Prasad S. Dalvi. Copyright by Prasad S. Dalvi 2012

Transcription

1 Molecular Mechanisms Involved in Insulin- and Leptinmediated Regulation of Hypothalamic Proglucagon Gene Expression and Action of Glucagon-like Peptides on Hypothalamic Neuropeptides by Prasad S. Dalvi A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Physiology University of Toronto Copyright by Prasad S. Dalvi 2012

2 Molecular Mechanisms Involved in Insulin- and Leptin-mediated Regulation of Hypothalamic Proglucagon Gene Expression and Action of Glucagon-like Peptides on Hypothalamic Neuropeptides Prasad S. Dalvi Doctor of Philosophy Department of Physiology University of Toronto 2012 Abstract The hypothalamus is a central regulator of energy homeostasis. Recently, proglucagonderived peptides have emerged as potential appetite regulators. The proglucagon gene is expressed in the periphery and also in selective hypothalamic neurons. The regulation of hypothalamic proglucagon by two key regulators of energy balance, insulin and leptin, remains unstudied. Central glucagon-like peptide (GLP)-1 receptor (GLP-1R) activation by exendin-4, a long-acting GLP-1R agonist, induces anorexia; however, the specific hypothalamic neuronal populations activated by exendin-4 remain largely unknown. The role of GLP-2 as a central appetite regulator is poorly understood. In this thesis, using murine hypothalamic cell lines and mice as experimental models, mechanisms involved in the direct regulation of proglucagon gene by insulin and leptin were studied, and the actions of exendin-4 and GLP-2 on hypothalamic neuropeptides were determined. It was found that insulin and leptin regulate hypothalamic proglucagon mrna by activating Akt and signal transducer and activator of transcription 3, respectively. Insulin and leptin did not regulate human proglucagon promoter regions, but affected proglucagon mrna stability. In mice, intracerebroventricular exendin-4 and GLP-2 induced anorexia, activated proopiomelanocortin- and neuropeptide Y-expressing neurons in the arcuate nucleus and ii

3 neurotensin- and ghrelin-expressing neurons in major hypothalamic appetite-regulating regions. In the hypothalamic neuronal models, exendin-4 and GLP-2 activated camp-response elementbinding protein/activating transcription factor-1, and regulated neurotensin and ghrelin mrna levels via a protein kinase A-dependent mechanism. Overall, the in vivo and in vitro findings suggest that these neuropeptides may serve as potential downstream mediators of exendin-4 and GLP-2 action. This research demonstrates direct regulation of hypothalamic proglucagon by insulin and leptin in vitro, and reports a previously unrecognized link between central GLP-1R and GLP-2R activation and regulation of hypothalamic neuropeptides. A better understanding of the regulation of hypothalamic proglucagon and central GLP-1R and GLP-2R activation is important to further expand our knowledge of feeding circuits. iii

4 This thesis is dedicated to my mother, Jayanti, and father, Shridhar, my sisters, Shailaja and Ranjana, and brothers-in-law, Bapu and Anant my mother-in-law, Leena, and late father-in-law, Ramchandra my daughter, Sanika, and son, Ojus, & my wife, Pooja for their endless love, support and encouragement iv

5 Preface This thesis is submitted for the degree of Doctor of Philosophy at the University of Toronto. The research described herein was conducted between January 2007 and April 2012 under the supervision of Dr. Denise D. Belsham in the Department of Physiology, University of Toronto, ON, Canada. Financial stipend was provided by Banting and Best Diabetes Centre, Ontario Government, Natural Sciences and Engineering Research Council of Canada and Dr. Denise D. Belsham. This research is, to the best of my knowledge, original, except where acknowledgements and references are made to previous work. This research has been presented in the following publications: Prasad S. Dalvi, Frederick D. Erbiceanu, David M. Irwin, Denise D. Belsham. Direct Regulation of the Proglucagon Gene by Insulin, Leptin and camp in Embryonic versus Adult Hypothalamic Neurons. Molecular Endocrinology (2012) 26: Prasad S. Dalvi, Anaies Nazarians-Armavil, Matthew J. Purser and Denise D. Belsham. Glucagon-like peptide-1 receptor agonist, Exendin-4, regulates feeding-associated neuropeptides in hypothalamic neurons in vivo and in vitro. Endocrinology (2012) 153: Prasad S. Dalvi and Denise D. Belsham. Glucagon-like peptide-2 regulates hypothalamic neurons expressing neuropeptides linked to appetite control in vivo and in vitro. Endocrinology (2012) 153: v

6 Acknowledgments First and foremost, my heart-felt gratitude goes to my supervisor, Dr. Denise D. Belsham, who hired me as a Master s student at a very critical period in my life and has ever since guided me through my doctorate degree. I am grateful to her for her time, constant support and guidance. Her enthusiasm and encouragement were a strong motivational force throughout the course of my graduate studies. The past five and a half years in the lab have been thoroughly enjoyable and an amazing learning experience that has been profoundly rewarding and has truly played an important part in shaping me as a cellular biology researcher. I also sincerely thank Dr. Patricia Brubaker, Dr. Tianru Jin and Dr. Allen Volchuk for being a part of my supervisory committee and for their invaluable insight towards the project. I am also grateful to Dr. Martin Ralph, Dr. David Irwin and their teams for the great collaborative efforts. I also greatly appreciate the support from the Department of Physiology, Banting and Best Diabetes Centre, Ontario Government, Natural Sciences and Engineering Research Council of Canada and other funding agencies for supporting my research. My deepest appreciation goes towards the past and present members of the Belsham Lab for their invaluable technical help at the bench and helping me orient to the Canadian way of life. I am forever grateful to Jennifer Chalmers, Margaret Koletar and Luisa Centeno for teaching me the ABCs of many scientific methods. It has been a great pleasure working with all. Last but definitely not least, this work would not have been possible without my incredibly supportive family, in-laws, teachers, mentors, friends and well-wishers spread across the globe. Their endless love, teachings, support and encouragement have meant the world to me and I thank them from the bottom of my heart for their endless love, support and encouragement! vi

7 Table of Contents List of Figures... xi List of Abbreviations... xiv Chapter 1: Background and Significance Introduction Regulation of Energy Homeostasis Hypothalamus Arcuate nucleus Paraventricular nucleus Ventromedial hypothalamic nucleus Dorsomedial hypothalamic nucleus Lateral hypothalamus The Hypothalamic Neuropeptides Neuropeptide Y Pro-opiomelanocortin Neurotensin Ghrelin Proglucagon-derived peptides (PGDPs) Biosynthesis Role of PGDPs as appetite regulators Expression of PGDPs and their receptors in the hypothalamus Signaling pathways activated by GLP-1R and GLP-2R stimulation Insulin and insulin receptor signaling Insulin Insulin receptor signaling Insulin regulation of proglucagon Leptin and leptin receptor signaling Leptin Leptin receptor signaling vii

8 1.5.3 Leptin regulation of proglucagon Hypothalamic Neuronal Cell Models Generation of immortalized hypothalamic cell lines Immortalized embryonic hypothalamic cell lines, mhypoe-xx Immortalized adult hypothalamic cell lines, mhypoa-xx Rationale, Hypotheses and Specific Aims Rationale Hypotheses Specific aims Chapter 2: Materials and Methods Cell culture and reagents One-step reverse transcriptase-polymerase chain reaction Radioimmunoassay for adenosine 3,5 -cyclic monophosphate Quantitative reverse transcription-polymerase chain reaction SDS-polyacrylamide gel electrophoresis and western blot analysis Reporter gene plasmids Transient transfections Animal (mouse) experiments Intracerebroventricular (i.c.v.) microinjections for feeding study Assessment of neuronal activation by c-fos immunohistochemistry Experimental normalization Statistical analysis Chapter 3: Regulation of the Proglucagon mrna levels by Insulin and Leptin in Embryonic versus Adult Hypothalamic Neurons Abstract Introduction viii

9 3.3 Results Characterization of the expression profile of the hypothalamic cell lines Activation of signaling pathways by insulin and leptin in the hypothalamic neuronal cells Regulation of proglucagon mrna transcript expression by insulin and leptin Reversal of insulin-mediated regulation of proglucagon mrna expression by PI3K inhibitors Reversal of leptin-mediated regulation of proglucagon mrna expression by JAK2/STAT3 inhibitors Regulation of human or rat proglucagon 5 flanking promoter constructs by insulin and leptin Regulation of mrna stability by insulin and leptin in mhypoa-2/10 and mhypoe-39 neuronal cells In silico analysis of murine proglucagon mrna sequence for microrna binding sites and RNA-binding protein sites Discussion Chapter 4: Glucagon-like Peptide-1 Receptor Agonist, Exendin-4, Regulates Feedingassociated Neuropeptides in Hypothalamic Neurons in vivo and in vitro Abstract Introduction Results Effect of i.c.v. exendin-4 on food and water intake, and animal weight Effect of i.c.v. exendin-4 on activation of hypothalamic nuclei and neuropeptidergic neurons Expression of GLP-1R and appetite-regulating neuropeptides in adult mhypoa-2/30 and embryonic mhypoe-36/1 neurons Activation of CREB/ATF-1 and c-fos by exendin-4 in the hypothalamic neuronal cells ix

10 4.3.5 Regulation of neurotensin and ghrelin mrna transcript expression by exendin Reversal of exendin-4 regulation of neurotensin and ghrelin by PKA inhibitors Discussion Chapter 5: Glucagon-like Peptide-2 Regulates Hypothalamic Neurons Expressing Neuropeptides Linked to Appetite Control in vivo and in vitro Abstract Introduction Results Effect of i.c.v. h(gly 2 )GLP-2 on food and water intake, and animal weight Effect of h(gly 2 )GLP-2 on activation of hypothalamic nuclei and neuropeptidergic neurons Expression of GLP-2R and appetite-regulating neuropeptides in adult mhypoa-2/30 cells Activation of CREB/ATF-1 and c-fos by h(gly 2 )GLP-2 in the hypothalamic neuronal cells Regulation of neurotensin and ghrelin mrna transcript expression by h(gly 2 )GLP PKA inhibitors reverse the h(gly 2 )GLP-2-induced up-regulation of neurotensin and ghrelin mrna transcript levels Discussion Chapter 6: Discussion Overall conclusions Limitations Future directions Chapter 7: References x

11 List of Figures Figure 1. Schematic diagrams of the hypothalamic and brainstem regions that express neuropeptides involved in energy homeostasis (coronal sections)... 8 Figure 2. The products of preproglucagon cleavage. Schematic diagrams of the central regions showing gene expression of proglucagon and GLP-1R and GLP-2R Figure 3. Summary of GLP-1R signaling Figure 4. Summary of GLP-2R signaling Figure 5. Schematic of the proposed regulation of proglucagon-expressing neurons by insulin and leptin in the hypothalamus Figure 6. Summary of the proposed signal transduction mechanisms triggered by GLP-1R and GLP-2R activation in the hypothalamus Figure 7. Characterization of the expression profile of the proglucagon-expressing hypothalamic cell lines Figure 8. Insulin activates Akt and leptin activates STAT3 in the hypothalamic neuronal cells Figure 9. Insulin and leptin regulate proglucagon mrna expression in the hypothalamic neuronal cells Figure 10. Regulation of proglucagon mrna expression by insulin via activation of the PI3K/Akt pathway Figure 11. Regulation of proglucagon mrna expression by leptin via activation of the JAK2/STAT3 pathway... Figure 12. Insulin and leptin do not affect the transcription of proglucagon promoter constructs, but regulate mrna stability... Figure 13. In silico analysis of murine proglucagon mrna sequence for mirna binding sites and RNA-binding protein sites... Figure 14. Intracerebroventricular injection of exendin-4 was effective to reduce food and water intake, and body weight in mice Figure 15. Exendin-4 activates hypothalamic neurons Figure 16. Acute exendin-4 treatment increases the number of hypothalamic neurons co- expressing c-fos-immunoreactivity (ir) with α-msh-, NPY-, neurotensin- or ghrelin-ir xi

12 Figure 17. Graphical representation showing the number of neurons expressing c-fos and neuropeptide-immunoreactivity in the ARC, VMH, DMH, PVN, LH, PeV, and the internuclear space between the VMH and DMH of the saline- or exendin-4-treated mouse hypothalamus Figure 18. Expression profile of GLP-1 receptor and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines Figure 19. Exendin-4 induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic neuronal cell lines Figure 20. Regulation of neurotensin and ghrelin mrna expression by exendin-4 in the mhypoa-2/30 and mhypoe36/1 neuronal cell lines Figure 21. Regulation of neurotensin and ghrelin mrna expression via activation of the protein kinase A pathway Figure 22. Intracerebroventricular injection of h(gly 2 )GLP-2 inhibits food and water intake, and induces weight loss in a dose-dependent manner in mice Figure 23. Acute h(gly 2 )GLP-2 treatment activates hypothalamic appetite-regulating nuclei Figure 24. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity (ir) in the hypothalamic neurons expressing α-msh-, NPY-, neurotensin- or ghrelin-ir Figure 25. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity in the hypothalamic neurons expressing NPY-, neurotensin- or ghrelin-ir Figure 26. Graphical representation of neurons expressing c-fos- and neuropeptideimmunoreactivity in the ARC, VMH, DMH, PVN, LH and internuclear space between the DMH and LH of the saline- or h(gly 2 )GLP-2-treated mouse hypothalamus Figure 27. Expression profile of GLP-2R and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines Figure 28. Acute h(gly 2 )GLP-2 treatment induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic GLP-2R-positive mhypoa-2/30 neuronal cells Figure 29. Regulation of neurotensin and ghrelin mrna expression by h(gly 2 )GLP-2 in the mhypoa-2/30 neuronal cell line in protein kinase A-dependent manner xii

13 Figure 30. Summary of the mechanisms activated by insulin and leptin to regulate proglucagon gene expression in the mhypoa-2/10 and mhypoe-39 neuronal cells Figure 31. Summary of the mechanisms activated by exendin-4 to regulate neurotensin and ghrelin gene expression in the mhypoa-2/30 and mhypoe-36/1 neuronal cells Figure 32. Summary of the mechanisms activated by GLP-2 to regulate neurotensin and ghrelin gene expression in the mhypoe-36/1 neuronal cells xiii

14 List of Abbreviations aa AgRP AMP camp ANOVA AP-1 ATF-1 ARC ATP AMPK BBB BCA BDNF CART CNS CNTF CRE CREB CRH DAB DMEM DMH DMSO DNA cdna amino acid agouti-related peptide adenosine monophosphate cyclic adenosine monophosphate analysis of variance activator protein-1 activating transcription factor-1 arcuate nucleus of the hypothalamus adenosine-5 - triphosphate AMP-dependent kinase blood-brain barrier bicinchoninic acid brain-derived neurotrophic factor cocaine- and amphetamine-regulated transcript central nervous system ciliary neurotrophic factor camp-response element camp-response element-binding protein corticotrophin-releasing hormone diaminobenzidine dulbecco s modified Eagle medium dorsomedial nucleus of the hypothalamus dimethyl sulfoxide deoxyribonucleic acid complementary deoxyribonucleic acid xiv

15 DPP-4 DRB Epac ERK FBS GABA Gβ GHS-R Ghrl / GLPs GLP-1 GLP-1R GLP-2 GLP-2R GnRH GRPP h(gly 2 )GLP-2 GUE HPA i.c.v. IGF-1 IP -ir IBMX IHC IR IRS dipeptidyl peptidase-4 5, 6-Dichlorobenzimidazole 1-β-D-ribofuranoside exchange protein activated by camp extracellular signal-regulated kinases fetal bovine serum -aminobutyric acid G protein beta subunit ghrelin receptor (growth hormone secretagogue receptor) ghrelin-deficient glucagon-like peptides glucagon-like peptide-1 GLP-1 receptor glucagon-like peptide-2 GLP-2 receptor gonadotropin-releasing hormone glicentin-related pancreatic peptide human (Gly 2 )GLP-2(1-33) proglucagon gene upstream enhancer element hypothalamic-pituitary-adrenal axis intracerebroventricular insulin-like growth factor-1 intervening peptide -immunoreactive 3-isobutyl -1-methylxanthine immunohistochemistry insulin receptor insulin receptor substrate xv

16 JAK LH MAPK MCH MC3R MC4R ME MPGF α-msh NPY NT-AS NTS Ntsr Ob-Rb PBS PC1/3 PC2 PeV PGDPs PI3K PIP3 PKA PKC PKI PVN mrna mirna janus kinase lateral hypothalamus mitogen-activated protein kinase melanin-concentrating hormone melanocortin-3 receptor melanocortin-4 receptor median eminence major proglucagon fragment α--melanocyte stimulating hormone Neuropeptide Y neurotensin antiserum nucleus of the solitary tract neurotensin receptor leptin receptor phosphate buffered saline prohormone convertase 1/3 prohormone convertase 2 periventricular nucleus of the hypothalamus proglucagon-derived peptides phosphoinositide-3-kinase phosphatidylinositol-3,4,5-triphosphate protein kinase A protein kinase C PKI amide, myristoylated paraventricular nucleus of the hypothalamus messenger ribonucleic acid microrna xvi

17 sirna RBPDB RT-PCR qrt-pcr SDS-PAGE STAT SV TRH VMH WHO small interfering RNA RNA-binding protein database reverse transcription-polymerase chain reaction quantitative reverse transcription-polymerase chain reaction sodium dodecyl sulfate polyacrylamide gel electrophoresis signal transducer and activator of transcription simian virus thyrotropin-releasing hormone ventromedial nucleus of the hypothalamus World Health Organization xvii

18 1 Chapter 1 Background and Significance

19 2 Published figure: Figure 1. Schematic diagrams of the hypothalamic and brainstem regions that express neuropeptides involved in energy homeostasis (coronal sections). Prasad S. Dalvi, Anaies Nazarians-Armavil, Stephanie Tung, and Denise D. Belsham. Immortalized neurons for the study of hypothalamic function. Am J Physiol Regul Integr Comp Physiol (2011) 300:R Permissions were obtained to reproduce the copyrighted material.

20 3 1 Background and Significance 1.1 Introduction Over the past two decades, an escalating pandemic of obesity has affected millions of people worldwide (1, 2). According to Statistics Canada, more than half of Canadians are overweight or obese, with almost 23% of the population falling in the category of "obese". Obesity leads to numerous complications, such as type 2 diabetes, hypertension, stroke and heart disease, that contribute to the increase in global burden of morbidity and mortality among all age groups of human population (3). Increased appetite, indiscriminate and enhanced high calorie consumption coupled with sedentary and stressful life-styles, and significant contributions of genetic factors are responsible for dysregulation of energy balance that leads to the development of obesity (4, 5). Since dieting alone is unable to control body weight in obese individuals, more research is focused on the development of appetite-regulating drugs for which it is imperative to understand mechanisms underlying central regulation of appetite and energy balance. The longterm objective of this research project is to find new insights into the control mechanisms that are critical to understand regulation of appetite and energy homeostasis. The main control centre of appetite and energy homeostasis in the brain is the hypothalamus, which is comprised of multiple neuronal subtypes expressing either appetite-stimulating (orexigenic) or appetite-suppressing (anorexigenic) neuropeptides. Among them, the proglucagon-derived peptides (PGDPs) have emerged as potential regulators of feeding behavior, particularly glucagon-like peptide (GLP)-1 and GLP-2, which induce anorexigenic effects via stimulation of GLP-1 receptor (GLP-1R) and GLP-2R, respectively (6, 7). Two key anorexigenic hormones, insulin, and leptin, are known as adiposity signals and maintain energy homeostasis through the regulation of hypothalamic neurons (8-10). Several key studies have demonstrated the importance of central insulin actions in regulating energy homeostasis (11-14). Similarly, leptin, an adipocyte-derived hormone acts in the hypothalamus to reduce feeding and body weight (15-17). Insulin and leptin regulation of the hypothalamic proglucagon gene expression, and also the cellular mechanisms triggered by GLP- 1 and GLP-2 in the process of appetite regulation are poorly understood, mainly due to complexity of the in vivo architecture of the hypothalamus and the lack of representative neuronal cell models required for these studies. Recently, our laboratory has generated an

21 4 extensive collection of immortalized, clonal hypothalamic neuronal cell lines that can be used to determine the effects of hormones and neuropeptides on specific neuroendocrine cell types (18, 19). Some of these cell models express the proglucagon gene and can be used to study mechanisms involved in proglucagon gene regulation. Importantly, we have the first hypothalamic cell lines that endogenously express GLP-1R and GLP-2R, and can be used to study hormone-responsiveness, at the level of the individual neuropeptide gene and through analysis of signal transduction events. Using these cells as in vitro models and mice as in vivo models in these studies, the short-term objectives of this research project were to study regulation of proglucagon gene expression in the hypothalamic neurons, to assess the effect of long-acting GLP-1R agonist exendin-4 and GLP-2R agonist h(gly 2 )GLP-2[1-33] on modulation of hypothalamic appetite-regulating neuropeptides and to dissect the mechanisms involved in this process. First, using the hypothalamic cell models, potential action of insulin and leptin, two key regulators of food intake and energy balance, on proglucagon mrna transcript levels in the hypothalamic cell models was investigated. Second, the transcriptional mechanisms involved in the regulation of proglucagon gene expression were defined. Finally, the action of exendin-4 and h(gly 2 )GLP-2[1-33] on the hypothalamic neuropeptides neurotensin and ghrelin, and the signaling mechanisms involved were investigated. 1.2 Regulation of energy homeostasis The central nervous system (CNS) regulates energy balance via an evolutionarily conserved homeostatic system. Over the last two decades, extensive research has been conducted to investigate neuronal pathways, neuropeptides, neurotransmitters and modulators, and signal transduction mechanisms involved in appetite regulation and energy balance. In the CNS, the hypothalamus is a primary control site that regulates central neuroendocrine functions including but not limited to thermoregulation, fluid and electrolyte balance, reproduction, circadian regulation, stress response, and energy homeostasis (20). Afferent signals from the periphery are integrated in the neuroendocrine hypothalamus and processed into efferent signals to modulate food intake and energy expenditure in order to maintain energy balance (21). Dysregulation of this homeostatic system in favor of increased energy intake may lead to metabolic disorders that further result in obesity and related complications such as type 2 diabetes (22).

22 Hypothalamus The hypothalamus plays a vital role in maintaining the energy homeostasis (23). It comprises of several nuclei that regulate energy intake and expenditure. The main nuclei involved in appetite regulation are the arcuate nucleus (ARC), paraventricular nucleus (PVN), ventromedial nucleus (VMH), dorsomedial nucleus (DMH), and lateral hypothalamus (LH) (Figure 1). The ARC, PVN, and DMH are the most important hypothalamic nuclei involved in the regulation of energy balance, and the hypothalamic VMH and LH are involved in the control satiety and hunger, respectively (24, 25). These hypothalamic nuclei contain specific neuronal phenotypes to form a complex network of orexigenic and anorexigenic circuits that regulate energy homeostasis (20). Numerous neuropeptides, including but not limited to neuropeptide Y (NPY), agouti-related peptide (AgRP), melanin-concentrating hormone (MCH), α-melanocyte stimulating hormone (α-msh)/pro-opiomelanocortin (POMC), corticotrophin-releasing hormone (CRH), cocaine- and amphetamine-regulated transcript (CART), brain-derived neurotrophic factor (BDNF), orexin A and B, glucagon-like peptides (GLPs), galanin, ghrelin, and neurotensin mediate orexigenic and anorexigenic processes in the hypothalamus (6). Recently, the PGDPs, specifically, glucagon-like peptide-1 (GLP-1), glucagon-like peptide-2 (GLP-2) and oxyntomodulin, have been investigated as potential regulators of feeding behavior (6). Several studies have shown that central administration of PGDPs controls satiety and reduces food intake (26-28), however further research is warranted to dissect the mechanism(s) utilized by PGDPs to induce anorectic signals and their exact role in appetite regulation (7) Arcuate nucleus The ARC encloses the third ventricle and lies immediately above the median eminence (ME). The ARC-ME area represents an important region because the lack of blood-brain barrier (BBB) in ME permits the interplay between peripheral organs and the brain to take place in this small area (29). Due to its multitude of neuropeptidergic neurons, it is a key hypothalamic nucleus in the regulation of appetite. The two major neuronal populations in the ARC implicated in the regulation of feeding are orexigenic NPY and AgRP and anorexigenic POMC. NPY is a powerful central appetite-stimulant and acts upon its own receptors, Y1, Y2, Y4, Y5 (30), but exerts its orexigenic effect predominantly via the Y1 and Y5 receptors. The majority of the

23 6 hypothalamic NPY neurons found within ARC also co-express AgRP (31, 32). AgRP is part of the melanocortin system and is an antagonist to the melanocortin-3 receptor (MC3R) and melanocortin-4 receptor (MC4R) (33). Thus, when the NPY/AgRP neuron is activated, NPY stimulates orexigenic pathways and AgRP inhibits the anorexigenic melanocortin pathway. Hypothalamic POMC neurons are primarily located in the lateral ARC region (34, 35). Cleavage of POMC by endoproteases yields active peptides including melanocortins that are essential for normal energy homeostasis (10). Feeding stimulates hypothalamic POMC mrna expression, and mice that are obese due to leptin deficiency or insensitivity exhibit decreased hypothalamic POMC mrna expression (35-37). Mice and humans entirely lacking POMC develop hyperphagia and severe early-onset obesity (38). Thus, hypothalamic POMC neurons mediate catabolic responses such as decreased food intake and increased energy expenditure. In fact, the POMC peptide product α-msh has been observed to have significant catabolic effects through its action on MC4R (39, 40). The majority of POMC neurons in the ARC also coexpress CART mrna. Animal studies have shown that central administration of CART inhibits food intake, whereas central injection of CART antiserum increases food intake (41). The analysis of the hormonal control of NPY/AgRP and POMC neurons has progressed immensely. Both neuronal populations express insulin and leptin receptors and are targeted by the respective hormones (12, 42, 43). A large body of evidence has revealed an important role for leptin-activated STAT3 signaling in the arcuate POMC/AgRP neurons in control of energy homeostasis. Disruption of neuron-specific STAT3 results in hyperphagia, obesity, diabetes and infertility (44). Further, it was demonstrated that insulin- and leptin-evoked phosphoinositide-3- kinase (PI3K) activation resulted in Akt-mediated disinhibition of POMC transcription (45, 46). The projections from the first order NPY/AgRP and POMC neurons extensively communicate with the second-order neurons, located in other hypothalamic areas involved in appetite regulation, such as the PVN, VMH, DMH, LH, and the nucleus of the solitary tract (NTS) in the brainstem (47). The axons of these neurons project to CRH-, thyrotropin-releasing hormone (TRH)- and oxytocin-expressing neurons located in the PVN, and MCH- and orexin/hypocretin-expressing neurons of the LH and perifornical area. The ARC is the chief hypothalamic area involved in the control of food intake, therefore, when satiety or adiposity

24 7 signals reach the ARC, anorexigenic peptides are released which activate a catabolic circuit to suppress feeding. In contrast, when satiety or adiposity signals are low in the brain, the activation of anabolic pathway leads to the release of orexigenic peptides indicating the urgency to replenish fuel stores by stimulating food intake Paraventricular nucleus The PVN is located on either side of the superior part of the third ventricle in the anterior hypothalamus. The PVN is a highly vascularised region of the hypothalamus and is protected by the BBB, although its neuroendocrine neurons extend to ME and the posterior pituitary which lack the BBB. It contains a dense cluster of heterogeneous neurons that are activated by various stimuli including stress and physiological changes (48, 49). The magnocellular neurosecretory neurons of the PVN project directly to the neurohypophysis (posterior pituitary) where they release oxytocin or vasopressin into the general circulation (50, 51). The parvocellular neurosecretory neurons of the PVN project to the ME and transmit CRH, TRH, gonadotropinreleasing hormone (GnRH), growth hormone-releasing hormone, somatostatin and dopamine into the portal blood vessels of the adenohypophysis (anterior pituitary) (52). There are other PVN neurons that express neuropeptides such as ghrelin, CART, nociceptin and neurotensin (53-56), which may be involved in the hypothalamic regulation of appetite and autonomic functions in the brainstem and spinal cord. Evidence from many experiments suggests that the PVN is a crucial site for the action of many peptides involved in appetite regulation including but not limited to NPY, ghrelin, orexin-a, leptin, GLP-1 (57). It has been shown that microinjection of CRH and leptin in the PVN attenuated fasting-induced feeding (58, 59). The NPY/AgRP neurons of the ARC innervate TRH neurons in the PVN and inhibit pro-trh gene expression, whereas α-msh/pomc projections stimulate TRH (60-62). Thus by integrating numerous neuronal pathways implicated in energy balance such as projections from the NPY/AgRP and POMC neurons of the ARC as well as projections from the suprachiasmatic nucleus and the NTS the PVN plays an important role in the integration of nutritional signals with the hypothalamopituitary axis and other downstream endocrine glands (49, 63).

25 Figure 1. Schematic diagrams of the hypothalamic and brainstem regions that express neuropeptides involved in energy homeostasis (coronal sections). AgRP, agouti-related peptide; ARC, arcuate nucleus; AP, area postrema; BBB, blood-brain barrier; CART, cocaine and amphetamine-regulated transcript; CB1, endocannabinoid receptor 1; CRH corticotropinreleasing factor; DMH, dorsomedial hypothalamic nucleus; GLP-1, glucagon-like peptide-1; GLP-2, glucagon-like peptide-2; LH, lateral hypothalamus; MCH, melanin-concentrating hormone; ME, median eminence; NPY, NPY; NT, neurotensin; NTS, nucleus tractus solitarius; ObRb, leptin receptor; OC, optic chiasm; OXM, oxyntomodulin; OXY, oxytocin; Pit, pituitary gland; POMC, pro-opiomelanocortin; PVN, paraventricular nucleus; SCN, suprachiasmatic nucleus; 3V, third cerebral ventricle; TRH, thyrotropin-releasing hormone; VMH, ventromedial hypothalamic nucleus. Orange arrows point to the coronal sections of the corresponding brain regions. Dashed arrows indicate the central sites of action for regulating factors. 8

26 Ventromedial hypothalamic nucleus The VMH of the hypothalamus, located adjacent to the ARC, receives NPY, AgRP, β- endorphin and POMC/CART projections from the ARC (41, 64-67), and sends efferent projections to the DMH, PVN and brain stem regions such as NTS. The neurons in the VMH have long been hypothesized to play a major role in metabolic regulation since bilateral VMH lesions cause hyperphagia and obesity (68-70). The VMH contains a large population of glucoresponsive neurons which show increased activity during oral hyperglycemia (71, 72). Recent work has demonstrated that BDNF is highly expressed in the VMH and central administration of BDNF reduces feeding and causes loss of body weight (73). The VMH BDNF neurons are regulated by food deprivation (74) and ARC POMC neurons via MC4R activate them to decrease food intake (74). It has been found that the VMH expresses receptors to several signalling molecules including leptin, which is a potent anorexigenic signal to inhibit appetite and feeding and stimulate energy expenditure and weight reduction (75) Dorsomedial hypothalamic nucleus The DMH receives a high level of ARC NPY and α-msh/pomc neuronal terminals (32, 76), whereas it sends projections to communicate with VMH, PVN and LH to integrate and process information from these regions (15, 77). It has been shown that the electrolytic destruction of the DMH disrupts feeding causing hyperphagia and obesity that suggests a role for DMH in appetite regulation (78). Other findings have confirmed that DMH may act as a site of action for leptin and interaction between NPY and leptin (79, 80). Normally, the DMH contains a few NPY neurons (81, 82), however in diet-induced obese, in genetic obese mice and during suckling-induced hyperphagia NPY gene expression is increased in the DMH (15, 83-85) Lateral hypothalamus The LH extends rostrally from the mesencephalic tegmentum to the lateral preoptic area in dorsal and lateral aspect to VMH (77). It contains sparsely distributed subpopulation of neurons expressing orexigenic MCH and orexins (orexin A and B or hypocretin-1 and -2) (86, 87) which are extensively synapsed by projections from NPY/AgRP and α-msh/pomc neurons from the ARC. MCH-immunopositive fibers further project to the cortex, brainstem and spinal

27 10 cord (88), and orexin/hypocretin neurons exert their effects via extensive communication with the ARC, PVN and NTS (86, 89). The mechanism by which MCH and orexin/hypocretin neurons regulate energy homeostasis is yet unclear. Apart from these neurons, large number of glucose-sensing neurons are present in the LH that may integrate peripheral signals and mediate the marked hyperphagia which is normally induced by hypoglycemia (24) The Hypothalamic Neuropeptides Neuropeptide Y NPY is a 36-amino acid (aa) peptide and is one of the most potent neuropeptides known to induce feeding in animals. NPY is one of the most abundant peptides found in the brain and NPY-expressing neurons are widely distrubuted throughout the CNS. In the hypothalamus, NPY is abundantly synthesized by the NPY-expressing neurons of the ARC and are known as the key orexigenic neurons (90). These arcuate NPY neurons send projections towards the PVN as well as the DMH (91). The widespread distribution of NPY facilitates functional diversity, including cardiovascular regulation, seizure and cognition, stress, modulation of neuroendocrine systems and appetite regulation. NPY mediates its biological actions through a portfolio of G-protein coupled receptors, Y1-Y6 and second messenger systems in different downstream effector neuronal types (30, 92, 93). Several studies suggest that the Y1 and Y5 receptors are important in mediating the orexigenic effects of NPY in rats (94). Negative energy states such as fasting, starvation, food deprivation, diet restriction or hunger induce an increase in NPY neurons in the hypothalamus and NPY mrna expression in various brain nuclei (77, ), whereas positive energy state following refeeding normalizes both NPY secretion and NPY mrna levels (102). NPY neurons are regulated by numerous metabolic factors such as glucose, free fatty acids, insulin, leptin, ghrelin and other hormones such as glucocorticoids ( ). It has been shown that the satiety hormones insulin and leptin decrease NPY expression (103, 104), and acute exposure to glucose inhibits NPY neurons (122). In contrast, the orexigenic hormone ghrelin directly stimulates glucose-sensitive ARC NPY neurons (103, ). Glucocorticoid receptors are found on almost all NPY neurons in the ARC (108), and glucocorticoids upregulate NPY expression in the hypothalamus (109).

