The Effects of Saturated Fatty Acid Palmitate on Neuropeptide Gene. Expression, Signal Transduction, and Insulin Signaling in an Immortalized

Transcription

1 The Effects of Saturated Fatty Acid Palmitate on Neuropeptide Gene Expression, Signal Transduction, and Insulin Signaling in an Immortalized Hypothalamic Neuronal Cell Model, mhypoa-npy/gfp By Brian Wong A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Physiology University of Toronto Copyright by Brian Wong 2015

2 The Effects of Saturated Fatty Acid Palmitate on Neuropeptide Gene Expression, Signal Transduction, and Insulin Signaling in an Immortalized Abstract Hypothalamic Neuronal Cell Model, mhypoa-npy/gfp Brian Wong Master of Science Department of Physiology University of Toronto 2015 Recent evidence suggests a role for hypothalamic insulin resistance in obesity pathogenesis, and that obesity-associated hypothalamic inflammation underlies this resistance. However, few studies have examined the direct effects of saturated fatty acids on specific hypothalamic neurons. Therefore, an immortalized hypothalamic neuronal cell model expressing NPY and AgRP was used to determine the effects of palmitate on neuropeptide gene expression, signal transduction events and insulin signaling. In the mhypoa-npy/gfp neuronal cell model, palmitate was found to upregulate the expression of NF-κB and IκBα within 4 hours of treatment, and upregulate expression of AgRP after 24 hours of treatment. Regulation of AgRP gene expression appeared to be palmitate metabolism-dependent. Palmitate also induced p38 MAPK phosphorylation, and prolonged palmitate pre-treatment decreased levels of phosphorylated Akt following insulin rechallenge. This is the first evidence of palmitate-mediated changes in AgRP gene expression and its signaling through p38 MAPK in a representative NPY/AgRP neuronal cell model. ii

3 Acknowledgements First and foremost, I owe my deepest gratitude to Dr. Denise Belsham. I am grateful to have had you as my mentor and supervisor. The last two years in the laboratory have been an incredible learning experience. Your continual guidance, support, and encouragement have enabled me to grow as a person, and have prepared me for the road ahead. Thank you Denise. I would also like to thank my committee members, Dr. Michael Wheeler, Dr. Amira Klip, and Dr. Adria Giacca. Your guidance, mentorship, and valuable insight have been crucial to the completion of this degree. I would like to thank my fellow lab mates, who made the laboratory an enjoyable place to work in. Whether it was generating discussion at lab meetings or simply lending a helping hand with a new experiment, you have all contributed to this experience. Finally, I would like to thank my family for their unwavering love and encouragement. The sacrifices you have made and the many opportunities you have provided me with do not go unnoticed. I would not be where I am today without you. iii

4 Table of Contents Acknowledgements... iii Table of Contents... iv List of Tables and Figures... viii List of Abbreviations... ix Chapter 1 Introduction 1.1 Preface The Central Melanocortin System and Energy Homeostasis Adiposity Negative Feedback Insulin as an Adiposity Signal The Hypothalamus The Role of Neuropeptides in the Brain The Melanocortin System Neuropeptide Y and Agouti-Related Peptide Glucose Sensing in Hypothalamic Neurons Glucose Entry into the Brain and the Discovery of Glucose Sensing Neurons Glucose-Excited Neurons Glucose-Inhibited Neurons Fatty Acid Sensing in Hypothalamic Neurons Fatty Acids and Metabolic Physiology Blood-Brain Barrier Permeability, Fatty Acid Uptake and Metabolic Fates...13 iv

5 1.4.3 Fatty Acids as Signaling Molecules in the Hypothalamus Obesity and Hypothalamic Inflammation High-Fat Feeding is Associated with Inflammation Evidence that Hypothalamic Inflammation Contributes to HFD-Induced Obesity Palmitate as a Mediator of Hypothalamic Inflammation Obesity and Hypothalamic Insulin Resistance Obesity, Diabetes and Insulin Resistance Hyperinsulinemia and the Development of Insulin Resistance High Fat Feeding and Neuronal Insulin Resistance FFA Metabolism and Insulin Resistance Cell Model The Need for Cell Lines Adult Hypothalamic Cell Lines (mhypoa-xx) mhypoa-npy/gfp Cell Line Hypothesis and Aims...26 Chapter 2 Materials and Methods 2.1 Cell Culture and Reagents Palmitate Preparation TNF-α Preparation Insulin Preparation Quantitative RT-PCR Western Blot Analysis...34 v

6 2.7 Statistical Analysis...35 Chapter 3 Results 3.1 Palmitate elicits an inflammatory response and upregulates Agrp gene expression in mhypoa-npy/gfp neurons TNF-α, a pro-inflammatory surrogate of palmitate, also upregulates AgRP gene expression in mhypoa-npy/gfp neurons Palmitate-mediated regulation of AgRP gene expression is metabolism-dependent Palmitate triggers the phosphorylation of p38 MAPK in mhypoa-npy/gfp neurons Palmitate pre-treatment dampens the mhypoa-npy/gfp neurons response to insulin...43 Chapter 4 Discussion 4.1 General Discussion Plasma Non-Esterified Fatty Acids and the Determination of Palmitic Acid Concentrations in the Brain Transcriptional Effects of Palmitate on mhypoa-npy/gfp Neurons Transcriptional Effects of TNF-α on mhypoa-npy/gfp Neurons Palmitate Metabolism and Signaling Dynamics in mhypoa-npy/gfp neurons Palmitate Impairs Insulin Signaling in mhypoa-npy/gfp Neurons Limitations Future Directions...63 vi

