%4G!(4$-.2! D-9B4('&9-!A422!D-92'F4-.&'9(!'(!&/4!F9-4O-.'(!%$O4D4(G<0.*!)#='4(-1&+!26"! a`gMYb!VLbbaX*!

! #+&! ga*! h4'%%"!d*!4&!.2*!5%!&/4-4!.!(4$-.2!%&40!A422!'(!&/4!0.00.2'.(!F9-4O-.'(t!J('%@1# ='4(-1&+!25"!e`aYbe!VLbbcX*! g`*! ?.204-"!;*6*"!;.3./.%/'"!7*!>!,.B4"!T*8*!;/4!.G$2&!-.&!/'DD9A.0D$%!A9(&.'(%! D-'09-G'.2!(4$-.2!%&40!A422%*!?-*#!'**#='4(-1&+!/"!e`bYfMf!VLbbaX*! gb*! d4.O4-B"!)*=*!>!C.(!G4-!+99<"!6*!#G$2&!-9G4(&!(4$-9B4('A!-4B'9(%Q!&/4! C4(&-'A$2.-!%$O4D4(G<0.!A9(&.'(%!(4$-.2!%&40!A422%"!O$&!&/4!G4(&.&4!B<-$%! A9(&.'(%!-4%&-'A&4G!D-9B4('&9-%*!)#='4(-1&+!00"!La`fYbe!V_MM_X*! cM*! 8.22"!?*#*!>!h.&&"!T*=*!d&40!A422%Q!&/4!B4(4-.&'9(!.(G!0.'(&4(.(A4!9F!A422$2.-! G'C4-%'&<*!K'9'*-GH'%0!237"!cLbYee!VLb`bX*! cL*! ?9&&4("!1*d*!>!:94FF24-"!=*!d&40!A422%Q!.&&-'O$&4%"!A!C.(!G4-!+99<"!6*!d&40!.(G!D-9B4('&9-!A422%Q!&/4!D-40.&$-4! G4%4-&'9(!9F!-'B9-9$%!G4F'('&'9(%*!J('%@1#='4(-1&+!07"!L_gYeL!V_MMeX*! ce*! h4'%%0.("!5*:*!d&40!A422%Q!$('&%!9F!G4C429D04(&"!$('&%!9F!-4B4(4-.&'9("!.(G! $('&%!'(!4C92$&'9(*!!'**!233"!LgaYc`!V_MMMX*! cf*! :9'%"!1*!>!#2C.-4lY@$<22."!#*!?-92'F4-.&'(B!%$OC4(&-'A$2.-!l9(4!A422%!'(!&/4! .G$2&!0.00.2'.(!F9-4O-.'(!A.(!G'FF4-4(&'.&4!'(&9!(4$-9(%!.(G!B2'.*!:(-&#=$0*# "&$@#C&+#D#C#"!53"!_MafYa!VLbbeX*! cg*! ,9$2G"!i*"!@4<2'("!#*"!;.(.D.&"!?*"!)44C4%"!#*!>!d/9-%"!;*7*!:4.-('(B!4(/.(A4%! .G$2&!(4$-9B4(4%'%!'(!&/4!/'DD9A.0D.2!F9-0.&'9(*!=$0#='4(-1&+!0"!_cMYg! VLbbbX*! cc*! j4%A9C'"!#*:*!4&!.2*!5%92.&'9(!.(G!A29('(B!9F!0$2&'D9&4(&'.2!%&40!A422%!F-90!&/4! 40O-<9('A!/$0.(!1Rd!.(G!4%&.O2'%/04(&!9F!&-.(%D2.(&.O24!/$0.(!(4$-.2! %&40!A422!2'(4%!OFG#='4(-*!2.7"!aLY`e!VLbbbX*! ca*! )9<"!R*d*!4&!.2*!5(!C'&-9!(4$-9B4(4%'%!O!?.-.&'"!i*#*!5%92.&'9(!.(G!'(&-.A4-4O-.2! B-.F&'(B!9F!(9(&-.(%F9-04G!0$2&'D9&4(&'.2!40O-<9('A!/$0.(!1Rd!%&40!A422%*! )#='4(-0($4H$!27"!c`bYbe!VLbbbX*! af*! =9-%/4.G"!1*=*!4&!.2*!R4$-.2!%&40!A422%!'(!&/4!.G$2&!0.00.2'.(!F9-4O-.'(Q!.! -42.&'C42

! #+'! ag*! :4(G./2"!s*"!q'004-0.("!:*@*!>!=A+.<"!)*6*!1Rd!%&40!A422%!4ED-4%%!.!(4H! A2.%%!9F!'(&4-04G'.&4!F'2.04(&!D-9&4'(*!!'**!73"!g`gYbg!VLbbMX*! ac*! 89A3F'42G"!d*!>!=A+.<"!)*6*!5G4(&'F'A.&'9(!9F!0.m9-!A422!A2.%%4%!'(!&/4! G4C429D'(B!0.00.2'.(!(4-C9$%!%<%&40*!)#='4(-1&+!."!eeLMY_`!VLb`gX*! aa*! @-.l42"!1*r*!4&!.2*!d9E_!4ED-4%%'9(!G4F'(4%!.!/4&4-9B4(49$%!D9D$2.&'9(!9F! (4$-9%D/4-4YF9-0'(B!A422%!'(!&/4!.G$2&!0$-'(4!O-.'(*!"3+%3#!'**!4"!LbaY_Ma! V_MMgX*! a`*! ,-./.0"!j*"!+/$G<.39C"!7*"!i22'%"!?*!>!?4C(<"!:*!dNu_!F$(A&'9(%!&9!0.'(&.'(! (4$-.2!D-9B4('&9-!'G4(&'&<*!='4(-%!15"!afbYcg!V_MMeX*! ab*! 6p#09$-"!+*#*!>!,.B4"!T*8*!,4(4&'A!.(G!F$(A&'9(.2!G'FF4-4(A4%!O4&H44(! 0$2&'D9&4(&!(4$-.2!.(G!D2$-'D9&4(&!40O-<9('A!%&40!A422%*!:(-&#=$0*#"&$@#C&+# D#C#"!233!89::;!2"!LL`ccYa_!V_MMeX*! `M*! 1.'"!7*!4&!.2*!?-9D4-&'4%!9F!.!F4&.2!0$2&'D9&4(&!(4$-.2!%&40!A422!VRi?!A422X*!K'9# E+-*!0.2"!__LYfM!V_MM_X*! `L*! 8'&9%/'"!d*"!;-9D4D4"!j*"!i334-"!=*!>!C.(!G4-!+99<"!6*!R4$-.2!%&40!A422!2'(4.B4%! .-4!-4B'9(.22!#2C.-4lY@$<22."!#*! i,T!A9(C4-&%!&-.(%'&Y.0D2'F<'(B!(4$-9B4('A!D-4A$-%9-%!'(!&/4!.G$2&!O-.'(! '(&9!0$2&'D9&4(&!%&40!A422%*!='4(-%!17"!LM_LYef!V_MM_X*! `e*! 694&%A/"!T*"!1.'224"!5*"!:'0"!6*#*"!,.-A'.Yj4-G$B9"!7*=*!>!#2C.-4lY@$<22."!#*! d$OC4(&-'A$2.-!l9(4!.%&-9A<&4%!.-4!(4$-.2!%&40!A422%!'(!&/4!.G$2&!0.00.2'.(! O-.'(*!!'**!56"!aMeYLc!VLbbbX*! `f*! =9-%/4.G"!1*=*"!,.-A'."!#*6*"!d9F-9('4H"!=*j*!>!C.(!64-!+99<"!6*!;/4!.O2.&'9(! 9F!B2'.2!F'O-'22.-4(#)#='4(-1&+!2/"!acY`f!V_MMeX*! `g*! =$m&.O."!;*!4&!.2*!:'(4.B4Y-4%&-'A&4G!(4$-.2!D-4A$-%9-%!A.(!O4!'%92.&4G!F-90! O9&/!&/4!09$%4!(4$-.2!&$O4!.(G!A$2&$-4G!id!A422%*!K'9#E+-*!024"!LLeY_a! VLbbbX*! `c*! =.2.&4%&."!?*"!8.-&F$%%"!i*!>!,9&l"!=*!5%92.&'9(!9F!-.G'.2!B2'.2!A422%!O!:4D-'(A4"!?*!).G'.2!B2'.! D/4(9&!84'(&l"!R*!@-.'(!2'D'GYO'(G'(B!D-9&4'(!V@:@?XQ!.!(9C42! %'B(.2'(B!%<%&40!'(!&/4!G4C429D'(B!0.00.2'.(!1Rd*!='4(-%!20"!`bgYbM`! VLbbfX*! bM*! 50$-."!;*"!+9-(O2$0"!8*5*!>!d9F-9('4H"!=*j*!;/4!D-4G90'(.(&!(4$-.2!%&40!A422! '%92.&4G!F-90!D9%&(.&.2!.(G!.G$2&!F9-4O-.'(!O$&!(9&!4.-2