28 11 In the genetically obese Zucker rats, NPY mrna is increased in the ARC nucleus causing hyperphagia in these rats (110). Similarly, hypothalamic NPY gene expression is upregulated and peptide content in the ARC nucleus is significantly increased in the obese leptindeficient ob/ob mice (102, 111). Importantly, in adult rats central administration of NPY stimulates food intake, whereas ablation of NPY/AgRP neurons in young mice reduces food intake and causes severe starvation highlighting the importance of these neurons to energy balance ( ). However, neonatal NPY/AgRP ablation does not elicit any changes in feeding behavior (112). Similarly, the phenotype of NPY knockout mice does not alter from that of wild-type mice (115). The absence of any evident effect of NPY deletion on phenotype of NPY knockout mouse model can be attributed to the compensatory adaptive mechanisms that may develop during early developmental stages. Nevertheless, it has been demonstrated that NPY ablation in ob/ob mice attenuates obesity through decreased feeding and weight gain suggesting that the NPY neurons play a crucial role in the development of obesity (116) Pro-opiomelanocortin POMC neurons are expressed in the ARC of the hypothalamus and the anterior and intermediate lobes of the pituitary gland. POMC is a precursor polypeptide that undergoes extensive, tissue-specific, post-translational processing via cleavage by enzymes, such as prohormone convertases, carboxypeptidase E, peptidyl α-amidating monooxygenase, N- acetyltransferase, and prolycarboxypeptidase. The POMC-derived peptides have important roles in the regulation of appetite and energy homeostasis (117, 118). Mutations in the POMC gene in humans have been associated with hyperphagia and early onset obesity ( ), and POMCdeficient mice are severely obese (38, 122). In the hypothalamus, enzymatic cleavage of POMC produces melanocortin peptides including α-msh which exert their effects via melanocortin receptors. The melanocortin receptors MC3R and MC4R are expressed in the brain, but the MC4R is highly expressed in the hypothalamus, particularly the PVN (123). It is now clear that the α-msh released from ARC POMC neurons activates downstream MC4Rs to inhibit food intake (124), whereas AgRP from the ARC, being the endogenous inverse agonist and antagonist of the MC3R and MC4R, reverses the α-msh-mediated inhibition of feeding and stimulates food intake and a lower

29 12 metabolic rate (31). The physiological role of POMC-derived peptides in the brain is of particular interest in obesity research as it has been shown that α-msh together with the MC3R and MC4R is central to the regulation of appetite and energy homeostasis. In vivo studies have shown that POMC-derived peptides have potent anorexigenic effects, but α-msh has the most potent anorexigenic effects as it induces severe hypophagia and causes the greatest reduction in body weight when administered to POMC-null mice (125). These findings highlight the importance of this precursor polypeptide in the homeostatic regulation of feeding and body weight (125). Further, mutations in the MC4R cause obesity, while agonists of the receptor function suppress feeding ( ). The MC4R-deficient patients are hyperphagic and up to 6% of early-onset morbidly obese patients in British Caucasians were found to have MC4R mutations (129, 130). Recently, it was found that POMC is colocalized with GLP-1R in ARC neurons, and another recent study indicates that the GLP-1R activation in the ARC POMC neurons is involved in regulation of glucose homeostasis, but not feeding (131). Whether GLP-1 has any direct effect on POMC expression or secretion of POMC-derived peptides is yet to be determined Neurotensin The coordinated regulation of food intake and energy expenditure takes place in the hypothalamus. Among the multitude of central signaling pathways and neuropeptides involved in appetite regulation, hypothalamic neurotensin inhibits feeding due to its anorexigenic action (132, 133). Neurotensin-expressing neurons are located mainly in the parvocellular portion of the PVN, in the ARC, PeV, and LH. Moderate numbers of cell bodies containing neurotensin-like immunoreactivity can also be found in the anterior hypothalamus, the DMH, and the posterior hypothalamus (56). Neurotensin mediates its effect through neurotensin receptors (Ntsr) (134), which are expressed in the hypothalamic ARC and DMH nuclei ( ). Ntsr deficiency moderately increases food intake and body weight, and blocks neurotensin-induced anorexia in mice, implicating neurotensin-ntsr signaling pathway in feeding and body weight regulation (139). Leptin is one of the most important adipose-derived hormones that reduces appetite by regulating several central and peripheral signaling pathways involved in the regulation of energy

30 13 homeostasis (140). Recently, it was found that leptin directly stimulates neurotensin gene expression in the hypothalamic cell lines that express both leptin receptor and neurotensin (141). Furthermore, it has been suggested that leptin s inhibitory regulation on feeding is mediated at least partly through neurotensin-ntsr signaling pathway (142, 143). All these findings confirm that leptin targets neurotensin neurons to mediate its effect on appetite regulation. Interestingly, in another study, it was found that when GLP-1R activation was blocked by exendin-(9-39), leptin s anorectic action was abolished (144). Further, exendin-4 treatment decreased food intake and body weight in leptin-deficient ob/ob mice and fatty Zucker rats, indicating that leptin deficiency is overcome by GLP-1R activation to induce leptin s anorectic effect (145). Based on these findings it can be strongly suggested that leptin may interact with intermediate proglucagon neurons that subsequently excite downstream negative energy balance by stimulating anorexigenic neurons to induce appetite suppression; neurotensin neurons can be the potential downstream targets for the action of central GLPs. As yet, no studies have been performed to link hypothalamic GLPs or their receptors with neurotensin Ghrelin Ghrelin is a potent orexigenic hormone that strongly influences generation of hunger and therefore is known as the hunger hormone. Its plasma concentrations increase before meals and during fasting, and decrease after ingestion of food. Ghrelin is predominantly produced in the stomach and stimulates appetite by its action on the ARC of the hypothalamus via activation of growth hormone secretagogue receptor (GHS-R). Two isoforms of ghrelin have been identified: active (acyl ghrelin) and inactive (des-acyl ghrelin). The enzyme ghrelin O- acyltransferase converts proghrelin peptide into active acyl ghrelin (146, 147). Des-acyl ghrelin does not bind to the GHS-R, and its biological roles are uncertain in the absence of an identified equivalent receptor (148). Apart from the production of mature ghrelin, the proghrelin peptide is also post-translationally processed in the stomach to generate an entirely different peptide, obestatin, that apparently activates an orphan G-protein-coupled receptor, GPR39 (149). However, the physiological function of obestatin and its receptor remains unclear. Although, initial research demonstrated that obestatin had a potent anorectic effect in mice and rats (149), subsequently these findings were disputed (150, 151).

31 14 Human and animal studies have demonstrated that the activation of the GHS-R results in increased food intake and increased adiposity ( ). Peripheral and central administration of ghrelin increases feeding and promotes weight gain (153, 154). It is reported that endogenous centrally released ghrelin is also involved in the regulation of food intake and body weight (153, 155). Existence of ghrelin-expressing neurons in the hypothalamus is debated, however, ghrelinexpressing neurons have been shown to be present in the internuclear spaces between the PVN, ARC, VMH, and DMH hypothalamic nuclei, the perifornical region, and the ependymal layer of the third ventricle (105). These neurons interact with key hypothalamic neurons involved in appetite regulation, mainly the NPY/AgRP and POMC neurons of the ARC suggesting that hypothalamic ghrelin may stimulate orexigenic or inhibit anorexigenic neuropeptides (105, 156). Although a series of pharmacological and clinical studies suggested that ghrelin was an endogenous regulator of energy balance, recent genetic studies using adult ghrelin-deficient (Ghrl / ) mice resulting from targeted ghrelin gene disruption fail to demonstrate any significant defects in normal energy balance, food intake, and adiposity on a standard diet (157, 158). Such lack of a metabolic phenotype may be attributed to compensatory processes during early developmental phases or existence of redundant pathways controlling energy balance. Nevertheless, absence of ghrelin in Ghrl / mice protects them from the early-onset obesity induced by early exposure to a high-fat diet (159). Importantly, simultaneous deletion of ghrelin and its receptor leads to decreased body weight, increased energy expenditure, and increased motor activity on a standard diet demonstrating that ghrelin has a physiological role in the regulation of energy homeostasis (160). Another finding that the mice lacking GHS-R are resistant to the development of diet-induced obesity demonstrates the importance of GHS-R in the regulation of energy expenditure and body weight (161). Regulation of ghrelin-expressing neurons in the hypothalamus is not well studied. Although it has been shown that GLP-1 inhibited ghrelin-induced feeding and exendin-4 reduced plasma levels of peripheral ghrelin (162, 163), whether hypothalamic ghrelin expression is regulated by GLP-1R or -2R activation is not known. It can be speculated that the anorexigenic action of the centrally-originated GLPs is mediated via inhibition of hypothalamic ghrelin neurons; however, no studies have been conducted to elucidate this mechanism. Further investigation is needed to explore the possibility of this interaction because inhibition of the

32 15 ghrelin-expressing neurons may lead to reduced food intake, and consequently to reductions in body weight and adiposity. Therapeutic intervention by inhibiting ghrelin/ghs-r pathway may be used to prevent or treat obesity (164). 1.3 Proglucagon-derived peptides Biosynthesis and regulation The proglucagon gene was isolated from rodents and humans in the early 1980s ( ). The proglucagon gene is expressed in the pancreatic α-cells, intestinal L-cells and a small number of neurons in the CNS, specifically the hypothalamus and the brain stem, and a single copy of proglucagon gene generates the PGDPs (166, 168, 171, 172). Identical preproglucagon mrna transcripts are expressed in these cells, however, the tissue- or cell-specific posttranslational processing of proglucagon leads to the biosynthesis of diverse PGDPs (173, 174) (Figure 2A). Glicentin, oxyntomodulin, GLP-1, GLP-2, intervening peptide (IP)-1 and IP-2 are synthesized in the intestinal endocrine L-cells which are located mainly in the distal ileum and colon, whereas glucagon, glicentin-related pancreatic peptide (GRPP), major proglucagon fragment (MPGF) and IP-1 are predominantly produced by pancreatic α-cells (171, ). This diversification is achieved through proteolytic processing by cell- or tissue-specific prohormone convertases (174, ). Currently, there are 9 known mammalian prohormone convertases, all of which are thought to be responsible for the cleavage of many prohormones and proneuropeptides to generate shorter forms by targeting the carboxyl terminus of a single or paired basic residues of these prohormones and proneuropeptides (177). In the case of precursor proglucagon, each PGDP is flanked by pairs of basic residues that present canonical prohoromone cleavage sites for different prohormone convertases to exert their catalytic functions (171). The tissue-specific expression of the different convertases is responsible for the unique protein profile of PGDPs. Prohormone convertase 2 (PC2) is the most abundant convertase in the pancreatic α-cells and co-localizes with glucagon in the islet secretory granules (180). In contrast, intestinal L-cells highly express prohormone convertase 1/3 (PC1/3) that is responsible for the generation of the intestinal PGDPs including GLP-1 and GLP-2 (174, ). Both

33 16 PC2 and PC1/3 are capable of cleaving proglucagon to generate the N-terminal intermediate glicentin and C-terminal intermediate MPGF (181); however PC2 is required for additional processing of glicentin to GRPP, glucagon and IP-1 (173, 181), and PC1/3 is required to produce GLP-1 and GLP-2 from MPGF (173, 174, ). The post-translational processing of proglucagon in the CNS is mostly similar to that observed in the intestinal L-cells (183). Original studies investigating expression pattern of hypothalamic PGDPs showed that glicentin, oxyntomodulin and glucagon are abundantly present in fetal rat hypothalamus, whereas in adult rat hypothalamus glicentin and oxyntomodulin are present in greater amounts than glucagon and GLP-1 suggesting that the processing of proglucagon in the hypothalamus varies with the brain development (183, 184). Regulation of proglucagon gene expression in the pancreas and intestine has been extensively studied (185). Extensive investigations of mechanisms underlying proglucagon gene transcription have led to the identification of a minimum promoter region (G1) and four enhancer elements (G2 G5) in the proglucagon gene promoter (186, 187). Transgenic mouse studies have indicated that approximately 1.3 kb of rat proglucagon gene 5 flanking sequences are sufficient to direct pancreatic α-cell- and brain-specific rat proglucagon gene expression (188). In contrast, similar studies suggest that a much larger region ( 2.3 kb) of rat proglucagon 5 -flanking sequences is required for proglucagon gene expression in the intestine (189), indicating that DNA sequences located between -2.3 and -1.3 kb in the rat proglucagon promoter contain specific elements required for intestinal proglucagon gene expression. These sequences situated between -2.3 and -1.3 kb have been designated as the proglucagon gene upstream enhancer element (GUE) that is composed of multiple positive and negative cis-acting DNA regulatory elements involved in regulating tissue-specific proglucagon gene transcription (190). A combination of cell transfection and transgenic reporter studies have reported that approximately 1.6 kb of human proglucagon gene 5 -flanking sequences can direct proglucagon gene transcription to the brain and intestine, but for pancreas-specific proglucagon gene expression the sequences within the first 6 kb of the human proglucagon gene 5 -flanking region are required (191). These studies indicate that specific trans- and cis-regulatory elements are required for tissue-specific proglucagon gene transcription in rodents and humans.

34 17 Several studies have demonstrated that proglucagon gene transcription can be regulated by cyclic adenosine monophosphate (camp), amino acids, and a number of homeodomain protein transcription factors such as isl-1, cdx-2/3, pax-6, HNF-3α, HNF-3β, and brn 4 (185, ). Further, FoxO1 has been identified as a critical regulator of proglucagon gene expression by insulin in pancreatic αtc1-9 cells (197). Recently, the Wnt/TCF-4 pathway was shown to be involved in the regulation of proglucagon gene expression by insulin in enteroendocrine cells (198). Furthermore, pathological levels of insulin were found to stimulate intestinal proglucagon mrna and GLP-1 production (199), in contrast to insulin s known inhibitory effect on proglucagon gene expression in pancreatic α-cells (200, 201). Although it is known that the activators of camp/protein kinase A (PKA) pathway stimulate synthesis and secretion of the PGDPs in the hypothalamus (202), and excitatory amino acid glutamate stimulates PGDP secretion through a protein kinase C-dependent pathway (203), other internal as well as external factors involved in regulating these neurons are not fully identified. At present, the mechanisms involved in hypothalamic proglucagon gene regulation remain largely unknown Role of PGDPs as appetite regulators The appetite-related effects of PGDPs are well documented; particularly GLP-1, GLP-2 and oxyntomodulin are involved in inhibition of food intake (26, 204, 205). The actions of GLP- 1, GLP-2 and oxyntomodulin are exerted through receptors that belong to the family of 7- transmembrane-spanning G-protein-coupled receptors, members of glucagon receptor superfamily. GLP-1 and oxyntomodulin act via the GLP-1R, and GLP-2 specifically activates the GLP-2R (206, 207). GLP-1, GLP-2 and oxyntomodulin are mainly released from intestinal epithelial endocrine L-cells in response to food intake. GLP-1 is inactivated by the ubiquitous enzyme dipeptidyl peptidase-4 (DPP-4) in less than 2 minutes. Therefore, the long acting DPP-4 resistant GLP-1R agonist exendin-4 has been developed (208). Acute administration of GLP-1 or exendin-4 suppresses food intake inducing satiety, and chronic administration results in weight loss (26, ). GLP-2, similar to GLP-1, is also inactivated by DPP-4 thereby having a relatively short biological half-life of approximately 7 minutes. Consequently, a degradation-

35 18 resistant analog (Gly 2 )GLP-2 (Teduglutide) is used in biomedical research (212). GLP-2 administration results in inhibition of gastric emptying and reduced intestinal motility (204). Oxyntomodulin also reduces gastro-intestinal motiliy contributing to the ileal brake effect. Acute central administration of oxyntomodulin inhibits food intake and reduces body weight in rodents (205). Thus, it can be concluded that central GLP-1R and GLP-2R agonism constitutes a potential pharmacological tool to reduce food intake and enhance energy expenditure Expression of PGDPs and their receptors in the hypothalamus In the brain, proglucagon neurons are localized in the NTS and the hypothalamus (Figure 2B). The NTS preproglucagon neurons that process proglucagon to GLP-1, GLP-2 and oxyntomodulin, project mainly to the two hypothalamic nuclei involved in appetite regulation - the PVN and the DMH (213, 214). GLP-1-immunoreactive (ir) nerve fibres are densely located in the ARC, PVN and DMH (215), whereas GLP-2-ir nerve fibers are found mainly in the ventral part of the DMH and also in the ARC and PVN (214). The GLP-1R is widely expressed in all hypothalamic areas receiving GLP-1-ir fibers i.e. the ARC, PVN and DMH nuclei and also the VMH and LH (216, 217), but GLP-2R expression in the hypothalamus is much more limited and confined to the DMH and VMH nuclei (214, 218). Thus, the localization of proglucagonneurons as well as GLP-1R- and GLP-2R-expressing neurons in the hypothalamus is very well defined (Figure 2B); however, it is not yet completely known which PGDPs, whether centrallyor peripherally-derived PGDPs, activate GLP-1R and GLP-2R in the hypothalamic nuclei that regulate energy homeostasis. It has been demonstrated that GLP-1 and long-acting GLP-1R agonist exendin-4, can cross BBB (219, 220), and it is possible that similar to GLP-1, GLP-2 can also gain access to the brain from the periphery. However, it is a possibility that due to their rapid inactivation by DPP-4, gut-derived GLP-1 and GLP-2 may not even cross the BBB, and therefore only centrally-derived GLP-1 and GLP-2 are involved in their action on the hypothalamus. Also it is yet to be confirmed whether differential distribution of GLP-1R and GLP-2R in the hypothalamus is responsible for different roles played by central GLP-1 and GLP-2 in appetite regulation and energy homeostasis.

36 19 A B Proglucagon Hypothalamus NTS 3V LH GLP-1R ARC VMH DMH PVN LH 3V LH GLP-2R VMH DMH Figure 2. A: The products of preproglucagon cleavage. GLP-1 and GLP-2, glucagon-like peptides 1 and 2; GRPP, glicentin-related pancreatic peptide; IP-1 and IP-2, intervening peptides 1 and 2; MPGF, major proglucagon fragment; PC, prohormone convertase. B: Schematic diagrams of the central regions showing gene expression of proglucagon and GLP-1 receptor (GLP-1R) and GLP-2R. ARC, arcuate nucleus; AP, area postrema; DMH, dorsomedial hypothalamic nucleus; LH, lateral hypothalamus; NTS, nucleus tractus solitarius; Pit, pituitary gland; PVN, paraventricular nucleus; 3V, third cerebral ventricle; VMH, ventromedial hypothalamic nucleus.

37 Signaling pathways activated by GLP-1R and GLP-2R stimulation The GLP-1R is expressed in GLP-1 responsive tissues, including pancreatic β-cells, the gastrointestinal tract and the brain. GLP-1R is activated not only by GLP-1, but also by oxyntomodulin to mediate its anorexigenic actions (221). GLP-1R activation is inhibited by exendin-9 39, a known GLP-1R antagonist that blocks actions of GLP-1 and oxyntomodulin. The signaling pathways triggered by GLP-1R activation in peripheral tissues have been extensively studied (Figure 3). In the pancreatic β-cells, GLP-1R stimulation by GLP-1 activated camp/pka, exchange protein activated by camp (Epac), PI3K and mitogen-activated protein kinase (MAPK) pathways ( ). Oxyntomodulin also mimics many of the actions of GLP-1 on the islet β-cells. Similar to GLP-1, oxyntomodulin, via functional GLP-1R, activates camp formation in murine islets and INS-1 cells (228). GLP-2R expression is restricted to the gastrointestinal tract and the CNS, with limited expression in other peripheral organs (229). GLP-2R recognizes specifically GLP-2 and is inhibited by GLP-2(3-33). Complete information about GLP-2R activation is not available due to the lack of cell models that endogenously express GLP-2R (230). In heterologous cell lines transfected with rat or human GLP-2R, camp-dependent pathways are activated by GLP-2 (207, 231) (Figure 4). Recently GLP-2 has been shown to activate Akt in the intestinal epithelium (232) and also the wnt/β-catenin signaling pathway in the intestinal crypt cells (233). The signaling pathways activated by GLP-1R and GLP-2R stimulation in the CNS, particularly in the hypothalamus are not well studied. The available majority of the data highlighting the effects of GLP-1R and GLP-2R stimulation is based on studies performed in cell lines derived from peripheral tissues. Until now, it is due to the lack of representative neuronal cell models that the signal transduction pathways activated by GLP-1R and GLP-2R stimulation in the hypothalamic neurons have not been studied completely. To address this issue, for the first time, clonal, immortalized, hypothalamic, neuronal cell lines that endogenously express GLP-1R and GLP-2R will be used as research models (18, 19, 234, 235). Recently, using these cell lines, it was shown that GLP-1 regulates expression of hypothalamic insulin (234), however, regulation of other neuropeptides by PGDPs remains largely unknown.

38 Figure 3. Proposed GLP-1R signaling in the hypothalamus. GLP-1R activation in the hypothalamic neurons may trigger key signal transduction pathways to regulate gene expression of hypothalamic neuropeptides. In the peripheral tissues, GLP-1R activates camp/pka/ CREB/ATF-1, c-fos and MAPK/ERK pathways, and other signaling pathways, such as camp/epac, Wnt/β-catenin and PI3K/Akt can also be activated. 21

39 Figure 4. Proposed GLP-2R signaling in the hypothalamus. GLP-2R activation in the hypothalamic neurons may trigger key signal transduction pathways to regulate gene expression of hypothalamic neuropeptides. In the peripheral tissues, GLP-2R activates camp/pka/creb pathway, and other signaling pathways, such as Wnt/β-catenin, c-fos and PI3K/Akt can also be activated. 22

40 Insulin and insulin receptor signaling Insulin The discovery and isolation of insulin at the University of Toronto by Banting, Best, Collip, and McLeod in the last century was one of the greatest events in the history of medicine. Since then insulin has become the life-saving therapy for patients suffering from type 1 diabetes and certain patients with type 2 diabetes (236). Insulin is a polypeptide hormone produced within the β-cells of the islets of Langerhans. Insulin is generated from a 110-aa single chain preprohormone, that is initially processed to proinsulin, through removal of 24 aa, then to the mature 51 aa insulin and C-peptide through enzymatic cleavage by the prohormone convertases PC1/3 and PC2, and carboxypeptidase E (237). Once mature insulin is synthesized, it is packaged into granules that are secreted in response to glucose entry into β-cells. Higher glucose metabolism in the β-cells increases cellular ATP that causes closing of ATP-sensitive potassium channels, preventing potassium ion outflow, leading to membrane depolarization and calcium influx via voltage dependent calcium channels (238). The increased cytosolic calcium leads to exocytosis of the secretory granules, releasing insulin into circulation (239). The liver, skeletal muscle and adipose tissue express insulin receptor (IR) and are the key peripheral targets of insulin that mediate its anabolic actions (238) Insulin receptor signaling The key actions of insulin in peripheral tissues include increased glucose and amino acid uptake, fatty acid synthesis, glycogen synthesis; and also decreased lipolysis, proteolysis, and gluconeogenesis (238). Insulin action is mediated by a membrane receptor that belongs to the tyrosine kinase receptor superfamily (240, 241). The IR protein is a heterotetrameric protein composed of two 135 kda alpha subunits that bind insulin and two 95 kda beta subunits containing the transmembrane and tyrosine kinase domains linked by disulfide bonds (242, 243). Through alternative splicing of the primary transcript, the IR protein has two isoforms known as the A and B isoforms (240, 241, 244). Isoform A lacks exon 11, which codes for a 12-aa fragment near the C-terminal end of the α-chain (245), and differential expression of the two

41 24 isoforms allows for preferential activation of either the IR or insulin-like growth factor-1 receptor pathways (246). Only the IR isoform A is found in the brain (247). It is well established that IRs are located in the hypothalamus (248, 249), and central injection of insulin potently reduces food intake and body weight (250). Insulin exerts its central effects predominantly through the PI3K pathway. The IR remains in an inactive dimerized state when insulin is not bound (243). Insulin binding to the receptor causes autophosphorylation of the tyrosine residues that allow docking of the insulin receptor substrate (IRS) family of proteins, which mediate the IR signaling (251, 252). The IRS proteins are then activated through tyrosine phosphorylation and can recruit PI3K to the cell membrane for its activation (253). Activation of PI3K leads to the phosphorylation of membrane bound phosphatidylinositol-3,4-bisphosphate to generate phosphatidylinositol-3,4,5-triphosphate (PIP3). The binding of PIP3 to the N-terminus pleckstrin homology domain anchors Akt to the plasma membrane and allows its phosphorylation and activation by phosphoinositide-dependent kinase-1. Activation of Akt can then regulate numerous downstream signaling molecules in the periphery as well as in the CNS (254, 255). Insulin action in the CNS is mediated via modulation of neuropeptide transcription and release (256). Although IRS1 and IRS2, are often linked to the PI3K-Akt pathway, which is responsible for the most of the metabolic actions of insulin, the MAPK pathway, works in conjunction with the PI3K pathway to control cell proliferation, apoptosis and differentiation (257, 258). Among multiple signalling pathways activated by insulin, the PI3K-Akt pathway remains the main focus in central nervous system studies. Several investigations have demonstrated that insulin activates PI3K in neurons (259, 260), and that PI3K inhibitors can block the ability of insulin to regulate food intake (260) Insulin regulation of proglucagon The regulatory mechanisms underlying the differential and opposite actions of insulin on the pancreatic and intestinal proglucagon gene expression have been extensively investigated (185). It was demonstrated that insulin inhibited islet proglucagon gene expression via regulation of gene transcription (200, 201), but another study found that proglucagon gene expression was insensitive to changes in plasma glucose and insulin concentrations, however, hyperinsulinemia in the presence of hyperglycemia lowered glucagon secretion from pancreatic α-cells (261).

42 25 Further, it was found that Akt was sufficient to mimic the inhibitory effect of insulin on proglucagon expression in the pancreatic α-cells (262). Insulin may inhibit pancreatic proglucagon by causing exit of FoxO1 from the nucleus to the cytoplasm (197). The repression of proglucagon is physiologically important because it may result in decreased production of glucagon, the major counter-regulatory hormone of insulin secreted by islet α-cells. In contrast to inhibitory effect of insulin on proglucagon expression in pancreatic α-cells (200, 201), pathological levels of insulin stimulate intestinal proglucagon mrna and GLP-1 production via activation of the bipartite transcription factor β-catenin/t cell factor, the major effector of the canonical Wnt signaling pathway (199). Additionally, in the MKR mice, a non-obese model of chronic insulin resistance and hyperinsulinemia, intestinal proglucagon mrna expression and GLP-1 content were found to be significantly increased (199). A recent finding indicated that insulin stimulated GLP-1 secretion from the intestinal L-cells in a glucose-dependent manner (263). Furthermore, similar to the proglucagon gene transcription in pancreatic α-cells, insulin was demonstrated to utilize the same cis- and trans-regulatory elements in stimulating intestinal proglucagon expression in the intestinal L-cells, supporting the existence of crosstalk between insulin and Wnt signaling pathways (199). In the pancreatic α-cells, the repressive effect of insulin on proglucagon gene expression relies on Akt activity (262), however, in the intestinal L cells, it is likely that the Wnt signaling pathway, but not the Akt pathway (Akt-independent mechanism), is involved in insulin-stimulated proglucagon promoter transcription (199). Recently, it was found that Epac is involved in camp-stimulated proglucagon expression and hormone production in pancreatic and intestinal endocrine cells thus demonstrating the PKAindependent regulation of proglucagon in the peripheral tissues (264, 265). Currently, no studies have been conducted to elucidate how insulin regulates hypothalamic proglucagon and the underlying mechanisms. 1.5 Leptin and leptin receptor signaling Leptin Leptin is secreted by adipocytes and was cloned from the ob (leptin) gene in 1994 by Friedman and his colleagues (17). Leptin is essential to maintain physiological energy homeostasis. Knockout mice with mutations in the ob gene or db (leptin receptor) gene are

43 26 morbidly obese. Further, congenital leptin deficiency is associated with severe early-onset obesity in humans (266). Administration of leptin to leptin deficient ob/ob mice and humans normalizes body weight and neuroendocrine status (39, 267, 268), indicating the importance of leptin in energy homeostasis. Leptin plasma concentrations are directly proportional to adiposity levels and therefore it is known as an adiposity signal (269, 270). Although leptin does not exert any demonstrable effect on energy metabolism in non-obese humans (271), recent studies involving leptin administration to humans support its critical role in regulating neuroendocrine and immune functions as well as insulin resistance in states of energy deficit (272). The gene for the leptin receptor (Ob-R) is encoded by the db gene and is spliced into several receptor isoforms. There is one long leptin receptor isoform: Ob-Rb, and several shorter isoforms spliced at the C-terminal coding region: Ob-Ra, Ob-Rc, Ob-Rd, Ob-Re and Ob-Rf ( ). All of the leptin receptor isoforms have an extracellular leptin-binding domain, but only Ob-Rb has a full-length intracellular domain required for signal transduction (269, 276). It is well established that leptin receptors are located in the CNS, and within the rodent and human hypothalamus, Ob-Rb, is highly expressed in the ARC, DMH, VMH, and is also detected in the hypothalamic PeV, LH and PVN ( , 279). In the ARC, Ob-Rb is predominately expressed in the NPY/AgRP and POMC neurons (43, 273, 280). Leptin gains access to the brain by crossing the BBB via a saturable transport mechanism (281). It is known that leptin regulates feeding and energy balance through interaction with complex neuronal circuits comprised of anorexigenic and orexigenic neuropeptides. Central administration of leptin potently reduces food intake and body weight in rhesus macaque (282). Lack of functional leptin receptors or signaling in mice results in severe obesity (283, 284), while brain administration of leptin to ob/ob mice can overturn the obese phenotype (37, 39, 267). These and other findings indicate that leptin action within the hypothalamus via NPY/AgRP and POMC neurons is integral to the maintenance of energy homeostasis (37, 285, 286) Leptin receptor signaling Leptin receptors belong to the class 1 cytokine receptor family that utilizes several cytosolic signaling proteins to induce changes in gene transcription and metabolism (287). Leptin activation of Ob-Rb predominantly triggers the janus kinase 2 (JAK2)/signal transducer

44 27 and activator of transcription 3 (STAT3) pathway. JAK2, a tyrosine kinase, is constitutively associated with Ob-Rb and is rapidly activated upon leptin binding to its receptor. Activation of JAK2 allows for the recruitment and phosphorylation of STAT3 by binding to tyrosine residue 1138 of Ob-Rb ( ). Phosphorylated and activated STAT3 then dissociates from the receptor, homodimerizes in the cytoplasm and ultimately translocates to the nucleus to regulate gene transcription (293). Apart from activation of JAK2/STAT3 pathway, it has also been demonstrated that leptin induces PI3K activation in the hypothalamus that is linked to the anorexigenic action of central leptin, although the mechanisms involved remain unknown (294). Another molecular target of leptin signaling is the AMP-dependent kinase (AMPK) (295, 296). AMPK is activated as a result of an increase in the AMP/ATP ratio during low cellular energy levels. In the hypothalamus, AMPK activation increases food intake and leptin has been shown to decrease the phosphorylation of hypothalamic AMPK (297). Finally, evidence indicates that leptin can also activate MAPK-extracellular signal-regulated kinases (ERK) pathway in the hypothalamus (291, ) Leptin regulation of proglucagon The leptin receptor Ob-Rb has been demonstrated to colocalize with neurons expressing GLP-1-ir (301). Thus, it has been postulated that leptin may exert feeding regulation in part through activation of GLP-1 neurons, possibly through stimulation of proglucagon gene expression and production of GLP-1 (302, 303). In support, a recent study concluded that intact GLP-1 signaling is necessary to mediate the effect of leptin on food intake in rats, indicating that leptin and GLP-1 act in concert to control the activity of feeding centers (144, 303, 304). A number of reports suggest that leptin interacts with proglucagon-expressing neurons in mice and increases hypothalamic GLP-1 content as well as proglucagon mrna levels in brainstem neurons (302, 303, 305, 306). Leptin likely regulates the expression of proglucagon mrna through the STAT signaling pathway in the NTS neurons (307). Recently, it was found that leptin downregulates proglucagon mrna in αtc1-9 cells via STAT3 pathway at 12 h posttreatment (308). Currently, the regulation of hypothalamic proglucagon neurons by leptin remains largely unknown.