7 4.9 Conclusion...64 References...66 vii

8 List of Tables and Figures Table 1.1 Characterization of the mhypoa-npy/gfp cell line...30 Fig. 1.1 Schematic illustrating the PI3K-Akt pathway...5 Fig. 1.2 NPY/AgRP and POMC neurons are directly regulated by insulin...9 Fig. 1.3 Metabolic fates of palmitate upon entering the cell...14 Fig. 1.4 Activation of the IKKβ/NF-κB pathway leads to inflammation and impaired insulin signaling...18 Fig. 1.5 Mechanisms of palmitate-mediated inhibition of insulin signaling...22 Fig. 1.6 Generation of the mhypoa-npy/gfp cell line...27 Fig. 3.1 Saturated fatty acid palmitate upregulates pro-inflammatory and Agrp gene expression...39 Fig. 3.2 TNF-α upregulates pro-inflammatory and Agrp gene expression...41 Fig. 3.3 Methyl palmitate does not regulate Agrp gene expression...42 Fig. 3.4 Palmitate induces phosphorylation of p38 MAPK...44 Fig.3.5 Prolonged palmitate or insulin exposure dampens the insulin-mediated increase in phospho-akt...46 viii

9 Abbreviations AgRP ARC β-oxidation BBB cdna CNS CNTF CPT-1 CREB DAG DIO DMEM DMN DNA agouti-related peptide arcuate nucleus beta oxidation blood-brain barrier complementary deoxyribonucleic acid central nervous system ciliary neurotrophic factor carnitine palmitoyltransferase-1 camp response element binding protein diacylglycerol diet-induced obesity Dulbecco s modified eagle medium dorsomedial nucleus deoxyribonucleic acid eif2 eukaryotic initiation factor 2 ELK-1 ER ERK FABP FATP FBS FFA ETS domain-containing protein-1 endoplasmic reticulum extracellular-related kinase fatty acid binding protein fatty acid transport protein fetal bovine serum free fatty acid ix

10 FOXO1 forkhead box protein 01 GLUT4 glucose transporter type 4 GPAT HFD ICC ICV IκBα IKK-β IL-1β IL-6 IR IRS JNK LHA LPL MAPK MC3/4R MPO mrna mirna α-msh NEFA NF-κB glycerol-3 phosphate acyltransferase high-fat diet immunocytochemistry intracerebroventricular inhibitor of nuclear factor kappa B alpha inhibitor of IkappaB kinase beta interleukin-1 beta interleukin-6 insulin receptor insulin receptor substrate c-jun N-terminal kinase lateral hypothalamic area lipoprotein lipase mitogen-activated protein kinase melanocortin 3/4 receptor medial preoptic area messenger ribonucleic acid microrna alpha-melanocyte stimulating hormone non-esterified fatty acid nuclear factor kappa B x

11 NPY NTS PBS PCR PFA PI3K PKB PKC POMC PTP1B PVN qrt-pcr RNA sirna SPT STAT neuropeptide Y nucleus of the solitary tract phosphate buffer saline polymerase chain reaction perifornical area phosphatidylinositol 3-kinase protein kinase B protein kinase C proopiomelanocortin protein tyrosine phosphatase 1 B paraventricular nucleus quantitative reverse transcriptase polymerase chain reaction ribonucleic acid small interfering RNA serine palmitoyltransferase signal transducer and activator of transcription SV40 simian virus 40 T-Ag TG TAG TLR TNF-α T2DM T-antigen triglyceride triacylglycerol toll-like receptor tumor necrosis factor-alpha type 2 diabetes mellitus xi

12 1 Chapter 1 Introduction

13 2 Introduction 1.1 Preface Most overweight or obese individuals develop hyperlipidemia, low-grade inflammation, and insulin resistance often leading to type 2 diabetes mellitus (T2DM). While obesity and T2DM may both originate from a primary hypothalamic disease, little is known about how specific neurons within the hypothalamus sense and respond to nutrient (particularly fat) excess. The Belsham laboratory has generated several immortalized, hypothalamic neuronal cell lines from primary fetal and adult hypothalamic neuronal cell cultures, which have already provided insight into the direct control of neuropeptide synthesis by nutrients at a mechanistic level not practical in the whole brain. The purpose of this thesis was to evaluate the effects of palmitate (the most abundant non-esterified saturated fatty acid) on neuropeptide gene expression, signal transduction events and insulin signaling in an immortalized, hypothalamic neuronal cell model representative of the NPY/AgRP neuron. Using the mhypoa-npy/gfp cell line, I provide evidence of palmitate-mediated changes in AgRP gene expression and the dependency of such changes on palmitate metabolism. In addition, these studies begin to elucidate palmitate-mediated signal transduction events and provide further evidence of palmitate s ability to impair insulin signaling in distinct hypothalamic neurons. Taken together, these studies have direct relevance to the molecular mechanisms involved in the overall development of complex metabolic disorders, such as obesity. 1.2 The Central Melanocortin System and Energy Homeostasis Adiposity Negative Feedback

14 3 Despite daily variations in energy intake, the body fuel stored in adipose tissue remains relatively constant over time (1). This observation suggests that short-term differences in energy balance (the difference between energy consumed and energy expended) may be offset in the long term by a mechanism that maintains overall energy homeostasis. Indeed, changes in body fat content through dieting (2), behavior modification (3) or experimental over-feeding (4) have been shown to induce compensatory responses that restore adiposity to homeostatic levels. To explain this phenomenon, Kennedy proposed that inhibitory signals were generated in proportion to body fat stores and acted in the brain to reduce food intake (5). Weight loss reduced the plasma levels of these inhibitory signals, causing food intake to increase until body fat stores returned to normal levels (6) Insulin as an Adiposity Signal Insulin, a peptide hormone produced by the pancreatic β-cells, was the first hormonal signal implicated in the central nervous system control of energy homeostasis (7). It provides information regarding the amount of body fat stored and causes a long-term catabolic response, decreasing food intake and increasing energy expenditure (8). Insulin is secreted acutely in response to increases in blood glucose (i.e. after consumption of a meal) and its levels are directly correlated to the extent of body adiposity (9). As in peripheral tissues, insulin binds to its cognate receptor in the CNS. The receptor belongs to the family of tyrosine kinase receptors, and binding of insulin to its receptor triggers an intracellular signaling cascade (10).