! #+(! bL*! sA/'G."!R*!4&!.2*!6'-4A&!'%92.&'9(!9F!/$0.(!A4(&-.2!(4-C9$%!%<%&40!%&40!A422%*! :(-&#=$0*#"&$@#C&+#D#C#"!56"!Lfa_MYg!V_MMMX*! b_*! r'("!#*8*!4&!.2*!#1Lee"!.!(9C42!0.-34-!F9-!/$0.(!/40.&9D9'4&'A!%&40!.(G! D-9B4('&9-!A422%*!E*--@!53"!gMM_YL_!VLbbaX*! be*! h4'B0.(("!#*"!19-O4'2"!6*"!8422H'B"!#*!>!8$&&(4-"!h*@*!?-90'('("!.!(9C42! 0'A-9C'22'Y%D4A'F'A!D92<&9D'A!040O-.(4!D-9&4'(!9F!&/4!.D'A.2!%$-F.A4!9F! 4D'&/42'.2!A422%"!'%!&.-B4&4G!&9!D2.%0.2400.2!D-9&-$%'9(%!9F!(9(Y4D'&/42'.2! A422%*!:(-&#=$0*#"&$@#C&+#D#C#"!54"!L_f_gYeM!VLbbaX*! bf*! d.H.09&9"!+*!4&!.2*!,4(4-.&'9(!9F!G9D.0'(4-B'A!(4$-9(%!'(!&/4!.G$2&!O-.'(! F-90!04%4(A4D/.2'A!D-4A$-%9-!A422%!2.O424G!H'&/!.!(4%&'(Y,T?!&-.(%B4(4*!)# ='4(-1&+!02"!e`bgYbMe!V_MMLX*! bg*! ='-.B2'."!d*!4&!.2*!#!(9C42!F'C4Y&-.(%040O-.(4!/40.&9D9'4&'A!%&40!A422! .(&'B4(Q!'%92.&'9("!A/.-.A&4-'l.&'9("!.(G!0924A$2.-!A29('(B*!E*--@!53"!gMLeY _L!VLbbaX*! bc*! ,.&4%"!=*#*!4&!.2*!1422!.(G!0924A$2.-!.(.2<%'%!9F!&/4!G4C429D'(B!.(G!.G$2&! 09$%4!%$OC4(&-'A$2.-!l9(4!9F!&/4!A4-4O-.2!/40'%D/4-4%*!)#!-HG#='4(-*!172"! _fbYcc!VLbbgX*! ba*! 19-O4'2"!6*!4&!.2*!;/4!/$0.(!#1Lee!/40.&9D9'4&'A!%&40!A422!.(&'B4(!'%!.2%9! 4ED-4%%4G!'(!4D'&/42'.2!A422%!.(G!&.-B4&4G!&9!D2.%0.!040O-.(4!D-9&-$%'9(%*!)# E+-*#!<'H!06."!ggL_Y_M!V_MMMX*! b`*! ;.0.3'"!d*!4&!.2*!i(B-.F&04(&!9F!%9-&4Gn4ED.(G4G!/$0.(!A4(&-.2!(4-C9$%! %<%&40!%&40!A422%!F-90!F4&.2!O-.'(*!)#='4(-1&+#8'1!75"!bacY`c!V_MM_X*! bb*! ,.22'"!)*!4&!.2*!5%92.&'9(!.(G!A/.-.A&4-'l.&'9(!9F!&$09-'B4('A"!%&40Y2'34!(4$-.2! D-4A$-%9-%!F-90!/$0.(!B2'9O2.%&90.*!!$%&'(#8'1!74"!aMLLY_L!V_MMfX*! LMM*! 8400.&'"!8*6*!4&!.2*!1.(A4-9$%!%&40!A422%!A.(!.-'%4!F-90!D4G'.&-'A!O-.'(! &$09-%*!:(-&#=$0*#"&$@#C&+#D#C#"!233"!LgLa`Y`e!V_MMeX*! LML*! 8$/("!d*:*!4&!.2*!5G4(&'F'A.&'9(!9F!D/4(9&!6'-3%"!?*@*!1.(A4-!%&40!A422%!'(!(4-C9$%! %<%&40!&$09-%*!I%&-3'%'!01"!a_caYae!V_MMfX*! LMc*! :44"!7*!4&!.2*!;$09-!%&40!A422%!G4-'C4G!F-90!B2'9O2.%&90.%!A$2&$-4G!'(!OT,T! .(G!i,T!09-4!A29%42!,.B4"!T*8*!T'O-9O2.%&! B-9H&/!F.A&9-Y_!.A&'C.&4%!.!2.&4(&!(4$-9B4('A!D-9B-.0!'(!(4$-.2!%&40!A422%! F-90!G'C4-%4!-4B'9(%!9F!&/4!.G$2&!1Rd*!)#='4(-1&+!25"!`f`aYba!VLbbbX*!

! #+)!

LMb*! j4%A9C'"!#*:*"!)4<(92G%"!@*#*"!T-.%4-"!6*6*!>!h4'%%"!d*!OT,T!-4B$2.&4%!&/4! D-92'F4-.&'C4!F.&4!9F!$('D9&4(&!V(4$-9(.2X!.(G!O'D9&4(&!V(4$-9(.2n.%&-9B2'.2X! i,TYB4(4-.&4G!1Rd!D-9B4('&9-!A422%*!='4(-%!22"!bgLYcc!VLbbeX*! LLM*! 1/9m(.A3'"!#*"!d/'0.l.3'"!;*"!,-4BB"!1*"!h4'(0.%&4-"!,*!>!h4'%%"!d*! ,2!)$'l!'!#2&.O."!#*!d9('A!/4GB4/9B!-4B$2.&4%!&/4!B-9H&/!.(G! D.&&4-('(B!9F!&/4!A4-4O422$0*!K'9'*-GH'%0!207"!eM`bYLMM!VLbbbX*! LLf*! 6./0.(4"!R*!4&!.2*!;/4!d9('A!84GB4/9BY,2'!D.&/H.!dA/.FF4-"!6*j*!d9('A!/4GB4/9B!-4B$2.&4%!.G$2&! (4$-.2!D-9B4('&9-!D-92'F4-.&'9(!'(!C'&-9!.(G!'(!C'C9*!=$0#='4(-1&+!7"!_LYa! V_MMeX*! LLc*! ?.20."!j*!4&!.2*!d9('A!/4GB4/9B!A9(&-92%!%&40!A422!O4/.C'9-!'(!&/4!D9%&(.&.2! .(G!.G$2&!O-.'(*!K'9'*-GH'%0!210"!eegYff!V_MMgX*! LLa*! ?.20."!j*!>!)$'l!'!#2&.O."!#*!84GB4/9BY,:5!%'B(.2'(B!-4B$2.&4%!&/4!O4/.C'9-!9F! A422%!H'&/!%&40!A422!D-9D4-&'4%!'(!&/4!G4C429D'(B!(49A9-&4E*!K'9'*-GH'%0!212"! eeaYfg!V_MMfX*! LL`*! )9H'&A/"!6*8*"!+'%D4-&"!#*!>!=A=./9("!#*?*!?.EY_!-4B$2.&9-!6./0.(4"!R*!84GB4/9BY,2'!%'B(.22'(B!.(G!&/4! B-9H&/!9F!&/4!O-.'(*!=$0#8'9#='4(-1&+!1"!_fYee!V_MM_X*! L_M*! 1/4(("!#*!>!h.2%/"!1*#*!)4B$2.&'9(!9F!A4-4O-.2!A9-&'A.2!%'l4!O!h.2%/"!1*#*!5(A-4.%4G!(4$-9(.2!D-9G$A&'9("!4(2.-B4G!F9-4O-.'(%! .(G!A<&9.-A/'&4A&$-.2!G'%&9-&'9(%!'(!O4&.YA.&4('(!9C4-4ED-4%%'(B!&-.(%B4('A! 0'A4*!!'('.#!-(0'F!21"!gbbYcMc!V_MMeX*! L__*! =4B.%9("!d*,*!>!=A=./9("!#*?*!#!0'&9B4(!B-.G'4(&!9F!G9-%.2!0'G2'(4!h(&%! 9-B.('l4%!B-9H&/!'(!&/4!1Rd*!K'9'*-GH'%0!205"!_M`aYb`!V_MM_X*! L_e*! :'4"!6*1*!4&!.2*!h(&!%'B(.22'(B!-4B$2.&4%!.G$2&!/'DD9A.0D.2!(4$-9B4(4%'%*! =$04('!416"!LeaMYg!V_MMgX*! L_f*! @-$BB40.("!d*h*!4&!.2*!5(3f.!.(G!#-F!G'FF4-4(&'.22

! #+*!