45 Hypothalamic neuronal cell models Over the past two decades, hormones and peptides involved in energy homeostasis and development of obesity have been investigated through the study of many animal models including lesion, diet and genetic models (309). However, the function of metabolic hormones, neuropeptides and neuromodulators in the developing and mature hypothalamus, and their role in the hypothalamic regulation in energy homeostasis remains far from clear, particularly in appetite control. The use of primary hypothalamic culture to study hypothalamic functions is quite challenging and disadvantageous due to their short lifespan and heterogeneous nature as primary cultures include a mixture of neuronal and glial cell populations, often with a minimal number of healthy, peptide-secreting neurons. Similarly, studies on the whole brain can be complicated due to the complex architecture of the brain and the heterogeneity of the hypothalamic neurons (310). Thus, many questions about the function of hypothalamic neuropeptidergic neurons in the central nervous system remain unanswered. Immortalized, clonal hypothalamic cell models may help to elucidate the action of neuropeptides in the hypothalamus to determine their role in energy homeostasis Generation of Immortalized hypothalamic cell lines The hypothalamic neurons have unique phenotypes and express single or multiple neuropeptides (310). These neuropeptides are controlled by endogenous or exogenous stimuli. In the past, bilateral stereotactic or electrical stimulation or lesioning were used to study the functions of specific hypothalamic neurons; however hypothalamic neuronal subtypes, and afferent or efferent neuronal extensions could be excessively stimulated or destroyed by these methods that could potentially lead to erroneous and unreliable outcomes. Due to the multitude of synaptic inputs received from other adjacent neurons, classical in vivo approach is not useful to investigate any direct action of an agent on specific hypothalamic neuronal subtypes, or on neuropeptide gene regulation, synthesis or secretion. More recently, genetically-modified rodent models with deletion of the hypothalamic neuropeptides or complete ablation of neuropeptidergic neurons have been used; however, these models are enormously challenging to generate as it is not quite feasible to knock-down exclusively hypothalamus-specific neurons. In contrast, cell lines originating from hypothalamic tumours or immortalized from primary

46 29 hypothalamic cultures, represent a simpler model that lacks the complex network of neuronal inputs, multitude of connections and signalling, and makes it feasible to investigate those areas found to be challenging to investigate in vivo. Immortalized cell lines can be homogeneous, clonal population of specific neuronal subtypes and can be maintained in a regulated environment with fewer unregulated variables present in animal models that may interfere with the direct action of the chemicals under investigation. Cell models can therefore be used to investigate regulation of specific genes or proteins, which is challenging to perform in vivo due to the numerous neuronal phenotypes present in a given hypothalamic area. To circumvent challenging issues with the contemporary experimental models, scientists have tried to generate immortalized cell models with moderate success ( ). Using the retroviral transfer of simian virus (SV) 40-T-Antigen (18, 313), our laboratory has successfully generated an array of immortalized cell models from the mouse hypothalamus (18, 313). Many of these clonal cell models express specific neuropeptides and receptors, and have been extensively used in biomedical research [reviewed in (235, )] Immortalized embryonic hypothalamic cell lines, mhypoe-xx Historically, there have been unsuccessful attempts to generate hypothalamic cell lines (317, 318). The lack of representative neuronal cell models encouraged our research laboratory to generate embryonic, clonal hypothalamic mouse cell lines (18). We utilized the retroviral SV40-T-Antigen transfer technique to immortalize primary hypothalamic cultures obtained from fetal mice on E15, E17 and E18 and generated a heterogeneous mixed population of neurons. These mixed neuronal cultures were further subcloned to obtain several single, homogeneous, clonal cell populations. These cell models are designated as mhypoe- clone number (mouse, hypothalamic, embryonic clone number ) to distinguish them from other newly created cell lines and to avoid any confusion. Each cell line, out of over 60 cell models, exhibits a distinct neuronal phenotype, although all of the cell models commonly express mature neuronal markers such as neuron-specific enolase and neurofilament, but not neuroglia-specific glial fibrillary acidic protein. The cells also possess markers of the neurosecretory machinery, such as syntaxin, a large number of neurosecretory granules and exhibit an intracellular calcium response after depolarization by potassium chloride (18). These cell lines endogenously express hormone

47 30 receptors and neuropeptides involved in many hypothalamic functions. These cell lines have been used as representative neuronal models to dissect complex cellular signaling mechanisms involved in hypothalamic control of physiological processes [reviewed in (235, ) ] Immortalized adult hypothalamic cell lines, mhypoa-xx The embryonic hypothalamic models provide a useful tool in understanding cellular biology of specific neuroendocrine neurons, however, it is imperative to known how the embryonic neurons differ from the adult neurons in their functions under normal and pathological circumstances. Historically, immortalization of adult hypothalamic neurons appeared to be impossible due to their fully differentiated, non-proliferating nature. Thus, only proliferating cells such as embryonic neurons could be immortalized using SV40-T-antigen transfer (312, 313). Recent findings indicate that neurogenesis occurs throughout the hypothalamus at low levels and is not restricted to a specific neuronal phenotype or nuclei in the CNS ( ). A significant finding by Kokoeva et al. that ciliary neurotrophic factor (CNTF) induces de novo neurogenesis in the hypothalamus enabled us to devise a novel method to immortalize primary hypothalamic culture (19). Our laboratory used CNTF to induce adult neuronal proliferation and successfully immortalized adult mouse hypothalamic neurons (235, 314, 315, 319). So far, we have generated over 50 cell lines from adult mice and labelled them as mhypoa- clone number (mouse, hypothalamic, adult- clone number ). 1.7 Rationale, Hypotheses and Specific Aims Rationale At present, the mechanisms underlying proglucagon gene expression in pancreas and intestine have been extensively studied (185), however the regulation of the hypothalamic proglucagon by insulin or leptin remains unstudied (Figure 5). It has been demonstrated that in the periphery as well as in the CNS, insulin mainly acts via classic PI3K/Akt pathway and leptin activates classic JAK2/STAT3 pathway. Currently, it remains unknown, whether both hormones utilize the same or different signaling pathways to regulate proglucagon mrna levels in the hypothalamic neurons. Therefore, in the first part of the present thesis, we investigated whether

48 31 hypothalamic proglucagon mrna transcript expression is regulated by insulin in an Aktdependent manner, and by leptin via activation of JAK2/STAT3 pathway. The exact mechanisms of action of GLP-1 and GLP-2 in the hypothalamus are not completely clear; particularly regulation of hypothalamic neuropeptides by GLP-1R and GLP-2R activation needs further attention (Figure 6). Even more importantly, this area needs to be thoroughly investigated as drugs preventing GLP-1 inactivation and agonists of GLP-1R or GLP- 2R are already approved for clinical use or undergoing clinical trials for metabolic or inflammatory disorders (325, 326). It is known that GLP-1R and GLP-2R agonists activate classic camp/pka pathway among several other pathways in the periphery and in the brainstem neurons (Figures 3 and 4) (327), however, the signaling pathways activated by GLP-1R and GLP-2R agonists in the hypothalamic neuropeptidergic neurons remain unknown. Therefore, in the second part of this thesis, we investigated whether hypothalamic neuropeptides are regulated by GLP-1R and GLP-2R agonists via activation of camp/pka pathway. We focused on the regulation of hypothalamic neurotensin and ghrelin, because, in contrast to the extensively investigated NPY- and POMC-mediated regulation of energy homeostasis, these neuropeptides have received relatively little attention, despite their potential role in the hypothalamic regulation of feeding and energy homeostasis (6, 139, , 153, 155). Many questions about the regulation and action of GLPs in the CNS remain unanswered due to the lack of representative neuronal experimental models. Recently generated embryonic and adult immortalized, clonal, hypothalamic mouse cell lines, endogenously expressing proglucagon and functional GLP-1R and GLP-2R, will serve as relevant neuronal models to investigate regulation and potential action of GLPs in the hypothalamus. Initially, using these novel murine hypothalamic cell models, insulin- and leptin-mediated regulation of the hypothalamic proglucagon-expressing neurons was studied. Subsequently, the effects of exendin-4 and h(gly 2 )GLP-2[1-33] on the hypothalamic neuropeptides that have remained elusive until now were investigated. Dissecting the mechanisms underlying proglucagon gene regulation and the effects of GLP-1R and GLP-2R activation on hypothalamic neuropeptides, such as neurotensin and ghrelin, will lead to a better understanding of the role of proglucagon gene and GLPs in hypothalamic control of many vital functions, including appetite regulation

49 32 and energy homeostasis. In the long run, these findings should facilitate the development of safe and effective drugs for the treatment of metabolic or inflammatory disorders, including obesity Hypotheses The first hypothesis is that hypothalamic proglucagon mrna transcript expression is upregulated by insulin in an Akt-dependent manner, and by leptin via activation of JAK2/STAT3 pathway in the hypothalamic proglucagon-expressing neuronal cell models (Figure 5). The second hypothesis is that long-acting GLP-1R agonist exendin-4 and GLP-2R agonist h(gly 2 )GLP-2[1-33] upregulate hypothalamic anorexigenic neurotensin and downregulate orexigenic ghrelin mrna expression in a PKA-dependent manner in the hypothalamic GLP-1R and -2R-expressing neuronal cell models (Figure 6) Specific aims The hypotheses were examined in three aims: Aim 1: To study the regulation of proglucagon mrna expression by insulin and leptin using hypothalamic proglucagon-, IR- and ObRb-expressing neuronal cell models. Western blot analysis was performed to study activation of signal-transduction pathways, real-time qrt-pcr to detect changes in mrna expression, and transfection of proglucagon promoter-reporter constructs to determine regulation of gene transcription. Aim 2: To determine potential neuropeptidergic neurons activated by exendin-4 in the mouse hypothalamus [by immunohistochemistry (IHC)] and to investigate downstream signal transduction pathways activated by exendin-4 to regulate neurotensin and ghrelin mrna expression in the hypothalamic GLP-1R-expressing neuronal cell models (by Western blot and real-time qrt-pcr analysis). Aim 3: To study the actions of h(gly 2 )GLP-2 on potential downstream neuropeptidergic neurons in vivo using mouse model (by IHC) and investigate downstream signal transduction pathways involved in regulation of neurotensin and ghrelin mrna in the hypothalamic GLP-2Rexpressing neuronal cell model (by Western blot and real-time qrt-pcr analysis).

50 Figure 5. Schematic of the proposed regulation of proglucagon-expressing (GLU) neurons by insulin and leptin in the hypothalamus. The first hypothesis is that insulin and leptin directly, via activation of specific signaling pathways, regulate proglucagon-expressing neurons located in the hypothalamic nuclei involved in energy homeostasis. This regulation itself, as well as further activation of second-order neurons via GLP-1 or GLP-2 action may lead to a decrease in food intake and an increase in energy expenditure. 33

51 Figure 6. Summary of the proposed signal transduction mechanisms triggered by GLP-1R and GLP-2R activation in the hypothalamus. The second hypothesis is that acute GLP-1R and GLP-2R activation directly regulates NPY-, POMC-, neurotensin- and ghrelin-expressing neurons located in the hypothalamic nuclei regulating energy homeostasis. This regulation itself, as well as further activation of second-order neurons leads to a decrease in food intake and an increase in energy expenditure. 34

52 35 Chapter 2 Materials and Methods

53 36 2 Materials and Methods 2.1 Cell culture and reagents Immortalized neuronal cells (mhypoa-2/10, mhypoa-2/30, mhypoe-36/1, and mhypoe-39) were grown in monolayer in Dulbecco s modified Eagle medium (DMEM, Sigma, Canada), supplemented with 4.5 mg/ml glucose, 5% fetal bovine serum (Hyclone Laboratories, USA), and 1% penicillin/streptomycin (Gibco, Canada), and maintained at 37 C in an atmosphere of 5% CO 2 (18). For the mrna expression study, cell culture medium was replaced with DMEM containing 0.5% fetal bovine serum (FBS) and 1% penicillin/streptomycin for a minimum of 12 h prior to treatments. For the protein expression study, cell culture medium was replaced with FBS-free DMEM containing 1% penicillin/streptomycin for a minimum of 12 h prior to treatments. Insulin was a generous gift from Novo Nordisk Canada Inc. (Mississauga, ON, Canada). Recombinant murine leptin was obtained from Dr. A. F. Parlow, National Hormone and Peptide Program, Torrance, CA. Exendin-4, GLP-2(1-33) and h(gly 2 )GLP-2 were purchased from American Peptide, USA. 3-isobutyl -1-methylxanthine (IBMX), actinomycin D, wortmannin, LY hydrochloride, cucurbitacin I and SD 1008 were obtained from Tocris Bioscience, USA. Forskolin and 5, 6-Dichlorobenzimidazole 1-β-D-ribofuranoside (DRB), an inhibitor of transcription, were purchased from Sigma-Aldrich, Canada. The protein kinase A (PKA) inhibitor H89 was the product of Calbiochem (EMD Biosciences), San Diego, CA or Tocris Bioscience, USA. Expression vectors pgl2 and prl-cmv were purchased from Promega (Madison, WI). The G protein beta (Gβ) antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies against total Akt, phospho-akt, total STAT3, phospho-stat3, total CREB, phospho CREB/ATF-1, c-fos (for Western blot analysis) and secondary antibodies were obtained from Cell Signaling Technology Inc. (Danvers, MA, USA). 2.2 One-step reverse transcriptase-polymerase chain reaction (RT- PCR) One-step RT-PCR was carried out to screen the cells for proglucagon, ghrelin, neurotensin, NPY, POMC, IR, ObRb, GLP-1R and GLP-2R mrna transcripts using One-Step

54 37 RT-PCR Kit (Qiagen, Mississauga, ON). The primer pairs used for the RT-PCR were designed using mouse sequences in GenBank to span introns to minimize the possibility of genomic DNA contamination. The primer pairs used are as follows: IR-F: 5 -gtgataccagagcataggag-3, IR-R: 5 -ctgttcggaacctgatgac-3 ; Ob-Rb-F: 5 -atgacgcagtgtactgctg-3, Ob-Rb-R: 5 - gtggcgagtcaagtgaacct-3, NPY-F: 5 -taggtaacaagcgaatgggg-3, NPY-R: 5 -acatggaagggtcttcaagc- 3, -actin-f: 5 -gctccggcatgtgca-3, -actin-r: 5 -aggatcttcatgaggtagt-3, Proglucagon-F: 5 - tgaagaccatttactttgtggct-3, Proglucagon-R: 5 -tggtggcaagattgtccagaat-3 (328), Ghrelin-F: 5 - agcatgctctggatggacatg-3, Ghrelin-R: 5 -aggcctgtccgtggttacttgt-3, Neurotensin-F: 5 - ataggaatgaaccttcagctg-3, Neurotensin-R: 5 -gtaggaggccctcttgagtat-3, POMC-F: 5 - atgccgagattctgctacagtcg-3, POMC-R: 5 -ttcatctccgttgccaggaaacac-3, GLP-1R-F: 5 - tttgatgactatgcctgctgg-3, GLP-1R-R: 5 -agcccatcccactggtgtt-3, GLP-2R-S: 5 - tgctggtttccatcaagcaa-3, GLP-2R-AS: 5 - atcagctgcaaggtggacaa-3, Histone-F: 5 - gcaagagtgcgccctctactg-3, Histone-R: 5 -ggcctcacttgcctcgtgcaa-3. Total RNA was isolated from the immortalized hypothalamic cells, 3T3, α-tc and GluTag cells using the guanidinium thiocyanate phenol chloroform extraction method (329). One step RT-PCR was performed using the one step RT-PCR kit. Briefly, total 200 ng of RNA was used from all samples with 1X one step RT-PCR buffer, one step enzyme mix, 0.4 mm dntps, 0.6 mm of sense primer and 0.6 mm of anti-sense primer in a final volume of 25 μl. The RT protocol used for all genes was 50 C for 30 min, 95 C for 15 min, followed by amplification at 95 C for 15 s, 60 C for 15 s and 72 C for 1 min for 40 cycles with final incubation for 7 min at 72 C. All PCR-amplified products were visualized on 2% agarose gel containing ethidium bromide under ultraviolet light with the Kodak IS2000 digital imaging centre. The PCR amplicons were verified by purification and sequencing (The Centre for Applied Genomics, Toronto, Canada). 2.3 Radioimmunoassay for camp analysis The mhypoa-2/30 and mhypoe-36/1 cells were split into 24-well plates until 80% to 90% confluent. After overnight incubation in serum-free DMEM, cells were washed twice with 1X phosphate buffered saline (PBS) and pretreated for 5 minutes with vehicle, 1 µm exendin-9-39 (a GLP-1R antagonist) or GLP-2 (3-33) (a GLP-2R antagonist) alone prior to a 10-minute

55 38 treatment with vehicle, forskolin (1 or 10 µm), exendin-4 (10 or 50 nm), GLP-2 (10 or 50 nm) or h(gly 2 )GLP-2 (10 nm) and incubated at 37 C. All drugs including vehicle were diluted in OptiMEM (Invitrogen Life Technologies, ON) containing 100 mm IBMX, a competitive nonselective phosphodiesterase inhibitor that promotes accumulation of intracellular camp. Forskolin was used as a positive control. After incubation at 37 C, 1 ml anhydrous ethanol at -20 C (final concentration, 77% ethanol) was added to the cells to terminate the reaction. The plates were allowed to sit at -20 C for 24 hours. The extracts were collected, centrifuged and the supernatant was analyzed for camp content by using camp-ria kit (Biomedical Technologies Inc., USA). camp levels were calculated relative to total amount of protein, as determined by the bicinchoninic acid (BCA) protein assay kit (Thermo Scientific, IL, USA). 2.4 Quantitative reverse transcription-polymerase chain reaction (qrt-pcr) The mhypoa-2/10 and mhypoe-39 cells were treated with vehicle, insulin (10 nm) or leptin (10 nm), and harvested at the indicated time points. The mhypoa-2/30 and mhypoe-36/1 cells were harvested at the indicated time points following treatment with vehicle, exendin-4 (10 nm) or h(gly 2 )GLP-2 (10 nm) over 24 h time period. For the regulation of gene expression study, long-acting GLP-2, h(gly 2 )GLP-2, was used. The EC 50 of h(gly 2 )GLP-2 for mouse GLP- 2R is identical to that of rat GLP-2 for both the rat GLP-2R and hglp-2r, suggesting a high degree of similarity between their biological actions in different species (328). For the experiments using inhibitors, the cells were pre-treated with either 25 µm LY294002, 1 µm wortmannin, 5 µm cucurbitacin I, 10 µm SD1008, 1 μm PKI amide, 5 μm H89, 10 µg/ml actinomycin D, 60 µm DRB or dimethyl sulfoxide (DMSO) vehicle for 45 min to 1 h, and then treated with either insulin (10 nm), leptin (10 nm), exendin-4 (10 nm) or h(gly 2 )GLP-2 (10 nm) prior to total RNA isolation at the indicated time points. Total RNA from the cells was isolated using guanidinium thiocyanate-phenol-chloroform extraction method (329), and real-time qrt-pcr was performed with 2.0 µg of total RNA using using SuperScript II and random primer as described in the Superscript II cdna Synthesis Kit

56 39 (Invitrogen). Real-time qrt-pcr was performed with SYBR green PCR master mix according to the manufacturer s instructions (Applied Biosystems Inc., Streetsville, Ontario, Canada), and run on the Applied Biosystems Prism 7000 real-time PCR machine (19). Total 200 ng of template was used in total 10 µl reaction mixture containing 0.3X SYBR green dye, 1X ROX, 1X buffer, 3 mm MgCl2, 0.2 mm dntp, and 0.5U Platinum Taq (all from Invitrogen Life Technologies, Burlington, ON). The real-time qrt-pcr conditions for all genes were as follows: 45 cycles, 15 sec at 95 C and 1 min at 65 C. The gene primer sequences were purchased from Integrated DNA Technologies (Coralville, IA) and are as follows: -actin-sybr-f: 5'-cttccccacgccatcttg-3', -actin-sybr-r: 5'-cccgttcagtcaggatcttcat-3', proglucagon-sybr-f: 5 -gaggagaaccccagatcattcc-3, proglucagon- SYBR-R: 5 -gtggcgtttgtcttcattcatc-3, ghrelin-sybr-f: 5 -ggaggagctggagatcaggtt-3, ghrelin- SYBR-R: 5 -ggcccggccatgctgct-3 (Primer Express software, Applied Biosystems). TaqMan Gene Expression assay containing neurotensin-specific primers and probe (Applied Biosystems) was used to determine neurotensin mrna expression levels. Data were represented as Ct values, defined as the threshold cycle of PCR at which amplified product was first detected, and analyzed using ABI Prism 7000 SDS software package (Applied Biosystems). Samples were run in triplicate and average Ct was analyzed. Copy number of amplified proglucagon, ghrelin or neurotensin gene was standardized to -actin using the standard curve method (ABI Prism 7000 Users Bulletin). The final fold differences in mrna expression were relative to the corresponding time-matched control. 2.5 SDS-polyacrylamide gel electrophoresis and western blot analysis The hypothalamic cells were grown to 80-90% confluency and serum-starved overnight. The mhypoa-2/10 and mhypoe-39 cells were treated with either vehicle, insulin (10 nm), leptin (10 nm) over a 60 minute time course. The mhypoa-2/30 and mhypoe-36/1 cells were treated with either vehicle, exendin-4 (10 nm) or h(gly 2 )GLP-2 (10 nm) over a 6 h time course. The cells were harvested at the indicated time points, and total protein was isolated using a 1X lysis buffer [20 mm Tris-HCl (ph 7.5), 150 mm NaCl, 1 mm Na 2 EDTA, 1 mm EGTA, 1% Triton, 2.5 mm sodium pyrophosphate, 1 mm β-glycerophosphate, 1 mm Na 3 VO 4, 1 μg/ml

57 40 leupeptin] supplemented with 1mM PMSF as described previously (141). Protein concentration was determined using the BCA protein assay kit (Thermo Scientific, IL, USA). Total protein (20 µg) was resolved on 8% SDS-PAGE gels and blotted onto Immun-Blot polyvinyl difluoride membrane (Bio-Rad, CA, USA). The blots were blocked with 5% bovine serum albumin in trisbuffered saline with 0.1% Tween-20 for 1 h, and then incubated overnight at 4ºC with rabbit polyclonal anti-mouse primary antibodies against phospho-akt (Ser473, 1:1000), phospho- STAT3 (Tyr705, 1:1000), total Akt (1:1000), total STAT3 (1:1000), Phospho-CREB (Ser133, 1:500), total CREB (1:500), c-fos (1:500) and Gβ (1:1000). Following incubation with horseradish peroxidase-labeled secondary goat anti-rabbit IgG at a 1:5000 dilution for 1 h, enhanced chemiluminescence (ECL Advance kit; GE Healthcare, USA) was added and the immunoreactive bands were visualized using Kodak IS2000 digital imaging system. The intensity of bands was quantified by densitometry analysis using Kodak 1D Image Analysis Software 3.6 (Eastman Kodak Company, Rochester, NY, USA). For the inhibitor experiments, the mhypoa-2/10 and mhypoe-39 cells were pre-treated with either 25 µm LY294002, 1 µm wortmannin, 5 µm cucurbitacin I, 10 µm SD1008, or DMSO for 1 h, and then treated with 10 nm insulin or leptin for 15 min. 2.6 Reporter gene plasmids Reporter constructs based on the pgl2 vector were constructed by transferring the inserts from the reporter gene plasmids based on the pgl3 vector containing 312 and 476 bases of the rat proglucagon promoter and human proglucagon promoter constructs containing 332 and 602 bases that have previously been described (330). Human proglucagon promoter construct containing 829 bases of promoter was generated using a PCR based strategy as previously described (330), for which the following human proglucagon gene primers were used (the numbers in the names of primers identify the location of the 5' end of each primer relative to the mrna start site): -829F: 5 -gcgggtaccgcctgtgtgtccagtcacaaaac-3, and +58R: 5 - gcaagcttagagcaagccctctttgggaac-3.

58 Transient transfections The mouse hypothalamic mhypoe-39 and mhypoe-20/2 cell lines were grown in 24-well plates in DMEM without antibiotics and supplemented with 4.5 mg/ml glucose and 5% fetal bovine serum to 70-80% confluency. Prior to the transfection, the medium was changed to medium with reduced serum (OptiMEM). The luciferase reporter gene constructs were transfected into the cell lines using Lipofectamine 2000 (Invitrogen) as per the manufacturer s protocol, with 0.8 µg of DNA per well in a 24-well plate. All luciferase assays were performed in triplicate. Using three different passages of the cell lines, at least three separate transfections per plasmid were performed. For each transfection experiment, individual plasmids were transfected in three separate wells. The cells were incubated for 4-6 h with DNA-Lipofectamine 2000 complex, and then washed and incubated in fresh media for 18 h before being treated with vehicle, insulin (10 nm), leptin (10 nm), or forskolin/ibmx (10 µm). The cells were harvested at 12 h post-treatment, and luciferase assay was performed to detect luciferase activity using Firefly & Renilla Luciferase Assay Kit (Biotum, Inc., USA) and a Lumat LB 9501 luminometer (EG&G Berthold, wellesley, MA). Protein concentrations were determined using the BCA Protein Assay Kit (Thermo Scientific, Rockford, IL). Controls for transfections included the promoterless pgl2 basic vector (Promega), and the internal control reporter prl-cmv vector (containing enhancer to express Renilla luciferase; Promega). 2.8 Animal (mouse) experiments Adult male C57BL/6 (8-10 wk old, g) mice were housed in individual polycarbonate cages in a temperature-controlled vivarium on a 12:12 hour light:dark cycle. All animals were given ad libitum access to standard rodent chow and water at all times. Using flatskull coordinates from bregma (antero-posterior mm, medio-lateral 0 mm, dorso-ventral mm) a 26-gauge stainless steel guide cannula (Plastics One, Roanoake, VA) projecting into the third cerebral ventricle was implanted into each mouse. Correct cannula placement and the patency were verified 1 wk after surgery by intracerebroventricular (i.c.v.) injection of 250 ng of angiotensin II in 0.5 μl 0.9% saline (19). Angiotensin II induces a drinking a response and only mice with positive drinking responses were included in subsequent analyses. After allowing the mice to recover for 7-14 days, they were handled daily for at least one week before beginning the

59 42 experiments. All procedures and manipulations were conducted in accordance with the protocols and guidelines approved by the University of Toronto Animal Care Committee. 2.9 Intracerebroventricular microinjections for feeding study Exendin-4 was obtained from American Peptide, USA. For the animal treatment, exendin-4 was dissolved in 0.9% saline on the day of treatment and microinjections were administered one hour prior to onset of the dark phase. The dose of exendin-4 used to induce anorexia was based on a previous report (221). Each ad libitum-fed mouse received either exendin-4 or 0.9% saline in a total volume of 2 μl by slow infusion over minutes into the third ventricle using a 30-gauge injector attached by polyethylene tubing to a 2 μl glass syringe (Hamilton, Reno, NV). The mice were returned to their home cages with free access to a premeasured amount of chow and water, and the effect of i.c.v. exendin-4 on feeding was determined (4 mice per treatment group). Changes in food and water intake were measured at 2, 4, 18 and 24 h post-treatment, and change in animal body weight was recorded at 24 h posttreatment. The degradation-resistant human analog of GLP-2, h(gly 2 )GLP-2, was purchased from American Peptide Company, USA. For the animal treatment, h(gly 2 )GLP-2 was dissolved in 0.9% saline on the day of treatment, and microinjections were administered 1 h prior to onset of the dark phase. In order to determine the effective dose of h(gly 2 )GLP-2 to study mouse hypothalamic neuronal activation, a peptide dose-response feeding study using a range of doses of h(gly 2 )GLP-2 (100 ng, 1µg, 5 µg per mouse) was conducted. Each ad libitum-fed mouse received either different doses of h(gly 2 )GLP-2 or 0.9% saline in a total volume of 2 μl by slow infusion into the third ventricle (4 mice per treatment group). Upon returning to their home cages, the mice were allowed to freely access premeasured amounts of chow and water. The effect of i.c.v. h(gly 2 )GLP-2 on feeding was determined by measuring changes in food and water intake and animal weight at 1 and 2 h post-treatment.