15 4 Binding of insulin leads to rapid autophosphorylation of its receptor, followed by tyrosine phosphorylation and recruitment of insulin receptor substrate (IRS) proteins. This leads to activation of downstream pathways such as the phosphatidylinositol 3 kinase (PI3K) and the mitogen-activated protein kinase (MAPK) cascades (11). Activation of PI3K results in activation of protein kinase B/Akt and subsequent phosphorylation of the transcription factor FOXO, which is a critical downstream regulator of energy homeostasis in the CNS (Figure 1.1) (12). The hypothalamus contains the highest concentration of insulin receptors (IR) in the central nervous system. However, IRs are also expressed in the olfactory bulb, cerebral cortex, cerebellum and hippocampus (13, 14). Neurons within the hypothalamus are capable of sensing circulating insulin because of their location near the third ventricle, where insulin can enter via a saturable transporter across the blood-brain barrier (15) The Hypothalamus The hypothalamus is a key brain region controlling energy homeostasis. Histological techniques reveal nuclei as clusters of neurons within the hypothalamus that have distinct neuronal phenotypes. These neurons express a specific complement of neuropeptides, neurotransmitters and receptors. Classical lesion studies have shown that some of these hypothalamic nuclei act as discrete feeding and satiety centres (16). Lesions of the ventromedial, paraventricular or dorsal medial hypothalamus lead to hyperphagia, while lesions of the lateral hypothalamus lead to hypophagia (17). Besides regulating energy homeostasis, the hypothalamus is also the control centre for many other endocrine processes. Physiological processes that are under hypothalamic control include: stress, growth, temperature regulation, water balance, sexual behavior and

16 5 Insulin Cell Membrane Insulin Receptor IRS-1 PI3K PIP3 PDK1 Nucleus AKT P FoxO1 P FoxO1 P POMC AgRP Fig. 1.1 Schematic illustrating the PI3K-Akt pathway. Insulin binds to the insulin receptor, which is a receptor tyrosine kinase that autophosphorylates itself. This allows IRS proteins to dock. IRS proteins are then activated, and can recruit PI3K which phosphorylates PIP2 to PIP3. PIP3 acts as a docking site for PDK1 and AKT, allowing for the phosphorylation of Akt by PDK1. Phosphorylation of Akt leads to its nuclear translocation where it phosphorylates FoxO1 transcription factor. Phosphorylated FoxO1 can no longer repress POMC expression and stimulate AgRP gene expression, which results in decreased feeding.

17 6 reproduction, and circadian rhythms. Situated below the thalamus, posterior to the optic chiasm and surrounding the third ventricle, the hypothalamus has access to circulating factors that cross the blood-brain barrier (BBB) via diffusion or saturable transport mechanisms The Role of Neuropeptides in the Brain Over 70 genes in the mammalian genome encode for neuropeptides (16). Neuropeptides are peptide molecules synthesized by neurons, are released in a regulated manner and act on receptors present on other neurons. Compared to some classical neurotransmitters, such as epinephrine, neuropeptides are large with nanomolar affinities for their receptors. Neuropeptides can also diffuse over larger distances within the CNS than some classical neurotransmitters. Indeed, neurotransmitters like glutamate have been shown to have extrasynaptic effects. However, they are more likely to travel only as far as their nearest neighbouring synapse. Glutamate, in particular, has been found to travel distances of less than half a micrometer. In contrast, oxytocin released from neurons in the supraoptic nucleus of the hypothalamus results in biologically relevant concentrations throughout the anterior hypothalamus. In having high receptor binding affinity and the ability to affect distant populations of neurons, neuropeptide release can mediate changes in neuronal activity across multiple brain regions (17). A growing number of neuropeptides and neurotransmitters have been implicated in the regulation of feeding behavior in vivo. These neuropeptides are expressed in distinct neuronal populations located in specific regions of the hypothalamus, including the arcuate, paraventricular, and ventromedial nuclei (18).

18 The Melanocortin System The melanocortin system is central to the neuronal control of energy homeostasis. Here, the arcuate nucleus (ARC) of the hypothalamus is particularly important (19). Neurons within the ARC are strategically located close to fenestrated capillaries at the base of the hypothalamus such that they have access to circulating humoral signals (20). These neurons are controlled by neurotransmitters that are released from neighbouring axons, express receptors for metabolic hormones (20) and respond rapidly to nutritional cues (21). At present, the mammalian central melanocortin system is defined as a collection of CNS circuits that include: ARC neurons expressing hypothalamic neuropeptide Y (NPY) and agouti-related peptide (AgRP) or proopiomelanocortin (POMC), brainstem POMC neurons within the nucleus of the solitary tract (NTS) and downstream targets of these POMC and AgRP neurons which express melanocortin 3 (MC3R) and melanocortin 4 (MC4R) receptors (22) Neuropeptide Y and Agouti-Related Peptide Neuropeptides involved in food intake can be grouped into one of two categories: orexigenic (appetite-stimulating) or anorexigenic (appetite-suppressing). The main orexigenic neuron in the ARC is the NPY/AgRP neuron. Neuropeptide Y is a 36 amino acid peptide that is expressed throughout the central nervous system (23), and has notably high expression in the ARC (24). Agouti-related peptide is a 132 amino acid peptide that, unlike NPY, is only found in the ARC. ARC NPY/AgRP neurons project to nearby hypothalamic areas such as the paraventricular nucleus (PVN), dorsomedial nucleus (DMN), perifornical