L_c*! :4$(B"!1*!4&!.2*!@0'L!'%!4%%4(&'.2!F9-!A4-4O422.-!G4C429D04(&!.(G!'%! 9C4-4ED-4%%4G!'(!/$0.(!04G$229O2.%&90.%*!=$04('!40/"!eeaYfL!V_MMfX*! L_a*! :'"!:*!4&!.2*!?;iR!'(!(4$-.2!D-4A$-%9-!A422%Q!-4B$2.&'9(!9F!0'B-.&'9("!.D9D&9%'%"! .(G!D-92'F4-.&'9(*!?-*#!'**#='4(-1&+!03"!_LYb!V_MM_X*! L_`*! =929F%3<"!#*j*!4&!.2*!@0'YL!G4D4(G4(A4!G'%&'(B$'%/4%!(4$-.2!%&40!A422!%42FY -4(4H.2!F-90!D-9B4('&9-!D-92'F4-.&'9(*!=$04('!40."!bc_Ya!V_MMeX*! L_b*! =929F%3<"!#*j*!4&!.2*!5(A-4.%'(B!DLc5R+f.!4ED-4%%'9(!G4A-4.%4%!F9-4O-.'(! D-9B4('&9-%!.(G!(4$-9B4(4%'%!G$-'(B!.B4'(B*!=$04('!441"!ff`Yg_!V_MMcX*! LeM*! ;.3./.%/'"!+*!4&!.2*!5(G$A&'9(!9F!D2$-'D9&4(&!%&40!A422%!F-90!.G$2&!/$0.(! F'O-9O2.%&%!O!r.0.(.3."!d*!5(G$A&'9(!9F!D2$-'D9&4(&!%&40!A422%!F-90!09$%4! 40O-<9('A!.(G!.G$2&!F'O-9O2.%&!A$2&$-4%!O!)4<(92G%"!@*#*!@-.'(!&$09$-!%&40!A422%*!=$0#8'9# !$%&'(!7"!f_gYec!V_MMcX*! Lee*! d.(.'"!R*"!#2C.-4lY@$<22."!#*!>!@4-B4-"!=*d*!R4$-.2!%&40!A422%!.(G!&/4!9-'B'(!9F! B2'90.%*!=#>%3*#)#?'@!1.1"!`LLY__!V_MMgX*! Lef*! 89D4H422"!7*h*!>!h-'B/&"!i*#*!;/4!'0D9-&.(A4!9F!'0D2.(&.&'9(!%'&4!'(!A4-4O-.2! A.-A'(9B4(4%'%!'(!-.&%*!!$%&'(#8'1!05"!Lb_aYeL!VLbcbX*! Leg*! j'A3"!R*#*"!:'("!=*7*!>!@'B(4-"!6*6*!;/4!-924!9F!&/4!%$O4D4(G<0.2!D2.&4!'(!B2'.2! &$09-'B4(4%'%*!"&0$#='4(-G$0<-*!43"!ceYaL!VLbaaX*! Lec*! 8922.(G"!i*1*!4&!.2*!190O'(4G!.A&'C.&'9(!9F!).%!.(G!#3&!'(!(4$-.2!D-9B4('&9-%! '(G$A4%!B2'9O2.%&90.!F9-0.&'9(!'(!0'A4*!=$0#,'%'0!0."!ggYa!V_MMMX*! Lea*! 8922.(G"!i*1*"!8'C42<"!h*?*"!,.229"!j*!>!j.-0$%"!8*i*!=9G42'(B!0$&.&'9(%!'(!&/4! ,L!.--4%&!D.&/H.[email protected]&24&&"!?*T*!;/4!.G$2&!09$%4!/'DD9A.0D.2!D-9B4('&9-!'%! (4$-9B4('A!O$&!(9&!.!%&40!A422*!)#='4(-1&+!0."!LM`LgY_L!V_MMgX*! LfL*! #$B$%&4"!?*!4&!.2*!5(/'O'&'9(!9F!F'O-9O2.%&!B-9H&/!F.A&9-nF'O-9O2.%&!B-9H&/! F.A&9-!-4A4D&9-!.A&'C'&

! #"+!

Lff*! ,99G-'A/"!:*j*"!='24(39C'A"!:*"!8'BB'(%"!+*=*!>!dA9&&"!=*?*!#2&4-4G!(4$-.2!A422! F.&4%!.(G!04G$229O2.%&90.!'(!09$%4!D.&A/4G!0$&.(&%*!C&+'%&'!066"!LLMbYLe! VLbbaX*! Lfg*! =.-'(9"!d*!=4G$229O2.%&90.Q!G4C429D04(&.2!04A/.('%0%!9$&!9F!A9(&-92*! J('%@1#?-*#?'@!22"!LaY__!V_MMgX*! Lfc*! ?904-9<"!d*:*!4&!.2*!?-4G'A&'9(!9F!A4(&-.2!(4-C9$%!%<%&40!40O-<9(.2!&$09$-! 9$&A904!O.%4G!9(!B4(4!4ED-4%%'9(*!=$04('!42."!fecYf_!V_MM_X*! Lfa*! )9&/"!h*!4&!.2*!d4A-4&4G!T-'ll24GY-42.&4G!D-9&4'(%!'(/'O'&!09&'2'&!dA9&&"!=*?*!;/4!G4C429D04(&.2!O'929B!).FF42"!1*!dD9-.G'A!04G$229O2.%&90.%! A9(&.'(!9(A9B4('A!O4&.YA.&4('(!0$&.&'9(%*!!$%&'(#8'1!./"!`bcYb!VLbb`X*! Lgf*! :'22'4("!:*!>!).D/.42"!8*!@=?!.(G!T,T!-4B$2.&4!&/4!G4C429D04(&!9F!i,TY -4%D9(%'C4!(4$-.2!D-9B4('&9-!A422%*!K'9'*-GH'%0!206"!fbbeYgMMg!V_MMMX*! Lgg*! ?'AA'-'229"!d*,*!4&!.2*!@9(4!09-D/9B4(4&'A!D-9&4'(%!'(/'O'&!&/4!&$09-'B4('A! D9&4(&'.2!9F!/$0.(!O-.'(!&$09$-Y'('&'.&'(B!A422%*!=$04('!444"!acLYg!V_MMcX*! Lgc*! 6'(B"!d*!>!dA/$2&l"!?*,*!#!-924!F9-!A/40'%&-!6'(B"!d*!#!A/40'A.2!.DD-9.A/!&9!%&40YA422!O'929B

! #""!

Lcg*! ,.O.<"!:*"!:9H422"!d*"!)$O'("!:*:*!>!#(G4-%9("!6*7*!64-4B$2.&'9(!9F!G9-%9C4(&-.2! D.&&4-('(B!O!6'-3%"!?*@*!R4H!G-$B%!F9-!O-.'(! &$09-%t!5(%'B/&%!F-90!A/40'A.2!D-9O'(B!9F!(4$-.2!%&40!A422%*!?'@#;AG-0<'1'1! 60"!c`eYa!V_MMbX*! LaM*! #/("!d*!>!79<(4-"!#*:*!5(!C'C9!.(.2<%'%!9F!U$'4%A4(&!.G$2&!(4$-.2!%&40!A422%! -4%D9(G'(B!&9!d9('A!/4GB4/9B*!=$04('!416"!`bfYa!V_MMgX*! LaL*! q/.(B"!7*8*"!1/$(B"!;*6*!>!N2G4(O$-B"!+*)*!#!d'0D24!d&.&'%&'A.2!?.-.04&4-!F9-! s%4!'(!iC.2$.&'9(!.(G!j.2'G.&'9(!9F!8'B/!;/-9$B/D$&!dA-44('(B!#%%.<%*!)# E+-H-*#C&(''%!4"!caYae!VLbbbX*! La_*! d4'F4-&"!,*"!dA/'22'(B"!+*!>!d&4'(/.$%4-"!1*!#%&-9A<&4!G<%F$(A&'9(!'(!(4$-929B'A.2! G'%9-G4-%Q!.!0924A$2.-!D4-%D4A&'C4*!=$0#8'9#='4(-1&+!6"!LbfY_Mc!V_MMcX*! Lae*! ?/'22'D%"!8*d*!4&!.2*!=924A$2.-!%$OA2.%%4%!9F!/'B/YB-.G4!B2'90.!D-4G'A&! D-9B(9%'%"!G42'(4.&4!.!D.&&4-(!9F!G'%4.%4!D-9B-4%%'9("!.(G!-4%40O24!%&.B4%!'(! (4$-9B4(4%'%*!!$%&'(#!'**!5"!LgaYae!V_MMcX*! Laf*! 19-A9-.("!)*@*!>!dA9&&"!=*?*!#!09$%4!09G42!F9-!04G$229O2.%&90.!.(G!O.%.2! A422!(4C$%!%<(G-904*!)#='4(--%&-*!.1"!eMaYL`!V_MMLX*! Lag*! =.A69(.2G"!=*:*!4&!.2*!5G4(&'F<'(B!9FFY&.-B4&!4FF4A&%!.(G!/'GG4(!D/4(9&!,-4.C4%"!=*!:99D%"!2'(4.B4"!.(G!24$340'.*!!'**!54"!bYL_!VLbb`X*! La`*! d/.-90"!7*)*"!@4229H%"!6*d*!>!;<4-%"!=*!T-90!2.-B4!(4&H9-3%!&9!%0.22! 0924A$24%*!!4((#IG+%#!<'H#E+-*!/"!`LYbM!V_MMfX*! Lab*! @$&&4-H9-&/"!)*T*!>[email protected].$"!#*!#D909-D/'(4Q!%&4-9&!,-$OO"!=*d*!#G$2&!(4$-9B4(4%'%!.(G!F$(A&'9(.2! D2.%&'A'&!d9(B"!8*!,#@#!%4&%!&/4!&40D9!F9-!.A&'C'&!=.3.-4(39C"!j*!d&.&'%&'A.2!.(.2<%'%!9F!%<%&40.&'A!4--9-%!'(! /'B/Y&/-9$B/D$&!%A-44('(B*!)#E+-H-*#C&(''%!23"!ggaYca!V_MMgX*! L`f*! 1/.0O4-%"!7*=*"!124C42.(G"!h*d*"!+24'(4-"!@*!>!;$34<"!?*#*!,($G<+&$*#?'0<-@1# L-(#K$0$#"%$*A1+1M#"!V1/.D0.(!.(G!8.22"!R4H!r9-3"!Lb`eX*! L`g*! +-$&l'3"!?*N*"!5-'%/"!7*=*"!R92.("!,*?*!>!?4-4l"!N*6*!#(.2<%'%!9F!D-9&4'(! D/9%D/9-<2.&'9(!.(G!A422$2.-!%'B(.2'(B!4C4(&%!O

! #"#!