60 Assessment of neuronal activation by c-fos immunohistochemistry To study the effect of acute stimulation of central GLP-1R by exendin-4 and GLP-2R by h(gly 2 )GLP-2 on the activation of hypothalamic neuropeptidergic neurons, ad libitum fed mice (4 mice per treatment group) were treated with either i.c.v. exendin-4, h(gly 2 )GLP-2 or 0.9% saline as described above for the feeding study. 2h following i.c.v. injections, mice were anesthetized using CO 2 and perfused transcardially with ice-cold 0.1 M PBS followed by freshly prepared 4% paraformaldehyde solution. Brains were removed immediately, post-fixed in 4% paraformaldehyde, serially cryoprotected in 15% and 30% sucrose solution, snap-frozen in a prechilled dry ice-isopentane bath, and stored at -80 C. For subsequent immunohistochemical analysis, frozen brains were cut with a cryostat (Leica CM1510S, Leica Microsystems, Canada) in a rostral to caudal direction in the coronal plane into 20 μm sections, and serial sections were stored at -20 C in a cryoprotectant solution (19). The number of c-fos-, α-msh/pomc-, NPY-, neurotensin-, or ghrelin-ir neurons in specific hypothalamic regions were quantitatively assessed. For the immunohistological analysis, every other section at 20 μm intervals through the hypothalamus was selected throughout the hypothalamic nuclei. The sections were allocated rostral to caudal to visualize the distribution of neuropeptide- or c-fos-ir neurons on each hemisphere throughout these nuclei. Evenly spaced sections covering the region mm to mm from bregma were defined according to the Mouse Brain Atlas of Paxinos and Franklin (331). The specific primary rabbit anti-mouse antibodies and their concentration used for detection of immunoreactivity are as follows: anti-c-fos (1:25,000; Calbiochem, Canada), antineurotensin (1:5000; Immunostar, USA), anti-ghrelin (1:1000; Phoenix Pharmaceuticals, Catalog No. H ; anti-ghrelin antibody to n-octanoyl ghrelin), anti-α-msh (1:200; Phoenix Pharmaceuticals, Catalog No. H ), and anti-npy (1:1000; Phoenix Pharmaceuticals, Catalog No. H ). The detection of c-fos, α-msh, NPY, neurotensin or ghrelin immunoreactivity was performed by conventional avidin-biotin-immunoperoxidase method using biotinylated goat anti-rabbit IgG secondary antibody and diaminobenzidine (DAB) as chromogen (Vectastain ABC Elite Kit; Vector Laboratories, Canada) (301). The brain sections

61 44 were processed for immunohistochemical staining by Tyramide Signal Amplification method (TSA; PerkinElmer). The IHC for c-fos was performed with DAB to yield a brown nuclear reaction product, whereas the IHC for α-msh/pomc, NPY, neurotensin and ghrelin was performed using DAB with Metal Enhancer (cobalt chloride), yielding a more intense blue/black cytoplasmic reaction product (Sigma, Catalog No. D0426). Immunostained sections were examined under a Zeiss Axioplan 2 microscope outfitted with an AxioCam HRc camera and AXIOVISION 4.2 imaging software. For the quantification of cells, every other section throughout the ARC, PVN, DMH and VMH was taken to visualize the distribution of α-msh/pomc, NPY, neurotensin or ghrelin neurons throughout these nuclei (total 3-4 sections/mouse). For each of the ARC, PVN, DMH and VMH from both sides, an image was captured in a single plane of focus at X40 magnification and a 0.2 mm 2 box was placed in the center of the selected hypothalamic regions. Cells with brown nuclear staining were considered c-fos-ir, whereas cells with dark blue/black cytoplasmic staining were considered α- MSH/POMC-, NPY-, neurotensin- or ghrelin-positive. The immunoreactive neuronal cells from both hemispheres of 4 to 6 sections per each animal were counted in a blind manner, and in each group the mean value of the cell counts per section of an individual animal was used for statistical analysis. The results were expressed as the ratio of cells co-expressing c-fos with either α-msh/pomc, NPY, neurotensin or ghrelin to the total number of respective neuropeptide-ir cells per 0.2 mm 2 area of the ARC, PVN, DMH, VMH, PeV or the internuclear regions Experimental normalization For the analysis of the protein expression, phospho-akt expression was normalized to total Akt, phospho-stat3 expression was normalized to total STAT3, and phospho-creb and phospho-atf-1 expression was normalized to total CREB (300); only c-fos was normalized to Gβ. Primary antibody against Gβ was used as an internal loading control for all protein expression assays. For the real time and semi-quantitative qrt-pcr experiments, the sample mrna expression values were divided by the expression values of the housekeeping gene - actin to obtain sample/housekeeping ratio. These values represented relative mrna expression of the sample gene and were utilized for statistical analysis.

62 Statistical analysis Data are presented as the mean ± the standard error of the mean (SEM) from at least three independent experiments. Data were analyzed using GraphPad Prism software (GraphPad Software, Inc., USA) or SigmaPlot (Systat Software Inc., USA). Statistical analysis was performed using one-way or two-way analysis of variance (ANOVA), and statistical significance was determined by post hoc analysis using Bonferroni test or Student s t-test with P < 0.05.

63 46 Chapter 3 Regulation of the Proglucagon mrna Levels by Insulin and Leptin in Embryonic versus Adult Hypothalamic Neurons

64 47 Publication: Prasad S. Dalvi, Frederick D. Erbiceanu, David M. Irwin, and Denise D. Belsham. Direct Regulation of the Proglucagon Gene by Insulin, Leptin and camp in Embryonic versus Adult Hypothalamic Neurons. Molecular Endocrinology (2012) 26: F.D.E. was a graduate student under the supervision of D.M.I. and performed the study on the camp regulation of proglucagon gene expression included in this manuscript for his M.Sc. thesis research. All other experiments included in the manuscript and presented in this thesis were designed by P.S.D. and D.D.B. P.S.D. executed all the experiments included in this thesis, analyzed all data, designed and created all figures, wrote and revised the manuscript under the supervision of D.D.B. D.M.I. provided the human and rat proglucagon promoters and revised the manuscript. Published figures: Figure 7. Characterization of the expression profile of the proglucagon-expressing hypothalamic cell lines. Figure 8. Insulin activates Akt and leptin activates STAT3 in the hypothalamic neuronal cells. Figure 9. Insulin and leptin regulate proglucagon mrna expression in the hypothalamic neuronal cells. Figure 10. Regulation of proglucagon mrna expression by insulin via activation of the PI3K/Akt pathway. Figure 11. Regulation of proglucagon mrna expression by leptin via activation of the JAK2/STAT3 pathway. Figure 12. Insulin and leptin do not affect the transcription of proglucagon promoter constructs, but regulate mrna stability. Figure 13. In silico analysis of murine proglucagon mrna sequence for mirna binding sites and RNA-binding protein sites. Permissions were obtained to reproduce the copyrighted material.

65 48 3 Regulation of the Proglucagon mrna Levels by Insulin and Leptin in Embryonic versus Adult Hypothalamic Neurons 3.1 Abstract The proglucagon gene is expressed not only in the pancreas and intestine, but also in the hypothalamus. PGDPs have emerged as potential regulators of energy homeostasis. Whether insulin or leptin activation controls proglucagon gene expression in the hypothalamus is not known. A key reason for this has been the inaccessibility of hypothalamic proglucagonexpressing neurons and the lack of suitable neuronal cell lines. Herein, the mechanisms involved in the direct regulation of the proglucagon gene by insulin and leptin in hypothalamic cell models are described. Insulin, through an Akt-dependent manner, significantly increased proglucagon mrna levels by 70% in adult-derived mhypoa-2/10 neurons, while significantly suppressing them by 45% in embryonic-derived mhypoe-39 neurons. Leptin, via the JAK2/STAT3 pathway, caused an initial increase by 66% and 43% at 1 h followed by a decrease by 45% and 34% at 12 h in mhypoa-2/10 and mhypoe-39 cells, respectively. Transient transfection analysis determined that human or rat proglucagon 5' flanking promoter regions were not regulated by insulin and leptin in the mouse embryonic cell lines, whereas RNA stability assay demonstrated that insulin and leptin increased proglucagon mrna stability in the mouse adult cell line at 4 h and 1 h post-treatment. These findings suggest that insulin and leptin act directly, on specific hypothalamic neurons to regulate proglucagon mrna expression. Since PGDPs are potential regulators of energy homeostasis, an understanding of regulation of hypothalamic proglucagon neurons is important to further expand our knowledge of alternative feeding circuits. 3.2 Introduction Among numerous appetite-regulating neuropeptides, the PGDPs, including glucagon, GLP-1, GLP-2, and oxyntomodulin have emerged as potential regulators of feeding behavior (6). Proglucagon is encoded by a single proglucagon gene and is expressed in pancreatic α-cells, intestinal endocrine L-cells, brain stem neurons and the hypothalamus (172). PGDPs are

66 49 synthesized by post-translational processing of precursor proglucagon in cell-specific manner by PC1/3 and PC2 (174). GLP-1, GLP-2, glicentin, and oxyntomodulin are synthesized in the intestinal endocrine L-cells which are located mainly in the distal ileum and colon, whereas glucagon is predominantly produced and secreted by pancreatic α-cells (174, 332). In the CNS, proglucagon is expressed mainly in the caudal brainstem and in selective hypothalamic neurons, and the processing of proglucagon in these neurons appears to mirror that of the intestine yielding GLP-1, GLP-2, glicentin, and oxyntomodulin as major products (172, 183, 333). Receptors for both GLPs have been found in several areas of the brain that regulate appetite and energy homeostasis (332, 333). The two key regulators of food intake and energy balance, insulin and leptin, are secreted in proportion to body fat mass. Both, insulin and leptin, cross the BBB and interact with key neurons in the hypothalamus that express both IR and ObRb (43, 248). Insulin is the main metabolic hormone that regulates glucose homeostasis, and is secreted by pancreatic β-cells. Peripheral actions of insulin are anabolic as it increases energy storage by increasing fatty acid synthesis, whereas central actions are catabolic as it reduces food intake and body weight (334). Neuron-specific insulin receptor knockout mice display an obese phenotype, indicating the importance of the central actions of insulin (12). Leptin is secreted by adipocytes and was cloned from the obese (ob) gene. Knockout mice with mutations in the ob gene or leptin receptor db gene are morbidly obese, and administration of leptin to ob/ob mice normalizes body weight and neuroendocrine status (39, 267), indicating the importance of leptin in energy homeostasis. It is well established that IR and ObRb are located in the hypothalamus (43, 248, 249), and central administration of insulin or leptin potently reduces food intake and body weight (250, 282). It is known that insulin and leptin regulate feeding and energy balance through interaction with complex neural circuits comprised of appetite repressing anorexigenic and appetite stimulating orexigenic neuropeptides. It is now well established that insulin and leptin act in the hypothalamus to regulate orexigenic NPY/AgRP and anorexigenic POMC expression (36, ). In contrast to the well-studied hypothalamic NPY/AgRP and POMC regulation, the regulation of hypothalamic proglucagon by insulin and leptin remains unknown. Few studies have examined changes in the local regulation of proglucagon and production of the GLPs in the

67 50 brain, mainly due to a very small number of proglucagon-expressing neurons existing in the brain. Further, the complexity of the in vivo architecture of the hypothalamus renders these studies very challenging to perform in intact brain or whole animal models. Therefore, using embryonic- and adult-derived immortalized, clonal, hypothalamic cell models generated in our laboratory, whether insulin and leptin regulate hypothalamic proglucagon gene expression was investigated. Also the signal transduction and transcriptional mechanisms involved in any such regulation were studied. 3.3 Results Characterization of the expression profile of the hypothalamic cell lines A series of hypothalamic cell lines has recently been developed which display a variety of hypothalamic phenotypes, and a few of them were reported to express the proglucagon mrna (18, 19). To confirm these reports, RT-PCR was conducted to show that mhypoa-2/10, mhypoe-39, and mhypoe-20/2 express both proglucagon mrna (Figure 7A and B) and the insulin and leptin receptors (Figure 7C). Another cell model mhypoe36/1 does not express the proglucagon gene. Proglucagon-positive pancreatic α-tc cells and intestinal GLUTag L-cells served as positive controls, and fibroblasts 3T3 cells served as a negative control. Further, the expression of other hypothalamic neuropeptides involved in appetite regulation and hormone receptors in these cell lines were analyzed (Figure 7C). All cell models express neurotensin, IR and ObRb, whereas only mhypoe36/1 cell model expresses endogenous GLP-1R and GLP-2R. Currently, there are no hypothalamic cell models reported with endogenous proglucagon expression and receptors for insulin or leptin; therefore, based on their unique neuropeptide and receptor profile, mhypoa-2/10 and mhypoe-39 cell models were selected to study insulin- and leptin-mediated regulation of proglucagon mrna.

68 Figure 7. Characterization of the expression profile of the proglucagon-expressing hypothalamic cell lines. Expression of proglucagon mrna transcripts in pancreatic α-tc cells, intestinal GLUTag L-cells, hypothalamic neuroendocrine cell lines mhypoa-2/10, mhypoe-39, mhypoe-20/2, mhypoe-36/1 and fibroblasts 3T3 cells. (A) RT-PCR using specific primers for mouse proglucagon and -actin genes. Total RNA was isolated from the indicated cell lines and used as template for RT-PCR using the One-Step RT-PCR kit. M, markers; NTC, non-template control; +/ RT. (B) Graphical representation of relative proglucagon mrna transcripts levels quantified by densitometry. Proglucagon mrna values were normalized to -actin levels. (C) RT-PCR analysis results for the mrna expression of neuropeptides and receptors in the indicated hypothalamic cells. + indicates that the gene is expressed; indicates that the gene is weakly expressed or not expressed. 51

69 Activation of signaling pathways by insulin and leptin in the hypothalamic neuronal cells The key signaling pathway that insulin activates is the PI3K/Akt pathway. Thus, it was investigated if this signaling pathway is activated by insulin in the hypothalamic adult and embryonic cells. The neuronal cells were treated with 10 nm insulin, and activation of Akt was analyzed over 60 minutes. By Western blot analysis, it was found that insulin induced phosphorylation of Akt at Ser473 in both adult mhypoa-2/10 and embryonic mhypoe-39 cell lines during the entire period of 60 minutes post-treatment (Figure 8). The maximum significant increase in Akt phosphorylation by 179% was observed at the 5 min time point in the mhypoa- 2/10 cell model [pakt/total Akt (5 min): vehicle (0.66 ± 0.05) versus 10 nm insulin (1.83 ± 0.21), P < 0.001]. Similarly, in the mhypoe-39 cells, the maximum activation of Akt was by 390% at 5 min post-treatment [pakt/total Akt (5 min): vehicle (0.56 ± 0.06) versus 10 nm insulin (2.74 ± 0.45), P < 0.001] (Figure 8A and B). Further, leptin was found to significantly increase phosphorylation of STAT3 at Tyr705 by 60% at 15 min in the adult neuronal cells [pstat3/total STAT3 (15 min): vehicle (0.71 ± 0.06) versus 10 nm leptin (1.13 ± 0. 19), P < 0.05] (Figure 8C). Similarly, leptin significantly increased phosphorylation of STAT3 at 5 min in the embryonic cells by 42% [pstat3/total STAT3 (5 min): vehicle (0.72 ± 0.03) versus 10 nm leptin (0.99 ± 0. 08), P < 0.01] (Figure 8D). Interestingly, these findings indicate a temporal difference in the activation of the two pathways, as Akt is strongly and continuously activated by insulin over 60 min, while STAT3 is only activated by leptin for 5 to 15 minutes. Akt can be activated by its phosphorylation in a PI3K-dependent or -independent manner. Activated Akt can then activate or deactivate its downstream substrates via its kinase activity such as mammalian target of rapamycin or other signaling pathways to regulate effector genes. Activated STAT3 forms homo- or heterodimers that translocate to the cell nucleus, where they act as transcription activators and bind to STAT elements within the promoter region of downstream target genes to regulate their expression. Overall, activation of Akt and STAT3 in the neuronal cell models suggests that insulin and leptin could potentially regulate downstream signaling proteins and target genes endogenously expressed in these neuronal cells.

70 Figure 8. Insulin activates Akt and leptin activates STAT3 in the hypothalamic neuronal cells. The mhypoa-2/10 and mhypoe-39 neurons were serum starved overnight and then treated with either insulin (10 nm), leptin (10 nm) or vehicle. Protein was isolated over 60 min at the indicated time points, resolved on 10% SDS-PAGE, transferred to polyvinyl difluoride membrane, and immunoblotted with antisera for phospho-akt, total Akt, phospho-stat3, total STAT3 and Gβ (G protein β subunit). Insulin increased phosphorylation of Akt in the mhypoa- 2/10 cells (A) and mhypoe-39 cell line (B). Leptin induced phosphorylation of STAT3 in both mhypoa-2/10 (C) and mhypoe-39 (D) cell lines. Phosphorylation of Akt and STAT3 was normalized to total Akt and total STAT3, respectively. Gβ was used as a loading control. Representative Western blots are shown. All results shown in the bar graphs are expressed as mean ± SEM (n = 4-6 independent experiments; *P < 0.05, **P < 0.01, ***P < vs. vehicle control). Statistical significance was calculated by two-way ANOVA. 53

71 Regulation of proglucagon mrna transcript levels by insulin and leptin Next, direct regulation of proglucagon mrna levels by insulin and leptin was investigated. The adult and embryonic hypothalamic neuronal cells were exposed to 10 nm insulin or leptin over a 24 h time course. Using real-time qrt-pcr, it was found that in the adult mhypoa-2/10 neurons proglucagon mrna levels were significantly upregulated by 70% by insulin at 4 h post-treatment [vehicle (0.66 ± 0.07) versus 10 nm insulin (1.12 ± 0.11), P < 0.05] (Figure 9A), whereas in the embryonic mhypoe-39 neurons insulin significantly suppressed the mrna expression by 45% and 43% at 4 h and 8 h post-treatment, respectively [4 h, vehicle (1.04 ± 0.10) versus 10 nm insulin (0.57 ± 0.08), P < 0.05; 8 h, vehicle (1.00 ± 0.16) versus 10 nm insulin (0.57 ± 0.05), P < 0.05] (Figure 9B). With respect to the leptin action on proglucagon mrna expression in the adult mhypoa-2/10 cell line, it was found that leptin caused significant upregulation of mrna levels by 66% at 1 h post-treatment followed by downregulation by 45% at 12 h [1 h, vehicle (1.00 ± 0.07) versus 10 nm leptin (1.66 ± 0.22), P < 0.05; 12 h, vehicle (1.47 ± 0.15) versus 10 nm leptin (0.81 ± 0.07), P < 0.05] (Figure 9C). Leptin induced a similar biphasic effect in the mhypoe-39 cell line by initially upregulating proglucagon mrna by 43% at 1 h, followed by its suppression by 34% at 12 h post-treatment [1 h, vehicle (1.15 ± 0.10) versus 10 nm leptin (1.64 ± 0.22), P < 0.05; 12 h, vehicle (1.38 ± 0.10) versus 10 nm leptin (0.91 ± 0.05), P < 0.05] (Figure 9D). These results clearly indicate that the proglucagon mrna levels are regulated by insulin and leptin in the hypothalamic neuronal cell models Reversal of insulin-mediated regulation of proglucagon mrna levels by PI3K inhibitors To investigate involvement of Akt, a PI3K-dependent protein kinase, in the regulation of proglucagon mrna expression by insulin, pharmacological inhibitors of PI3K (LY at 25 µm and wortmannin at 1 µm concentrations) were used (32). First, the efficacy and specificity of each inhibitor was determined. The adult mhypoa-2/10 and embryonic mhypoe-39 cells were pretreated for 1 h with either of the inhibitors before exposure to 10 nm insulin. Total protein was isolated at 15 minutes after insulin treatment for the analysis using phospho-specific antibody for Akt. DMSO was used as a vehicle control, as the inhibitors were dissolved in it.

72 Figure 9. Insulin and leptin regulate proglucagon mrna levels in the hypothalamic neuronal cells. The mhypoa-2/10 (A, C) and mhypoe-39 (B, D) cells were exposed to either insulin (10 nm) (A, B), leptin (10 nm) (C, D) or vehicle over a 24 h time course; total RNA was extracted at the indicated time points and used as a template for real-time RT-PCR with primers specifically designed to amplify proglucagon mrna. Proglucagon mrna levels from both cell lines were quantified using the Ct method and normalized to the internal control (γ-actin). All results shown are relative to corresponding vehicle-treated mrna levels at each time point and are expressed as mean ± SEM (n = 4-6 independent experiments; *P < 0.05 vs. vehicle control). Statistical significance was calculated by two-way ANOVA. 55

73 Figure 10. Regulation of proglucagon mrna levels by insulin via activation of the PI3K/Akt pathway. mhypoa-2/10 (A, C) and mhypoe-39 (B, D) neuronal cells were pretreated with a PI3K inhibitor, either LY (25 μm) or wortmannin (1 μm), for 1 h followed by either vehicle (water or DMSO) or insulin (10 nm) treatment. To determine the efficacy and specificity of the inhibitors, total protein was isolated at 15 min following insulin treatment and analyzed using Western blot analysis with phospho-specific antibodies against Akt (A, B). Phosphorylation of Akt was normalized to total Akt. Gβ was used as a loading control. Representative Western blots are shown. To investigate involvement of PI3K/Akt pathway in proglucagon mrna regulation by insulin, cells were harvested for total RNA isolation at 4 h following insulin exposure, and proglucagon mrna levels were determined by real-time RT- PCR (C, D). mrna levels are shown relative to water only-treated mrna levels (set to 1.0). All results are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05). Statistical significance was calculated by two-way ANOVA. White bars = control (water or DMSO with or without PI3K inhibitor); black bars = treatment (insulin with or without PI3K inhibitor). 56

74 57 Using Western blot analysis, it was found that wortmannin and LY inhibited the phosphorylation of Akt by insulin in both cell models [mhypoa-2/10 cell line (15 min): DMSO (0.55 ± 0.13) versus DMSO + 10 nm insulin (1.90 ± 0.31), P < 0.05; LY (0.38 ± 0.18) versus LY nm insulin (0.34 ± 0.20), P > 0.05; wortmannin (0.36 ± 0.19) versus wortmannin + 10 nm insulin (0.34 ± 0.20), P > 0.05; mhypoe-39 cell line (15 min): DMSO (0.64 ± 0.27) versus DMSO + 10 nm insulin (2.16 ± 0.33), P < 0.05; LY (0.37 ± 0.24) versus LY nm insulin (0.48 ± 0.23), P > 0.05; wortmannin (0.41 ± 0.26) versus wortmannin + 10 nm insulin (0.53 ± 0.17), P > 0.05] (Figure 10A and B). The next step was to analyze whether the PI3K/Akt pathway is involved in the regulation of proglucagon by insulin. The adult mhypoa-2/10 and embryonic mhypoe-39 cells were pretreated with the inhibitors for 1 h, followed by insulin treatment. Total RNA was isolated at 4 h after insulin treatment and analyzed using real-time qrt-pcr. As compared to the vehicle treatment, insulin treatment alone significantly induced an increase in the proglucagon mrna levels in mhypoa-2/10 neuronal cells, whereas pretreatment with wortmannin or LY attenuated insulin-induced increase in proglucagon mrna expression [mhypoa-2/10 cell line (4 h): water (0.60 ± 0.13) versus 10 nm insulin (1.10 ± 0.15), P < 0.05; DMSO (0.60 ± 0.04) versus DMSO + 10 nm insulin (1.02 ± 0.12), P < 0.05; LY (1.10 ± 0.16) versus LY nm insulin (0.87 ± 0.08), P > 0.05; wortmannin (1.30 ± 0.23) versus wortmannin + 10 nm insulin (1.09 ± 0.22), P > 0.05] (Figure 10C). In the mhypoe-39 cells, pretreatment with PI3K inhibitors attenuated insulin-mediated repression of proglucagon mrna expression [mhypoe- 39 cell line (4 h): water (1.15 ± 0.14) versus 10 nm insulin (0.63 ± 0.11), P < 0.05; DMSO (1.02 ± 0.07) versus DMSO + 10 nm insulin (0.50 ± 0.08), P < 0.05; LY (1.33 ± 0.07) versus LY nm insulin (1.29 ± 0.16), P > 0.05; wortmannin (0.98 ± 0.16) versus wortmannin + 10 nm insulin (0.91 ± 0.15), P > 0.05] (Figure 10D). These findings indicated that the PI3K/Akt pathway is involved in the regulation of hypothalamic proglucagon mrna levels by insulin. It was also found that PI3K inhibitors increase basal proglucagon mrna levels in the adult cells, suggesting that the PI3K pathway may regulate basal proglucagon mrna turnover, probably by increasing the basal rate of mrna degradation (Figure 10C).

75 Reversal of leptin-mediated regulation of proglucagon mrna levels by JAK2/STAT3 inhibitors It was further investigated whether activation of STAT3 by leptin is involved in the regulation of proglucagon mrna levels, for which pharmacological inhibitors of JAK2/STAT3 (cucurbitacin I at 5 µm and SD1008 at 10 µm concentrations) were used (338, 339). In order to determine the efficacy and specificity of each inhibitor, the adult mhypoa-2/10 and embryonic mhypoe-39 cells were pretreated for 1 h with either of the inhibitors before treatment with 10 nm leptin. Total protein was isolated at 15 minutes after the leptin treatment and analyzed using a phospho-specific antibody for STAT3. As the inhibitors were dissolved in DMSO, it was used as a vehicle control. It was found that cucurbitacin I and SD1008 inhibited the phosphorylation of STAT3 by leptin in both cell models [mhypoa-2/10 cell line (15 min): DMSO (0.19 ± 0.02) versus DMSO + 10 nm leptin (0.38 ± 0.04), P < 0.05; cucurbitacin I (0.14 ± 0.06) versus cucurbitacin I + 10 nm leptin (0.13 ± 0.06), P > 0.05; SD1008 (0.11 ± 0.08) versus SD nm leptin (0.08 ± 0.05), P > 0.05; mhypoe-39 cell line (15 min): DMSO (0.17 ± 0.01) versus DMSO + 10 nm leptin (0.29 ± 0.01), P < 0.05; cucurbitacin I (0.15 ± 0.03) versus cucurbitacin I + 10 nm leptin (0.18 ± 0.01), P > 0.05; SD1008 (0.11 ± 0.05) versus SD nm leptin (0.10 ± 0.04), P > 0.05] (Figure 11A and B). Next, involvement of the JAK2/STAT3 pathway in the regulation of proglucagon mrna levels by leptin was investigated. The adult mhypoa-2/10 and embryonic mhypoe-39 cells were pre-treated with the inhibitors for 1 h, followed by leptin treatment and total RNA was isolated at 1 h and 12 h. Using real-time qrt-pcr, it was found that as compared to the vehicle treatment, leptin treatment alone significantly upregulated proglucagon mrna levels in both mhypoa-2/10 and mhypoe-39 neuronal cells, whereas pretreatment with cucurbitacin I or SD1008 attenuated the leptin-induced increase in proglucagon mrna levels [mhypoa-2/10 cell line (1 h): PBS (1.26 ± 0.10) versus 10 nm leptin (2.15 ± 0.07), P < 0.05; DMSO (0.88 ± 0.28) versus DMSO + 10 nm leptin (1.83 ± 0.25), P < 0.05; cucurbitacin I (0.47 ± 0.10) versus cucurbitacin I + 10 nm leptin (0.39 ± 0.11), P > 0.05; SD1008 (0.75 ± 0.28) versus SD nm leptin (0.67 ± 0.23), P > 0.05; mhypoe-39 cell line (1 h): PBS (1.27 ± 0.15) versus 10 nm leptin (1.88 ± 0.36), P < 0.05; DMSO (0.91 ± 0.17) versus DMSO + 10 nm leptin (1.61 ± 0.05), P < 0.05;

76 59

77 Figure 11. Regulation of proglucagon mrna levels by leptin via activation of the JAK2/STAT3 pathway. mhypoa-2/10 (A, C, E) and mhypoe-39 (B, D, F) neuronal cells were pre-treated with a JAK2/STAT3 inhibitor, either cucurbitacin I (5 μm) or SD1008 (10 μm), for 1 h followed by either vehicle (PBS or DMSO) or leptin (10 nm) treatment. To determine the efficacy and specificity of the inhibitors, total protein was isolated at 15 min following leptin treatment and analyzed using Western blot analysis with phospho-specific antibodies against STAT3 (A, B). Phosphorylation of STAT3 was normalized to total STAT3. Gβ was used as a loading control. Representative Western blots are shown. To investigate involvement of JAK2/STAT3 pathway in proglucagon mrna regulation by leptin, cells were harvested for total RNA isolation at 1 h (C, D) and 12 h (E, F) following leptin treatment, and proglucagon mrna expression was determined by real-time RT-PCR (C-F). mrna levels are shown relative to PBS only-treated mrna levels (set to 1.0). All results are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05). Statistical significance was calculated by two-way ANOVA. White bars = control (PBS or DMSO with or without JAK2/STAT3 inhibitor); black bars = treatment (leptin with or without JAK2/STAT3 inhibitor). 60

78 61 cucurbitacin I (0.57 ± 0.08) versus cucurbitacin I + 10 nm leptin (0.79 ± 0.26), P > 0.05; SD1008 (0.72 ± 0.11) versus SD nm leptin (0.80 ± 0.21), P > 0.05] (Figure 11C and D). At 12 h post-treatment, it was found that the JAK2/STAT3 inhibitors reversed the downregulation of proglucagon mrna levels caused by leptin [mhypoa-2/10 cell line (12 h): PBS (1.30 ± 0.27) versus 10 nm leptin (0.71 ± 0.11), P < 0.05; DMSO (0.88 ± 0.03) versus DMSO + 10 nm leptin (0.47 ± 0.04), P < 0.05; cucurbitacin I (1.12 ± 0.15) versus cucurbitacin I + 10 nm leptin (0.87 ± 0.08), P > 0.05; SD1008 (0.84 ± 0.14) versus SD nm leptin (1.06 ± 0.32), P > 0.05; mhypoe-39 cell line (12 h): PBS (1.27 ± 0.19) versus 10 nm leptin (0.68 ± 0.05), P < 0.05; DMSO (1.27 ± 0.13) versus DMSO + 10 nm leptin (0.84 ± 0.05), P < 0.05; cucurbitacin I (0.51 ± 0.09) versus cucurbitacin I + 10 nm leptin (0.51 ± 0.06), P > 0.05; SD1008 (1.28 ± 0.10) versus SD nm leptin (1.50 ± 0.16), P > 0.05] (Figure 11E and F). These findings indicate that the JAK2/STAT3 pathway is involved in the regulation of hypothalamic proglucagon mrna levels by leptin. Further, it was found that JAK2/STAT3 inhibitors decrease basal proglucagon mrna levels suggesting that JAK2/STAT3 pathway may regulate basal proglucagon mrna turnover, probably by decreasing the basal rate of mrna degradation (Figure 11F) Regulation of human or rat proglucagon 5 flanking promoter constructs by insulin and leptin In order to determine if insulin and leptin regulate proglucagon at the level of gene transcription, proglucagon promoter reporter plasmids were transiently transfected into the mhypoa-2/10 and mhypoe-39 cells. The human proglucagon constructs consisted of three lengths of human proglucagon 5 flanking region, 332 to +58, 602 to +58, and 829 to +58, inserted into luciferase reporter vectors. The rat proglucagon constructs consisted of two lengths of rat proglucagon 5 flanking region, 312 to +58, and 476 to +58, inserted into luciferase reporter vectors. The transfection efficiency in the adult hypothalamic cell model mhypoa-2/10 was less than 10% in contrast to more than 70% observed in the mhypoe-39 cell line. Therefore, the embryonic cell model was selected for the transient transfection analysis. 24 h after transfection, the cells were treated with 10 nm of insulin or leptin and harvested at 12 h post-treatment as the positive control forskolin/ibmx treatment was observed to increase proglucagon promoter