19 8 area (PFA), lateral hypothalamic area (LHA) and the medial preoptic area (MPO), which are integrative centers for the regulation of both feeding and energy expenditure (25). NPY acts at multiple sites to increase food intake. Locally, NPY released from the ARC acts to inhibit neighbouring POMC neurons by activation of Y 1 and Y 2 receptors (26). NPY also acts on neurons in the PVN to stimulate food intake, and this effect appears to be mediated by both Y 1 and Y 5 receptors (27, 28). However, unlike NPY, AgRP acts to increase food intake by acting as an endogenous antagonist to the melanocortin 3 and 4 receptors (31). This prevents the constitutive activity of these receptors (32), resulting in an inhibition of the anorexigenic melanocortin pathway and an increase in food intake (Figure 1.2). Insulin, among other hormones, regulates feeding and energy balance by modulating the expression of these hypothalamic neuropeptides. Insulin may have anorexigenic effects by increasing Pomc and decreasing Agrp gene expression (33), and this effect is mediated by the phosphorylation of forkhead transcription factor 1 (FOXO1). FOXO1 is a transcription factor that represses POMC gene expression and stimulates Agrp gene expression. Thus, insulin-mediated phosphorylation of FOXO1 leads to its export from the nucleus which relieves the repression on the POMC promoter (34). Concomitantly, FOXO1-induced expression of Agrp in NPY/AgRP neurons is inhibited (35). The importance of NPY and AgRP in the regulation of food intake and energy homeostasis has been well documented. Central administration of either NPY (36) or AgRP (29) increases food intake and body weight, and chronic administration results in obesity. A single dose of AgRP results in an increase in food intake that is sustained for 7 days, indicating its potency as an orexigenic neuropeptide (37). Inhibiting AgRP with arcuate-

20 9 AgRP = agouti-related peptide MC3R = melanocortin 3 receptor MC4R = melanocortin 4 receptor α-msh = alpha-melanocyte stimulating hormone NPY = neuropeptide Y Y1 Receptor = neuropeptide Y Y1 receptor Figure 1.2 NPY/AgRP and POMC neurons are directly regulated by insulin. A representative diagram of the NPY/AgRP and POMC neurons, and how insulin regulates these neurons. Insulin exerts its anorexigenic effects by increasing Pomc and decreasing AgRP gene expression. The overall effect is a reduction in food intake and an increase in energy expenditure.

21 10 specific sirna leads to decreases in both food intake and body weight (38). Furthermore, ablation of these neurons in adult mice leads to extreme starvation (39, 40). 1.3 Glucose Sensing in Hypothalamic Neurons Glucose Entry into the Brain and the Discovery of Glucose Sensing Neurons Glucose is the primary energy substrate of the brain, and glucose metabolism accounts for the majority of brain oxygen consumption. Stereospecific, but insulin-independent, GLUT-1 glucose transporters are highly expressed in brain capillary endothelial cells of the blood brain barrier. GLUT-1 mediates the facilitated diffusion of glucose through the blood-brain barrier, and can transport two to three times more glucose than is actually metabolized in the brain (139). The stereospecificity of the GLUT-1 transporter allows D-glucose, but not L- glucose, to pass into the brain. Brain glucose varies depending on blood glucose, and declines to approximately 0.7 mm after an overnight fast. During peripheral hypoglycemia, hypothalamic glucose concentrations have been shown to fall to as low as 0.3 mm (140). These and other studies have indicated that hypothalamic glucose levels may range anywhere from 0.2 to 4.5 mm as blood glucose levels vary from pathological hypoglycemia to hyperglycemia. In 1964, two independent groups suggested the existence of glucose sensing neurons (141, 142). In these studies, reciprocal changes in activity were measured in the ventromedial hypothalamus (VMH) and lateral hypothalamus (regions referred to as the satiety and feeding centers of the brain, respectively) following intravenous glucose or insulin injections. In the VMH, glucose increased neuronal activity. In the lateral hypothalamus, however, the opposite occurred. Later, Oomura et al. demonstrated that hypothalamic

22 11 neurons were directly regulated by glucose in vitro. This finding led to the terms glucose responsive for neurons that increased their activity in response to increased glucose, and glucose sensitive for neurons that decreased their activity in response to increased glucose. Today, glucose sensing neurons are more commonly referred to as either glucose-exicted (GE) or glucose-inhibited (GI) based on their physiological response to changes in extracellular glucose (143) Glucose-Excited Neurons The expression of glucokinase (GK) (144) and ATP-sensitive potassium (K ATP ) channels composed of Kir6.2 and SUR1 subunits (145) has led to the idea that GE neurons sense changes in extracellular glucose concentrations via a mechanism that is similar to that which operates in pancreatic β-cells. In this proposed model, increased glucose concentrations are detected primarily through increased oxidation of glucose and generation of ATP. The subsequent changes in electrical activity are mediated by closure of K ATP channels. Studies using transgenic POMC-green fluorescent protein (GFP) mice have shown that ARC POMC neurons exhibit typical GE responses and express the K ATP channel (146). However, recent studies have suggested an additional population of GE neurons that sense glucose independently of changes in K ATP channel activity. A K ATP channelindependent glucose-sensing mechanism has been identified in a population of ARC GE neurons, which is believed to involve cellular depolarization from the opening of a nonspecific cation channel in response to elevated glucose concentrations (147).