L`c*! 8.%%.("!=*"!@'42.H%3'"!7*?*"!840D42"!7*1*!>!h.2G0.("!=*!ND&'0'l.&'9(!.(G! C'%$.2'l.&'9(!9F!0924A$2.-!G'C4-%'&!)9$%%44$H"!?*!/+%@+%3#3(-4G1#+%#@$0$M#$%#+%0(-@4&0+-%#0-#&*410'(# $%$*A1+1M#"!Vh'24!79/.(%4("!1*! 64D-4%%'9(!.(G!A.(A4-!-'%3Q!.!-4B'%&4-YO.%4G!%&$GG+@'H+-*!2.."!LM``Ybg! V_MM_X*! LbM*! ?.ll.B2'."!d*!4&!.2*!:'(3'(B!6R#!G.0.B4!&9!04G$229O2.%&90.!&$09-'B4(4%'%!'(! D.&A/4G!/4&4-9l!)9&/"!@*:*!;/4!'(&4-.A&'9(!9F!.! A9(%&'&$&'C42!,.-2.(G"!;*!6'FF4-4(&'.2!%4(%'&'C'&!=9229<"!@*@*!14(&-.2!%4-9&9('(!.B9('%&!.A&'9(%!9F! :r!LcgLce"!LYV0Y&-'F2$9-904&/<2D/4(<2XYfYVDY.0'(9D/4(<24&/<2X! D'D4-.l'(4"!'(!-.&%*!)#:<$(H$&-*#>FG#J<'(!015"!fgfYb!VLb`cX*! Lb`*! @.-.3"!r*"!#A/'-9("!#*"!=.(G42"!=*"!='-4A3'"!5*!>!#'l4(O4-B"!6*!)4G$A4G!A.(A4-! '(A'G4(A4!.09(B!D.&'4(&%!H'&/!%A/'l9D/-4('.*!!$%&'(!234"!_`LaY_L!V_MMgX*! Lbb*! 1.-(4<"!1*?*"!h992%9("!)*T*"!79(4%"!:*"!R9<4%"!)*"!7-*!>!694OO42'(B"!@*R*! NAA$--4(A4!9F!A.(A4-!.09(B!D49D24!H'&/!04(&.2!/4.2&/!A2.'0%!'(!.(!'(%$-4G! D9D$2.&'9(*!:1A&<-1-H#?'@!77"!aegYfe!V_MMfX*! _MM*! 6.2&9("!d*N*"!=422403m.4-"!:*"!;/90.%%4("!:*"!=9-&4(%4("!?*@*!>!79/.(%4("!1*! )'%3!F9-!A.(A4-!'(!.!A9/9-&!9F!D.&'4(&%!/9%D'&.2'l4G!F9-!%A/'l9D/-4('.!'(! 64(0.-3"!LbcbYLbbe*!C&<+P-G<(#8'1!6."!eLgY_f!V_MMgX*! _ML*! ,92G.A-4"!=*7*"!+$-'(."!:*=*"!h9&&9("!1*7*"!r4.&4%"!6*!>!d4.B-9.&"!j*! dA/'l9D/-4('.!.(G!A.(A4-Q!.(!4D'G40'929B'A.2!%&$G<*!E(#)#:1A&<+$0(A!2/6"! eefY`!V_MMgX*!

! #"$!

_M_*! ,-'(%/D99("!#*!4&!.2*!1.(A4-!'(!%A/'l9D/-4('.Q!'%!&/4!-'%3!/'B/4-!9-!29H4-t! C&<+P-G<(#8'1!61"!eeeYfL!V_MMgX*! _Me*! :'A/&4-0.(("!6*"!i342$(G"!7*"!?$33.2."!i*"!;.(%3.(4("!#*!>!:9((UC'%&"!7*! 5(A'G4(A4!9F!A.(A4-!.09(B!D4-%9(%!H'&/!%A/'l9D/-4('.!.(G!&/4'-!-42.&'C4%*! "(&<#,'%#:1A&<+$0(A!./"!gaeY`!V_MMLX*! _Mf*! N2%4("!7*8*!4&!.2*!#&!642.&&-4"!7*r*!?-'0.-%3*#)#?'@!1.1"!_ML_Y_f!V_MMgX*! _Ma*! +$/("!8*,*"!h'(324-"!7*"!+40D4-0.(("!,*"!;/.2"!:*7*!>!,.B4"!T*8*!iD'G4-0.2! B-9H&/!F.A&9-!.(G!F'O-9O2.%&!B-9H&/!F.A&9-Y_!/.C4!G'FF4-4(&!4FF4A&%!9(!(4$-.2! D-9B4('&9-%!'(!&/4!.G$2&!-.&!O-.'(*!)#='4(-1&+!26"!g`_MYb!VLbbaX*! _M`*! ,'2Y?4-9&'("!d*!4&!.2*!:9%%!9F!Dge!'(G$A4%!A/.(B4%!'(!&/4!O4/.C'9-!9F! %$OC4(&-'A$2.-!l9(4!A422%Q!'0D2'A.&'9(!F9-!&/4!B4(4%'%!9F!B2'.2!&$09-%*!)# ='4(-1&+!07"!LLMaYLc!V_MMcX*! _Mb*! R9l.3'"!=*!4&!.2*!)924%!9F!&/4!F$(A&'9(.2!29%%!9F!Dge!.(G!9&/4-!B4(4%!'(! .%&-9A<&90.!&$09-'B4(4%'%!.(G!D-9B-4%%'9(*!='4(-#I%&-*!2"!L_fYea!VLbbbX*! _LM*! q/$"!r*!4&!.2*!i.-2!C.(!G4-!+99<"!6*!69D.0'(4!%D4A'F'A.22!:'(GC.22"!N*!T9-4O-.'(!.A4&<2A/92'(4! -4B$2.&4%!.G$2&!/'DD9A.0D.2!(4$-9B4(4%'%!.(G!24.-('(B*!='4(-.+-*#"3+%3!07"! bebYfc!V_MMgX*! _Lg*! :9;$-A9"!7*7*"!NH4(%"!6*T*"!84.&/"!=*7*"!6.C'%"!=*@*!>!+-'4B%&4'("!#*)*!,#@#!.(G! B2$&.0.&4!G4D92.-'l4!A9-&'A.2!D-9B4('&9-!A422%!.(G!'(/'O'&!6R#!%<(&/4%'%*! ='4(-%!2."!L_`aYb`!VLbbgX*! _Lc*! ;9l$3."!r*"!T$3$G."!d*"!R.0O."!;*"!d43'"!;*!>!8'%.&%$(4"!;*!,#@#4-B'A! 4EA'&.&'9(!D-909&4%!(4$-9(.2!G'FF4-4(&'.&'9(!'(!.G$2&!/'DD9A.0D.2! D-9B4('&9-!A422%*!='4(-%!46"!`MeYLg!V_MMgX*! _La*! @.(.%-"!=*"!84-<"!=*"!?-'(&40D%"!)*!>!6.%l$&."!#*!d4-9&9('(Y'(G$A4G!'(A-4.%4%! '(!.G$2&!A422!D-92'F4-.&'9(!.(G!(4$-9B4(4%'%!.-4!04G'.&4G!&/-9$B/!G'FF4-4(&! .(G!A9009(!gY8;!-4A4D&9-!%$O&

! #"%!