79 62 reporter activity at that time point. All three human proglucagon promoter constructs had basal luciferase expression fold, fold, and fold higher than the promoterless pgl2 plasmid (Figure 12A). However, the basal activity of rat proglucagon promoter constructs was much lower than the basal activity of the human proglucagon plasmids as it was only 2.74-fold and 2.66-fold higher than the promoterless pgl2 plasmid (Figure 12B). Forskolin/IBMX treatment was used as a positive control as camp activation has been demonstrated to modulate expression of proglucagon expression in islet and intestinal cells (185). It was found that forskolin/ibmx treatment affected transcription of only one plasmid that contained 829 bases of the human proglucagon promoter [human 332/+58: PBS (22.51 ± 5.42), 10 µm forskolin/ibmx (24.10 ± 0.43), P > 0.05; human 602/+58: PBS (18.15 ± 0.33), 10 µm forskolin/ibmx (23.34 ± 1.71), P > 0.05; human 829/+58: PBS (24.36 ± 1.63), 10 µm forskolin/ibmx (35.35 ± 2.49), P < 0.05] (Fig. 12A). Interestingly, insulin or leptin treatments did not affect the human plasmid transcription [human 332/+58: PBS (22.51 ± 5.42), 10 nm insulin (23.79 ± 0.49), 10 nm leptin (19.94 ± 1.43), P > 0.05 (PBS versus leptin or insulin); human 602/+58: PBS (18.15 ± 0.33), 10 nm insulin (22.34 ± 1.32), 10 nm leptin (17.80 ± 0.41), P > 0.05 (PBS versus leptin or insulin); human 829/+58: PBS (24.36 ± 1.64), 10 nm insulin (26.43 ± 2.28), 10 nm leptin (22.87 ± 1.02), P > 0.05 (PBS versus leptin or insulin)] (Figure 12A), or the rat plasmid transcription [rat 312/+58: PBS (2.74 ± 0.33), 10 nm insulin (2.20 ± 0.27), 10 nm leptin (2.42 ± 0.27), P > 0.05 (PBS versus leptin or insulin); rat 476/+58: PBS (2.66 ± 0.21), 10 nm insulin (1.97 ± 0.19), 10 nm leptin (2.06 ± 0.16), P > 0.05 (PBS versus leptin or insulin)] (Figure 12B). These results indicate that insulin and leptin do not affect the transcription of the human or rat proglucagon 5 flanking gene promoter regions in the mouse mhypoe-39 cell line Regulation of mrna stability by insulin and leptin in mhypoa-2/10 and mhypoe-39 neuronal cells Due to the lack of changes in human proglucagon promoter activity in the mouse cell lines, the regulation of proglucagon mrna stability in the presence of insulin or leptin was investigated using RNA polymerase II gene transcription inhibitors (actinomycin D at 10 µg/ml and DRB at 60 µm concentrations). The adult mhypoa-2/10 and embryonic mhypoe-39

80 63

81 Figure 12. Insulin and leptin do not affect the transcription of proglucagon promoter constructs, but regulate mrna stability. mhypoe-39 cells were transfected with proglucagon 5 flanking plasmids or promoterless control plasmid pgl2 (A, B), incubated for 24 h, then treated with either vehicle, insulin (10 nm), leptin (10 nm), or forskolin/ibmx (10 µm). Cells were harvested 12 h after treatment and a luciferase assay was performed. Data were normalized to protein concentration. To investigate regulation of mrna stability by insulin (C, F) and leptin (D, E, G, H), mhypoa-2/10 cells (C-E) and mhypoe-39 (F-H) were serum-starved overnight, pre-treated with a RNA polymerase II gene transcription inhibitor, either actinomycin D (Act D) (10 µg/ml) or DRB (60 µm), for 1 h followed by either vehicle (DMSO), insulin (10 nm) or leptin (10 nm) treatment. Total RNA was isolated at 1 h (D, G), 4 h (C, F) and 12 h (E, H) posttreatment, and proglucagon mrna expression was determined by real-time RT-PCR. mrna expression data are shown relative to DMSO with or without RNA polymerase II gene transcription inhibitor-treated mrna levels (set to 1.0). All results are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05). Statistical significance was calculated by two-way ANOVA. (C-H) White bars = control (DMSO with or without RNA polymerase II gene transcription inhibitor); black bars = treatment (insulin or leptin with or without RNA polymerase II gene transcription inhibitor). 64

82 65 cells were pre-treated with the inhibitors for 1 h, followed by insulin or leptin treatment. It was observed that actinomycin D was not tolerated well by embryonic mhypoe-39 cells, as this RNA polymerase II gene transcription inhibitor caused significant cell death at the selected dose within 12 h. Therefore, the embryonic mhypoe-39 cells were treated with DRB only, as it was well tolerated by these cells up to 24 h. The mechanisms utilized by actinomycin D and DRB to inhibit transcription are different and unlike DRB that induces slow and reversible inhibition of transcription, actinomycin D induces fast and irreversible global transcription inhibition (340), therefore, it seems that actinomycin D induces transcriptional stress response in the embryonic cells that may result in their death post-treatment (341). The mhypoa2/10 and mhypoe-39 cells were harvested for RNA isolation at 4 h after insulin treatment and at 1 h and 12 h following leptin treatment. Total RNA was isolated and analyzed using real-time qrt-pcr. As compared to the vehicle DMSO treatment, insulin treatment alone significantly induced an increase in the proglucagon mrna levels in mhypoa- 2/10 neuronal cells that was not attenuated by the pretreatment with either actinomycin D or DRB [mhypoa-2/10 cell line (4 h): DMSO (0.23 ± 0.04) versus DMSO + 10 nm insulin (0.39 ± 0.06), P < 0.05; actinomycin D (0.19 ± 0.04) versus actinomycin D + 10 nm insulin (0.37 ± 0.06), P < 0.05; DRB (1.88 ± 0.13) versus DRB + 10 nm insulin (2.95 ± 0.12), P < 0.05] (Figure 12C). Similar effects were observed at 1 h post-treatment with leptin, as the pretreatment with either RNA polymerase II gene transcription inhibitor did not suppress the leptin-induced upregulation of proglucagon mrna expression; however, the effect was significant only with DRB pre-treatment [mhypoa-2/10 cell line (1 h): DMSO (0.23 ± 0.05) versus DMSO + 10 nm leptin (0.60 ± 0.17), P < 0.05; actinomycin D (0.33 ± 0.07) versus actinomycin D + 10 nm leptin (0.53 ± 0.09), P > 0.05; DRB (1.45 ± 0.17) versus DRB + 10 nm leptin (2.86 ± 0.26), P < 0.05] (Figure 12D). In contrast, at 12 h post-treatment with leptin, the pretreatment with the RNA polymerase II gene transcription inhibitors reversed the leptin-induced suppression of proglucagon mrna expression [mhypoa-2/10 cell line (12 h): DMSO (0.55 ± 0.13) versus DMSO + 10 nm leptin (0.23 ± 0.06), P < 0.05; actinomycin D (0.24 ± 0.05) versus actinomycin D + 10 nm leptin (0.28 ± 0.09), P > 0.05; DRB (2.39 ± 0.17) versus DRB + 10 nm leptin (2.32 ± 0.21), P > 0.05] (Figure 12E). These findings suggest that leptin and insulin increase proglucagon mrna stability in the adult cell line at 1 h and 4 h post-treatment, and that the mrna stability remains unaffected at a later period following leptin treatment. The raw data

83 66 indicates that DRB, but not actinomycin D, increases basal proglucagon mrna levels in the adult cells, suggesting that DRB may inhibit transcription of proglucagon gene suppressor factors that could regulate basal proglucagon mrna levels. In the embryonic mhypoe-39 neuronal cells, as compared to the vehicle DMSO treatment, insulin treatment alone significantly decreased proglucagon mrna levels that was not reversed by the pretreatment with DRB [mhypoe-39 cell line (4 h): DMSO (0.26 ± 0.04) versus DMSO + 10 nm insulin (0.10 ± 0.03), P < 0.05; DRB (1.32 ± 0..47) versus DRB + 10 nm insulin (2.32 ± 0.51), P > 0.05] (Figure 12F). At 1 h post-treatment with leptin, the pretreatment with DRB suppressed the leptin-induced upregulation of proglucagon mrna expression [mhypoe-39 cell line (1 h): DMSO (0.49 ± 0.14) versus DMSO + 10 nm leptin (1.31 ± 0.31), P < 0.05; DRB (1.38 ± 0.36) versus DRB + 10 nm leptin (0.82 ± 0.27), P > 0.05] (Figure 12G). At 12 h post-treatment with leptin, the pretreatment with DRB reversed the leptin-induced suppression of proglucagon mrna expression [mhypoe-39 cell line (12 h): DMSO (0.15 ± 0.04) versus DMSO + 10 nm leptin (0.04 ± 0.01), P < 0.05; DRB (1.82 ± 0.13) versus DRB + 10 nm leptin (1.99 ± 0.15), P > 0.05] (Figure 12H). These findings suggest that insulin and leptin do not regulate proglucagon mrna stability in the embryonic cell line In silico analysis of murine proglucagon mrna sequence for microrna binding sites and RNA-binding protein sites Since leptin and insulin affect mrna stability in the adult hypothalamic cell model, we decided to analyze proglucagon mrna sequence for putative microrna (mirna) binding sites and RNA-binding proteins. Putative mirna binding sites on proglucagon mrna template were searched using a web tool MicroInspector that analyzed proglucagon mrna sequence for the occurrence of binding sites for known and registered mouse-specific mirnas (342) (Figure 13A and B). Next, using the RNA-binding protein database (RBPDB), proglucagon mrna sequence was scanned for putative mrna-binding protein sites that may be involved in regulation of proglucagon mrna stability (343) (Figure 13C). The in silico analysis results indicate that there are binding sites for several mirnas and RNA-binding proteins in proglucagon mrna sequence, suggesting that stability of proglucagon mrna and thereby increase or decrease in its levels can be regulated via activation or inactivation of mirnas or RNA-binding proteins.

84 67 Whether insulin or leptin utilize this mechanism to regulate proglucagon mrna stability needs to be investigated further, as it is likely that the regulation of mirnas or RNA-binding proteins by insulin and leptin may be involved in the rapid proglucagon mrna stabilization observed in the adult hypothalamic neuronal cell line (Figure 12C and D). 3.4 Discussion It is unknown if insulin and leptin have any direct action on the hypothalamic proglucagon neurons. The lack of knowledge in this area is mostly due to inaccessibility to the hypothalamic proglucagon-expressing neurons. The previous studies on the regulation of hypothalamic PGDPs were conducted using fetal rat hypothalamic primary cell cultures (183); however, these cultures are quite challenging to generate and maintain. In order to circumvent this issue, the novel cell lines generated from the embryonic and adult mouse hypothalamii that endogenously express proglucagon mrna, insulin and leptin receptors were used (18, 19). Insulin-mediated proglucagon gene regulation in the pancreas and gut has been extensively studied (185). This is the first study to demonstrate a direct role for insulin in the control of hypothalamic proglucagon gene expression. It was demonstrated that insulin upregulated proglucagon mrna expression in the adult hypothalamic cell model in contrast to its downregulating action in the embryonic cell model. This variable regulation of proglucagon gene expression can be explained by the age-dependent differences in the expression profile of PGDPs that are apparently inherent to the cell models generated and immortalized from adult and embryonic hypothalamii. The developmental studies on the hypothalamic PGDPs found that fetal rat hypothalamus predominantly expresses glicentin, oxyntomodulin and glucagon, whereas adult rat hypothalamus expresses glicentin and oxyntomodulin in greater amounts than glucagon and GLP-1; this suggests that processing of proglucagon in the hypothalamus is age-dependent and changes with development (183, 184). Thus, the differential profile of PGDPs may underlie the differential control of proglucagon gene expression by insulin in these neuronal cells and further investigations are required to confirm this hypothesis. Furthermore, tissue-specific variation in the regulation of proglucagon is quite possible, as insulin was demonstrated to inhibit islet proglucagon gene expression via regulation of gene transcription (200), whereas in GLUTag L-cells, a murine enteroendocrine cell line, insulin has been reported to stimulate

85 68 A Proglucagon mrna-binding micro-rnas C Proglucagon mrna-binding proteins Name of mirna Name of mirna mmu-mir-494* mmu-mir p mmu-mir-3070a mmu-mir p mmu-mir-5109 mmu-mir p mmu-mir-2182 mmu-mir-423-3p mmu-mir-770-5p mmu-mir-5110 mmu-mir-3960 mmu-mir-221 mmu-mir-615-5p mmu-mir p mmu-mir-185* mmu-mir-3070b-3p mmu-mir-770-3p mmu-mir-1224 mmu-mir-1956 mmu-mir-762 mmu-mir-329* mmu-mir-23b* mmu-mir-669c mmu-mir-2861 mmu-mir-27a* mmu-mir-465c-5p mmu-mir-3103* mmu-mir-574-5p mmu-mir-187 mmu-mir-1934 mmu-mir-667* mmu-mir-146b* mmu-mir p mmu-mir-346 mmu-mir-465b-5p mmu-mir-344c* mmu-mir-212-5p mmu-mir-547 mmu-mir-743a* mmu-mir-1188 mmu-mir-671-5p mmu-mir-5125 mmu-mir-705 mmu-mir-149* mmu-mir-19b-2* mmu-mir-877 mmu-mir-328* mmu-mir-1946b mmu-mir-678 mmu-mir-323-5p mmu-mir-1932 mmu-mir-23a* mmu-mir-341 mmu-mir-34a mmu-mir-3067 mmu-mir-743b-5p mmu-mir-365-1* mmu-mir p mmu-mir-1892 mmu-mir-128-2* mmu-mir-1901 mmu-mir-3090* mmu-mir-696 mmu-mir-19b-1* mmu-mir-128-1* mmu-mir-342-5p mmu-mir-1199 mmu-mir-3063* mmu-mir-3102* mmu-mir-3092 Figure 13. In silico analysis of murine proglucagon mrna sequence for mirna binding sites and RNA-binding protein sites. (A, B) Nucleotide sequences for the mouse proglucagon mrna were obtained from NCBI-GeneBank and analyzed using a web tool MicroInspector that analyzed proglucagon mrna sequence for the occurrence of binding sites for known and registered mouse-specific (mmu) mirnas. The diagram illustrates putative mirna binding sites on mouse proglucagon mrna template (A). The chart lists potential mirnas that may bind to mouse proglucagon mrna (B). (C) The mouse proglucagon mrna-binding proteins were obtained using the RNA-Binding Protein DataBase.

86 69 proglucagon gene expression (199). In the pancreatic α-cells, insulin represses proglucagon gene expression via activation of Akt (262), however, in the intestinal L-cells, it is likely that insulin utilizes the Wnt signaling pathway, but not the Akt pathway, to stimulate proglucagon promoter activity (199). Recently, it was reported that the camp activator forskolin stimulated proglucagon gene expression and hormone production in pancreatic and intestinal endocrine cells via activation of Epac, suggesting a PKA-independent regulation of proglucagon (264, 265). Similarly, although we found that Akt activation is required for insulin to variably regulate proglucagon mrna levels in the hypothalamic adult and embryonic cell models, we do not rule out involvement of other signaling pathways in the differential regulation of adult versus embryonic hypothalamic proglucagon mrna by insulin. Among multiple signalling pathways activated by insulin, the PI3K-Akt pathway remains the main focus in many studies on the CNS. Several investigations have demonstrated that insulin activates PI3K in neurons (259, 260), and that PI3K inhibitors can block the ability of insulin to regulate food intake (260). In the present study, it was confirmed that in the proglucagon-expressing selective neuronal cell models insulin activates Akt, as evidenced by an increase in the phosphorylation of Akt. Interestingly, treatment with the PI3K inhibitors reversed the insulin-induced changes in proglucagon mrna levels in both neuronal cell models. This indicates that similar to peripheral tissues, insulin mediates its action via Akt activation in the selective hypothalamic proglucagon neurons. A number of reports suggest that leptin interacts with proglucagon-expressing neurons in mice and increases hypothalamic GLP-1 content as well as proglucagon mrna levels in brainstem neurons through the STAT signaling pathway (302, 303, ). Similar to these findings, it was detected that leptin regulated proglucagon mrna in a STAT3-dependent manner in the hypothalamic neuronal cell models. It was found that leptin induced a rapid increase at 1 h in the proglucagon mrna followed by a decrease at later time points. The leptinmediated early proglucagon mrna upregulation is consistent with the action of CNTF, a cytokine that induced proglucagon mrna in mhypoa-2/10 cells while simultaneously activating the JAK2/STAT3 pathway (19). In contrast, leptin-mediated delayed downregulation is in accordance with the suppressive effect of other pro-inflammatory cytokines on proglucagon gene expression (344). The effect of leptin detected in this study is, however, difficult to interpret

87 70 in the physiological context, due to the biphasic regulation of proglucagon mrna. This action also warrants further investigation on the regulation of synthesis of PGDPs in the hypothalamic neuronal cells that seems to be challenging at present, because the quantity of PGDPs, particularly GLP-1 and GLP-2, synthesized or secreted by the hypothalamic neuronal cells appears to be quite low, most probably in picomolar or femtomolar range, and currently available detection assays, such as EIA, ELISA, RIA or Western blot analysis, are not sensitive enough for these studies. Notwithstanding, changes observed at the proglucagon mrna level are reflected at the translational level (345); and therefore, the mrna changes can be used as a reliable indicator of changes in PGDP production. Thus, it can be hypothesized that leptin may increase biosynthesis of PGDPs at early time points followed by their decrease at later time points as a compensatory response in the hypothalamic neuronal cells, and future studies to examine this possibility will be required. The level of mrna in a cell is determined by either changes in mrna stability or the direct regulation of gene transcription at the 5 regulatory region. The regulation of mrna halflife, and the cis-regulatory sequences and trans-acting factors necessary for the expression and regulation of the proglucagon gene have been determined in islet α-cells and intestinal L-cells (185, 194, 332, ). Most studies on the proglucagon promoter have focused on the rat gene, although some studies with promoters from other species, including a few studies using the human gene promoter have been conducted (191, 330, 349). To complement studies with cell lines, transgenic mice have been generated which have largely been concordant with results based on the cell lines (188, 191, 330). Yet, little is known about the sequences or factors required for the proglucagon gene expression in select neurons of the hypothalamus (332, 346, 347). A key reason for this limited knowledge has been the lack of a suitable neuronal cell line. Since the embryonic and adult hypothalamic cell lines express endogenous proglucagon gene, the regulation of proglucagon gene expression in these neuronal models was studied by transfecting available rat and human proglucagon promoter reporter genes due to the current unavailability of mouse proglucagon promoter reporter plasmids. It was found that the transfection efficiency in the adult hypothalamic cell model mhypoa-2/10 was very low when compared to more than 70% in the mhypoe-39 cell line; thus the transient transfection analysis was continued using the embryonic cell model.

88 71 Previous transfection of the reporter constructs containing 5 flanking regions of the human proglucagon gene demonstrated that a short 312-base rat proglucagon promoter region is sufficient for activation by camp, CREB and other transcription factors (193, 350, 351). By deletion analysis performed in this project, a novel area between bases -602 and -829 in human proglucagon promoter was identified, which is required for the promoter-driven induction by forskolin in the hypothalamic cell lines. Unlike forskolin, insulin and leptin did not directly affect human or rat proglucagon gene transcription at the level of the promoter in the mouse hypothalamic cell lines. There could be two possible reasons for the lack of an effect. First, there could be low sequence similarity between the human, rat and mouse proglucagon 5 regulatory regions. However, it is unlikely as human proglucagon promoter transgenic mice show that the human proglucagon promoters are expressed in mouse neurons (191, 330). Further, the proximal promoter elements are highly conserved in the rodent and human proximal proglucagon promoter sequences (191). The homology between human and rat proglucagon promoters shows that they are highly related, however, there can be several differences in the nucleotide sequence (191). The G1 promoter is highly conserved as compared to the less well conserved G2-G4 enhancer sequences. These elements have binding sites for several transcription factors including isl-1, cdx-2/3, Brn-4, pax-6, CREB and AP-1 (185, 191). Although using transcription factor search programs, STAT3 binding sequences within 4000 bp upstream of the transcription initiation site were not found, further search and investigation into distal promoter regions is required to detect STAT binding sequences and also other cis-elements for downstream transcription factors activated by insulin and leptin. A 300 bp sequence that contains G2, G3 and G4 enhancer elements of rat proglucagon promoter is sufficient for proglucagon gene expression in the mouse pancreatic islet cells, therefore, two reporter plasmids containing -476 to +58 and -312 to +58 from rat proglucagon promoter were generated (191, 330). Previously, these promoter constructs were used to study proglucagon gene expression in hamster islet cell line InR1-G9 and mouse islet cell line αtc-1 (330). As is evident from Figure 6A, B, transfection of embryonic hypothalamic cells with these two rat proglucagon promoter constructs containing -476 to +58 and -312 to +58 bp sequences from transcription initiation site resulted in a much lower basal activity of these plasmids than the basal activity of the human proglucagon plasmids, and no significant changes were detected in the activity of rat proglucagon promoter plasmids by insulin or leptin treatment. Again, the

89 72 low basal transcription activity of rat promoters in the mouse cell line could due to low sequence similarity between the rat and mouse proglucagon 5 regulatory regions. Second, the 5 flanking sequences used for the proglucagon constructs were only 312 and 476 bp long, and thus cisacting responsive elements in the neuronal proglucagon gene promoter could be located further upstream of these regions. Previously, it was found that unlike the rat proximal proglucagon promoter elements, the equivalent human proglucagon proximal promoter sequences containing G2, G3 and G4 enhancer elements did not induce transcriptional activity in transfected mouse islet or intestinal cells; therefore, plasmids containing additional elements up to -602 bp, -829 bp or larger segments of human proglucagon gene promoter were generated (191). These larger human reporter plasmids supported transcriptional activity in the mouse pancreatic islet or intestinal cells, and also in selected neurons of transgenic mice that contained 602 bases of human proglucagon 5 flanking sequence (191, 330). Thus, the use of non-species specific promoters can confound the analysis, and therefore, the lack of insulin or leptin transcriptional effects using the rat or human promoter in the mouse hypothalamic cell lines does not conclusively exclude transcriptional regulation of mouse hypothalamic proglucagon gene. Generation of mouse proglucagon promoter plasmids is required to conduct further studies. Because of the lack of changes in the activity of proglucagon reporter gene constructs, the stability of the proglucagon mrna in the presence of insulin and leptin was analyzed. Using RNA polymerase II gene transcription inhibitors actinomycin D and DRB in the hypothalamic neuronal cells, it was detected that the changes in mrna expression at early time points induced by both insulin and leptin were due to increased mrna stability, whereas leptin-induced changes in the mrna expression at the later time point do not seem to be controlled by changes in the rate of mrna decay in the adult neuronal cells. The decrease in mrna expression caused by leptin treatment at 12 h may occur due to suppressed transcription, for which further investigation of distal promoter regions for putative cis- and trans-elements is required. Mechanisms involved in proglucagon mrna stability in islet α-cells and intestinal L-cells are known to some extent (194, 345), however, currently, no information is available upon regulatory mechanisms for proglucagon mrna stability in the hypothalamus. Regulation of gene expression at the post-transcriptional level by mrna transcript stability is widespread in eukaryotes (352). Among several factors that regulate mrna stability,

90 73 mirna and mrna-binding proteins play an important role is altered turnover of mrna. For example, for the rapid decay of TNF-α mrna, which contains an adenylate-uridylate-rich elements. mir16 is required alongwith RNA-binding proteins in HeLa cells (353). mirnas are small ( 22 nucleotide), noncoding RNAs that act as negative regulators of gene expression at a post-transcriptional level (354). They regulate either the degradation of their specific target mrnas or the inhibition of mrna translation (355, 356). In this thesis, as it was found that insulin and leptin affected hypothalamic proglucagon mrna stability, putative mirna binding sites on proglucagon mrna template were searched. It was detected that there are binding sites for several mirnas including mirna128, a brain-enriched microrna, that downregulates doublecortin gene expression by inducing degradation of doublecortin mrna in neuroblastoma cells (357), and mirna494 that downregulates PTEN in the bronchial epithelial cells (358). Because insulin and leptin have been shown to regulate mirna expression (359, 360), it is possible that they may regulate proglucagon mrna stability as well via mirna activation or suppression. Next, using the RBPDB program, the proglucagon mrna sequence was scanned for putative mrna-binding protein sites that may also regulate proglucagon mrna stability. Several RNA-binding proteins were detected, including nervous system-specific RNA-binding protein ELAVL2 or QKI that have been found to regulate gene expression in the nervous system ( ). Further studies to determine role of insulin, leptin and RNA-binding proteins in hypothalamic proglucagon mrna regulation are warranted. Taken together, these experiments indicate that insulin and leptin can act directly upon proglucagon neuronal cells to regulate proglucagon mrna levels. Insulin, through an Aktdependent mechanism, and leptin, through JAK2/STAT3 activation, regulate proglucagon mrna levels, although transfections with human proglucagon promoter reporter gene constructs indicate that insulin or leptin may not act directly at the level of transcription, but may instead act to increase the stability of proglucagon mrna. The PGDPs are key regulators of feeding behavior, thus a better understanding of the mechanisms through which insulin and leptin regulate hypothalamic proglucagon neurons is important to further expand and challenge our current knowledge of feeding circuits.

91 74 Chapter 4 Glucagon-like Peptide-1 Receptor Agonist, Exendin-4, Regulates Feeding-associated Neuropeptides in Hypothalamic Neurons in vivo and in vitro

92 75 Publication: Prasad S. Dalvi, Anaies Nazarians-Armavil, Matthew J. Purser, and Denise D. Belsham. Glucagon-like peptide-1 receptor agonist, Exendin-4, regulates feeding-associated neuropeptides in hypothalamic neurons in vivo and in vitro. Endocrinology (2012) 153: A.N.A. was a graduate student in the D.D.B. lab and assisted in the treatment of cells and RNA isolation in the inhibitor study included in the publication and presented in this thesis. M.J.P. was a summer research and work-study student in the D.D.B. lab working directly under supervision of P.S.D. and performed the immunohistological study on the hypothalamic sections of ghrelin knockout mice included in the original publication, but not presented in this thesis. All other experiments included in the publication and presented in this thesis were designed by P.S.D. and D.D.B. P.S.D. executed all the experiments in this thesis, analyzed all data, designed and created all figures, wrote and revised the manuscript under the supervision of D.D.B. Published figures: Figure 14. Intracerebroventricular injection of exendin-4 was effective to reduce food and water intake, and body weight in mice. Figure 15. Exendin-4 activates hypothalamic neurons. Figure 16. Acute exendin-4 treatment increases the number of hypothalamic neurons coexpressing c-fos-ir with α-msh-, NPY-, neurotensin- or ghrelin-ir. Figure 17. Graphical representation showing the number of neurons expressing c-fos and neuropeptide-ir in the ARC, VMH, DMH, PVN, LH, PeV, and the internuclear space between the VMH and DMH of the saline- or exendin-4-treated mouse hypothalamus. Figure 18. Expression profile of GLP-1R and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines. Figure 19. Exendin-4 induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic neuronal cell lines. Figure 20. Regulation of neurotensin and ghrelin mrna expression by exendin-4 in the mhypoa-2/30 and mhypoe36/1 neuronal cell lines. Figure 21. Regulation of neurotensin and ghrelin mrna expression via activation of the PKA pathway. Permissions were obtained to reproduce the copyrighted material.

93 76 4 Glucagon-like Peptide-1 Receptor Agonist, Exendin-4, Regulates Feeding-associated Neuropeptides in Hypothalamic Neurons in vivo and in vitro 4.1 Abstract Exendin-4, a long-acting GLP-1R agonist, is a potential regulator of feeding behavior through its ability to inhibit gastric emptying, reduce food intake and induce satiety. GLP-1R activation by exendin-4 induces anorexia; however, the specific populations of neuropeptidergic neurons activated by exendin-4 within the hypothalamus, the central regulator of energy homeostasis, remain unclear. This study determines whether exendin-4 regulates hypothalamic neuropeptide expression and explores the signaling mechanisms involved. The distribution and quantity of exendin-4-induced c-fos immunoreactivity were evaluated to determine activation of α-msh/pomc, NPY, neurotensin and ghrelin neurons in hypothalamic nuclei during exendin-4- induced anorexia in mice. Additionally, exendin-4 action on neurotensin and ghrelin transcript regulation was examined in immortalized hypothalamic neurons. With anorexia induced by intracerebroventricular exendin-4, α-msh/pomc and NPY neurons were activated in the ARC, with simultaneous activation of neurotensin-expressing neurons in the PVN, and ghrelinexpressing neurons in the ARC, PVN and periventricular hypothalamus, suggesting that neurons in one or more of these areas mediate the anorexic action of exendin-4. In the hypothalamic neuronal cell models, exendin-4 increased camp, CREB/ATF-1 and c-fos activation, and via a PKA-dependent mechanism regulated neurotensin and ghrelin mrna expression, indicating that these neuropeptides may serve as downstream mediators of exendin-4 action. These findings provide a previously unrecognized link between central GLP-1R activation by exendin-4 and the regulation of hypothalamic neurotensin and ghrelin. Further understanding of this central GLP- 1R activation may lead to safe and effective therapeutics for the treatment of metabolic disorders.

94 Introduction The hypothalamus integrates peripheral and central signals to regulate appetite and energy homeostasis. Recently, GLP-1 has emerged as a potential negative regulator of feeding behavior (6). The GLP-1R is widely expressed in the hypothalamus (217). GLP-1 plays multiple roles in metabolic homeostasis due to its glucoregulatory function and inhibitory action on food and water intake (26, 364). Endogenous GLP-1 has a short plasma half-life because of its rapid degradation by DPP-4, therefore longer-acting GLP-1 analogs, GLP-1R agonists or DPP-4 inhibitors are under investigation (365). Exendin-4 is a long-acting GLP-1R agonist peptide that was originally isolated from the salivary gland of the Gila monster, Heloderma suspectum. Currently, exendin-4 and several DPP-4 inhibitors that increase the concentration of intact endogenous GLP-1 by preventing its degradation are used to treat type 2 diabetes. Exendin-4 has been shown to induce weight loss in overweight patients, suggesting that it can regulate food intake and body weight by either directly interacting with peripheral GLP-1R-expressing vagal afferents to affect gastro-intestinal motility, or readily crossing the BBB to directly interact with hypothalamic appetite-regulating centers (219, 366, 367). However, other clinical trials indicate that DPP-4 inhibitors are weight neutral (368). These findings suggest that the hypothalamic GLP-1R activation mediated by exendin-4 may differ from that mediated by increased concentrations of intact endogenous GLP-1 achieved through DPP-4 inhibition. That is, while both GLP-1 and exendin-4 act on the hypothalamus to modulate feeding behavior, key differences exist between the anorectic actions of endogenous GLP-1 and exendin-4 (369). Considering the widespread clinical use of exendin-4, further research is warranted to study the mechanisms underlying the anorectic actions of exendin-4 within the hypothalamus (7). The hypothalamus is subdivided into several nuclei consisting of groups of neurons with specific functions. The ARC, PVN, VMH, DMH, LH, and PeV play an important role in the regulation of energy intake and expenditure by integrating central and peripheral orexigenic and anorexigenic signals. It is well established that there are two distinct neuronal populations in the ARC: NPY and α-msh/pomc neurons. Additionally, anorexigenic neurotensin is expressed within the ARC (105, 106, 156), PVN, DMH and LH (56), whereas orexigenic ghrelin neurons are present in the ARC, PVN, the internuclear spaces between hypothalamic nuclei, the perifornical region, and the ependymal layer of the third ventricle (105).