23 Glucose-Inhibited Neurons In contrast to GE neurons, changes in AMP-activated protein kinase (AMPK) activity are likely to mediate the inhibitory effects of glucose on ARC hypothalamic GI neurons (148). AMPK is an evolutionarily conserved enzyme that acts as an intracellular energy sensor to regulate fuel availability within a cell. AMPK is a heterotrimeric protein that becomes activated allosterically by an increase in the intracellular AMP/ATP ratio. It has been proposed that at low glucose concentrations, the rate of glucose uptake through GLUT3 and metabolism through GK and the glycolytic pathway are low. The resulting increase in the AMP:ATP ratio would lead to activation of AMPK, which may then act to directly phosphorylate and inactivate different ion channels leading to cellular depolarization (143). While the mechanism of glucose-induced inhibition remains unclear, much more is known about the physiological identities of these GI neurons. In the ARC, GI neurons were found to co-express NPY and AgRP. Similarly, 94% of rat ARC neurons that were stimulated by lowering extracellular glucose concentrations contained NPY immunoreactivity (149). By switching extracellular glucose between 0.5 and 5 mm, 40% of ARC NPY neurons were reversibly hyperpolarized and inhibited. Since NPY and AgRP are orexigenic in nature, their co-localization in GI neurons implicates these neurons in the mechanisms which lead to a stimulation of feeding. 1.4 Fatty Acid Sensing in Hypothalamic Neurons Fatty Acids and Metabolic Physiology The function of non-esterified fatty acids (NEFAs) was elucidated in the 1950 s through the work of Vincent Dole (41) and Robert Gordon (42). Gordon demonstrated that

24 13 plasma NEFAs originate from adipose tissues, and elucidated their use by tissues such as the liver and myocardium. We now understand that NEFAs are the primary fuel for most tissues under fasting conditions (43). The release of NEFAs into the circulation results partly from the hydrolysis of triacylglycerol-rich lipids via the action of lipoprotein lipase (LPL) (43). In addition to being an important source of energy, NEFAs are also necessary for membrane lipid synthesis and lipid signaling (44). Although mostly bound to albumin, NEFA turnover is fast. The circulating half-life of NEFAs is only 3-4 minutes (43). In the fasting state, plasma NEFAs arise almost entirely from the hydrolysis of trigylcerides (TG) in adipocytes (45). However, after a meal, there is an additional source of plasma NEFA. LPL in the capillaries of adipose tissue hydrolyzes circulating TG, which constitutes much of the dietary fat carried in chylomicrons. Though fatty acids thereafter become taken up by adipocytes for storage, there is always a proportion that escapes and joins the plasma NEFA pool (46). Therefore, the plasma NEFA pool composition changes in accordance with the composition of meal fat (47). Fat mobilization is rapidly suppressed by insulin. Therefore, plasma NEFA concentrations fall after any meal containing carbohydrates. Typical plasma NEFA concentrations range from µmol/l in an overnight fasting state to approximately 1,300 µmol/l after a 72 hour fast (48) Blood-Brain Barrier Permeability, Fatty Acid Uptake and Metabolic Fates Once in the plasma, free fatty acids are bound by the carrier protein albumin. Albumin increases the solubility of these FFAs and facilitates their transport across membranes (44). To cross the blood-brain barrier, FFAs readily desorb from albumin and are

25 14 rapidly taken up by a flip-flop diffusion process (49) and/or transport proteins. Transport proteins include CD36, fatty acid transport protein (FATP) and plasma membrane fatty acid binding protein (FABP) (50). Upon entry into the cell, FFAs become coupled to FABPs, which carry FFAs from the plasma membrane to their target organelles. After cellular uptake, fatty acids become rapidly esterified to a fatty acyl-coenzyme A (fatty acyl-coa). This reaction is catalyzed by the enzyme acyl-coa synthetase (51). In this activated aycl-coa form, fatty acids can be (i) degraded by mitochondrial β-oxidation to provide cellular energy, (ii) esterified to membrane lipids or (iii) enter the sphingolipid pathway and contribute to the generation of ceramide metabolites (Figure 1.3) Fatty Acids as Signaling Molecules in the Hypothalamus Fatty acyl-coa s and the pathways regulating fatty acyl-coa metabolism have been implicated in the hypothalamic control of feeding behavior and energy homeostasis. One hypothesis is that circulating lipids regulate feeding behavior by generating an increase in the hypothalamic fatty acyl-coa pool. In turn, these fatty acyl-coa s signal an energy surplus within the hypothalamus, which activates neuronal circuits to decrease both food intake and liver glucose production (52). Indeed, intracerebroventricular (icv) administration of the monounsaturated fatty acid oleic acid is sufficient to inhibit food intake and liver glucose production. Furthermore, icv oleic acid inhibits the expression of orexigenic NPY and AgRP in the hypothalamus (53). Given these findings, it was hypothesized that similar metabolic and behavioral effects would be seen by increasing fatty acyl-coa availability. Under genetic or pharmacological inhibition of hypothalamic CPT1 (an enzyme which facilitates the transport

26 15 Fig. 1.3 Metabolic fates of palmitate upon entering the cell. Once palmitate enters the cell, it becomes primed in order to cross the mitochondrial membrane. This occurs in the peroxisome, where peroxisomal acyl-coa synthetase catalyzes the reaction between the fatty acid and CoA. The resultant palmitoyl-coa can then: 1) enter the mitochondria where it undergoes β-oxidation, 2) participate in protein palmitoylation via the actions of protein acyltransferase or 3) contribute towards de novo ceramide synthesis.

27 16 of fatty acyl-coa molecules into the mitochondria for β-oxidation), the concentration of hypothalamic fatty acyl-coa s increased, whereas the expression of orexigenic NPY and AgRP decreased (54). Therefore, these data lend support to the idea that fatty acids and the availability of fatty acyl-coas are important components of hypothalamic lipid sensing. 1.5 Obesity and Hypothalamic Inflammation High-Fat Feeding is Associated with Inflammation Excessive caloric intake is a primary risk factor for the development of obesity. Epidemiological studies have shown that individuals consuming high fat diets are particularly prone to gaining body mass (55). In peripheral tissues, the deleterious metabolic consequences of obesity arise, in part, from cellular inflammation triggered by this nutrient excess. Excess visceral adiposity is accompanied by chronic low grade inflammation in the liver, adipose tissue, skeletal muscle and vasculature. This inflammation is associated with increased circulating levels of pro-inflammatory cytokines (56). Circulating saturated fatty acids are also capable of triggering Toll-like receptor (TLR) signaling, which results in subsequent activation of intracellular inflammatory signals such as inhibitor of kb-kinase-β (IKKβ)/nuclear factor-kb (NF-κB) and c-jun N-terminal kinase (JNK) (57). The end result is a vicious cycle of inflammation that produces progressive, systemic metabolic impairment. In 2005, evidence emerged that inflammatory changes could be detected in the brains of high fat diet-fed animals. A 20-week HFD-feeding study found increased NF-κB signaling in the rat cerebral cortex (58). Honing in on the hypothalamus, De Souza et al. (59) tested the hypothesis that high fat consumption could modulate gene expression in the hypothalamus. Using a macroarray, the expression of more than 1,000 hypothalamic genes was