__M*! ,/.%/B/.4'"!8*;*!4&!.2*!;/4!-924!9F!(4$-4B$2'(Yi-O@f!'(&4-.A&'9(%!9(!&/4! D-92'F4-.&'9(!.(G!9-B.('l.&'9(!9F!A422%!'(!&/4!%$OC4(&-'A$2.-!l9(4*!:(-&#=$0*# "&$@#C&+#D#C#"!231"!LbeMYg!V_MMcX*! __L*! 89B2'(B4-"!,*s*!4&!.2*!69D.0'(4!G4D24&'9(!'0D.'-%!D-4A$-%9-!A422!D-92'F4-.&'9(! '(!?.-3'(%9(!G'%4.%4*!=$0#='4(-1&+!6"!a_cYeg!V_MMfX*! ___*! ,9&l"!=*!>!8$&&(4-"!h*@*!;/4!A422!O'929B!+9%9G9"!r*!d<004&-'A!C4-%$%!.%<004&-'A!A422!G'C'%'9(!G$-'(B! (4$-9B4(4%'%!'(!&/4!G4C429D'(B!C4-&4O-.&4!A4(&-.2!(4-C9$%!%<%&40*!!4((#IG+%# !'**#E+-*!26"!cf`Yga!V_MMgX*! __f*! =9--'%9("!d*7*!>!+'0O24"!7*!#%<004&-'A!.(G!%<004&-'A!%&40YA422!G'C'%'9(%!'(! G4C429D04(&!.(G!A.(A4-*!=$04('!442"!LMc`Yaf!V_MMcX*! __g*! :'$"!@*r*"!=A64-09&&"!d*?*"!+/H.m."!d*d*!>!#24E.(G4-"!1*=*!;/4!&-.(%F9-0'(B! .A&'C'&!:'C4-&9("!R*!;.-B4&'(B!&/4!04&.O9&-9D'A!B2$&.0.&4! -4A4D&9-!0,2$)f!F9-!&/4!&-4.&04(&!9F!G'%4.%4%!9F!&/4!A4(&-.2!(4-C9$%!%<%&40*! !4((#J-G#?'@#!<'H!."!``gYbg!V_MMgX*! __b*! =.-'(9"!=*7*!4&!.2*!#229%&4-'A!09G$2.&'9(!9F!B-9$D!555!04&.O9&-9D'A!B2$&.0.&4! -4A4D&9-!fQ!.!D9&4(&'.2!.DD-9.A/!&9!?.-3'(%9(p%!G'%4.%4!&-4.&04(&*!:(-&#=$0*# "&$@#C&+#D#C#"!233"!Lecc`Yae!V_MMeX*! _eM*! 6$(.<4C'A/"!i*!4&!.2*!iFF'A.A!?.-(.C42.%"!7*,*!#A&'C.&'9(!9F!&/4!,#@##! -4A4D&9-!'(/'O'&%!&/4!D-92'F4-.&'C4!4FF4A&%!9F!OT,T!'(!A9-&'A.2!D-9B4('&9-!A422%*! >4(#)#='4(-1&+!5"!_bLY`!VLbbaX*! _ee*! 8.B422"!?*!>!NG'("!?*!#D909-D/'(4!'(!&/4!&-4.&04(&!9F!?.-3'(%9(p%!G'%4.%4*!)# ='4(-1&+#=4(1!11"!_LYef"!eaY`!V_MMLX*! _ef*! 1.'22.-G"!?*"!=9$-4("!u*"!@.'22'.-&"!N*!>!129.-4A"!=*!iC.2$.&'9(!9F!'F4(D-9G'2! 4FF'A.A

! #"&!

_ec*! 1/422<"!7*"!19(A9-G4&"!7*?*"!+.D2.("!7*1*!>!+./("!#*!5224B'&'0.&4!&-.(%A-'D&'9(Q! &-.(%A-'D&'9(!9F!.(!;<.B'"!d*!d&9A/.%&'A!0)R#! %<(&/4%'%!'(!0.00.2'.(!A422%*!:Q-C#E+-*!4"!4eMb!V_MMcX*! _e`*! R.'-"!j*6*!>!d4.2F9("!d*1*!#B9('%&Y%D4A'F'A!&-.(%.A&'C.&'9(!9F!D/9%D/9'(9%'&'G4! eY3'(.%4!%'B(.2'(B!D.&/H.!;/90D%9("!1*:*!='A-9&$O$24Y .%%9A'.&4G!D-9&4'(!2'B/&!A/.'(!_!'%!.!%&.-B.l'(Y#=?#!-4A4D&9-!A90D24EY '(&4-.A&'(B!D-9&4'(!'(!C'C9*!)#E+-*#!<'H!065"!eLMM_Yb!V_MMfX*! _f_*! iO'/.-."!;*"!,$9"!T*"!q/.(B"!:*"!+'0"!7*r*!>!d.FF4("!6*!=$%A.-'('A!.A4&<2A/92'(4! -4A4D&9-%!%&'0$2.&4!1._o!'(F2$E!'(!?1L_6!A422%!D-4G90'(.(&2!d.FF4("!6*!=$%A.-'('A!.A4&<2A/92'(4!-4A4D&9-Y04G'.&4G!'(G$A&'9(! 9F!l'F_c`!0)R#!'(!?1L_6!A422%!-4U$'-4%!D-9&4'(!3'(.%4!1!.(G!&/4!'(F2$E!9F! 4E&-.A422$2.-!A.2A'$0*!)#='4(-&<'H!7/"!LMMLYLM!VLbbaX*! _ff*! :.H"!d*h*"!19((442<"!N*=*"!64=.<9"!T*7*!>!Np=.224<"!@*h*!5G4(&'F'A.&'9(!9F!.! (4H!O-.'(Y%D4A'F'A!&-.(%A-'D&'9(!F.A&9-"!Rs))L*!?-*#>%@-&(+%-*!7"!_L_bYeg! VLbb_X*! _fg*! d.3$-.G."!+*"!N/%/'0.Yd.3$-.G."!=*"!?.204-"!;*6*!>!,.B4"!T*8*!R$--L"!.(! 9-D/.(!($A24.-!-4A4D&9-"!'%!.!&-.(%A-'D&'9(.2!.A&'C.&9-!9F!4(G9B4(9$%! &<-9%'(4!/!d$-.('"!=*#*!6<(.0'A!4U$'2'O-'$0!.(G! /4&4-9B4(4'&!8$.(B"!d*! ;-.(%A-'D&904YH'G4!(9'%4!A9(&-92%!2'(4.B4!A/9'A4!'(!0.00.2'.(!D-9B4('&9-! A422%*!=$04('!4.1"!gffYa!V_MM`X*! _gM*! 8$"!=*!4&!.2*!=$2&'2'(4.B4!B4(4!4ED-4%%'9(!D-4A4G4%!A900'&04(&!'(!&/4! /409D9'4&'A!%<%&40*!,'%'1#K'9!22"!aafY`g!VLbbaX*! _gL*! =4-324"!T*;*"!='-l.G4/"!q*!>!#2C.-4lY@$<22."!#*!=9%.'A!9-B.('l.&'9(!9F!(4$-.2! %&40!A422%!'(!&/4!.G$2&!O-.'(*!C&+'%&'!126"!e`LYf!V_MMaX*! _g_*! h.(B"!8*!4&!.2*!R'A9&'('A!.A4&<2A/92'(4!-4A4D&9-!.2D/.a!%$O$('&!'%!.(!4%%4(&'.2! -4B$2.&9-!9F!'(F2.00.&'9(*!=$04('!402"!e`fY`!V_MMeX*!

! #"'!