95 78 Exendin-4 has been shown to activate neurons in the PVN via GLP-1R-dependent networks (221). Compelling evidence demonstrates that the GLP-1Rs located in the PVN, but not in the ARC, reduce food intake (131), implying that GLP-1Rs in specific hypothalamic regions have discrete effects on appetite regulation. Furthermore, the finding that GLP-1Rs expressed on the ARC POMC neurons do not induce anorexia raises the possibility that exendin- 4 may stimulate or inhibit other, yet to be identified downstream effectors within the hypothalamus to mediate its anorectic effect (131), such as neurotensin and ghrelin neurons, localized to similar regions as the GLP-1R (216). Importantly, neurotensin and ghrelin have been shown to play a significant role in feeding and energy homeostasis (6). Whether exendin-4 exerts its effect directly on neurotensin or ghrelin neurons via GLP-1R activation remains unknown. The GLP-1R is a G-protein-coupled receptor acting through adenylate cyclase, camp and PKA activation (370). Recently, exendin-4 was demonstrated to suppress appetite and reduce body weight by PKA activation in brainstem GLP-1R-expressing neurons (327). However, in the hypothalamus the exact neurons activated by exendin-4 to induce anorexia are unknown, and whether appetite-regulating neuropeptide gene expression is altered within GLP- 1R-expressing hypothalamic neurons remains to be determined. The present study proposes that hypothalamic GLP-1R activation by exendin-4 regulates feeding-related neurons, and modulates neuropeptide expression via camp/pka activation. Using in vivo and in vitro models, the hypothalamic neuronal activation by exendin-4 was mapped, and the modulation of neurotensin and ghrelin gene expression following exendin-4 treatment and the signaling mechanisms involved were studied. 4.3 Results Effect of i.c.v. exendin-4 on food and water intake, and animal weight In order to test the efficacy of the i.c.v. injections, a short-term analysis of exendin-4 effects on food and water intake was performed. The dose of exendin-4 (100 ng/mouse) used to induce anorexia was based on a previous report (221). It was found that i.c.v. exendin-4 significantly reduced food in ad libitum-fed wild type mice, and this effect lasted for at least 24 h

96 Figure 14. Intracerebroventricular (i.c.v.) injection of exendin-4 was effective to reduce food and water intake, and body weight in mice. To determine the efficacy of i.c.v. exendin-4 on food intake, ad libitum-fed mice received injection of 100 ng of exendin-4 dissolved in 2 μl of 0.9% normal saline 1 h before the onset of the dark cycle. Mice were returned to their home cages with pre-weighed amount of chow and water. Change in food (A) and water (B) intake was measured at 2, 4, 8, and 24 h post-injection, and the body weight change (C) was recorded over 24 h. 0.9% normal saline solution was used as control treatment. All results are expressed as mean ± SEM (n = 3-4 mice/group); *P < 0.01, **P < 0.001). 79

97 80 [cumulative food intake (g): 1 h, saline (0.67 ± 0.14) versus exendin-4 (0.13 ± 0.03), P < 0.01; 4 h, saline (1.67 ± 0.07) versus exendin-4 (0.50 ± 0.05), P < 0.01; 8 h, saline (4.63 ± 0.17) versus exendin-4 (1.23 ± 0.07), P < 0.001; 24 h, saline (5.00 ± 0.25) versus exendin-4 (1.73 ± 0.11), P < 0.001] (Figure 14A). Similarly, water intake was found to be suppressed over 24 h time period [cumulative water intake (ml): 1 h, saline (2.67 ± 0.27) versus exendin-4 (0.08 ± 0.07), P < 0.001; 4 h, saline (4.00 ± 0.47) versus exendin-4 (0.08 ± 0.07), P < 0.001; 8 h, saline (8.67 ± 0.27) versus exendin-4 (0.33 ± 0.27), P < 0.001; 24 h, saline (10.33 ± 0.27) versus exendin-4 (0.67 ± 0.54), P < 0.001] (Figure 14B). Furthermore, significant weight loss at 24 h in exendin-4- treated mice compared with the saline-treated control mice was observed [change in body weight (g): 24 h, saline 1.00 ± 0.17 versus exendin ± 0.79, P < 0.01] (Figure 14C). The dose of exendin-4 used to induce anorexia was based on a previous report (221). Using an equivalent central dose, Baggio et al. showed that the anorectic action of central exendin-4 was not simply a nonspecific response, as anorexia was absent in GLP-1R knock-out mice, which lack functional GLP-1R (371). This suggests that the anorexia induced by central exendin-4 may be mediated via activation of GLP-1Rs expressed in hypothalamic appetite-regulating nuclei located in the vicinity of the third ventricle. Nevertheless, as peptides delivered into the third cerebral ventricle can access GLP-1Rs not only in the hypothalamus but also in the brainstem and the central nucleus of the amygdala, any indirect activation of hypothalamic nuclei through activation of these extra-hypothalamic regions that can also induce suppression of food and water intake by their own mechanisms cannot be ruled out (209) Effect of i.c.v. exendin-4 on activation of hypothalamic nuclei and neuropeptidergic neurons As assessed by IHC for c-fos, a protein encoded by the immediate-early gene c-fos, a distinctive pattern of neuronal activation was noted after the central injection of exendin-4 into the third cerebral ventricle compared with the saline controls (Figure 15B, C, E, and F). Significant increases in the number of c-fos-positive neurons were detected in the hypothalamic ARC (increase by 315%), DMH (increase by 214%), PVN (increase by 467%), and PeV (increase by 1927%), but not in the LH and VMH [c-fos-positive nuclei: ARC, saline (12.77 ± 1.55) versus exendin-4 (53.00 ± 5.35), P < 0.001; DMH, saline (13.62 ± 1.39) versus exendin-4

98 Figure 15. Exendin-4 activates hypothalamic neurons. Immunohistochemistry was performed to assess neuronal activation by c-fos-immunoreactivity (ir) in wild-type mice treated with saline or exendin-4. A-F: Representative photomicrographs showing expression of c-fos-ir in the hypothalamic ARC (A-C), VMH (A-C), DMH (A-C), PVN (D-F), LH (D-F) and PeV (A-F) regions in coronal sections of the mouse hypothalamii (as indicated on the images). Scale bar: 1 mm. Inset in each image represents a higher magnification of the boxed area. Scale bar: 100 μm. A and D represent negative control images for the c-fos antibody. DAB staining: nuclear brown (c-fos). 3V, third ventricle. G: Bar graph showing the number of c-fos-ir neurons in the hypothalamic regions at 2 h post-treatment. Data in the bar graph are expressed as mean ± SEM (n = 3-4 animals/group; *P < 0.05, **P < 0.01, ***P < 0.001). 81

99 82 (42.78 ± 3.93), P < 0.001; PVN, saline (12.06 ± 1.97) versus exendin-4 (68.37 ± 13.37), P = 0.002; PeV, saline (2.37 ± 0.42) versus exendin-4 (48.03 ± 11.52), P = 0.017; LH, saline (4.25 ± 1.48) versus exendin-4 (9.52 ± 2.30), P = 0.082; VMH, saline (1.71 ± 0.45) versus exendin-4 (3.08 ± 0.94), P = 0.227] (Figure 15G). Activation of these hypothalamic regions was expected, as these regions widely express GLP-1Rs (216). In this study, the potential activation of other CNS areas known to express the GLP-1R, such as certain brainstem nuclei and the central nucleus of the amygdala, was not studied as the main aim of this study was to investigate the activation of hypothalamic nuclei involved in feeding regulation. The next goal was to determine the neuropeptidergic neurons activated in these regions by performing double-staining IHC for c-fos and neuropeptide co-expression (Figs. 16 and 17). It was found that exendin-4 significantly activated α-msh/pomc-ir neurons (by 129%) in the hypothalamic ARC [saline (12.51 ± 1.98) versus exendin-4 (28.69 ± 3.33), P = 0.002] (Figure 17B), and also robustly activated neurotensin neurons (by 488%) in the PVN [saline (6.62 ± 0.98) versus exendin-4 (38.92 ± 1.42), P < 0.001] (Figure 17F). Both α-msh/pomc and neurotensin neurons, being anorexigenic, are implicated in the inhibition of appetite. Surprisingly, it was found that orexigenic NPY-ir neurons were significantly increased in the ARC [saline (10.75 ± 2.35) versus exendin-4 (42.45 ± 5.91), P = 0.004] (Figure 17D), whereas ghrelin-ir neurons were activated in the ARC, PVN and PeV of exendin-4-treated mice [ARC, saline (6.15 ± 2.14) versus exendin-4 (22.48 ± 1.16), P = 0.005; PVN, saline (7.16 ± 0.90) versus exendin-4 (22.52 ± 2.67), P = 0.011; PeV, saline (3.33 ± 0.53) versus exendin-4 (34.11 ± 8.51), P = 0.042] (Figure 17H). Importantly, no changes were noticed in the number of neurons expressing only neuropeptide-immunoreactivity in the hypothalamic regions of saline- or exendin-4-treated animals (Figure 17A, C, E, and G). Since there is still some controversy regarding the synthesis of ghrelin in the hypothalamus, the specificity of the ghrelin antibody was confirmed by using the same antibody in ghrelin knockout mouse in another control experiment (372). The antibody did not detect ghrelin-ir staining in the PVN of the ghrelin knockout sections versus the litter-matched controls (372). These findings suggest that the induced anorexia may be a function of complex interactions occurring between appetiteregulating anorexigenic and orexigenic neurons in the hypothalamus.

100 Figure 16. Acute exendin-4 treatment increases the number of hypothalamic neurons coexpressing c-fos-immunoreactivity (ir) with α-msh-, NPY-, neurotensin- or ghrelin-ir. A- P: Bright-field photomicrographs showing neurons that co-express c-fos-ir (brown nuclei) and neuropeptide-ir (blue-black cytoplasm) in the coronal sections of the hypothalamic regions from wild-type mice at 2 h after intracerebroventricular administration of saline (A, C, E, G, I, K, M, O) or exendin-4 (B, D, F, H, J, L, N, P) (n = 3-4 animals/group). A and B: Co-expression of c- Fos-ir with α-msh-ir in the ARC. C and D: Co-expression of c-fos-ir with NPY-ir in the ARC. E-H: Co-expression of c-fos-ir with neurotensin-ir in the ARC (E, F) and PVN (G, H). I-P: Coexpression of c-fos-ir with ghrelin-ir in the ARC (I, J), PVN (K, L), PeV (M, N), and the internuclear space between DMH and VMH (O, P). Insets in each image represent a higher magnification of the adjacent boxed areas. Black arrowheads represent neurons expressing only nuclear c-fos-ir, yellow arrowheads represent neurons expressing only cytoplasmic perinuclear neuropeptide-ir, and red arrowheads represent double-labeled neurons with co-expression of c- Fos-ir and neuropeptide-ir. Scale bars: 100 μm (A-P) and 10 μm (Insets). 83

101 Figure 17. Graphical representation showing the number of neurons expressing c-fos and neuropeptide-immunoreactivity (ir) in the ARC, VMH, DMH, PVN, LH, PeV, and the internuclear space between the VMH and DMH of the saline- or exendin-4-treated mouse hypothalamus. Double-labeled immunohistochemistry for c-fos-ir and α-msh-ir (B), c-fos-ir and NPY-ir (D), c-fos-ir and neurotensin-ir (F), or c-fos-ir and ghrelin-ir (H) indicates that intracerebroventricular exendin-4 activates hypothalamic neuropeptidergic neurons. Note that there was no change in the number of neurons expressing only α-msh-ir (A), NPY-ir (C), Neurotensin-ir (E) or ghrelin-ir (G) in the hypothalamic regions of saline- or exendin-4-treated animals. Data are represented as mean ± SEM (n = 3-4 mice/group); *P < 0.05, **P < 0.01, ***P < vs. saline treatment. Statistical significance was calculated by two-tailed, unpaired t-test. 84

102 Expression of GLP-1R and appetite-regulating neuropeptides in adult mhypoa-2/30 and embryonic mhypoe-36/1 neurons Based on the in vivo findings that exendin-4 activates neurotensin and ghrelin neurons in the hypothalamus, direct regulation of hypothalamic neurotensin and ghrelin mrna expression by exendin-4 was investigated using hypothalamic cell models. Our lab has previously reported the generation of embryonic and adult mouse hypothalamic cell lines (18, 19). Using RT-PCR, the presence of GLP-1R mrna in both adult mhypoa-2/30 and embryonic mhypoe-36/1 neuronal cell models was confirmed (Figure 18A). GLP-1R-positive mouse jejunal tissue served as a positive control. Further, the expression of hypothalamic neuropeptides involved in appetite regulation in these cell lines was analyzed. Both cell models express ghrelin and neurotensin, and the adult mhypoa-2/30 cells express POMC as well (Figure 18B). Currently, there are no hypothalamic cell models reported with a functional endogenous GLP-1R. Therefore, we sought to determine if the GLP-1R was functionally active in the adult and embryonic neurons by measuring GLP-1R-mediated camp activation following exendin-4 treatment. Forskolin was used as a positive control. Using camp-ria, it was found that exendin-4 increased camp levels in mhypoa-2/30 neuronal cell line [camp content (pmol/µg protein): vehicle, 0.49 ± 0.08; 1 µm forskolin, 2.35 ± 0.17; 10 nm exendin-4, 0.79 ± 0.16; 50 nm exendin-4, 0.99 ± 0.07; forskolin versus saline, P < 0.01; exendin-4 (10 nm and 50 nm) versus saline, P < 0.05] (Figure 18C). Also in mhypoe-36/1 neuronal cells exendin-4 significantly increased camp content compared to vehicle control [camp content (pmol/µg protein): vehicle (0.54 ± 0.03), 1 µm forskolin (2.02 ± 0.39), 10 nm exendin-4 (0. 90 ± 0.06), 50 nm exendin-4 (0.81 ± 0.06); forskolin versus saline, P < 0.01; exendin-4 (10 nm and 50 nm) versus saline, P < 0.05, (Figure 18D)]. These data indicate that the GLP-1Rs in these cells are responsive to exendin-4. Further, using exendin-9-39, a GLP-1R antagonist, we studied whether exendin-4 directly activates GLP- 1R to stimulate camp production [16]. It was found that 1 µm exendin-9-39 alone did not stimulate camp production in both cell models, but completely attenuated the stimulatory effect of exendin-4 on camp levels, suggesting that exendin-4 stimulates camp production via GLP- 1R activation in both cell lines [camp content (pmol/µg protein): mhypoa-2/30, vehicle, 0.49 ± 0.08; 10 µm forskolin, ± 0.79; 10 nm exendin-4, 0.89 ± 0.14; 1 µm exendin-9-39, 0.41 ± 0.02; 1 µm exendin nm exendin-4, 0.55 ± 0.02; mhypoe-36/1, vehicle, 0.38 ± 0.13;

103 86

104 Figure 18. Expression profile of GLP-1 receptor (GLP-1R) and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines. A: Expression of GLP-1R mrna transcript in jejunum and the mhypoa-2/30 and mhypoe-36/1 neuronal cell lines by RT-PCR using specific primers for mouse GLP-1R gene. Total RNA, isolated from mouse jejunum and the indicated cell lines, was used as template for RT-PCR using One-Step RT-PCR kit. Jejunal RNA was used as a positive control for GLP-1R expression. M, markers; NTC, non-template control. B: RT-PCR analysis results for the mrna expression of neuropeptides in mhypoa-2/30 and mhypoe-36/1 cells. + indicates that the gene is expressed; - indicates that the gene is not expressed. C and D: Expression of functional GLP-1R by camp-ria. E and F: Exendin-4 (Ex- 4) stimulates camp production via the GLP-1R activation. camp is activated by Ex-4 in mhypoa-2/30 (C, E) and mhypoe-36/1 (D, F) neuronal cells. The cells were pretreated for 5 minutes with 1 µm exendin (9-39) [Ex(9-39)] or vehicle alone prior to a 10-minute treatment with vehicle, forskolin (1 or 10 µm) or Ex-4 (10 or 50 nm). The amount of intracellular camp was determined in triplicate by RIA. Forskolin was used as a positive control. All results are expressed as mean ± SEM (n = 4-6; *P < 0.05, **P < 0.01 vs. vehicle control). 87

105 88 10 µm forskolin, 8.69 ± 0.71; 10 nm exendin-4, 0.89 ± 0.14; 1 µm exendin-9-39, 0.35 ± 0.01; 1 µm exendin nm exendin-4, 0.41 ± 0.01; forskolin versus saline, P < 0.01; exendin-4 (10 nm) versus vehicle, P < 0.05] (Fig 18E and F) Activation of c-fos and CREB/ATF-1 by exendin-4 in the hypothalamic neuronal cells The key signaling pathway that exendin-4 activates is the camp/pka pathway. Therefore, it was next determined if the downstream effectors of this pathway, such as CREB/ATF-1 and c-fos, are activated by exendin-4 in the hypothalamic adult and embryonic cells. The neuronal cells were treated with 10 nm exendin-4, and activation of CREB/ATF-1 and c-fos was analyzed over 6 h. By Western blot analysis, exendin-4 significantly induced c-fos activation in the adult mhypoa-2/30 cell line from 30 min through 4 h post-treatment [c-fos/gβ (30 min): vehicle (0.73 ± 0.10) versus 10 nm exendin-4 (1.14 ± 0.05), P < 0.01] (Figure 19A). Further, exendin-4 significantly increased phosphorylation of CREB at Ser 133 at 15 and 30 min in the adult neuronal cells, by 53% and 58%, respectively, [pcreb/total CREB: 15 min, vehicle (0.74 ± 0.08) versus 10 nm exendin-4 (1.13 ± 0.08), P < 0.01; 30 min, vehicle (0.83 ± 0.05) versus 10 nm exendin-4 (1.31 ± 0.22), P < 0.01] (Figure 19B). Similarly, exendin-4 significantly increased phosphorylation of CREB at 5min in the embryonic cells by 42%, [pcreb/total CREB: 5 min, vehicle (0.74 ± 0.07) versus 10 nm exendin-4 (1.05 ± 0.15), P < 0.05] (Figure 19D). Simultaneously, exendin-4 induced phosphorylation of ATF-1 from 15 to 60 min in the adult neuronal cells with the maximal phosphorylation at 30 min [patf-1/total CREB: 30 min, vehicle (0.87 ± 0.08) versus 10 nm exendin-4 (1.34 ± 0.25), P < 0.01] (Figure 19C). In the embryonic cells, exendin-4 induced significant increase in ATF-1 phosphorylation at 5 and 30 min [patf-1/total CREB: 5 min, vehicle (1.05 ± 0.14) versus 10 nm exendin-4 (1.83 ± 0.16), P < 0.05; 30 min, vehicle (1.84 ± 0.13) versus 10 nm exendin-4 (2.55 ± 0.11), P < 0.05] (Figure 19E). Activated CREB or ATF-1 bind to a camp response element (CRE) within the promoter region of camp/creb/atf-1 downstream target genes. This leads to the recruitment of CREB binding protein (CBP/p300), and possibly other nuclear co-activators, to enhance transcription of

106 89

107 Figure 19. Exendin-4 induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic neuronal cell lines. The mhypoa-2/30 and mhypoe-36/1 cells were serum starved overnight and then treated with exendin-4 (10 nm) or vehicle. Protein was isolated over 6 h at the indicated time points, resolved on 10% SDS-PAGE, transferred to nitrocellulose membrane, and immunoblotted with antisera for c-fos, phospho-creb/atf-1, total CREB and Gβ (G-protein β subunit). Exendin-4 significantly increased c-fos expression in the mhypoa- 2/30 cells (A), and induced phosphorylation of CREB (B, D) and ATF-1 (C, E) in both mhypoa-2/30 (B, C) and mhypoe-36/1 (D, E) cell lines. c-fos expression was normalized to Gβ, and phosphorylation of CREB and ATF-1 was normalized to total CREB. Representative Western blots are shown. All results shown in the bar graphs are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05, **P < 0.01 vs. vehicle control). Statistical significance was calculated by two-way ANOVA. 90

108 91 immediate-early or late-response genes. Thus, phosphorylation of CREB may directly, via CRE, activate c-fos, which then heterodimerizes with members of the c-jun, ATF and JDP families to form activator protein-1 (AP-1), a transcription factor that in turn regulates target gene expression. Overall, activation of CREB/ATF-1 and c-fos in the neuronal cells suggests that exendin-4 could potentially regulate downstream target genes that express camp-response element or AP-1 consensus cis-elements in their promoter region Regulation of neurotensin and ghrelin mrna transcript expression by exendin-4 Next, regulation of neurotensin and ghrelin mrna transcript levels by exendin-4 was investigated. The adult and embryonic hypothalamic neuronal cells were exposed to 10 nm exendin-4 over a 24 h time course. Using real-time qrt-pcr, it was found that in the adult mhypoa-2/30 neurons neurotensin mrna levels were significantly up-regulated by 52% at 4 h post-treatment [vehicle (0.79 ± 0.07) versus 10 nm exendin-4 (1.20 ± 0.13), P < 0.05] (Figure 20A), whereas in the embryonic mhypoe-36/1 neurons the effect was more prominent as the mrna expression was increased by 64% at 6 h and remained significantly higher by 47% until 24 h [6 h, vehicle (0.80 ± 0.02) versus 10 nm exendin-4 (1.31 ± 0.06), P < 0.05; 24 h, vehicle (0.79 ± 0.07) versus 10 nm exendin-4 (1.16 ± 0.07), P < 0.05] (Figure 20C). With respect to the ghrelin mrna expression, it was found that exendin-4 caused significant attenuation of mrna levels by 34% at 1 h post-treatment in the adult mhypoa-2/30 cell line [1 h, vehicle (1.56 ± 0.16) versus 10 nm exendin-4 (1.03 ± 0.08), P < 0.05] (Figure 20B). Similarly, exendin-4 suppressed ghrelin mrna transcript levels by 29% at 4 h post-treatment in the embryonic mhypoe-36/1 cell line [4 h, vehicle (1.12 ± 0.10) versus 10 nm exendin-4 (0.79 ± 0.08), P < 0.05] (Figure 20D). However, in the adult mhypoa-2/30 neurons, ghrelin mrna levels were significantly increased by 78% at the 24 h time point [vehicle (0.82 ± 0.20) versus 10 nm exendin-4 (1.46 ± 0.31), P < 0.05] (Figure 20B). These results clearly indicate that neurotensin and ghrelin mrna levels are regulated by GLP-1R activation in the hypothalamic neuronal models, and complement the in vivo findings. The regulation of POMC expression in the adult mhypoa-2/30 neurons was not investigated due to failure of SYBR or TaqMan primers and probe to detect the message by real-time qrt-pcr.

109 Figure 20. Regulation of neurotensin (NT) (A, C) and ghrelin (B, D) mrna expression by exendin-4 in the mhypoa-2/30 (A, B) and mhypoe36/1 (C, D) neuronal cell lines. The cells were exposed to exendin-4 (10 nm) or vehicle over a 24 h time course; total RNA was extracted at the indicated time points and used as a template for real-time RT-PCR with primers specifically designed to amplify neuropeptide mrna. NT (A, C) and ghrelin (B, D) mrna levels were quantified using the CT method, normalized to the internal control (γ-actin) and are shown relative to vehicle only-treated mrna levels set to 1.0 at 1 h time point. All results shown are relative to corresponding control mrna levels at each time point and are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05 vs. vehicle control). Statistical significance was calculated by two-way ANOVA. 92

110 Reversal of exendin-4 regulation of neurotensin and ghrelin by PKA inhibitors The next step was to investigate whether PKA, a camp-dependent protein kinase, was involved in the regulation of neurotensin and ghrelin mrna expression by exendin-4. Pharmacological inhibitors of PKA (H-89 at 5 µm and PKI (14-22) amide at 1 µm concentration) were used to pre-treat the adult and embryonic cells for 1 h before exposure to 10 nm exendin-4. PBS and DMSO were used as vehicle controls, as exendin-4 was dissolved in PBS and PKA inhibitors were dissolved in DMSO. Total RNA was isolated at 1, 4 and 24 h after the exendin-4 treatment and analyzed using real-time qrt-pcr. As compared to the vehicle treatments, exendin-4 treatment significantly induced the neurotensin mrna levels in mhypoa- 2/30 neuronal cells [4 h: PBS (0.66 ± 0.08) versus PBS + exendin-4 (1.22 ± 0.06), P < 0.05; DMSO (0.85 ± 0.06) versus DMSO + exendin-4 (1.42 ± 0.15), P < 0.05] as well as in mhypoe- 36/1 neuronal cells [24 h: PBS (1.02 ± 0.08) versus PBS + exendin-4 (1.61 ± 0.09), P < 0.05; DMSO (0.91 ± 0.04) versus DMSO + exendin-4 (1.27 ± 0.09), P < 0.05] (Figure 21A and C). In contrast, exendin-4 treatment significantly decreased ghrelin mrna levels in both cell lines [mhypoa-2/30 cell line (1 h): PBS (1.52 ± 0.09) versus PBS + 10 nm exendin-4 (0.97 ± 0.13), P < 0.05; DMSO (1.12 ± 0.05) versus DMSO + 10 nm exendin-4 (0.68 ± 0.10), P < 0.05; mhypoe-36/1 cell line (4 h): PBS (1.13 ± 0.05) versus PBS + 10 nm exendin-4 (0.85 ± 0.07), P < 0.05; DMSO (1.15 ± 0.16) versus DMSO + 10 nm exendin-4 (0.76 ± 0.05), P < 0.05] (Figure 21B and D). Further, it was found that both PKA inhibitors attenuated the exendin-4-induced increase in neurotensin mrna expression in both neuronal cells models [mhypoa-2/30 cell line (4 h): PKI (0.99 ± 0.10) versus PKI + 10 exendin-4 (0.96 ± 0.08), P < 0.05; H89 (0.98 ± 0.09) versus H nm exendin-4 (1.06 ± 0.01), P < 0.05; mhypoe-36/1 cell line (24 h): PKI (0.99 ± 0.08) versus PKI + 10 nm exendin-4 (0.74 ± 0.07), P < 0.05; H89 (0.80 ± 0.04) versus H exendin-4 (0.66 ± 0.06), P > 0.05] (Figure 21A and C). On the other hand, PKA inhibition reversed the attenuation of ghrelin mrna transcripts caused by exendin-4 [mhypoa-2/30 cell line (1 h): PKI (0.89 ± 0.06) versus PKI + 10 exendin-4 (0.91 ± 0.02), P < 0.05; H89 (0.98 ± 0.11) versus H nm exendin-4 (1.19 ± 0.07), P > 0.05; mhypoe-36/1 cell line (4 h):

111 Figure 21. Regulation of neurotensin (NT) (A, C) and ghrelin (B, D) mrna expression via activation of the protein kinase A (PKA) pathway. Following overnight serum starvation mhypoa-2/30 (A, B) and mhypoe-36/1 (C, D) neuronal cells were pre-treated with PKA inhibitors PKI amide (1 μm) or H89 (5 μm) for 1 h followed by either vehicle (PBS or DMSO) or exendin-4 (10 nm) treatment. At 1, 4 and 24 h following exendin-4 exposure, cells were harvested for total RNA isolation, and NT (A, C) and ghrelin (B, D) mrna expression was determined by real-time RT-PCR. All results are relative to PBS only-treated mrna levels (set to 1.0) and are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05). Statistical significance was calculated by two-way ANOVA. White bars = control (PBS or DMSO with or without PKA inhibitor); black bars = treatment (exendin-4 with or without PKA inhibitor). 94

112 95 PKI (0.84 ± 0.17) versus PKI + 10 nm exendin-4 (0.90 ± 0.09), P > 0.05; H89 (0.98 ± 0.09) versus H exendin-4 (1.39 ± 0.11), P < 0.05] (Figure 21B and D). It was also observed that DMSO appears to regulate basal levels of ghrelin mrna in the mhypoa-2/30 line, as DMSO treatment alone resulted in attenuation of ghrelin mrna levels as compared to PBS treatment [PBS (1.52 ± 0.09) versus DMSO (1.12 ± 0.05), P < 0.05] (Figure 21B). Overall, these results indicate that PKA activation is involved with the regulation of neurotensin and ghrelin mrna by exendin Discussion Exendin-4 has been shown to regulate food intake and energy expenditure via GLP-1Rdependent mechanisms (371). Using an equivalent central dose, Baggio et al. showed that the anorectic action of central exendin-4 was not simply a nonspecific response, as anorexia was absent in GLP-1R knock-out mice, which lack functional GLP-1R (371). In the present study, using single-label IHC with c-fos, a marker of neuronal activation, it was found that exendin-4, at a dose that induced anorexia, activated hypothalamic ARC, PVN, DMH and PeV, regions that widely express GLP-1Rs along with several neuropeptides involved in energy metabolism. Furthermore, using double-label IHC, it was detected that central exendin-4 significantly activated α-msh/pomc and NPY neurons in the ARC, neurotensin-expressing neurons in the PVN, and ghrelin-expressing neurons in the ARC, PVN and PeV. Finally, using the adult mhypoa-2/30 and embryonic mhypoe-36/1 hypothalamic neuronal cell models, it was found that exendin-4, in a PKA-dependent manner, increased neurotensin mrna expression while attenuating ghrelin mrna levels. Overall, the in vivo findings suggest that complex interactions may occur between satiety- and hunger-related neuropeptides in one or more hypothalamic nuclei to mediate the overall anorexic action of exendin-4. The in vitro findings suggest that regulation of hypothalamic neurotensin and ghrelin may lie downstream of exendin-4-mediated GLP-1R activation. However, the present results must be interpreted with caution as these data may not be relevant to endogenous GLP-1 action due to the distinct mechanisms of receptor activation by endogenous GLP-1 and exendin-4, as exendin-4 is speculated to activate an unknown receptor apart from GLP-1R activation (163, 369, 373). In support, both central and peripheral administration of exendin-4 suppressed ghrelin levels, but the suppression of ghrelin was neither mimicked by GLP-1 nor blocked by the GLP-1R antagonist exendin-9-39,

113 96 suggesting that ghrelin inhibition appears to correspond to an exendin-specific mechanism, which might be independent of GLP-1R activation (163). Several other studies have provided evidence that the actions of exendin-4 cannot be adequately explained by activation of the GLP- 1R, and therefore, presence of another unknown receptor is suggested. Indeed, exendin-4, but not GLP-1, increases insulin sensitivity and glucose transport in cultured L6 myotubules in vitro (374), and the effects of exendin-4 on camp levels in 3T3-L1 adipocytes are not mediated by the GLP-1R (375). Significantly, unlike exendin-4, GLP-1 activation was unable to modify energy expenditure in mice (221). Moreover, GLP-1 could not reproduce other endocrine effects elicited by exendin-4, such as the reduction of thyrotropin secretion in rats (376). Thus, because exendin-4 exerts some of its effects through a mechanism that may not involve the activation of the GLP-1R, further investigations using GLP-1R inhibitors or GLP-1R knockout mice are required to investigate endogenous GLP-1 action and distinguish GLP-1R-dependent and - independent actions of exendin-4 on neuropeptidergic neuronal activation. Recently, it was found that GLP-1R activation in the ARC does not induce anorexia via α-msh/pomc or NPY neurons (131), suggesting that other mediators, such as anorexigenic neurotensin and orexigenic ghrelin neurons, may be involved. Despite their potential role in the hypothalamic regulation of energy homeostasis (6, 139, , 153, 155), neurotensin and ghrelin have received relatively little attention with respect to appetite regulation. While it is well established that neurotensin-expressing neurons are expressed in the hypothalamus and exert an anorexigenic action (56, 143), the existence of central ghrelin-expressing neurons is still controversial and a general consensus does not exist concerning the source of central ghrelin (377). Although ghrelin-positive neurons are located within the intact hypothalamus (105, 148, 378, 379), and display synaptic interactions with POMC, NPY and other ghrelin-containing neurons (156, 380), at present the spectrum of functions and physiological role of ghrelinexpressing neurons in the hypothalamus are yet to be fully elucidated. The present study found that exendin-4 induced almost a 6-fold increase in neurotensin neuronal activation in the PVN, compared to only less than a 2-fold increase in the ARC. This might explain why the PVN GLP- 1R system, but not that of the ARC, is predominant in mediating anorexia, as reported by Sandoval et al. (131).