28 17 simultaneously measured. Of the 1,000 genes examined, more than 15% were modulated by diet. Grouping the genes based on function revealed that inflammatory genes were most affected after 16 weeks of HFD feeding (59) Evidence that Hypothalamic Inflammation Contributes to HFD-Induced Obesity Consistent with a role for hypothalamic inflammation in diet-induced obesity, neuron-specific disruption of TLR4 or IKKβ/NF-κB pathways protected against diet-induced obesity and its associated metabolic consequences (60). Viral deletion of IKKβ or overexpression of a dominant-negative IKKβ isoform in the mediobasal hypothalamus also reduced food intake and weight gain during HFD feeding (60). Moreover, in genetically normal animals, central infusion of an IKKβ inhibitor or antibodies to TLR4 can reduce food intake in diet-induced obese (DIO) mice (61). Taken together, these studies demonstrate the causal role of hypothalamic inflammation in HFD-induced weight gain. Complementing these findings is the fact that augmented hypothalamic inflammation is associated with HFD-induced obesity. For example, neuronal expression of a constitutively active IKKβ isoform increases food intake (60). Furthermore, infusion of the cytokine IL-4 directly into the brain of HFD-fed rats exerts a pro-inflammatory effect on the hypothalamus that exacerbates weight gain in an IKKβ-dependent manner (62). These data suggest that hypothalamic inflammation is both necessary and sufficient for initial and sustained weight gain during HFD feeding. Yet, if hypothalamic inflammation is to be implicated in obesity pathogenesis, it must occur prior to obesity onset. Indeed, hypothalamic inflammation is observed weeks before

29 18 peripheral cytokines are produced in the liver and adipose tissue, and before alterations in body weight occur (63) Palmitate as a Mediator of Hypothalamic Inflammation Whether hypothalamic inflammation is a consequence of excess caloric intake irrespective of diet composition has been the subject of considerable debate. One mechanism that has received attention is the ability of saturated versus unsaturated fatty acids to activate TLR4/NF-κB signaling. A predominant saturated fatty acid in our diet, palmitic acid (16:0), is found in high concentrations in all animal products and accounts for approximately 20-30% of the total FFAs in humans. Palmitic acid enters the brain linearly with time and is rapidly incorporated into brain lipids (64). Many studies that have investigated the role of FFAs in HFD-induced hypothalamic inflammation have utilized palmitic acid. Recent work demonstrates that saturated fatty acid palmitate (16:0) induces NF-κB signaling through a TLR4-dependent mechanism when administered in neuronal cell culture and after infusion into the brain (65). Signaling through NF-κB leads to the induction of cytokine gene expression, which causes local levels of TNF-α, IL-1β and IL-6 to rise and exacerbate the inflammatory state (Figure 1.4) (66). 1.6 Obesity and Hypothalamic Insulin Resistance Obesity, Diabetes and Insulin Resistance Currently, more than one third of U.S. adults are obese (which is defined as having a BMI >30 kg/m 2 ) and over 11% of individuals aged 20 or over have diabetes (67). Due to the strong association between T2DM and obesity, Zimmet et al. coined the term diabesity (68). However, only 20% of obese individuals develop T2DM, and this is thought to be due

30 19 AP-1 = activator protein 1 IKKβ = inhibitor of IkappaB kinase beta Ins = insulin IR = insulin receptor IRS = insulin receptor substrate protein JNK = c-jun N-terminal kinase Pal = palmitate TNF = tumor necrosis factor TLR4 = toll-like receptor 4 TNFR = tumor necrosis factor receptor TAK1 = transforming growth factor-β-activated kinase 1 NF-κB = nuclear factor kappa B Fig. 1.4 Activation of the IKKβ/NF-κB pathway leads to inflammation and impaired insulin signaling. Activation of TLR4 or TNF-α receptor by palmitate and TNF-α, respectively, stimulates downstream NFκB and AP-1 transcription factors to upregulate gene expression of pro-inflammatory cytokines. Insulin signaling is inhibited by chronic receptor stimulation by insulin itself or by stimulation of the TLR4 and/or TNF-α receptors. IKKβ and JNK, in particular, can inhibit insulin signaling by phosphorylating serine residues on IRS proteins.

31 20 to a compensatory response by pancreatic β-cells to increase insulin secretion. Therefore, T2DM involves both decreased insulin sensitivity and a loss of compensatory insulin secretion. Insulin resistance, then, is defined as a diminished cellular response to insulin, resulting in the inability to increase glucose uptake leading to increased blood glucose (69). Obesity results from an imbalance between energy intake and energy expenditure. The result is adipocyte hypertrophy, along with increased lipolysis (70). Increased adiposity can lead to insulin resistance through increased adipocyte-derived FFAs and increased adipokines (71). Excess FFAs become stored in non-adipose cells such as muscle, where they may be catabolized into lipid metabolites such as fatty acyl-coas, diacylglycerol (DAG), triacylglycerol (TAG) and ceramide. These lipid metabolites are capable of inhibiting insulin signal transduction leading to insulin resistance (71). Moreover, in an obese state, there is an increase in peripherally-derived pro-inflammatory cytokines such as tumor necrosis factoralpha (TNF-α), which have also been shown to induce insulin resistance in mice and humans (72). Another contributing factor to the development of insulin resistance is hyperinsulinemia, which will be discussed in the following section. Finally, impaired insulin signaling has the ability to potentiate obesity pathogenesis due to the importance of central insulin action in regulating energy balance. Therefore, the maintenance of proper insulin action is critical for maintaining energy homeostasis Hyperinsulinemia and the Development of Insulin Resistance Insulin resistance is a hallmark feature of obesity. There are numerous etiologies for insulin resistance, including lipotoxicity, inflammation and hyperinsulinemia (73). Hyperinsulinemia reflects the compensation by insulin-secreting β-cells to systemic insulin