_ge*! i(C4-"!;*"!?4-."!=*"!?4&4-%9("!1*!>!#(G-4H%"!?*h*!d&40!A422!%&.&4%"!F.&4%"!.(G! &/4!-$24%!9F!.&&-.A&'9(*!!'**#C0'H#!'**!4"!e`aYba!V_MMbX*! _gf*! #A.-"!=*"!=4&&4&.2"!7*;*!>!C.(!N$G4(..-G4("!#*!d&9A/.%&'A!%H'&A/'(B!.%!.! %$-C'C.2!%&-.&4B!+.4-("!=*!#!A/.(A4!.&!%$-C'C.2Q!B4(4!4ED-4%%'9(!(9'%4!.(G! D/4(9&!#-3'("!#*?*!6'C4-%'&[email protected]"!8*T*!T.A&9-%!'(C92C4G!'(!&/4!G4&4-0'(.&'9(!9F!&/4! (4$-9&-.(%0'&&4-!D/4(9&[email protected]"!8*T*!;/4! G'FF4-4(&'.&'9(!D9&4(&'.2!9F!/$0.(!F94&.2!(4$-9(.2!D-9B4('&9-!A422%!'(!C'&-9*! E($+%#8'1#K'9#E($+%#8'1!2.1"!ebYgL!V_MMfX*! _gb*! ).m"!#*!>!C.(!N$G4(..-G4("!#*!R.&$-4"!($-&$-4"!9-!A/.(A4Q!%&9A/.%&'A!B4(4! 4ED-4%%'9(!.(G!'&%!A9(%4U$4(A4%*!!'**!21."!_LcY_c!V_MM`X*! _cM*! 1/.0O4-%"!5*!4&!.2*!R.(9B!%.F4B$.-G%!D2$-'D9&4(A!d9-B4-"!?*+*!R9(YB4(4&'A! 9-'B'(%!9F!A422Y&9YA422!C.-'.O'2'&!j.'G<."!j*#*!=924A$2.-!.(G!A422$2.-!.A&'9(%!9F!A/-9('A! 424A&-9A9(C$2%'C4!%4'l$-4%*!)#>!J!24"!L`LYbe!VLbb`X*! _cf*! ;-'C4G'"!=*8*!4&!.2*!iC.2$.&'9(!9F!9$&A904%!H'&/!A'&.29D-.0!F9-!G4D-4%%'9(! $%'(B!04.%$-404(&YO.%4G!A.-4!'(!d;#)v6Q!'0D2'A.&'9(%!F9-!A2'('A.2!D-.A&'A4*! "H#)#:1A&<+$0(A!271"!_`YfM!V_MMcX*! _cg*! C.(!?-..B"!8*"!+40D4-0.(("!,*!>!,.B4"!T*8*!)$(('(B!'(A-4.%4%!A422! D-92'F4-.&'9(!.(G!(4$-9B4(4%'%!'(!&/4!.G$2&!09$%4!G4(&.&4!B<-$%*!=$0# ='4(-1&+!0"!_ccYaM!VLbbbX*! _cc*! @-9H("!7*!4&!.2*!i(-'A/4G!4(C'-9(04(&!.(G!D/<%'A.2!.A&'C'&4(#)#='4(-1&+!26"!_Mf_Yc! V_MMeX*! _ca*! d'0D%9("!+*:*"!T'%/4-"!;*=*"!h.&4-/9$%4"!@*6*!>!:'("!)*1*!?-9m4A&'9(!D.&&4-(%! F-90!&/4!-.D/4!($A24.-!A90D24E!&9!&/4!4D4(G<0.2!H.22!9F!&/4!C4(&-'A$2.-! %<%&40!'(!&/4!-.&*!)#!-HG#='4(-*!155"!cLYa_!VLbb`X*! _c`*! R.A/4-"!7*"!#29(%9Y:29%."!,*"!)9%422"!6*)*!>!=AiH4("!@*d*!R=6#!-4A4D&9-! .(&.B9('%&!&-4.&04(&!'(A-4.%4%!&/4!D-9G$A&'9(!9F!(4H!(4$-9(%!'(!&/4!.B4G! -.&!/'DD9A.0D$%*!='4(-.+-*#"3+%3!04"!_aeY`f!V_MMeX*! _cb*! ='-.(G.Y19(&-4-.%"!:*"!@4('&4lY6'.l"!?*"!?4(.Y19(&-4-.%"!q*"!=4(G9l.Y@-'A4(9"! )*j*!>!?.2.A'9%Y?-$"!i*!:4C42%!9F!.0'(9!.A'G!(4$-9&-.(%0'&&4-%!G$-'(B! (4$-9B4(4%'%!.(G!'(!/'%&9&

! #"(!

_aM*! ='-.(G.Y19(&-4-.%"!:*!4&!.2*!:4C42%!9F!.0'(9!.A'G!(4$-9&-.(%0'&&4-%!G$-'(B! 09$%4!92F.A&9-!?.2.A'9%Y?-$"!i*:*!:4C42%!9F!.0'(9!.A'G!(4$-9&-.(%0'&&4-%!G$-'(B! 09$%4!A4-4O422.-!(4$-9B4(4%'%!.(G!'(!/'%&9&!?.2.A'9%Y?-$"!i*:*!:4C42%!9F! 09(9.0'(4!.(G!.0'(9!.A'G!(4$-9&-.(%0'&&4-%!'(!&/4!G4C429D'(B!0.24!09$%4! /4(#)#='4(-1&+!20"!ebLYc!V_MMMX*! _af*! 1.04-9("!8*#*"!8.l42"!;*,*!>!=A+.<"!)*6*!)4B$2.&'9(!9F!(4$-9B4(4%'%!O!R4.24"!7*8*!,#@#!'(G$A4%!D-92'F4-.&'9(!9F! '00.&$-4!A4-4O422.-!B-.($24!A422%!B-9H(!'(!C'&-9*!E($+%#8'1#K'9#E($+%#8'1! 22."!LY`!VLbbbX*! _ac*! =."!h*!>[email protected]"!7*:*!190D2404(&.-4(#)#='4(-1&+!20"!L__aYfM!V_MMMX*! _ab*! h4'%%"!i*)*"!=.(4%%"!?*!>!:.$G4-"!7*=*!h/!,.B4"!T*8*!#G$2&!O-.'(!(4$-9B4(4%'%!.(G!D%!=.2O4-B"!7*!)4B$2.&'9(!9F!.G$2&!(4$-9B4(4%'%!O!6$0.("!)*d*!1/-9('A!.(&'G4D-4%%.(&! &-4.&04(&!'(A-4.%4%!(4$-9B4(4%'%!'(!.G$2&!-.&!/'DD9A.0D$%*!)#='4(-1&+!03"! bLMfYLM!V_MMMX*! _`e*! @-4l$("!7*=*!>!6.%l$&."!#*!64D24&'9(!'(!%4-9&9('(!G4A-4.%4%!(4$-9B4(4%'%!'(! &/4!G4(&.&4!B<-$%!.(G!&/4!%$OC4(&-'A$2.-!l9(4!9F!.G$2&!-.&%*!='4(-1&+'%&'!/5"! bbbYLMM_!VLbbbX*! _`f*! @.34-"!d*#*"[email protected]"!+*#*!>!8.BB"!;*!6e!G9D.0'(4!-4A4D&9-%!G9!(9&!-4B$2.&4! (4$-9B4(4%'%!'(!&/4!%$OC4(&-'A$2.-!l9(4!9F!.G$2&!0'A4*!='4(-.+-*#K+1!2/"!g_eY a!V_MMgX*!

! #")!

_`g*! @.34-"!d*#*"[email protected]"!+*#*!>!8.BB"!;*!69D.0'(4-B'A!('B-9%&-'.&.2!D-9m4A&'9(%! -4B$2.&4!(4$-.2!D-4A$-%9-!D-92'F4-.&'9(!'(!&/4!.G$2&!09$%4!%$OC4(&-'A$2.-! l9(4*!>4(#)#='4(-1&+!03"!gagYb!V_MMfX*! _`c*! =4-A92."!=*!1/40'A.2!D-9O4%!9F!(4$-.2!%&40!A422!%42FY-4(4H.2*!=$0#!<'H#E+-*! 1"!_fcYa!V_MMaX*! _`a*! +$23.-('"!j*#*"!7/."!d*!>!j.'G<."!j*#*!64D24&'9(!9F!(9-4D'(4D/-'(4!G4A-4.%4%! &/4!D-92'F4-.&'9("!O$&!G94%!(9&!'(F2$4(A4!&/4!%$-C'C.2!.(G!G'FF4-4(&'.&'9("!9F! B-.($24!A422!D-9B4('&9-%!'(!&/4!.G$2&!-.&!/'DD9A.0D$%*!>4(#)#='4(-1&+!27"! _MM`YL_!V_MM_X*! _``*! T2'(&"!#*1*"!:'$"!u*!>!+-'4B%&4'("!#*)*!R9(%<(.D&'A!B2!R.3.B.H."!d*!)4B$2.&'9(!9F!.G$2&!(4$-9B4(4%'%!OFG#J<'(!055"!fMLYa!V_MMLX*! _b_*! 1/4("!,*"!=.%.(."!=*5*!>!=.(m'"!8*+*!:'&/'$0!-4B$2.&4%!?+1Y04G'.&4G! '(&-.A422$2.-!A-9%%Y&.23!.(G!B4(4!4ED-4%%'9(!'(!&/4!1Rd!'(!C'C9*!E+G-*$(#K+1-(@! 0"!_LaYec!V_MMMX*! _be*! =99-4"!,*7*"!@4OA/$3"!7*=*"!h'2G%"!5*@*"!1/4("!,*!>!=.(m'"!8*+*!:'&/'$0Y'(G$A4G! '(A-4.%4!'(!/$0.(!O-.'(!B-4!=.(m'"!8*+*! i(/.(A404(&!9F!/'DD9A.0D.2!(4$-9B4(4%'%!O!T.$22"!)*:*!?-9B4('&9-!A422%!.(G!.G$2&! (4$-9B4(4%'%!'(!(4$-9G4B4(4-.&'C4!G'%4.%4%!.(G!'(m$-'4%!9F!&/4!O.%.2!B.(B2'.*! !*+%#>FG#:<$(H$&-*#:!j4-3/-.&%3<"!#*!50D.'-4G!A422!D-92'F4-.&'9(!'(!&/4! %$OC4(&-'A$2.-!l9(4!'(!.(!#2l/4'04-p%!G'%4.%4!09G42*!='4(-('G-(0!03"!bMaYL_! V_MMbX*! _bb*! )9G-'B$4l"!7*7*!4&!.2*!50D.'-4G!.G$2&!(4$-9B4(4%'%!'(!&/4!G4(&.&4!B<-$%!9F!.! &-'D24!&-.(%B4('A!09$%4!09G42!9F!#2l/4'04-p%!G'%4.%4*!:Q-C#I=>!1"!4_beg! V_MM`X*! eMM*! )4'F"!#*!4&!.2*!R4$-.2!%&40!A422!D-92'F4-.&'9(!'%!G4A-4.%4G!'(!%A/'l9D/-4('."!O$&! (9&!'(!G4D-4%%'9(*!?-*#:1A&<+$0(A!22"!gLfY__!V_MMcX*! eML*! =.&&%9("!=*?*"!=.$G%24<"!d*!>!=.-&'("!@*!@6RT!.(G!gY8;Q!.!G<(.0'A!G$9!'(! .B4Y-42.&4G!(4$-9(.2!D2.%&'A'&