114 97 Although ghrelin neurons are orexigenic, it was observed that exendin-4 significantly activated these neurons in the ARC, PVN and PeV. However, whether this neuronal activation translates into stimulation or inhibition is not completely clear. It is possible that the activation of hypothalamic ghrelin neurons may be complementary or compensatory to exert a function other than appetite regulation. Exendin-4 has been shown to induce anti-inflammatory action in the peripheral tissues (381). It also exerts neuroprotective effects on dopaminergic neurons by inhibiting pro-inflammatory mediators (382). In this context, as ghrelin is found to exert antiinflammatory and neuroprotective effects (383), the ghrelin neuronal activation observed in this study may be involved in mediating anti-inflammatory effects of exendin-4 (381). Interestingly, both orexigenic ghrelin and anorexigenic neurotensin share a common characteristic: to stimulate release of CRH (an anorexigenic and anxiogenic factor in the PVN) resulting in hypothalamicpituitary-adrenal axis (HPA) activation and stimulation of glucocorticoid release (384, 385). As exendin-4 has been demonstrated to stimulate the HPA axis in rodents and humans (386), it is important to explore the role of hypothalamic ghrelin and neurotensin in the HPA regulation by GLP-1R agonists. This could elucidate possible reasons for why the GLP-1R knockout mice exhibit paradoxically increased corticosterone responses to stress, even though they appear to have a normal HPA axis (387). Activation of ARC and PeV neurons (particularly of ghrelin neurons) suggests a possible involvement of ghrelin in glucoregulation and neurogenesis induced by GLP-1R activation (19, 106, 107). Additionally, significant activation of α-msh/pomc and NPY neurons in the ARC was detected. The finding that ARC α-msh/pomc neurons are activated by exendin-4 agrees with that reported by Sandoval et al. (131). However, it is not clear whether exendin-4-activated α- MSH/POMC neurons actually induce anorexia, as measuring the changes in the α-msh synthesis and secretion or POMC mrna levels in the ARC in vivo was beyond the scope of this study. Nevertheless, given the fact that GLP-1R-expressing ARC POMC neurons are not involved in appetite suppression, as reported by Sandoval et al. (131), it would be appropriate to further investigate if these neurons play any role in mediating the central glucoregulatory action of exendin-4 or native GLP-1 (131, 369). Given that ARC NPY neurons do not express GLP-1Rs, the observed significant activation of these neurons is quite intriguing. Absence of GLP-1R expression on ARC NPY

115 98 neurons excludes any direct action of exendin-4, suggesting their indirect activation by other neurons, such as POMC neurons or -aminobutyric acid (GABA) cells in the ARC (388). As neurotensin was demonstrated to inhibit feeding via Ntsr expressed in the hypothalamic ARC (134), it is possible that the activated PVN neurotensin neurons regulate ARC NPY neurons to ultimately inhibit them. However, whether ARC NPY neurons express Ntsr is not known. Another possibility is that the activated ghrelin-immunopositive neurons in the ARC could influence NPY neurons via synaptic transmission (156). Given that the ARC GLP-1R system regulates glucose homeostasis (131), and that ARC NPY neurons are glucose-sensitive neurons directly regulated by exogenous ghrelin (106, 107), the possible interaction between ARC NPY and ghrelin neurons in exendin-4-mediated glucose regulation needs to be analyzed further. A striking finding in this study was the absence of c-fos activation in the VMH, the nucleus involved in the control of satiety (25), and in the LH, the region known to control hunger (24). However, the lack of c-fos activation in the hypothalamic VMH was in accordance with very low GLP-1R expression in this region (216). The lack of c-fos expression in the LH suggests that the LH is not activated by exendin-4, despite moderate expression of GLP-1R in this region (216). This result may also indicate that orexigenic MCH and orexin/hypocretin neurons in the LH are not activated by exendin-4. It has been demonstrated that leptin inhibits LH indirectly via the activation of ARC POMC neurons, and that when leptin-mediated suppression of LH neurons was reversed via ablation of the ARC, c-fos expression was increased in LH neurons (389). This suggests that the lack of c-fos activation in the LH neurons observed in this study could be due to inhibitory quiescence via exendin-4-activated ARC POMC neurons. However, caution must be exercised in interpreting these results based on the absence of c-fos activation in these regions, for the c-fos activation itself does not indicate neuronal stimulation or inhibition unless advanced techniques, such as electrophysiological methods or magnetic resonance imaging, are simultaneously employed. The in vitro part of this study characterized the signaling pathways and changes in neurotensin and ghrelin gene expression that occur as a direct result of exendin-4 action on the hypothalamic adult mhypoa-2/30 and embryonic mhypoe-36/1 neuronal cell models (18, 19). These cell models were used in the present study because it is speculated that they originate from the PVN (given the co-expression of galanin and arginine-vasopressin) (18, 316, 390, 391).

116 99 Using both adult and embryonic neuronal cell lines, it was observed that mature neurons do not differ from neurons of embryonic origin in the signaling mechanisms triggered by GLP-1R activation. The intracellular signaling pathways regulating hypothalamic neurotensin and ghrelin gene expression following GLP-1R activation have not been previously established. In pancreatic β-cells, GLP-1R activation stimulates adenylyl cyclase, thereby increasing camp, and subsequently activating PKA. Similarly, GLP-1 is shown to activate adenylyl cyclase in rodent brain (392), and in hypothalamic POMC neurons, GLP-1R activation stimulates camp/pka pathway to induce anorectic activity (393). In the present study, hypothalamic adult mhypoa- 2/30 and embryonic mhypoe-36/1 neuronal cell models were used because they are the only hypothalamic cell models that express functional GLP-1R as well as neurotensin and ghrelin (18, 19). It was found that GLP-1R activation lead to up-regulation of intracellular camp in these neuronal cell models, and that transcription factors CREB and ATF-1 were activated upon exendin-4 stimulation. Furthermore, activation of c-fos in the mhypoa-2/30 line suggests that the in vivo activation of c-fos found in the present study could be mediated through direct activation of GLP-1R-expressing hypothalamic neurons. Further studies are required to determine whether exendin-4 regulates gene expression at the promoter region of these genes. The importance of camp/pka activation in appetite regulation has been demonstrated in several studies. Recently, hindbrain GLP-1R activation by exendin-4 was shown to induce anorexia in a PKA-dependent manner (327). Hypothalamic activation of camp/pka mediates central regulation of satiety by inhibiting NPY-induced feeding (336). Furthermore, PKA and CREB signaling have been shown to negatively regulate NPY gene expression (337), whereas transcription of CART, a potent appetite-suppressing peptide, is upregulated by camp/pka/creb activation (394). Overall, it appears that stimulation of camp/pka activity inhibits feeding behavior by increasing anorexigenic and decreasing orexigenic neuropeptide expression. These studies are therefore consistent with the findings that exendin-4 induced neurotensin and suppressed ghrelin expression via PKA activation. However, regulation of hypothalamic neurotensin and ghrelin gene expression was shown to occur through other pathways as well. Neurotensin gene expression was augmented by leptin via activation of STAT3 and MAPK ERK1/2 (300), whereas ghrelin was repressed by insulin via both PI3K/Akt

117 100 and MAPK ERK1/2 pathways (395). The MAPK ERK1/2 plays an important role in food intake control, and exendin-4 has recently been shown to induce anorexia partially by activation of MAPK ERK1/2 in hindbrain GLP-1R neurons (327). Activation of MAPK ERK1/2 by exendin-4 in hypothalamic neurons has yet to be studied. The induction of either PKA or CREB signaling normally serves to activate gene transcription. Therefore, the upregulation of neurotensin mrna expression by exendin-4 may reflect stimulation of gene transcription at the promoter level. Neurotensin gene promoter analysis indicates that synergistic activation of AP-1 and CRE elements is important in hormonal regulation of neurotensin gene expression (396). Furthermore, it was demonstrated that leptin upregulated neurotensin gene expression by activation of ATF-1 and c-fos (300). However, it cannot be ruled out that the increase in neurotensin mrna was due to an increase in mrna stability, similar to GLP-1-induced increase in insulin mrna stability to augment insulin mrna levels in β-cells (397). The suppression in ghrelin mrna levels observed following GLP-1R activation does not seem to be due to a decrease in transcription at the promoter level, at least in the adult cell line, as the mrna transcript levels decrease rapidly within 1 h post-treatment. This down-regulation likely occurred due to destabilization of the ghrelin mrna by activated PKA, similar to the PKA-mediated increased decay of angiotensin II type 1 receptor mrna (398). In another study it was proposed that insulin-triggered changes in mrna half-life could potentially suppress orexigenic NPY and AgRP gene expression (399). Overall, further studies are required to determine whether exendin-4 regulates neurotensin and ghrelin mrna expression through the AP-1 or CRE consensus sites, or through other cis-elements within the promoter or enhancer region of these genes, or via specific molecular components induced by PKA that mediate mrna decay. Recently, the ability of mirnas to regulate mrna has been addressed as a powerful mechanism to control neuronal functions. In this context, it was demonstrated that CREBregulated mirna132 regulates neuronal morphogenesis by decreasing mrna transcript levels of the GTPase-activating protein p250gap (400). The induction phase of the mirna132 is rapid and parallels c-fos induction (400), which raises the possibility for CREB-induced mirna132 to target ghrelin mrna and rapidly suppress its expression, as seen in this study. Whether the presence of a mir132 recognition element in ghrelin mrna or lack of it in

118 101 neurotensin mrna is responsible for differential regulation of these neuropeptide transcripts by exendin-4 needs to be studied further. In summary, the data presented here demonstrate that exendin-4 activates multiple hypothalamic sites where complex interactions may occur between appetite-regulating neuropeptidergic neurons to induce anorexia. The findings from the in vitro model of direct action of exendin-4 on neurotensin and ghrelin mrna regulation complement the in vivo findings that neurotensin and ghrelin neurons, in addition to α-msh/pomc and NPY neurons, can be activated and regulated by central exendin-4. Ultimately, this study provides a previously unrecognized link between exendin-4 action and neurotensin and ghrelin regulation.

119 102 Chapter 5 Glucagon-Like Peptide-2 Directly Regulates Hypothalamic Neurons Expressing Neuropeptides Linked to Appetite Control in vivo and in vitro

120 103 Publication: Prasad S. Dalvi and Denise D. Belsham. Glucagon-like peptide-2 regulates hypothalamic neurons expressing neuropeptides linked to appetite control in vivo and in vitro. Endocrinology (2012) 153: All the experiments included in the publication and in this thesis were designed by P.S.D. and D.D.B. P.S.D. executed all the experiments in this thesis, analyzed all data, designed and created all figures, wrote and revised the manuscript under the supervision of D.D.B. Published figures: Figure 22. Intracerebroventricular (i.c.v.) injection of h(gly 2 )GLP-2 inhibits food and water intake, and induces weight loss in a dose-dependent manner in wild-type mice. Figure 23. Acute h(gly 2 )GLP-2 treatment activates hypothalamic appetite-regulating nuclei. Figure 24. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity (ir) in the hypothalamic neurons expressing α-msh-, NPY-, neurotensin- or ghrelin-ir. Figure 25. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity (ir) in the hypothalamic neurons expressing NPY-, neurotensin- or ghrelin-ir. Figure 26. Graphical representation showing the number of neurons expressing c-fos- and neuropeptide-immunoreactivity (ir) in the ARC, VMH, DMH, PVN, LH and internuclear space between the DMH and LH of the saline- or h(gly 2 )GLP-2-treated mouse hypothalamus. Figure 27. Expression profile of GLP-2 receptor (GLP-2R) and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines. Figure 28. Acute h(gly 2 )GLP-2 treatment induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic GLP-2R-positive mhypoa-2/30 neuronal cells. Figure 29. Regulation of neurotensin (A, C) and ghrelin (B, D) mrna expression by h(gly 2 )GLP-2 in the mhypoa-2/30 neuronal cell line in protein kinase A (PKA)-dependent manner. Permissions were obtained to reproduce the copyrighted material.

121 104 5 Glucagon-Like Peptide-2 Directly Regulates Hypothalamic Neurons Expressing Neuropeptides Linked to Appetite Control in vivo and in vitro 5.1 Abstract GLP-2 has been postulated to affect appetite at the level of the hypothalamus. To gain better insight into this process, a degradation-resistant GLP-2 analog, human (Gly 2 )GLP-2(1-33) [h(gly 2 )GLP-2] was intracerebroventricularly injected into mice to examine its action on food and water intake and also activation of hypothalamic anorexigenic α-msh/pomc, neurotensin, and orexigenic NPY, and ghrelin neurons. Central h(gly 2 )GLP-2 administration significantly suppressed food and water intake with acute weight loss at 2 h. Further, central h(gly 2 )GLP-2 robustly induced c-fos activation in the hypothalamic ARC, DMH, VMH, PVN, and LH nuclei. Differential colocalization of neuropeptides with c-fos in specific regions of the hypothalamus was observed. To assess whether hypothalamic neuropeptides are directly regulated by GLP-2 in vitro, an adult-derived clonal, immortalized hypothalamic cell line, mhypoa-2/30, that endogenously expresses functional GLP-2R and two of the feeding-related neuropeptides linked to GLP-2R activation in vivo: neurotensin and ghrelin was used in the present study. Treatment with h(gly 2 )GLP-2 stimulated c-fos expression and phosphorylation of CREB/ATF-1 in this neuronal cell model. In addition, treatment with h(gly 2 )GLP-2 significantly increased neurotensin and ghrelin mrna transcript levels by 50 and 95%, respectively, at 24 h after treatment in a PKA-dependent manner. Taken together, these findings implicate the PKA pathway as the means by which GLP-2 can upregulate hypothalamic neuropeptide mrna levels and provide an evidence for a link between central GLP-2R activation and specific hypothalamic neuropeptides involved in appetite regulation. 5.2 Introduction The glucagon gene encodes two homologous peptides, GLP-1 and GLP-2. These peptides are generated by post-translational processing mainly in the enteroendocrine L cells of the intestine, and specific neurons of the caudal brainstem (171, 172, 215, 401). Both GLPs play

122 105 important roles in energy homeostasis by regulating nutrient ingestion and disposal. GLP-2 is a hormone that mainly stimulates crypt cell proliferation and inhibits apoptosis of the intestinal epithelium to exert trophic effects ( ). GLP-2 also regulates energy homeostasis by regulating intestinal motility, permeability, nutrient ingestion and absorption (204, 405, 406). In the CNS, GLP-1 and GLP-2 are synthesized in the caudal brainstem and hypothalamus (172, 215, 333, 401). The GLP-1/GLP-2-expressing preproglucagon neurons of the NTS send extensive projections to the hypothalamic PVN and DMH nuclei, both regions known to be involved in the regulation of feeding behavior (213). It has been speculated that GLP-2 also acts as a neurotransmitter and exerts similar actions like GLP-1 in the CNS (214, 333). The GLP-1 and GLP-2 mediate their actions via unique G protein-coupled receptors. The GLP-1R is widely expressed in the hypothalamus as compared with the restricted expression of the GLP-2R in the VMH and DMH nuclei (214, , 231). Thus, they may play distinct roles in appetite regulation despite being synthesized in the same neurons (213). However, little is known of GLP-2-mediated signaling in appetite-regulation. Tang-Christensen et al. discovered that central administration of GLP-2 is involved in regulation of food intake in rats (214). Similarly, pharmacological doses of central long-acting GLP-2 inhibited short-term food intake in mice (218). Therefore, the mechanisms of central GLP-2R action to induce reduction in food intake require further characterization. The GLP-2R is predominantly expressed in the gastrointestinal tract and the CNS, with limited expression in lung, cervix, vagal afferents and other peripheral organs (229). The signaling mechanisms of the GLP-2R activation are not completely understood due to the lack of cell models that endogenously express GLP-2R. Data collected from heterologous cell lines transfected with GLP-2R have demonstrated that camp/pka- and AP-1-dependent pathways are activated by GLP-2 (231). Also fetal rat intestinal cell cultures treated with h[gly 2 ]-GLP-2 were shown to have increased camp concentration (407). Further, the PI3K/Akt pathway has been shown to be implicated in the stimulatory effects of GLP-2 to enhance intestinal IGF-I mrna transcript levels in the intestinal subepithelial myofibroblasts (408). However, few studies have addressed central GLP-2R activation and the target hypothalamic neurons largely remain elusive. Thus, clonal, hypothalamic cell lines that endogenously express functional GLP- 2R were used. The present study investigated activation of hypothalamic α-msh/pomc-,

123 106 neurotensin-, NPY-, and ghrelin-expressing neurons following GLP-2 exposure in vivo, and determined potential signaling pathways involved in vitro. 5.3 Results Effect of i.c.v. h(gly 2 )GLP-2 on food/water intake, and animal weight To establish the efficacy of the h(gly 2 )GLP-2, the effect on food and water intake over a short-term exposure was studied. For the dose-response study, the anorexigenic doses of h(gly 2 )GLP-2 (100 ng, 1 µg and 5 µg) were selected based on previously conducted studies in rodents (214, 218). It was found that the cumulative food intake was significantly reduced by 66.67% by i.c.v. dose of 5 µg of h(gly 2 )GLP-2, as compared with saline at 1 h post-treatment in ad libitum-fed wild type mice [cumulative food intake 1 h (g): saline, 0.15 ± 0.03; 100 ng h(gly 2 )GLP-2, 0.15 ± 0.03; 1 µg h(gly 2 )GLP-2, 0.13 ± 0.03; 5 µg h(gly 2 )GLP-2, 0.05 ± 0.03; 5 µg h(gly 2 )GLP-2, versus saline, P < 0.05] (Figure 22A). At 2 h post-treatment, no additional food intake suppression was detected. On the contrary, the suppression of cumulative food intake was attenuated and it did not differ across all treatment groups [food intake 2 h (g): saline, 0.37 ± 0.03; 100 ng h(gly 2 )GLP-2, 0.33 ± 0.03; 1 µg h(gly 2 )GLP-2, 0.33 ± 0.03; 5 µg h(gly 2 )GLP-2, 0.28 ± 0.08] (Figure 22A). Further, it was found that different doses of h(gly 2 )GLP-2 did not have any significant effect on cumulative water intake at 1 h post-treatment [cumulative water intake 1 h (ml): saline, 1.75 ± 0.25; 100 ng h(gly 2 )GLP-2, 1.67 ± 0.29; 1 µg h(gly 2 )GLP-2, 1.33 ± 0.29; 5 µg h(gly 2 )GLP-2, 1.00 ± 0.58] (Figure 22B). In contrast, at 2h post-treatment, cumulative water intake was significantly reduced by 41.75% and 43.75% by both i.c.v. doses of 1 and 5 µg of h(gly 2 )GLP-2, respectively, as compared with saline treatment [cumulative water intake 2 h (ml): saline, 4.00 ± 0.41; 100 ng h(gly 2 )GLP-2, 3.00 ± 0.87; 1 µg h(gly 2 )GLP-2, 2.33 ± 0.76; 5 µg h(gly 2 )GLP-2, 2.25 ± 0.48; 1 and 5 µg h(gly 2 )GLP-2 versus saline, P < 0.05] (Figure 22B). Based on the anorexia induced, the dose of 5 µg of h(gly 2 )GLP-2 was selected to determine its effect on body weight. It was found that the weight changes following single i.c.v. dose of h(gly 2 )GLP-2 were remarkable because the weight loss was significant within 1 and 2 h [change in weight (g): saline (1h), ± 0.01 versus 5 µg h(gly 2 )GLP-2 (1h), ± 0.18, P < 0.05; saline (2h), ± 0.16 versus 5 µg h(gly 2 )GLP-2 (2h), ± 0.06, P < 0.05] (Figure 22C). Further duration of these rapid weight losses and the recovery period were not studied.

124 Figure 22. Intracerebroventricular (i.c.v.) injection of h(gly 2 )GLP-2 inhibits food and water intake, and induces weight loss in a dose-dependent manner in wild-type mice. To determine the efficacy of i.c.v. h(gly 2 )GLP-2 to induce anorexia, ad libitum-fed mice received injection of either 100 ng, 1 µg or 5 µg of h(gly 2 )GLP-2 dissolved in 2 μl of 0.9% normal saline 1 h prior to the onset of the dark cycle. Mice were returned to their home cages with pre-weighed amount of chow and water. Changes in food (A) and water (B) intake, and animal weight (C) were measured at 1 and 2 h post-injection. 0.9% normal saline solution was used as control treatment. All results are expressed as mean ± SEM (n = 3-4 mice/group); *P < 0.05 vs. saline). 107

125 Effect of h(gly 2 )GLP-2 on activation of hypothalamic nuclei and neuropeptidergic neurons As assessed by IHC for c-fos expression, a distinctive pattern of neuronal activation was noted after the central injection of h(gly 2 )GLP-2 compared with the saline controls (Figure 23 A-F). Significant increases in the number of c-fos-positive neurons were detected in the hypothalamic ARC (increase by 152%), DMH (increase by 342%), VMH (increase by 99%), LH (increase by 411%) and PVN (increase by 429%) [c-fos-positive nuclei: ARC, saline, ± 1.42 versus 5 µg h(gly 2 )GLP-2, ± 2.28, P < 0.001; DMH, saline, ± 1.04 versus 5 µg h(gly 2 )GLP-2, ± 4.83, P < 0.001; VMH, saline, 3.24 ± 0.85 versus 5 µg h(gly 2 )GLP-2, 6.45 ± 1.14, P = 0.046; LH, saline, 4.04 ± 0.81 versus 5 µg h(gly 2 )GLP-2, ± 2.61, P < 0.001; PVN, saline, ± 1.50 versus 5 µg h(gly 2 )GLP-2, ± 10.44, P < 0.001] (Figure 23G). The next goal was to determine the neuropeptidergic neurons activated in the hypothalamic nuclei by performing double-staining IHC for c-fos and neuropeptide coexpression (Figs. 24 and 25). It was found that in the ARC, single dose of i.c.v. h(gly 2 )GLP-2 significantly increased the number of c-fos-expressing α-msh/pomc neurons by 105% [saline, ± 1.92 versus 5 µg h(gly 2 )GLP-2, ± 2.55, P < 0.001] (Figure 26B), NPY neurons by 140% [saline, ± 2.53 versus 5 µg h(gly 2 )GLP-2, ± 4.69, P < 0.001] (Figure 26D), neurotensin neurons by 332% [saline, 6.90 ± 2.59 versus 5 µg h(gly 2 )GLP-2, ± 1.67, P < 0.001] (Figure 26F) and ghrelin neurons by 87% [saline, ± 1.72 versus 5 µg h(gly 2 )GLP- 2, ± 0.70, P = 0.003] (Figure 26H). Further, it was found that in the hypothalamic PVN, h(gly 2 )GLP-2 robustly increased the number of c-fos-positive nuclei with neurotensin coexpression by 504% [saline, 9.07 ± 2.85 versus 5 µg h(gly 2 )GLP-2, ± 8.67, P = 0.007] (Figure 26F) and ghrelin co-expression by 233% [saline, 8.92 ± 1.66 versus 5 µg h(gly 2 )GLP-2, ± 0.96, P < 0.001] (Figure 26H). In the hypothalamic DMH, a significant increase in c-fos and neurotensin and NPY co-expressing neurons by 416% and 334%, respectively, was detected [neurotensin: saline 9.10 ± 0.92 versus 5 µg h(gly 2 )GLP-2, ± 4.38, P= 0.003; NPY: saline 9.11 ± 1.22 versus 5 µg h(gly 2 )GLP-2, ± 7.45, P = 0.001] (Figure 26H).

126 Figure 23. Acute h(gly 2 )GLP-2 treatment activates hypothalamic appetite-regulating nuclei. Immunohistochemistry was performed to assess neuronal activation by c-fosimmunoreactivity (ir) in wild-type mice treated with intracerebroventricular (i.c.v.) saline or h(gly 2 )GLP-2 (hglp-2). Representative photomicrographs showing expression of c-fos-ir in the hypothalamic ARC, DMH, VMH, LH, and PVN regions in coronal sections of the mouse hypothalamii. A-F: low magnification (X50) images of the hypothalamic regions (scale bars, 1 mm). A -F : High magnification (X400) images representative of the regions indicated by arrows in images A-F, respectively (scale bars, 100 μm). c-fos DAB-positive neurons are visible as brown nuclei. G: Bar graph showing the number of c-fos-ir neurons in the hypothalamic regions at 2 h post-treatment. Data in the bar graph are expressed as mean ± SEM (n = 3-4 animals/group; *P < 0.05, **P < 0.001). 109

127 Figure 24. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity (ir) in the hypothalamic neurons expressing α-msh-, NPY-, neurotensin- or ghrelin-ir. A-L: Brightfield photomicrographs showing neurons that coexpress c-fos-ir (brown nuclei) and neuropeptide-ir (blue-black cytoplasm) in the coronal sections of the hypothalamic arcuate (ARC) and paraventricular (PVN) nuclei from wild-type mice at 2h after intracerebroventricular administration of saline (A, C, E, G, I, K) or h(gly 2 )GLP-2 (B, D, F, H, J, L). A and B: Coexpression of c-fos-ir with a-msh-ir in the ARC. C and D: Coexpression of c-fos-ir with NPY-ir in the ARC. E-H: Coexpression of c-fos-ir with neurotensin-ir in the ARC (E, F) and PVN (G, H). I-L: Coexpression of c-fos-ir with ghrelin-ir in the ARC (I, J) and PVN (K, L). Black arrowheads represent neurons expressing only nuclear c-fos-ir, white arrowheads represent neurons expressing only cytoplasmic perineuclear neuropeptide-ir, and black arrows represent double-labeled neurons with coexpression of c-fos-ir and neuropeptide-ir. 3v, third cerebral ventricle. Original magnification: X400, scale bar: 100 μm. 110

128 Figure 25. Acute h(gly 2 )GLP-2 treatment induces c-fos-immunoreactivity (ir) in the hypothalamic neurons expressing NPY-, neurotensin- or ghrelin-ir. A-J: Bright-field photomicrographs showing neurons that co-express c-fos-ir (brown nuclei) and neuropeptide-ir (blue-black cytoplasm) in the coronal sections of the hypothalamic dorsomedial (DMH), ventromedial (VMH), lateral hypothalamic nucleus (LH) and internuclear space between DMH and LH (INS), from wild-type mice at 2h after intracerebroventricular administration of saline (A, C, E, G, I) or h(gly 2 )GLP-2 (B, D, F, H, J). A and B: Coexpression of c-fos-ir with NPY-ir in the DMH. C and D: Coexpression of c-fos-ir with NT-ir in the DMH. E and F: Coexpression of c-fos-ir with NT-ir in the VMH. G and H: Coexpression of c-fos-ir with NT-ir in the LH. I and J: Coexpression of c-fos-ir with Ghr-ir in the INS. Black arrowheads represent neurons expressing only nuclear c-fos-ir, white arrowheads represent neurons expressing only cytoplasmic perineuclear neuropeptide-ir, and black arrows represent double-labeled neurons with coexpression of c-fos-ir and neuropeptide-ir. f, fornix. Original magnification: X400, scale bar: 100 μm. 111

129 Figure 26. Graphical representation of neurons expressing c-fos- and neuropeptideimmunoreactivity (ir) in the ARC, VMH, DMH, PVN, LH and internuclear space between the DMH and LH of the saline- or h(gly 2 )GLP-2-treated mouse hypothalamus. Doublelabeled immunohistochemistry for c-fos-ir and α-msh-ir (B), c-fos-ir and NPY-ir (D), c-fos-ir and neurotensin-ir (F), or c-fos-ir and ghrelin-ir (H) indicates that intracerebroventricular h(gly 2 )GLP-2 activates hypothalamic neuropeptidergic neurons. Note that there was no change in the number of neurons expressing only α-msh-ir (A), NPY-ir (C), neurotensin-ir (E) or ghrelin-ir (G) in the hypothalamic regions of saline- or h(gly 2 )GLP-2-treated animals. Data are represented as mean ± SEM (n = 3-4 mice/group); *P < 0.05, ** P < 0.01, *** P < vs. saline treatment. Statistical significance was calculated by two-tailed, unpaired t-test. 112

130 113 The number of c-fos-positive nuclei co-expressing neurotensin-immunoreactivity were significantly increased by 404% in the VMH [saline 4.01 ± 1.67 versus 5 µg h(gly 2 )GLP-2, ± 5.02, P = 0.045], and by 535% in the LH [saline 4.01 ± 1.67 versus 5 µg h(gly 2 )GLP-2, ± 5.02, P = 0.031]. Also, the number of c-fos-positive nuclei co-expressing ghrelinimmunoreactivity were significantly increased by 242% in the internuclear space between DMH and LH [saline 5.28 ± 1.76 versus 5 µg h(gly 2 )GLP-2, ± 2.09, P = 0.003]. The number of neurons expressing only neuropeptide-immunoreactivity remained unchanged in these hypothalamic regions of saline- or h(gly 2 )GLP-2-treated animals (Figure 26A, C, E, and G) Expression of GLP-2R and appetite-regulating neuropeptides in adult mhypoa-2/30 cell line Direct regulation of hypothalamic neurotensin and ghrelin mrna expression by h(gly 2 )GLP-2 was studied using a clonal, immortalized hypothalamic neuronal cell model, adult mhypoa-2/30 neuronal cell line. Prior to using this cell model, the presence of GLP-2R mrna in the mhypoa-2/30 neuronal cells was confirmed using RT-PCR (Figure 27A). In another cell model, embryonic mhypoe-36/1 cell line, GLP-2R gene expression was not confirmed. GLP- 2R-positive mouse jejunal tissue mrna was used as a positive control. Also, the expression of hypothalamic neuropeptides involved in appetite regulation was analyzed and it was found that neurotensin and ghrelin genes are expressed in this cell line (Figure 27B). Currently, there are no hypothalamic cell models reported with a functional endogenous GLP-2R. Therefore, to determine if the GLP-2R is functionally active in the adult mhypoa-2/30 neuronal cells, activation of camp was measured by camp-ria following GLP-2 treatment using two different concentrations (10 and 50 nm). Forskolin (1µM) was used as a positive control for camp activation. Using camp-ria, it was found that both doses of GLP-2 significantly increased camp levels in the neuronal cell line [camp content (pmol/µg protein): vehicle, 0.49 ± 0.08; 1 µm forskolin, 2.35 ± 0.17; 10 nm GLP-2, 0.79 ± 0.11; 50 nm GLP-2, 0.92 ± 0.06; forskolin versus saline, P < 0.01; GLP-2 (10 nm and 50 nm) versus vehicle, P < 0.05] (Figure 27C). Increases in the camp content indicate that the GLP-2R expressed in these cells is functional and responsive to GLP-2. Further, using GLP-2(3-33), a GLP-2R antagonist, it was studied whether h(gly 2 )GLP-2 directly activates GLP-2R to stimulate camp production. It was