32 21 resistance. In vivo studies indicate that prolonged exposure to high levels of insulin can lead to insulin resistance (74). Indeed, plasma insulin levels are increased in obese states and this increase occurs prior to a reduction in insulin sensitivity (75). At the cellular level, hyperinsulinemia impairs insulin signal transduction through homologous desensitization. The insulin receptor itself is involved in negative feedback involving a reduction in (i) receptor affinity, (ii) the number of receptors expressed on the cell surface and (iii) the effectiveness of the receptor as a transmitter of stimulatory signals (76). Continual exposure to insulin can also lead to serine phosphorylation of downstream IRS proteins, which reduces its ability to activate downstream elements in the insulin signaling pathway (76) High Fat Feeding and Neuronal Insulin Resistance The growing trend towards a more sedentary lifestyle combined with the consumption of fat-rich foods play an important role in the current obesity epidemic (34). Consumption of HFD for as little as 72 hours is sufficient to reduce hypothalamic insulin sensitivity in rats (77). Elevated saturated fatty acids not only increase body weight, but also chronically reduce hypothalamic insulin sensitivity (77). At a molecular level, saturated fatty acids such as palmitate cross the blood-brain barrier and accumulate in the hypothalamus. Here, they activate pro-inflammatory signaling pathways including toll-like receptor 4 (TLR4) signaling, resulting in central insulin resistance (65). Palmitate-mediated central insulin resistance is due, at least in part, to the activation of protein kinase Cθ (PKCθ). Subsequent translocation of PKCθ to the cell membrane prevents insulin-mediated activation of PI3K via direct interaction with insulin receptor and insulin receptor substrate proteins (78). On the other hand, palmitate can also activate NF-κB,

33 22 which subsequently induces suppressor of cytokine signaling (SOCS) 3 expression (60). SOCS3 is one of the principal negative regulators of insulin signaling, interfering with insulin-mediated phosphorylation of IR and its downstream molecules. It also targets IRS proteins for proteasomal degradation (79). Elevated NF-κB signaling also triggers endoplasmic reticulum stress leading to increased activity of c-jun N-terminal kinase (JNK). In turn, JNK mediates inhibitory phosphorylation events on IRS serine residues which contribute to the development of insulin resistance (80,81). During obesity progression, elevated levels of pro-inflammatory cytokines exacerbate central insulin resistance by further activating NF-κB and JNK signaling. TNF-α, like palmitate, can induce expression of protein tyrosine phosphatase 1B (PTP1B), potentially through transactivation of NF-κB (82). Elevated levels of PTP1B in the ARC, as seen after 20 weeks of HFD feeding, inhibit insulin signaling by direct dephosphorylation of the IR (Fig. 1.5) (83) FFA Metabolism and Insulin Resistance Dysregulated FFA metabolism is thought to play a causal role in the development of insulin resistance (84). The large majority of FFAs enter the glycerolipid pathway, where they become substrates for membrane glycerophospholipids and TAG, the primary form of stored fat (86). Glycerol-3-phosphate acyltransferase (GPAT) is the enzyme that regulates entrance of FFAs into this pathway, and it transfers the acyl group from fatty acyl-coa to glycerol-3- phosphate (86). The saturated fatty acid palmitate increases intracellular levels of DAG, the precursor to TAG, and TAG itself in rat islets (87). Interestingly, the GPAT knockout mouse is protected from insulin resistance on a HF diet (88).

34 23 AKT = protein kinase B DAG = diacylglycerol ER = endoplasmic reticulum Pal = palmitate Pal-CoA = palmitoyl-coa PKCθ = protein kinase C θ PP2A = protein phosphatase 2A PTP1B = protein tyrosine phosphatase 1B Fig. 1.5 Mechanisms of palmitate-mediated inhibition of insulin signaling. Palmitate has been shown to inhibit insulin signaling through a variety of mechanisms. 1) Palmitate may enter the cell and contribute to an increase in de novo ceramide synthesis. Ceramides are known to interact with and activate the phosphatase enzyme PP2A. PP2A dephosphorylates Akt, leading to impaired insulin signaling. 2) Palmitate can increase intracellular levels of diacylglycerol, leading to activation of PKCθ and subsequent serine phosphorylation of IRS-1. 3) Palmitate (and inflammation) induces the expression of PTP1B, which is a protein tyrosine phosphatase enzyme. PTP1B inhibits insulin signaling by removing phosphate groups from the tyrosine residues of the activated insulin receptor.