! #"*! eM_*! +'&.<.0."!;*"!r9(4<.0."!=*"!;.0.3'"!+*!>!r9(4G."!r*!)4B$2.&'9(!9F!(4$-9(.2! G'FF4-4(&'.&'9(!O!#(G4-%9("!6*7*!)4B$2.&9-!6'd&4F.(9"!?*d*!R4$-9&-9D/'A!F.A&9-%Q! F-90!0924A$24!&9!0.(*!J('%@1#='4(-1&+!26"!L`_YbM!VLbbfX*! eMc*! :'(G4F9-%"!R*"!@-9G'("!i*!>!=4&%'%"!=*!dD.&'9&40D9-.2!%424A&'C4!4FF4A&%!9(! O-.'(YG4-'C4G!(4$-9&-9D/'A!F.A&9-!.(G!&-3@!04%%4(B4-!)R#!'(!-.&! /'DD9A.0D$%!O!6$0.("!)*d*!)4B$2.&'9(!9F!@6RT!.(G!&-3@!0)R#! '(!-.&!O-.'(!O!:$%3'("!=*@*!5(&-.C4(&-'A$2.-! .G0'('%&-.&'9(!9F!@6RT!'(A-4.%4%!&/4!($0O4-!9F!(4H2!,92G0.("!d*#*!#G4(9C'-.2! O-.'(YG4-'C4G!(4$-9&-9D/'A!F.A&9-!'(G$A4%!O9&/!(49%&-'.&.2!.(G!92F.A&9-!:$%3'("!=*@*!5(F$%'9(!9F!O-.'(Y G4-'C4G!(4$-9&-9D/'A!F.A&9-!'(&9!&/4!2.&4-.2!C4(&-'A24!9F!&/4!.G$2&!-.&!24.G%!&9! (4H!(4$-9(%!'(!&/4!D.-4(A/<0.!9F!&/4!%&-'.&$0"!%4D&$0"!&/.2.0$%"!.(G! /!,92G0.("!d*#*!#G4(9C'-.22!?4-%%9("!8*!6'FF4-4(&'.2!$%.B4!9F! 0$2&'D24!O-.'(YG4-'C4G!(4$-9&-9D/'A!F.A&9-!D-909&4-%!'(!&/4!-.&!O-.'(! F9229H'(B!(4$-9(.2!.A&'C.&'9(*!:(-&#=$0*#"&$@#C&+#D#C#"!53"!``M_Yc!VLbbeX*! eLf*! R.<29-"!?*"!d&4H.-&"!1*#*"!h-'B/&"!d*)*"!?4.-%9("!)*1*!>!)4'G"!5*1*!)4D4.&4G!i1d! '(G$A4%!,2$)L!0)R#!O$&!(9&!R=6#)L#Y,!0)R#!'(!&/4!-.&!/'DD9A.0D$%*! E($+%#8'1#?-*#E($+%#8'1!1."!efbYge!VLbbcX*! eLg*! @.(.%-"!=*"!d9$0'4-"!#*"!84-<"!=*"!=9A.4-"!i*!>!6.%l$&."!#*!#B9042.&'(4"!.!(4H! .(&'G4D-4%%.(&"!'(G$A4%!-4B'9(.2!A/.(B4%!'(!/'DD9A.0D.2!(4$-9B4(4%'%*!E+-*# :1A&<+$0(A!.5"!LM`aYbc!V_MMcX*! eLc*! q.F-."!T*"!:'(G/920"!6*"!1.%&-4("!i*"!8.-&'33."!7*!>!;/94(4("!8*!)4B$2.&'9(!9F! O-.'(YG4-'C4G!(4$-9&-9D/'A!F.A&9-!.(G!(4-C4!B-9H&/!F.A&9-!0)R#!'(!D-'0.-

! ##+! eLa*! q.F-."!T*"!1.%&-4("!i*"!;/94(4("!8*!>!:'(G/920"!6*!5(&4-D2.!6$0.("!)*d*!1/-9('A!.(&'G4D-4%%.(&!.G0'('%&-.&'9(! '(A-4.%4%!&/4!4ED-4%%'9(!9F!A#=?!-4%D9(%4!42404(&!O'(G'(B!D-9&4'(!V1)i@X! '(!-.&!/'DD9A.0D$%*!)#='4(-1&+!27"!_ecgYa_!VLbbcX*! eLb*! i0%24<"!7*,*!>!8.BB"!;*!i(G9B4(9$%!.(G!4E9B4(9$%!A'2'.-FG#='4(-*!2/1"!_b`YeLM! V_MMeX*! e_M*! #(G.(B"!=*!4&!.2*!8'%&9(4!8_#uYG4D4(G4(&!,#@#V#X!-4A4D&9-!-4B$2.&'9(!9F! %&40!A422!D-92'F4-.&'9(*!=$04('!4.2"!fcMYf!V_MM`X*! e_L*! d&$G4-"!:*"!;.O.-"!j*!>!=A+.<"!)*6*!;-.(%D2.(&.&'9(!9F!4ED.(G4G! 04%4(A4D/.2'A!D-4A$-%9-%!24.G%!&9!-4A9C4-[email protected]"!8*T*!;/4!A9(&-9224G!A9(C4-%'9(! 9F!/$0.(!(4$-.2!D-9B4('&9-!A422%!G4-'C4G!F-90!F94&.2!C4(&-.2!04%4(A4D/.29(! '(&9!G9D.0'(4-B'A!(4$-9(%!'(!C'&-9*!E($+%#8'1#K'9#E($+%#8'1!217"!_aYef! V_MM_X*! e_e*! 6$"!u*!>!5.A9C'&&'"!:*!=$2&'D24!%'B(.2'(B!D.&/H.<%!G'-4A&!&/4!'('&'.&'9(!9F! &<-9%'(4!/!5.A9C'&&'"!:*!?-9&4'(!3'(.%4!1!.A&'C.&9-%!H9-3!'(!%<(4-B!5.A9C'&&'"!:*!d<(4-B!5.A9C'&&'"!:*!@-.'(YG4-'C4G!(4$-9&-9D/'A!F.A&9-!H9-3%! A99-G'(.&42!5.A9C'&&'"!:*!=4A/.('%0%!B9C4-('(B!&/4!G'FF4-4(&'.&'9(!9F!.! %4-9&9(4-B'A!D/4(9&!#-3'("!#*!d&9A/.%&'A!04A/.('%0%!'(!B4(4!4ED-4%%'9(*!:(-&# =$0*#"&$@#C&+#D#C#"!54"!`LfYb!VLbbaX*! e_b*! ,.22<"!6*:*"!@9B.("!7*#*"!i'%4(%&4'("!@*5*!>!@290F'42G"!5*1*!i(C'-9(04(&.2! -4B$2.&'9(!9F!&/4!F'0!%H'&A/!A9(&-922'(B!&!T.239H"!d*!5G4(&'F'A.&'9(!.(G!A/.-.A&4-'l.&'9(!9F!.!B4(4! D-9G$A&!&/.&!-4B$2.&4%!&!T.239H"!d*!N-B.('l.&'9(!.(G!4ED-4%%'9(!9F!B4(4%!-4%D9(%'O24! F9-!&