131 Figure 27. Expression profile of GLP-2R and appetite-regulating neuropeptides in the hypothalamic neuronal cell lines. A: Expression of GLP-2R mrna transcript in jejunum and the hypothalamic neuronal cell lines mhypoa-2/30 and mhypoe-36/1 by RT-PCR using specific primers for mouse GLP-2R gene. Total RNA, isolated from mouse jejunum and the indicated cell lines, was used as template for RT-PCR using One-Step RT-PCR kit. Mouse jejunal RNA was used as a positive control for GLP-2R expression. M, markers; NTC, no template control. B: RT-PCR analysis results for the mrna expression of neuropeptides in mhypoa-2/30 neuronal cells. + indicates that the gene is expressed; - indicates that the gene is not expressed. C: Expression of functional GLP-2R by camp-ria in mhypoa-2/30 neuronal cells. D: h(gly 2 )GLP-2 (hglp-2) stimulates camp production via the GLP-2R activation. The cells were pretreated for 5 minutes with 1 µm GLP-2 (3-33) or vehicle alone prior to a 10-minute treatment with vehicle, forskolin (1 or 10 µm), GLP-2 (10 or 50 nm) or hglp-2 (10 nm). The amount of intracellular camp was determined in triplicate by RIA. Forskolin was used as a positive control. All results are expressed as mean ± SEM (n = 4; *P < 0.05, **P < 0.01 vs. vehicle control). 114

132 115 found that 1 µm GLP-2(3-33) alone did not stimulate camp production, but completely attenuated the stimulatory effect of h(gly 2 )GLP-2, suggesting that h(gly 2 )GLP-2 stimulates camp production via GLP-2R activation [camp content (pmol/µg protein): vehicle, 0.40 ± 0.02; 10 µm forskolin, 9.79 ± 0.28; 10 nm h(gly 2 )GLP-2, 0.79 ± 0.17; 1 µm GLP-2(3-33), 0.41 ± 0.01; 1 µm GLP-2(3-33) + 10 nm h(gly 2 )GLP-2, 0.45 ± 0.01; forskolin versus saline, P < 0.01; h(gly 2 )GLP-2 (10 nm) versus vehicle, P < 0.05] (Figure 27D) Activation of CREB/ATF-1 and c-fos by h(gly 2 )GLP-2 in the hypothalamic neuronal cells The key signaling pathway that GLP-2 activates is the camp/pka pathway. Therefore, the next step was to determine downstream effectors of camp/pka pathway activated by h(gly 2 )GLP-2 in the hypothalamic adult neuronal cells. Therefore, neuronal cells were treated with 10 nm h(gly 2 )GLP-2 and activation of CREB/ATF-1 and c-fos was analyzed over 6 h. By western blot analysis, it was found that h(gly 2 )GLP-2 significantly induced c-fos activation by 62% in the adult mhypoa-2/30 cell line at 2 h [c-fos/gβ: vehicle, 0.91 ± 0.07 versus 10 nm h(gly 2 )GLP-2, 1.69 ± 0.14, P < 0.05] (Figure 28A). Similarly, h(gly 2 )GLP-2 significantly increased phosphorylation of CREB at Ser 133 at 15 min, 1 h and 2 h by 63%, 33% and 59%, respectively, in the adult neuronal cells (Figure 28B). Simultaneously, h(gly 2 )GLP-2 induced significant increase in phosphorylation of ATF-1 from 5 min to 6 h and the maximum increase was by 72% at 2 h post-treatment in these neuronal cells (Figure 28C) Regulation of neurotensin and ghrelin mrna transcript levels by h(gly 2 )GLP-2 Next, neurotensin and ghrelin mrna transcript regulation by h(gly 2 )GLP-2 was investigated. The adult hypothalamic neuronal cells were exposed to 10 nm h(gly 2 )GLP-2 over a 24 h time course. Using real-time qrt-pcr, it was found that in the adult mhypoa-2/30 neurons, neurotensin mrna levels were significantly up-regulated by 50% at 24h post-treatment [neurotensin mrna expression (fold change relative to -actin): vehicle, 1.02 ± 0.11 versus 10 nm h(gly 2 )GLP-2, 1.53 ± 0.17, P < 0.05] (Figure 29A). Similarly, it was observed that h(gly 2 )GLP-2 significantly increased ghrelin mrna levels by 95% at 24 h post-treatment

133 Figure 28. Acute h(gly 2 )GLP-2 treatment induces c-fos activation and CREB/ATF-1 phosphorylation in the hypothalamic GLP-2R-positive mhypoa-2/30 neuronal cells. The mhypoa-2/30 neuronal cells were serum starved overnight and then treated with h(gly 2 )GLP-2 (10 nm) or vehicle. Protein was isolated over 6 h at the indicated time points, resolved on 10% SDS-PAGE, transferred to PVDF membrane, and immune-blotted with antisera for c-fos, phospho-creb/atf-1, total CREB and Gβ (G-protein β subunit). h(gly 2 )GLP-2 significantly increased c-fos expression (A), and induced phosphorylation of CREB (B) and ATF-1 (C) in the mhypoa-2/30 cells. c-fos expression was normalized to Gβ, and phosphorylation of CREB and ATF-1 was normalized to total CREB. Representative Western blots are shown. All results shown in the bar graphs are expressed as mean + SEM (n = 4 independent experiments; *P < 0.05, **P < 0.01 vs. vehicle control). Statistical significance was calculated by two-way ANOVA. 116

134 Figure 29. Regulation of neurotensin (A, C) and ghrelin (B, D) mrna levels by h(gly 2 )GLP-2 in the mhypoa-2/30 neuronal cell line in protein kinase A (PKA)-dependent manner. Following overnight incubation with DMEM containing 0.5% FBS, the cells were exposed to h(gly 2 )GLP-2 (10 nm) or vehicle over a 24h time course and at the indicated time points RNA was isolated (A, B). For the use of PKA inhibitors, the neuronal cells were pretreated with PKA inhibitor PKI (1 mm) or H89 (5 mm) for 1h followed by either vehicle (PBS or DMSO) or h(gly 2 )GLP-2 (10 nm) treatment and the RNA was isolated at 24h time point (C, D). Total RNA was used as a template for real-time RT-PCR with primers specifically designed to amplify neurotensin or ghrelin mrna. Neurotensin (A, C) and Ghrelin (B, D) mrna levels were quantified using the standard curve method and normalized to the internal control (γ-actin). All results shown are relative to corresponding control mrna levels at each time point (A, B) or to PBS only-treated mrna levels (set to 1.0) (C, D), and are expressed as mean ± SEM (n = 4 independent experiments; *P < 0.05, **P < 0.01). White bars (C, D) represent vehicle (PBS or DMSO)-treated samples (with or without PKA inhibitor); black bars represent h(gly 2 )GLP-2-treated samples (with or without PKA inhibitor). 117

135 118 in the adult mhypoa-2/30 cell line [ghrelin mrna levels (change relative to -actin): vehicle, 0.80 ± 0.19 versus 10 nm h(gly 2 )GLP-2, 1.56 ± 0.28, P < 0.05] (Figure 29B). These results clearly indicate that neurotensin and ghrelin mrna levels are regulated by GLP-2R activation in the hypothalamic neuronal models, and complement the findings within the in vivo setting PKA inhibitors reverse the h(gly 2 )GLP-2-induced up-regulation of neurotensin and ghrelin mrna transcript levels Further, the role of PKA in the regulation of neurotensin and ghrelin mrna transcript levels by h(gly 2 )GLP-2 was determined. Pharmacological inhibitors of PKA (H-89 at 5 µm and PKI (14-22) amide at 1 µm concentration) were used to pre-treat the adult neuronal cells for 1 h before exposure to 10 nm h(gly 2 )GLP-2. PBS and DMSO were used as vehicle controls, as h(gly 2 )GLP-2 was dissolved in PBS and PKA inhibitors were dissolved in DMSO. Total RNA was isolated at 24 h after the h(gly 2 )GLP-2 treatment and analyzed using real-time qrt-pcr. As compared to the vehicle treatments, at 24 h, h(gly 2 )GLP-2 treatment significantly induced increase in the neurotensin mrna levels [PBS, 0.91 ± 0.02 versus PBS + 10 nm h(gly 2 )GLP-2, 1.61 ± 0.19, P < 0.05; DMSO, 0.68 ± 0.03 versus DMSO + 10 nm h(gly 2 )GLP-2, 1.12 ± 0.12, P < 0.05] (Figure 29C). Further, it was found that both PKA inhibitors did not affect the basal neurotensin mrna levels, but significantly attenuated the h(gly 2 )GLP-2-induced increase in neurotensin mrna levels in the mhypoa-2/30 neuronal cell model [PKI, 0.98 ± 0.13 versus PKI + 10 nm h(gly 2 )GLP-2, 0.69 ± 0.10, P < 0.05; H89, 0.69 ± 0.10 versus H nm h(gly 2 )GLP-2, 0.50 ± 0.04, P < 0.05] (Figure 29C). Next, the role of PKA in the ghrelin mrna upregulation caused by h(gly 2 )GLP-2 was investigated. At 24 h, h(gly 2 )GLP-2 significantly increased ghrelin mrna levels as compared to the vehicle treatment [PBS, 0.72 ± 0.10 versus PBS + 10 nm h(gly 2 )GLP-2, 1.42 ± 0.06, P < 0.05; DMSO, 0.72 ± 0.02 versus DMSO + 10 nm h(gly 2 )GLP-2, 1.09 ± 0.07, P < 0.05] (Figure 29D). Similar to the suppression of neurotensin mrna levels detected at 24 h, both PKA inhibitors attenuated the h(gly 2 )GLP-2-induced upregulation in ghrelin mrna levels [PKI, 0.98 ± 0.12 versus PKI + 10 nm h(gly 2 )GLP-2, 1.13 ± 0.15; H89, 0.82 ± 0.06 versus H nm h(gly 2 )GLP-2, 1.05 ± 0.16] (Figure 29D). Overall, these results indicate that PKA activation is involved with the regulation of neurotensin and ghrelin mrna by h(gly 2 )GLP-2.

136 Discussion The proglucagon system plays an important role in the central regulation of energy homeostasis (7, 333). Many studies have described the hypothalamic actions of GLP-1 on energy homeostasis. However, the potential central effects of GLP-2 still remain largely unknown. The hypothalamus is comprised of various nuclei, a subdivision of which are known to be involved in the regulation of feeding behavior. Besides the well-documented NPY and POMC neurons, two other major regulators of energy homeostasis, neurotensin and ghrelin neurons are also expressed within the ARC (56, 105, 106, 156). The neurotensin system in the hypothalamus is extensive, and strong neurotensin immunoreactivity is found in the ARC, PVN, DMH and LH, with low immunoreactivity in the VMH (56, 409). Ghrelin neurons are also located in the PVN, the perifornical region and the internuclear spaces between hypothalamic nuclei (105). At present, the neuronal phenotypes expressing GLP-2R in the DMH and VMH are not clear. It was found that h(gly 2 )GLP-2 remarkably activated main hypothalamic regions involved in regulation of energy homeostasis. Furthermore, it was detected that central h(gly 2 )GLP-2 differentially activated α-msh/pomc, NPY, neurotensin, and ghrelin neurons in a number of these regions. These findings indicate that GLP-2R activation involves complex interactions between several hypothalamic nuclei, activation of feeding-related neuropeptidergic neurons, and transcriptional regulation of downstream neuropeptides. Central h(gly 2 )GLP-2 administration activated several hypothalamic regions with a distinct pattern of c-fos expression in comparison with the long acting GLP-1R agonist, exendin- 4 (372). Not only the major nuclei that are located in the vicinity of the third ventricle were activated, but also the lateral hypothalamus was activated. As the DMH and VMH have been demonstrated to express GLP-2R, activation of c-fos in these areas suggests direct action of the h(gly 2 )GLP-2, however activation of ARC, PVN and LH could be due to indirect activation, as GLP-2R expression is not found in these regions (214, 218). Because these regions function together to regulate energy balance (410), their indirect activation by GLP-2R-expressing neurons is quite possible. Interestingly, except for the VMH, these hypothalamic regions highly express GLP-1R (217). Thus, there is a remote possibility of cross-reactivity of h(gly 2 )GLP-2 with GLP-1R at the pharmacological dose used in the present study. However, it has been demonstrated that h(gly 2 )GLP-2 activates only the GLP-2R-expressing cells, but not those

137 120 expressing GLP-1R (218). Moreover, the effects of h(gly 2 )GLP-2 on food intake in GLP-1R knockout mice were not attenuated by disruption of GLP-1R signaling, but rather potentiated, suggesting that GLP-2 does not mediate its effects on feeding through GLP-1R (218). However, based on the activation of several hypothalamic nuclei by pharmacological h(gly 2 )GLP-2, specificity of GLP-2 action in the hypothalamus needs to be investigated further using GLP-2R antagonists or GLP-2R knock-down strategy. The detection of significant activation of NPY and neurotensin neurons in the DMH and neurotensin neurons in the VMH coordinates well with the expression of GLP-2R in these areas. However, whether NPY or neurotensin neurons express GLP-2R in vivo is not known. The significant increase in the number c-fos-immunopositive α-msh/pomc, NPY, neurotensin and ghrelin neurons in those areas that do not express GLP-2R is intriguing. c-fos immunostaining may have arisen from activation of projections from the GLP-2R-expressing neurons of the DMH or VMH, or from activation of brainstem GLP-2R-expressing neurons by h(gly 2 )GLP-2 carried from the third to the fourth ventricle. These findings suggest that during the central GLP- 2R-induced anorexia not only anorexigenic neurons, but also orexigenic neurons are activated in multiple hypothalamic nuclei. Furthermore, activation of all these areas by h(gly 2 )GLP-2 implies a complex interaction between these nuclei and neurons that ultimately enhances the perception of satiety and thereby facilitates anorexia and decreased body weight. It is not quite clear whether endogenous GLP-2 or peripherally administered h(gly 2 )GLP-2 cross the BBB. Because GLP-2 and GLP-1 have nearly 50% sequence homology and are secreted in parallel from intestinal L-cells (411), and also because GLP-1, glucagon and oxyntomodulin cross the BBB (219, 412, 413), it is likely that GLP-2 can also cross the BBB. Therefore, as the long acting GLP-2, teduglutide, is currently undergoing clinical trials for short bowel syndrome, careful consideration should be given to the novel findings on the central action of GLP-2. It must be noted that although recent studies in humans found no effect of intravenous GLP-2 on food intake (414), it was found that GLP-2 decreases fat mass and increases lean mass in short-bowel patients (415). The present data show that pharmacological dose of h(gly 2 )GLP-2 decreases food and water intake, and induces body weight loss within 2 h post-i.c.v. administration. It is possible that the body weight loss could be entirely due to significant water intake suppression due to GLP-2 action on the central angiotensin II receptors

138 121 or vasopressin-expressing neurons that were not examined in the present study (416). On the other hand, the fact that GLP-2 reduces food intake yet increases the CNS ghrelin levels suggests a role for ghrelin signaling to decrease water intake (417). Further, it can be speculated that the weight loss was due to the stimulation of diuresis or defecation that also needs further attention. Apart from appetite regulation, other potential actions from the pharmacological activation of central GLP-2R have yet to be fully delineated. GLP-2 has been shown to reduce glutamate-induced cell death in cultured murine hippocampal and cortical cells (418). Similarly, ghrelin is also found to exert neuroprotective effects by its anti-inflammatory action in the CNS (383). Thus, ghrelin may potentially mediate anti-inflammatory action and thereby neuroprotective effects of central GLP-2R action (418). Further, GLP-2 has been shown to stimulate proliferation of cultured rat astrocytes (419). A synthetic neurotensin agonist was found to induce an increase in the proliferation of the astrocytic cell lines (420), whereas ghrelin was demonstrated to promote neurogenesis (421, 422). Whether neurotensin and ghrelin act as downstream effectors of GLP-2 action in cell proliferation is not known and requires further study. Furthermore, GLP-2-mediated regulation of murine hippocampal neurons by increasing glucose uptake implicates an important role for GLP-2 in neurotransmitter release and hormone secretion from GLP-2R-positive neurons and endocrine cells (423). As hypothalamic α- MSH/POMC, NPY and ghrelin neurons are implicated in glucose regulation (106, 107, 131), whether GLP-2-mediated activation plays any role in peripheral glucose metabolism needs to be explored further. This study focused exclusively on the action of GLP-2 in the hypothalamus, but extrahypothalamic action of GLP-2 also needs to be investigated. The direct action of GLP-2 on the regulation of hypothalamic neuropeptides remains unstudied due to the lack of endogenous GLP-2R-expressing neuronal cells. Thus, the adult mouse-derived hypothalamic mhypoa-2/30 neuronal cell model that expresses functional GLP- 2R was used in the current study. The hypothalamic cell model represents a clonal, homogeneous population of a single neuron, and provides a useful tool to perform gene regulation and mechanistic studies that are quite challenging to perform in vivo (19). GLP-2 stimulates camp production and activates PKA to exert its CNS actions (392, 418, 423). Therefore, the present study focused on the camp/pka pathway in the regulation of hypothalamic neurotensin and ghrelin mrna expression. Direct GLP-2R activation lead to an increase in intracellular camp,

139 122 and activation of transcription factors CREB and ATF-1. Furthermore, the observed increase in c-fos expression in these neuronal cells correlates well with the in vivo c-fos activation found in the GLP-2R-expressing DMH and VMH regions, suggesting that the DMH and VMH neurons could be directly activated by central h(gly 2 )GLP-2. Similarly, GLP-2 was found to up-regulate c-fos mrna in BHK-GLP-2R cells as well as in astrocytes to promote cellular proliferation (231, 419), again pointing to the involvement of GLP-2 in cell proliferation. Neurotensin and Ntsr are widely expressed in the CNS controlling a number of physiological functions, such as regulation of circadian rhythm, anti-psychotic action, analgesia, thermoregulation, regulation of HPA axis, and neuromodulation of dopamine neurotransmission. Particularly, hypothalamic neurotensin neurons are known to regulate feeding behavior (143). Ntsr deficiency moderately increases food intake and body weight, and blocks neurotensininduced anorexia in mice, implicating neurotensin-ntsr signaling pathway in feeding and body weight regulation (139). The observed increase in neurotensin transcript levels in the present study suggests that GLP-2 may activate neurotensin in favor of anorexigenic action quite similar to other satiety-inducing neuropeptides and hormones. Central ghrelin-expressing neurons are the main source of ghrelin in the CNS (105, 424). Apart from the peripheral ghrelin, it is reported that central ghrelin is also involved in the regulation of food intake and body weight (153, 155). In the hypothalamic neuronal cell model, a stimulatory effect on ghrelin mrna expression was observed. Recently, it was found that GLP- 1R activation by exendin-4 induced a decrease in the ghrelin mrna levels in agreement with the anorexigenic action of GLP-1R agonism (372). Also, insulin, a satiety factor and adiposity signal, was shown to directly inhibit ghrelin expression in another hypothalamic cell model (395). Similar to the exendin-4 or insulin action, negative regulation of ghrelin mrna levels by GLP-2 was expected due to its orexigenic characteristics. The positive regulation may suggest that ghrelin could play a role in the promotion of hypothalamic neuronal survival, as ghrelin is known to exert anti-inflammatory and neuroprotective effects in rat brain (383). Although, ghrelin and neurotensin exert opposite effects on food intake and energy homeostasis, they exert comparable actions in improving memory and learning, as is evident from decreased ghrelin and neurotensin mrna transcript expression in the brains of patients suffering from Alzheimer's

140 123 disease (425). Thus, it is possible that GLP-2-activated hypothalamic ghrelin and also neurotensin neurons may potentially exert other functions unrelated to appetite regulation. The camp/pka pathway has been demonstrated to play an important role in appetite regulation. Hypothalamic activation of camp/pka mediates central regulation of satiety by inhibiting NPY-induced feeding (336). Recently, it was shown that exendin-4 activated hindbrain GLP-1R to induce anorexia in a PKA-dependent manner (327). Furthermore, camp/pka/creb signaling upregulates transcription of CART, a potent appetite-suppressing peptide (394), but PKA and CREB have been shown to negatively regulate NPY gene expression (337). Thus, it appears that stimulation of camp/pka activity inhibits feeding behavior by increasing anorexigenic and decreasing orexigenic neuropeptide expression. This agrees with the finding that the h(gly 2 )GLP-2 induced anorexigenic neurotensin mrna transcript levels via PKA activation, but does not correlate with the increase in the orexigenic ghrelin mrna transcript levels. In fact, at present, the exact role of ghrelin expression in the hypothalamus is not clear. In addition to mature ghrelin, the ghrelin gene also encodes an entirely different peptide, obestatin (149). Although the physiological relevance of obestatin remains unclear, initially it was found to exert anorectic effect that was subsequently disputed (149, 150). Further, circulating ghrelin exists in two forms, acylated ghrelin and des-acyl ghrelin. Acylated ghrelin acts as an orexigenic peptide to increase food intake, whereas des-acyl ghrelin, although its function is not quite clear, has been shown to decrease food intake and gastric emptying (426). Thus, further investigation is needed to ascertain the correlation between GLP-2-mediated PKA-dependent stimulation of ghrelin mrna transcript levels and post-translational processing of the hypothalamic precursor polypeptide proghrelin. Apart from camp/pka pathway, regulation of hypothalamic neurotensin and ghrelin gene expression can occur through alternative pathways. Neurotensin gene expression may be augmented by leptin via activation of STAT3 and MAPK ERK1/2 (300), whereas ghrelin gene expression can be repressed by insulin via both PI3-K/Akt and MAPK ERK1/2 pathways (395). In a similar fashion, GLP-2 has been shown to activate other pathways to mediate intestinal cell survival and proliferative actions i.e. Akt in the intestinal epithelium (232), and the wnt/β-catenin signaling pathway in the intestinal crypt cells through an indirect mechanism requiring IGF- 1R/IGF-1 signaling (233). As well, h(gly 2 )GLP-2 was found to stimulate IGF-1 mrna through

141 124 PI3K/Akt pathway in intestinal subepithelial fibroblasts (408). Likewise, whether GLP-2R activates MAPK ERK1/2 and PI3K/Akt pathways in hypothalamic neurons remains to be investigated. It is normally observed that the activation of PKA or CREB induces gene transcription. Therefore, the h(gly 2 )GLP-2-induced increase in neurotensin and ghrelin mrna transcript levels could suggest stimulation of 5 regulatory promoter elements, such as CRE, to induce gene transcription. Previous neurotensin gene promoter analysis study indicates that AP-1 site and CRE element are involved in hormonal regulation of this gene (396). Recently, it was demonstrated that neurotensin gene expression was induced by leptin via activation of transcription factors ATF-1 and c-fos (300). However, it is also possible that the increase in neurotensin mrna could be due to an increase in mrna stability. Contrary to neurotensin gene promoter region, mouse ghrelin promoter region is not well characterized; whether it contains regulatory AP-1 site and CRE element remains unknown. Although human ghrelin promoter does not have AP-1 site or CRE element, glucagon and its second messenger camp enhanced the ghrelin promoter activity, suggesting that ghrelin gene transcription may be regulated by some cell-specific transcription factors and cofactors to integrate camp stimulation to activate ghrelin gene transcription (427). Thus, further studies are required to determine whether GLP-2 regulates hypothalamic neurotensin and ghrelin mrna transcript levels through the AP-1 or CRE consensus sites or through other cis-elements within the promoter or enhancer region of these genes, or via specific molecular components induced by PKA to increase mrna stability. In summary, the results of the present study demonstrate that pharmacological action of central GLP-2 results in activation of multiple hypothalamic regions to trigger complex interactions between appetite-regulating neuropeptidergic neurons to induce transient anorexigenic effect. Using a novel GLP-2R-positive hypothalamic cell line, it was found that h(gly 2 )GLP-2, in a PKA-dependent manner, directly regulates mrna transcript levels of feeding-related neuropeptides neurotensin and ghrelin. Overall, these findings suggest that central GLP-2R may serve as a potential target for pharmacological intervention to modulate neuropeptides involved in appetite regulation.

142 125 Chapter 6 Discussion

143 126 6 Discussion 6.1 Overall conclusions The PGDPs are generated from a single common proglucagon precursor expressed in pancreatic islet α-cells, gut enteroendocrine L-cells, and in selective brainstem and hypothalamic neurons. Until now, the vast majority of research has focused only on the regulation of peripheral proglucagon gene expression; therefore, much remains unknown about the regulation of hypothalamic proglucagon gene expression. Specifically, the mechanisms by which two major hormones, insulin and leptin, regulate proglucagon-expressing hypothalamic neurons have not been elucidated. For over last two decades, a wealth of data has widened our understanding of the role of PGDPs in the periphery and the CNS. The GLPs serve important roles in the control of energy balance, gastro-intestinal motility, nutrient absorption and glucose homeostasis. Particularly, GLP-1 has been well established as a central regulator of physiological functions ranging from energy homeostasis, fluid homeostasis, memory and neuronal regeneration. However, we are still far from having clear knowledge of potential central actions of other PGPDs including GLP-2. Although PGDPs have emerged as potential regulators of feeding behavior (1), the exact mechanisms of action of GLP-1R and GLP-2R stimulation in the hypothalamus are not completely clear; particularly how GLP-1R and GLP-2R stimulation regulates hypothalamic appetite-regulating neuropeptides has not been investigated. In this thesis, the hypothalamic control mechanisms at the level of proglucagon-specific neurons were defined; specifically, the molecular mechanisms utilized by insulin and leptin to regulate expression of proglucagon in the hypothalamic neurons were studied using novel neuronal cell models. Further, the actions of long-acting GLP-1R and GLP-2R agonists on mrna transcript levels of appetite-regulating hypothalamic neuropeptides that have remained elusive until now were investigated (372, 428). Central actions of insulin and leptin are critical in the control of appetite regulation and maintenance of energy homeostasis. Insulin and leptin activate catabolic POMC neurons (36, 335), while inhibiting anabolic NPY/AgRP neurons in the hypothalamus (429, 430). Apart from these neuronal circuits that have been well established as critical appetite regulators, PGDPs also play important roles in regulation of feeding behavior (6). Similar to insulin and leptin, glucagon

144 127 plays an important role in energy homeostasis, and its central administration suppresses food intake in rats and chicks (431, 432). The available data suggest that centrally administered GLP- 1 acts on the hypothalamus to regulate food intake and body weight (26, 209, 210). Although GLP-2 has intestinotrophic activity in rodents, it also functions as a hormonal ileal brake since it was demonstrated that GLP-2 dose-dependently inhibited centrally-induced antral motility in pigs (204). Further, central administration of GLP-2 is involved in regulation of food intake in rats and mice (214, 218). Acute central or peripheral administration of another PGDP, oxyntomodulin, inhibits food intake in rodents, whereas chronic administration reduces body weight (205, 433). Thus, it is evident that exogenous PGDPs act as central regulators of appetite, and hence it is imperative to study the regulation of hypothalamic proglucagon-expressing neurons and action of hypothalamic PGDPs. At present, it is not known whether central, peripheral, or a combination of both sources of PGDPs signal hypothalamic nuclei to regulate energy homeostasis. For example, it is possible that due to its very short plasma half-life, gutderived GLP-1 may not even reach the hypothalamus or other parts of the central nervous system. Therefore, only centrally-derived GLP-1 may be functional in these central regions or peripheral GLP-1 may act only on the circumventricular regions of the brain which have an incomplete BBB, such as ME or area postrema. The ambiguity and lack of knowledge is mostly due to inaccessibility to the hypothalamic proglucagon-, GLP-1R- or -2R-expressing neurons. Previous studies on the regulation of hypothalamic PGDPs were conducted on fetal rat hypothalamic primary cell cultures (183, 202); however, these cultures are quite challenging to generate and maintain. In order to circumvent this technical issue, the present research used phenotypically distinct cell lines generated from embryonic and adult mouse hypothalamii that endogenously express proglucagon mrna, insulin and leptin receptors, and receptor signaling proteins (18, 19). At present, it is unknown if insulin and leptin have any direct action on the hypothalamic proglucagon neurons, and the regulatory mechanisms utilized by these molecules remain unstudied. This thesis research characterized the signaling pathways and changes in proglucagon mrna levels that occur as a direct result of insulin and leptin administration to adult and embryonic hypothalamic, clonal cell lines (Figure 30). A new evidence that insulin and leptin can act directly upon proglucagon neurons to regulate proglucagon mrna levels was demonstrated. It was found that insulin activates an Akt-dependent pathway and leptin triggers

145 Figure 30. Summary of the mechanisms activated by insulin and leptin to regulate proglucagon mrna transcript levels in the mhypoa-2/10 and mhypoe-39 neuronal cells. Exposure of insulin and leptin leads to the activation of classic signal transduction pathways in the novel hypothalamic proglucagon-expressing neurons: induction of Akt by insulin and JAK2/STAT3 by leptin to regulate proglucagon mrna levels. Insulin and leptin regulate proglucagon mrna stability through unidentified mechanisms and/or may trigger yet unknown transcriptional mechanisms to modulate proglucagon mrna levels in the hypothalamic neuronal cells. TF, transcription factor; upward pointing solid green arrows represent upregulation and downward pointing red arrows represent downregulation of mrna levels or stability; solid blue arrows represent interactions by known mechanisms, whereas dashed blue arrows and question marks represent yet to be identified mechanisms. 128

146 129 activation of STAT3 to regulate hypothalamic proglucagon mrna levels in adult and embryonic cell models models. It was further determined that insulin and leptin do not regulate activity of the transfected human or rat proglucagon promoter reporter constructs in the mouse embryonic cells, but rather affect proglucagon mrna stability. Overall, these findings suggest that insulin and leptin can act directly on specific hypothalamic neurons to regulate proglucagon mrna levels. A better understanding of the mechanisms through which insulin and leptin regulate hypothalamic proglucagon neurons will further enable us to understand roles of PGDPs in energy homeostasis. GLP-1R activation by exendin-4 has been shown to regulate food intake and energy expenditure (371). The widespread clinical use of exendin-4 as an anti-diabetic drug warrants further research on the mechanisms underlying the anorectic actions of exendin-4 within the hypothalamus (7). The present study identified hypothalamic neuropeptidergic neurons responsive to central GLP-1R activation in vivo, and elucidated the direct action of exendin-4 on neurotensin and ghrelin mrna transcript regulation in vitro (Figure 31). It was found that exendin-4 activated hypothalamic ARC, PVN, DMH and PeV, regions that widely express GLP- 1R along with several neuropeptides involved in energy metabolism. Furthermore, it was detected that central exendin-4 significantly activated α-msh/pomc and NPY neurons in the ARC, neurotensin-expressing neurons in the PVN, and ghrelin-expressing neurons in the ARC, PVN and PeV. Overall, the in vivo findings suggest that complex interactions may occur between satiety- and hunger-related neuropeptides in one or more hypothalamic nuclei to mediate the overall anorexic action of exendin-4. Finally, using the hypothalamic neuronal cell models, it was found that exendin-4, in a PKA-dependent manner, increased neurotensin mrna levels while attenuating ghrelin mrna levels. These in vitro findings complement the in vivo findings and suggest that regulation of hypothalamic neurotensin and ghrelin may lie downstream of exendin-4 GLP-1R activation. Given that central exendin-4 induces net inhibition of both drinking and feeding that results in significant weight loss, it is likely that hypothalamic anorexigenic neurotensin neurons are stimulated and orexigenic ghrelin neurons are inhibited in vivo. Ultimately, this study provides a previously unrecognized link between exendin-4 action and neurotensin and ghrelin regulation.

147 Figure 31. Summary of the mechanisms activated by exendin-4 to regulate neurotensin and ghrelin mrna transcript levels in the mhypoa-2/30 and mhypoe-36/1 neuronal cells. Exendin-4 activates classic camp-pka signal transduction pathway in the novel hypothalamic GLP-1R-expressing neuronal cells to regulate neurotensin and ghrelin mrna levels. Exendin-4 may trigger yet unknown transcriptional mechanisms and/or modify mrna stability through unidentified mechanisms to regulate neurotensin and ghrelin mrna levels in the hypothalamic neuronal cells. NT, neurotensin; upward pointing solid green, yellow and black arrows represent upregulation and downward pointing red arrows represent downregulation of mrna levels or stability; solid blue arrows represent interactions by known mechanisms, whereas dashed blue arrows and question marks represent yet to be identified mechanisms. 130