35 24 Rates of ceramide synthesis depend on the availability of long-chain saturated fatty acids, which participate in the initial rate-limiting step of de novo ceramide synthesis (86). In this reaction, serine palmitoyltransferase (SPT) catalyzes the condensation of palmitoyl-coa and serine to produce 3-ketosphinganine. Subsequent reactions lead to the eventual synthesis of ceramide, which serves as a precursor for more complex sphingolipids (86). Ceramides with different fatty acid and long-chain base compositions can be formed in different compartments or membranes of the cell, each with potentially distinct functions. Interestingly, ceramide levels are elevated in rodent and human insulin-resistant tissues (89). As demonstrated in numerous reports, elevated ceramide levels may inhibit insulinstimulated glucose uptake, GLUT4 translocation and/or glycogen synthesis (90). This dysregulation of insulin signaling has been linked to ceramide-mediated regulation of IRS-1 and Akt/PKB. Three independent groups found that treating cultured cells with short-chain ceramide analogs blocked insulin-stimulated tyrosine phosphorylation of IRS-1 and its subsequent activation of PI3K (91). These groups proposed that ceramide may promote the phosphorylation of IRS-1 on inhibitory serine/threonine residues. In various cell types, ceramides have been shown to activate extracellular signal-regulated kinase 2 (ERK2), p38, JNK and IkB kinases (IKKs) (92), which have been implicated in serine/threonine phosphorylation of IRS-1 (93). Many groups have demonstrated that ceramide also inhibits phosphorylation and activation of Akt/PKB. It is now understood that ceramide inhibits activation of Akt/PKB through two distinct mechanisms. First, ceramide promotes the dephosphorylation of Akt/PKB through direct activation of protein phosphatase 2A (PP2A) (94). Indeed, the PP2A

36 25 inhibitor okadaic acid was sufficient to prevent the inhibitory ceramide effects on Akt/PKB in C2C12 myotubes and brown adipocytes (95, 96). Second, ceramide may activate PKCζ, which inhibits Akt/PKB translocation to the membrane by phosphorylating threonine-34 (97). 1.7 Cell Model The Need for Cell Lines In vivo experimentation is important for determining the overall function of a molecule (i.e. hormone, receptor or structural protein). Indeed, the use of animal models has enhanced our basic understanding of physiological processes, such as energy homeostasis. These studies have elucidated the role of the brain in overall metabolism, and have triggered the development of brain-specific and neuron-specific mouse models. Yet, despite new technologies allowing for closer examination of intracellular workings, animal experimentation has its limits. This is especially true in the context of the hypothalamus, where a collection of cell phenotypes exist. Therefore, classical in vivo approaches cannot establish the molecular mechanisms involved in gene regulation and cellular signaling. Moreover, it is difficult to determine the direct actions of agents, such as nutrients or hormones, on specific cell types. Given these limitations, researchers have turned to the use of cell lines (98). Cell lines allow for experimentation with homogeneous populations of cells in a controlled environment. However, one cannot state for certain that cell lines function exactly as the native cells would. For this reason, caution must be taken before extrapolating findings from cell lines to an in vivo model. When working with neuronal cell lines, it is particularly

37 26 important to remember that these models lack the complexity and integrated network of neurons found in vivo. Regardless, recent studies looking at molecular events in vitro have found that the results from cell lines mirror that of in vivo studies (98). Basic tissue culture techniques were established in 1885 by Wilhelm Roux, and it was not until 1940 that the first immortal cell line was developed (99). Since that time, cell lines have been produced from many different tissues although the first attempt at immortalizing neurons was not performed until 1974 by Shaw et al. (100). Shaw et al. transfected primary hypothalamic cells from embryonic mice with simian virus 40 T-Ag to create an immortalized cell population labeled HT9. Years later, Cepko et al. developed retroviral shuttle vectors which would allow researchers to retrovirally infect primary cells with an immortalizing oncogene and selectively propagate them (101) Adult Hypothalamic Cell Lines (mhypoa-xx) Non-transformed primary hypothalamic cultures are difficult to maintain, have a short lifespan and represent a heterogeneous population of neurons. Contrarily, immortalized, clonal cell lines represent an unlimited and homogeneous population of specific neuronal cell types. Since the hypothalamus contains a wide range of cell types, Belsham et al. recognized the need for mouse cell lines representative of many unique hypothalamic neurons. Belsham et al. initially developed 38 embryonic, clonal hypothalamic mouse cell lines (98). However, to understand key molecular mechanisms involved in adult neuroendocrine cells, the Belsham group also immortalized adult hypothalamic cell models. In order to immortalize the adult neurons, cells were treated with ciliary neurotrophic factor (CNTF) to induce cell proliferation. This would render the cell cultures amenable to

38 27 retroviral transfer of the SV40 T-antigen oncogene (Figure 1.6) (102). Over 50 adult mouse cell lines were eventually established, and labeled in the form mhypoa- clone number. Like the embryonic cell lines before them, the adult cell lines also express mature neuronal markers, display typical neuronal morphology and have been characterized for the expression of necessary neuropeptides and receptors. The hypothalamic cell lines made available from our lab and others allow for the study of molecular events involved in nutrient sensing in distinct neuronal populations. These novel, representative cell models put us in an advantageous position to determine the direct effects of nutrients (i.e. palmitate) on neuropeptide gene expression and signaling events in neuropeptide-expressing neurons mhypoa-npy/gfp Cell Line To immortalize NPY/AgRP-expressing neurons from the adult hypothalamus, the Belsham group dissected hypothalamii from NPY-GFP transgenic mice and immortalized as described above. NPY/AgRP neurons were then selected using flow cytometry. The NPY- GFP cell line has been thoroughly characterized using RT-PCR, ICC and NPY secretion assays, and the phenotypic characterization is described in Table 1.1. In this thesis, the mhypoa-npy/gfp cell model is used to describe palmitate-mediated regulation of AgRP gene expression and to elucidate the effects of palmitate on signaling events in representative NPY/AgRP-expressing neurons. 1.8 Hypothesis and Aims Neuronal circuits within the hypothalamus form the homeostatic control mechanism that controls food intake and energy balance. It has been well established that coordinated regulation

39 28 1) 2) 5) 4) 3) Fig. 1.6 Generation of the mhypoa-npy/gfp cell line. Adult NPY/GFP-expressing transgenic mice were generated (1). Cells were harvested from the GFP-expressing mouse hypothalamus (2). These cells were then treated with CNTF to induce neurogenesis, and transfected with SV-40 T antigen for immortalization (3). Cells were FAC sorted for GFP fluorescence with greater than 95% purity (4). The fluorescent cells represent the cells of interest, mhypoa-npy/gfp (5).