! ##"! ee_*! +2400"!?*!;H9!-4B$2.&9-?EI#)!."!Le`bYbe!VLb`cX*! eee*! ;<%9("!7*7*!>!8.((%B4("!+*@*!;/4!G'%&-'O$&'9(%!9F!A422!%'l4!.(G!B4(4-.&'9(!&'04! '(!.!09G42!9F!&/4!A422!A!8.-C4<"!7*6*!,4(4-.&'9(!&'04!%&.&'%&'A%!9F!i%A/4-'A/'.!A92'!@! 04.%$-4G!O!:'(Y1/.9"!d*!#(.2<%'%!9F!&/4!D/<%'929B'A.2!A9(&-92!9F!-4D2'A.&'9(! 9F!192iLY&!+$O'%&."!=*!,4(4!4ED-4%%'9(! D-9F'2'(B!'(!%'(B24!A422%!F-90!&/4!D.(A-4.&'A!'%24&%!9F!:.(B4-/.(%!-4C4.2%! 29B(9-0.2!G'%&-'O$&'9(!9F!0)R#!24C42%*!,'%-H'#8'1!2."!Le``Yb_!V_MMgX*! ee`*! R9--'%"!#*7*!4&!.2*!6<(.0'A!D.&&4-(%!9F!B-9H&/!/9-09(4!B4(4!&-.(%A-'D&'9(!'(! '(G'C'G$.2!2'C'(B!D'&$'&.-%@-&(+%-*!26"!LbeY_M_!V_MMeX*! eeb*! h.2&4-%"!=*1*!4&!.2*!i(/.(A4-%!'(A-4.%4!&/4!D-9O.O'2'&!+'9$%%'%"!6*!).(G90!.A&'C.&'9(!9F!.! &-.(%B4(4!$(G4-!&/4!A9(&-92!9F!.!/?EI#)!24"!gagY`f!VLbbgX*! efL*! +9-9O39C."!i*"!i09(4&"!;*"!j'2.-"!7*=*"!d/'0'l$"!;*d*!>!12$l42"!?*!T-90! 0924A$2.-!(9'%4!&9!O4/.C'9$-.2!C.-'.O'2'&!C.(!N$G4(..-G4("!#*!d&9A/.%&'A!B4(4!4ED-4%%'9(!'(!F2$A&$.&'(B! 4(C'-9(04(&%*!,'%'0+&1!276"!g_eYeM!V_MMfX*! efe*! d.09'29C"!=*"!?2<.%$(9C"!d*!>!#-3'("!#*?*!d&9A/.%&'A!.0D2'F'A.&'9(!.(G! %'B(.2'(B!'(!4(l<0.&'A!F$&'24!A!i2F"!7*!R9'%4Y'(G$A4G!='(!D/4(9&!6$O(.$"!6*!R9'%4!'(!B4(4!4ED-4%%'9(!G4&4-0'(4%!A422! F.&4!'([email protected]'22$%!%$O&'2'%*!C&+'%&'!126"!g_cYb!V_MMaX*! efc*! @2.34"!h*7*"!="!+*#*"!1.(&9-"!1*)*!>!1922'(%"!7*7*!R9'%4!'(!4$3.-<9&'A!B4(4! 4ED-4%%'9(*!=$04('!400"!ceeYa!V_MMeX*! efa*! 1/.D9("!1*!iED-4%%'9(!9F!0.2;"!&/4!-4B$2.&9-!B4(4!9F!&/4!0.2&9%4!-4B'9(!'(! i%A/4-'A/'.!A92'"!'%!2'0'&4G!O9&/!.&!&-.(%A-'D&'9(!.(G!&-.(%2.&'9(*!>?EI#)!2"! ecbYaf!VLb`_X*! ef`*! r.-A/$3"!N*"!7.AU$4%"!R*"!,$'224-4l"!7*!>!6-4!dH.'("!?*d*!d&9A/.%&'A!B4(4!4ED-4%%'9(! '(!.!%'(B24!A422*!C&+'%&'!056"!LL`eYc!V_MM_X*!

! ###! egM*! dH.'("!?*d*"!i29H'&l"!=*@*!>!d'BB'."!i*6*!5(&-'(%'A!.(G!4E&-'(%'A!A9(&-'O$&'9(%!&9! %&9A/.%&'A'&!C.(!N$G4(..-G4("!#*!19(&-'O$&'9(%!9F!29H! 0924A$24!($0O4-!.(G!A/-909%90.2!D9%'&'9('(B!&9!%&9A/.%&'A!B4(4! 4ED-4%%'9(*!=$0#,'%'0!16"!beaYff!V_MMgX*! eg_*! ).%4-"!7*=*!>!Npd/4."!i*+*!19(&-92!9F!%&9A/.%&'A'&!i(C4-"!;*!;/4!2'(4.B4!A900'&04(&!9F!/.409D9'4&'A!D-9B4('&9-! A422%*!!4((#IG+%#,'%'0#K'9!6"!cMbYLe!VLbbaX*! egf*! h4'(&-.$O"!8*!#%%40O2!8$-%&"!:*6*!iC92$&'9(!9F!A/-909%904!9-B.('l.&'9(!G-'C4(!O!d'4B0$(G"!6*N*!d&9A/.%&'A!09G42!9F!D-9&4'(Y D-9&4'(!'(&4-.A&'9(Q!H/!79/(%&9("!=*j*!#2&4-4G!G4C429D04(&!9F!B2$&.0.&4!.(G! ,#@#!-4A4D&9-%!'(!&/4!O.%.2!B.(B2'.!9F!B'-2%!H'&/!)4&&!%<(G-904*!>FG#='4(-*! 2.7"!efgYg_!VLbbbX*! egb*! #-0%&-9(B"!6*"!6$(("!7*+*"!#(&.2FF<"!@*!>!;-'C4G'"!)*!d424A&'C4!G4(G-'&'A! .2&4-.&'9(%!'(!&/4!A9-&4E!9F!)4&&!%<(G-904*!)#='4(-G$0<-*#>FG#='4(-*!.4"!LbgY _ML!VLbbgX*! ecM*! #0'-"!)*i*!4&!.2*!=$&.&'9(%!'(!4E9(!L!9F!=i1?_!.-4!.!-.-4!A.$%4!9F!)4&&! %<(G-904*!)#?'@#,'%'0!40"!4Lg!V_MMgX*! ecL*! +$G9"!d*!=4&/<2Y1D,YO'(G'(B!D-9&4'(!=41?_!-4D-4%%4%!dDLY.A&'C.&4G! &-.(%A-'D&'9(!9F!&/4!/$0.(!24$39%'.2'(!B4(4!H/4(!&/4!D-909&4-!'%! 04&/<2.&4G*!?-*#!'**#E+-*!2/"!gfb_Yb!VLbb`X*! ec_*! R.("!u*!4&!.2*!;-.(%A-'D&'9(.2!-4D-4%%'9(!O!q9B/O'"!8*r*!)4&&!%<(G-904!.(G!=41?_Q!2'(3'(B! 4D'B4(4&'A%!.(G!(4$-9(.2!F$(A&'9(*!"H#)#;4H#,'%'0!62"!L_gbYa_!V_MM_X*! ecf*! r$"!T*"!;/'4%4("!7*!>!d&-.&2'(B"!h*8*!8'%&9(4!G4.A4&<2.%4Y'(G4D4(G4(&! &-.(%A-'D&'9(.2!-4D-4%%'9(!O!@'-G"!#*!=41?_!'%!.!&-.(%A-'D&'9(.2!-4D-4%%9-!H'&/! .O$(G.(&!O'(G'(B!%'&4%!'(!B4(90'A!A/-90.&'(*!!'**!//"!faLY`L!VLbbaX*! ec`*! 89-'34"!d*"!1.'"!d*"!='<.(9"!=*"!1/4(B"!7*T*!>!+9/H'Yd/'B40.&%$"!;*!:9%%!9F! %'24(&YA/-90.&'(!299D'(B!.(G!'0D.'-4G!'0D-'(&'(B!9F!6:ug!'(!)4&&!%<(G-904*! =$0#,'%'0!16"!eLYfM!V_MMgX*!

! ##$! ecb*! ?4-4-."!=*!4&!.2*!64F4A&'C4!(4$-9(9B4(4%'%!'(!&/4!.O%4(A4!9F!62Eg*!?-*#!'**# ='4(-1&+!0."!LgeYcL!V_MMfX*! eaM*! d&$/04-"!;*"!#(G4-%9("!d*#*"!i334-"!=*!>!)$O4(%&4'("!7*:*!iA&9D'A!4ED-4%%'9(!9F! &/4!62E!B4(4%!'(G$A4%!B2$&.0'A!.A'G!G4A.-O9E<2.%4!.(G!62E!4ED-4%%'9(*! K'9'*-GH'%0!205"!_fgYg_!V_MM_X*! eaL*! ?.(B.('O.("!,*!>!)$O4(%&4'("!7*:*!64C429D04(&.2!F$(A&'9(%!9F!&/4!6'%&.2Y 24%%n62E!/9049O9E!B4(4%*!K'9'*-GH'%0!205"!feaLY`c!V_MM_X*! ea_*! iO4-&"!=*8*"!dA/0'G&"!6*i*"!;/90D%9("!;*!>!@$&24-"!=*,*!i24C.&4G!D2.%0.! B.00.Y.0'(9O$&<-'A!.A'G!V,#@#X!24C42%!'(!'(G'C'G$.2%!H'&/!4'&/4-!?-.G4-Y h'22'!%<(G-904!9-!#(B420.(!%<(G-904*!)#='4(-G1A&<+$0(A#!*+%#='4(-1&+!5"! agY`M!VLbbaX*! eae*! @9-B.&&'"!)*!4&!.2*!?4-'D/4-.2!0.-34-%!9F!&/4!B.00.Y.0'(9O$&<-'A!.A'G! V,#@#X4-B'A!%<%&40!'(!#(B420.(p%!%<(G-904*!)#!<+*@#='4(-*!2/"!_LYg!V_MMeX*! eaf*! #3.%/'"!+*!4&!.2*!;-.(%A-'D&'9(.2!.AA4%%'O'2'&!=42&9("!6*#*! wd&40(4%%wQ!&-.(%A-'D&'9(.2!D-9F'2'(B!9F!40O-<9('A!.(G!.G$2&!%&40!A422%*! C&+'%&'!05/"!gbaYcMM!V_MM_X*! eac*! :$(<.3"!j*j*!4&!.2*!19-4D-4%%9-YG4D4(G4(&!%'24(A'(B!9F!A/-909%90.2!-4B'9(%! 4(A9G'(B!(4$-9(.2!B4(4%*!C&+'%&'!05/"!LafaYg_!V_MM_X*! eaa*! i%A.

!